Report No. 47349-CN China From Poor Areas to Poor People China's Evolving Poverty Reduction Agenda An Assessment of Poverty and Inequality in China March 5, 2009 Poverty Reduction and Economic Management Department East Asia and Pacific Region Document of the World Bank RCRL Rural Land Contracting Law RHS Rural Household Survey SEAC State Ethnic Affairs Commission SOE State-Owned Enterprises SPA Social Pool Account SWOT Strength, Weakness, Opportunity, Threat TVE Township and Village Enterprise UDB Urban Di Bao UHS Urban Household Survey UI Unemployment Insurance UMT Unverified Means Testing URBMI Urban Resident Basic Medical Insurance USFHS Urban Short-Form Household Survey VLS Village-Level Survey VMT Verified Means Testing WTO World Trade Organization Vice President: James Adams Country Director: David Dollar Sector Director: Vikram Nehru Task Team Leaders: Gaurav Datt and Shubham Chaudhuri Table of contents Preface and acknowledgements.............................................................................................. i Executive summary ............................................................................................................... iii China's record of poverty reduction and growth is enviable iii But the task of poverty reduction continues and in some respects has become harder iv Policy initiatives in response to these challenges suggest a broader poverty reduction agenda is evolvingx What the review suggests about the main priorities for poverty reduction and what is needed to implement them xii Adopting a broader conception of poverty and an adequate threshold for identifying and targeting the poor .............................................................................................................................................................. xii Retaining rural poverty reduction as the top priority................................................................................... xiii Promoting opportunity by raising the returns to labor................................................................................. xiv Enhancing security by expanding and improving the coverage of the social protection system in rural and urban areas.......................................................................................................................................... xix Fostering equity and reducing poverty by ensuring secondary school education and basic healthcare for all ........................................................................................................................................................... xxiv Supplementing area-based poverty reduction efforts with a household-oriented approach.....................xxvii Providing an adequate and equitable allocation of resources for local governments..............................xxviii Strengthening institutional arrangements to promote participation, enhance accountability and improve coordination......................................................................................................................................... xxx Enhancing statistical monitoring and evaluation capacity ........................................................................xxxii PART I WHY REVIEW CHINA'S POVERTY REDUCTION AGENDA? ............. 1 1. China's poverty reduction record over the last quarter century has been remarkable .... 2 a. The incidence, depth and severity of poverty have declined dramatically both according to official poverty estimates and by international standards 2 Between 1981 and 2004, the fraction of the population consuming less than a dollar-a-day fell from 65% to 10%.........................................................................................................................................................3 ...and half a billion people were lifted out of poverty.....................................................................................5 The depth and severity of poverty also declined considerably......................................................................6 b. Between 2001 and 2004 the pace of poverty reduction accelerated 6 c. ...and there are indications that poverty has continued to decline up to 2007 9 d. But the progress in poverty reduction has been temporally uneven... 11 e. ...and, at the household level, income growth has been uneven 11 2. But the task of poverty reduction is not complete and in some respects has become harder................................................................................................................................16 a. Measured by international standards, the number of poor in China remains high 16 China's official poverty line is low by international standards......................................................................17 China's official poverty line seems low relative to rising incomes and growing aspirations.........................18 China's poverty line may not cover essential food and non-food needs .....................................................21 b. Vulnerability to poverty because of income shocks remains widespread 22 Nearly a third of China's rural population was poor at least once between 2001 and 2004........................23 A large part of the severity of poverty in rural China is attributable to risk ..................................................25 Exposure to risk may be one reason why China's poor have unusually high savings rates........................28 c. Eliminating the remaining poverty has become harder as the poverty rate has declined 29 ...because the remaining poor are harder to reach.....................................................................................29 ...and because the responsiveness of poverty reduction to economic growth has fallen ...........................30 d. Income inequality has increased significantly 32 China is no longer the low-inequality country it was a quarter century ago.................................................32 The rural-urban income gap has grown......................................................................................................34 ...and inequality within both rural and urban areas has risen .....................................................................35 ...while the contribution of inter-provincial income differences to overall inequality has narrowed .............35 e. Disparities in other aspects of human development remain and in some cases have grown 37 Because of the marketization of public services, incomes now matter more for access to health and education than they used to ..................................................................................................................37 ...and the burden of health and education expenditures has increased for households.............................37 f. Restructuring of the urban labor market and the dismantling of the iron rice bowl has led to new challenges 38 Unemployment has risen in urban areas while labor force participation has decreased.............................38 Growing informalization of the urban labor market has raised concerns about the welfare of urban workers ..............................................................................................................................................................39 g. A large "floating" population of rural migrants has emerged in urban areas 39 The nature of migration is changing and integrating migrant workers and their families into urban areas poses new challenges...........................................................................................................................40 3. A poverty reduction agenda consistent with the vision of a xiaokang society ...............42 a. The challenges that remain as well as those that have emerged suggest a case for reviewing and broadening China's poverty reduction agenda 42 b. Recent policy initiatives suggest that a broader poverty reduction agenda is indeed evolving 42 PART II REVIEWING CHINA'S POVERTY REDUCTION AGENDA ................ 46 4. Establishing the basic facts about poverty and the poor in China ..................................47 a. The poor in China are predominantly from rural areas 47 Virtually all of the China's poor are in or from rural areas ...........................................................................47 The numbers of non-hukou migrants from rural areas have grown dramatically .......................................48 Excluding migrant workers from the rural population, 90% of poverty is still rural.......................................49 Only with a 50% or higher urban-rural cost of living differential is there any noticeable poverty amongst urban residents .....................................................................................................................................50 b. Geography and ethnicity matter but are not the only determinants of poverty 52 The incidence and severity of poverty is highest in the western region but nearly half of the poor are elsewhere..............................................................................................................................................52 Poverty is most severe in mountainous and minority areas, but more than half the poor are in non- mountainous non-minority areas...........................................................................................................54 c. The poor are dispersed throughout China's villages 55 Measuring the extent to which poverty is concentrated within villages .......................................................55 The concentration of poverty within villages is highest in the southwestern region and least in the coastal region ....................................................................................................................................................56 The concentration of poverty tends to decline as the incidence of poverty itself declines...........................57 d. Most of China's poor are able to work 58 Nearly three-quarters of China's rural poor live in households where no-one lacks work capacity and 97% live in households with two or more productive members .....................................................................58 What the poor lack is human capital ...........................................................................................................60 e. More than the elderly, children, and in particular, girls, are likely to be poor 60 f. The poor have some characteristics in common, but differ in other ways 62 Constructing a taxonomy of the rural poor in China....................................................................................62 The rural poor can be usefully grouped into seven clusters, each distinguished by a particular disadvantage... .....................................................................................................................................63 The regional distribution of the clusters varies, as does the importance of individual clusters within each region ....................................................................................................................................................64 g. A profile of the disadvantaged in urban areas 67 There is no single threshold for identifying the urban "poor" .......................................................................67 Nearly 20 million people in urban areas consume less than twice the World Bank poverty line and may be considered the "urban disadvantaged"..................................................................................................69 The urban disadvantaged are relatively evenly spread across regions but heavily concentrated in smaller cities......................................................................................................................................................70 The elderly in urban areas are no more likely to be disadvantaged than other age groups........................71 Households without high-school educated workers constitute most of the disadvantaged in urban areas .72 Ethnic minorities and those without local urban hukou are more likely to be disadvantaged but are relatively few in number.........................................................................................................................72 More than half of the urban disadvantaged live in households with unemployed workers..........................73 5. Promoting opportunity: helping the poor move out of poverty .......................................76 a. China's poverty reduction efforts have been development-oriented 76 Rural Reforms and the introduction of the Household Responsibility System were central to poverty reduction during the early 1980s...........................................................................................................76 Area-based efforts began with the National Poverty Reduction and Development Programs during 1986- 1993 ......................................................................................................................................................77 The 8-7 Poverty Reduction Plan continued and intensified these efforts during 1994-2000.......................77 Further refinements were made in the New Century Poverty Reduction Strategy for 2001-2010...............78 The funding for poverty alleviation programs comes from multiple sources................................................79 b. Raising Incomes through poor-area development: the Integrated Village Development Program 80 A reasonably good effort was made to designate poor villages according to specified criteria...................80 But more than half the poor live outside the designated villages ................................................................81 The program raised incomes in designated villages but the poorest households have failed to benefit .....85 Program impacts hampered by difficulties in inter-agency coordination, inadequate resources and limited participation of poor households ...........................................................................................................87 c. Raising Incomes through other developmental poverty reduction programs: training for labor transfer, subsidized credit and agribusiness 89 Training for labor transfer............................................................................................................................89 Subsidized credit and agribusiness.............................................................................................................90 d. How can developmental poverty reduction programs better help the poor? 91 e. Rural-to-urban migration as a path out of poverty 93 Large and growing scale of migration though full-family migration remains limited.....................................93 Most migration is inter-provincial, from central and southwest to coastal areas, and to medium and large cities......................................................................................................................................................95 Male young adults with junior high school education are more likely to migrate .........................................99 Migration reduces poverty.........................................................................................................................100 But many of the poor are unable to migrate..............................................................................................101 There remain several impediments to migration .......................................................................................104 What do potential migrants say?...............................................................................................................106 The downside of migration in source areas ..............................................................................................106 Policy implications.....................................................................................................................................108 6. Enhancing security: protecting households from falling into poverty ..........................110 a. Land allocations alone are an inadequate basis for the rural safety net 110 There are large variations in the productivity of land.................................................................................111 The share of farm income in rural income continues to decline ................................................................112 Despite recent improvements, security of tenure is still limited by land readjustments and requisitions...113 b. Are rural social protection programs adequate? 115 Features of key social assistance programs in rural areas .......................................................................116 Coverage of rural social assistance programs has expanded but remains uneven relative to the levels of poverty ................................................................................................................................................119 Benefit levels for recipients are low for several programs especially in comparison with similar urban programs.............................................................................................................................................121 Benefit levels and coverage vary from areas to area, not with need, but with local fiscal resources ........122 Coverage of rural health insurance has improved dramatically but challenges of limited coverage of rural pensions and low benefit levels of all programs remain ......................................................................123 Policy implications.....................................................................................................................................126 c. Promoting re-employment in a changing urban labor market 128 Nearly half of all urban employment is informal in nature .........................................................................128 High unemployment and declining labor force participation......................................................................130 Urban poverty is linked to the work status of household members ...........................................................132 Promoting re-employment in urban areas.................................................................................................133 Policy implications.....................................................................................................................................134 d. Replacing the iron rice bowl: urban social insurance programs 134 Features of the main urban social insurance programs ............................................................................135 Policy issues .............................................................................................................................................138 e. Replacing the iron rice bowl: the urban Di Bao program 142 Key features of the program......................................................................................................................142 Characteristics of urban Di Bao beneficiaries resemble those of the relatively poor in urban areas.........144 Assessing Di Bao's performance ..............................................................................................................146 By international standards, urban Di Bao has excellent targeting performance........................................147 But the coverage of the program could be increased................................................................................149 Eligibility thresholds vary with local fiscal resources rather than with need...............................................150 At least in terms of its design, if not in practice, work disincentive effects of the program are a concern .152 Policy issues .............................................................................................................................................153 7. Fostering equity: reducing disparities and building human capacities.........................157 a. The rise in income inequality is a concern but it is important to be clear on the reasons why 157 b. Inequalities in human capital underlie the three components of income inequality that have risen the most158 Differences in the human capital endowments of rural and urban workers explain much of the rural-urban labor productivity gap ..........................................................................................................................158 Differential access to off-farm employment due, increasingly, to differences in schooling is the most visible source of inequality within rural areas .................................................................................................160 Income inequality within urban areas stems primarily from disparities in labor incomes attributable in large part to differences in educational attainment.......................................................................................161 c. Disparities in other aspects of human development mirror, and can be traced to disparities in income, raising the possibility of an inequality trap 165 Changes in institutional arrangements have led to a marketization of public service delivery, particularly in rural areas...........................................................................................................................................165 Incomes matter more now than they used to for access to health and education.....................................165 ...and the burden of health and education expenditures has increased for households...........................166 Because health and education determine income prospects, this raises the possibility of an inequality trap ............................................................................................................................................................166 d. Compulsory education finance reform has helped improve access to affordable basic education but many challenges remain 167 e. Publicly-financed interventions can ensure equity of opportunity but under China's current fiscal system do not adequately do so 169 Expenditures on poverty-reduction-related activities are primarily the responsibility of local governments ............................................................................................................................................................169 Despite recent initiatives to enhance equalizing transfers, public expenditures tend to favor richer provinces and localities .......................................................................................................................171 Disparities in spending can be even more pronounced at lower levels of government.............................173 f. Group-based inequities are particularly visible along certain axes 174 The rural-urban income gap is only one manifestation of the great divide that separates China's urban and rural areas...........................................................................................................................................174 The gender gap in China begins at birth and continues through the working years and into old age .......175 Integrating rural migrant workers and their families in urban areas...........................................................177 PART III WHAT THE REVIEW SUGGESTS ABOUT THE POVERTY REDUCTION AGENDA ............................................................................. 186 8. What the priorities are and what is needed to implement them.....................................187 a. Adopting a broader conception of poverty and an adequate threshold for identifying and targeting the poor 187 b. Retaining rural poverty reduction as the top priority 188 c. Promoting opportunity by raising the returns to labor 189 Realizing the potential of migration for poverty reduction .........................................................................189 Promoting agricultural and rural development in poor areas.....................................................................191 d. Enhancing security by expanding and improving the coverage of the social protection system in rural and urban areas 193 Improving rural social protection ...............................................................................................................193 Improving urban social protection .............................................................................................................195 Harmonizing the rural and urban systems ................................................................................................197 e. Fostering equity and reducing poverty by ensuring secondary school education and basic healthcare for all 197 Improving access to affordable education.................................................................................................198 Improving access to affordable healthcare ...............................................................................................199 f. Supplementing area-based poverty reduction efforts with a household-oriented approach 200 g. Providing an adequate and equitable allocation of resources for local governments 201 h. Strengthening institutional arrangements to promote participation, enhance accountability and improve coordination 203 i. Enhancing statistical monitoring and evaluation capacity 204 Data appendix......................................................................................................................206 Rural and Urban Household Surveys (NBS) 206 2004 World Bank-NBS Special Purpose Village-Level Survey 207 China Urban Labor Survey (CULS) 208 2006 Qualitative Study of Poverty 208 Data and information from secondary sources 209 References...........................................................................................................................210 List of tables Table 0.1 : Responsiveness of poverty to growth has declined, 1981-2005 .................................................. vii Table 0.2 : Share of informal employment for local residents in urban China in 2002 .................................... ix Table 0.3 : Rural-urban distribution of poverty in China, 2003.......................................................................xiii Table 0.4 : A rough estimate of the additional cost of scaling up poverty reduction efforts.......................... xxx Table 1.5 : Poverty lines for China...................................................................................................................4 Table 1.6 : Half a billion Chinese have been lifted out of poverty in the last quarter century ...........................6 Table 1.7 : The incidence, depth and severity of poverty: 1981-2001..............................................................6 Table 1.8 : Recent changes in poverty, 2001-2004 .........................................................................................7 Table 1.9 : Official estimates of recent changes in poverty in rural areas, 2001-2004.....................................8 Table 1.10: Rural taxes and fees have been reduced significantly since 2001..............................................10 Table 1.11: The reduction in taxes and fees likely reduced poverty and inequality in rural China .................11 Table 2.12 : The incidence of persistent and transient income poverty in rural China: 2001-2004................23 Table 2.13 : The incidence of persistent and transient consumption poverty in rural China: 2001-2004 .......24 Table 2.14 : The contribution of risk to income and consumption poverty in rural China ...............................26 Table 2.15 : Even the poor in China tend to save and household saving rates are high ..............................29 Table 2.16 : Responsiveness of poverty to growth has declined, 1981-2005 ................................................31 Table 4.17 : Rural-urban distribution of poverty in China, 2003.....................................................................47 Table 4.18 : Share of rural poverty in China, 1981-2001 ...............................................................................48 Table 4.19 : Poverty estimates for rural residents, migrants and urban residents in China, 2003 .................50 Table 4.20 : Poverty estimates for rural residents, migrants and urban residents in China, 2003, assuming a higher (50%) urban-rural cost of living differential ..................................................................................51 Table 4.21 : Is poverty in China a Western province phenomenon? Evidence for 2003................................53 Table 4.22 : Poverty in China by topography-minority area categories and by type of city, 2003..................54 Table 4.23 : Village-level concentration of rural poverty in China, 2003 ........................................................57 Table 4.24 : Poverty by the number of laborers in the household, based on household per capita income before receiving public transfers, in rural China, 2003............................................................................58 Table 4.25 : Poverty and lack of work capacity within households in rural China, 2003 ................................59 Table 4.26 : Income poverty in China by level of education, 2003.................................................................60 Table 4.27 : The incidence of consumption poverty amongst women and the elderly in China, 2003...........61 Table 4.28 : Household characteristics used in constructing a taxonomy of rural poor in China ...................63 Table 4.29 : A taxonomy of the rural poor in China, 2003..............................................................................64 Table 4.30 : Regional shares of different categories of rural poor in China, 2003 .........................................65 Table 4.31 : Types of rural poor in different regions, China 2003--a region-specific taxonomy ....................66 Table 4.32 : Urban poverty lines in China (in Yuan per person per year) ......................................................68 Table 4.33 : Food poverty lines and implied calorie prices in urban and rural areas .....................................69 Table 4.34 : The urban disadvantaged in China, 2003 ..................................................................................70 Table 4.35 : Urban disadvantaged in China by region and city-type, 2003....................................................71 Table 4.36 : Urban disadvantaged in China by age and sex, 2003................................................................72 Table 4.37 : Education and urban disadvantaged in China, 2003..................................................................72 Table 4.38 : Hukou and minority status and urban disadvantaged in China, 2003 ........................................73 Table 4.39 : Employment status and relative poverty in urban areas, 2003...................................................73 Table 5.40 : Central government funding for poverty alleviation programs, 2001 to 2007 .............................79 Table 5.41 : Number of designated poor villages, by region..........................................................................80 Table 5.42 : Targeting outcomes under poor village and poor county designations ......................................82 Table 5.43 : Coverage rates adjusted for concentration of poverty and scale of program .............................84 Table 5.44 : Comparing per-capita income growth in designated poor villages.............................................85 Table 5.45 : Household-level impacts of the Integrated Village Development Program ................................86 Table 5.46 : Basic information on migration from rural areas in China, 2003.................................................94 Table 5.47 : The extent of family out-migration from rural areas, 2000-2004.................................................95 Table 5.48 : Main working place of rural workers, by region, 2003 and 2006 ................................................96 Table 5.49 : Destinations of migrants by source region, 2003 .......................................................................97 Table 5.50 : Gender, age and education level of migrant workers, 2003 and 2006.......................................99 Table 5.51 : Rural households with migrant workers are less likely to be poor, 2003..................................100 Table 5.52 : Distribution of rural workers among farm, local non-farm and migration activities, by poverty status, 2003..........................................................................................................................................103 Table 5.53 : Work activity of rural workers from poor households in different regions, 2003 .......................103 Table 5.54 : Non- migrant workers from poor households without migrants, 2003 ......................................104 Table 5.55 : Select factors influencing migration from rural areas, 2003 .....................................................104 Table 5.56 What do potential migrants in poor areas say?..........................................................................106 Table 6.57 : Income from agriculture for the rural population, 2004.............................................................112 Table 6.58 : Risk of land expropriation faced by migrant households, 2004................................................113 Table 6.59 : Prevalence of land taking and the extent of compensation paid, 2000-2004...........................115 Table 6.60 : Number of beneficiaries of social assistance programs in China, 2004-7................................116 Table 6.61 : Official statistics on natural disasters and disaster relief in China............................................118 Table 6.62 : Village-Level evidence on natural disasters and disaster relief, 2004......................................119 Table 6.63 : Coverage of social assistance programs and poverty in rural China, 2004 .............................121 Table 6.64 : Benefit levels of social assistance programs in rural China, 2004 ...........................................121 Table 6.65 : Rural pension insurance system, 1997-2007...........................................................................126 Table 6.66 : Share of informal employment for local residents in urban China in 2002 ...............................129 Table 6.67 : Share of urban workers with formal contracts, 2005................................................................130 Table 6.68 : Alternative estimates of China's urban unemployment rate .....................................................131 Table 6.69 : Disaggregated urban labor force participation and unemployment rates in 2003 ....................132 Table 6.70 : Urban relative poverty rate with unemployed and out of labor force family members.............133 Table 6.71 : Coverage of urban social insurance programs in China, 1995-2007 .......................................136 Table 6.72 : Basic information on the urban Di Bao program in China, 1996-2007 .....................................142 Table 6.73 : Composition of urban Di Bao beneficiaries, 2002-04...............................................................144 Table 6.74 : Di Bao beneficiaries have different characteristics from non-beneficiaries ..............................145 Table 6.75 : Profile of urban Di Bao beneficiaries, 2003..............................................................................146 Table 6.76 : Targeting and coverage under Di Bao .....................................................................................149 Table 6.77 : Variables commonly used for proxy means targeting ..............................................................155 Table 7.78 : Rising wage inequality in urban China.....................................................................................162 Table 7.79 : Estimates of the returns to education in urban China in 2003..................................................163 Table 7.80 : Analysis of variance of urban wages in 2003...........................................................................165 Table 7.81 : Expenditures on poverty-reduction related activities are primarily borne by local governments, 2006......................................................................................................................................................170 Table 7.82 : Women fare worse than men in education, health and employment........................................176 Table 7.83 : Village policies in relation to women's land rights, 2004 ..........................................................176 Table 7.84 : Migrant population and workers in urban and rural China........................................................177 Table 7.85 : Urban poverty headcount rates for local residents and migrants (%).......................................178 Table 7.86 : Income poverty headcount rates of local residents and migrants, by city (using twice the World Bank poverty line).................................................................................................................................179 Table 7.87 : Factors explaining relatively low poverty rate of migrants in 5 large cities...............................179 Table 7.88 : Characteristics, occupations and sectors of local residents and migrants in Chinese cities, 2000180 Table 7.89 : Occupations of local residents and migrants in 5 large cities in 2005......................................181 Table 7.90 : Decomposition of hourly earnings differences between migrants and local residents .............181 Table 7.91 : City educational costs of migrant children with and without local hukou..................................182 Table 7.92 : School fees of migrant children in place of hukou and in cities................................................183 Table 8.93 : A rough estimate of the additional cost of scaling up poverty reduction efforts........................203 List of figures Figure 0.1 : China's record of poverty reduction and growth over the last quarter century has been remarkable............................................................................................................................................... iii Figure 0.2 : Progress in poverty reduction during 2005-2007 (World Bank poverty line: 888 yuan per person per year at 2003 rural prices)................................................................................................................... iv Figure 0.3 : China's poverty line in comparison with other countries ...............................................................v Figure 0.4 : Almost a third of China's rural population was consumption poor in at least one year between 2001 and 2004......................................................................................................................................... vi Figure 0.5 : Poverty concentration declines at lower levels of poverty across region-topography-ethnicity categories in rural China, 2003................................................................................................................ vi Figure 0.6 : The evolution of income inequality in China since the start of reforms ....................................... vii Figure 0.7 : Health and education outcomes at the provincial level are more closely tied to incomes than they used to be .......................................................................................................................................viii Figure 0.8 : Estimates of urban unemployment rates and labor force participation rate ................................. ix Figure 0.9 : Rising tide of rural migrants, 2001-2007 .......................................................................................x Figure 0.10 : The relationship between migration probability and log income per capita at the household level, 2003 .............................................................................................................................................. xv Figure 0.11 : Fraction of the income poor living in designated poor villages.................................................xvii Figure 0.12 : Social assistance benefits depend on village revenues, 2004.................................................. xx Figure 0.13 : Lower coverage of social insurance for the poor in Chinese cities, 2005 ................................xxii Figure 0.14 : Differences in human capital endowments and returns underlie much of income inequality between and within rural and urban areas, 2003..................................................................................xxiv Figure 0.15 : The provincial distribution of expenditures in 2004 favored the richer provinces....................xxix Figure 1.16 : China's record of poverty reduction and growth over the last quarter century has been remarkable................................................................................................................................................2 Figure 1.17 : China's poverty reduction using the new 2005 international poverty standard is no less impressive than using the old 1993 international standard.......................................................................5 Figure 1.18 : Progress in poverty reduction during 2005-2007 ........................................................................9 Figure 1.19: The poor proportionally benefited the most from the reduction in taxes and fees......................10 Figure 1.20 : Progress in poverty reduction was temporally uneven..............................................................11 Figure 1.21 : Between 1990 and 2004 incomes grew throughout the distribution, but grew more at the upper end..........................................................................................................................................................13 Figure 1.22 : Income growth incidence curves for China for various sub-periods between 1990 and 2004 ..14 Figure 1.23 : The unevenness of growth between 2001 and 2003 ................................................................15 Figure 2.24 : China's poverty line in comparison with other countries ...........................................................17 Figure 2.25 : The relative decline of China's poverty line, 1980-2007 ...........................................................19 Figure 2.26 : The subjective poverty line .......................................................................................................20 Figure 2.27 : The relative contribution of risk rises as poverty becomes less severe ....................................27 Figure 2.28 : Even for the persistently poor, risk adds to the severity of poverty...........................................28 Figure 2.29 : Many households are not income-poor but are consumption-poor...........................................29 Figure 2.30 : The Western region's share of China's rural poor has fallen to about half in the last decade...30 Figure 2.31 : The poverty-growth scissors: 1981-2005..................................................................................31 Figure 2.32 : The evolution of income inequality in China since the start of reforms .....................................32 Figure 2.33 : The distribution of income in China in 2003..............................................................................33 Figure 2.34 : Income inequality in China from a cross-country perspective...................................................33 Figure 2.35 : China stands out globally in the magnitude and pace with which incomes and income inequality have increased .......................................................................................................................................34 Figure 2.36 : The rural-urban income gap has grown in recent years............................................................34 Figure 2.37 : Income inequality within rural and urban areas has increased .................................................35 Figure 2.38 : Inequality within rural and urban areas has become an increasingly important component of overall inequality.....................................................................................................................................36 Figure 2.39 : The burden of health and education expenditures has increased for rural households............38 Figure 2.40 : Estimates of urban unemployment rates and labor force participation rate ..............................38 Figure 2.41 : Recent trends in the number of rural migrant workers.............................................................40 Figure 2.42 : Migrants are living in the cities for long periods of time ...........................................................41 Figure 4.43 : The growing difference between rural registered and rural resident population in China, 1978- 2005........................................................................................................................................................49 Figure 4.44 : Regional shares in the incidence, depth and severity of consumption poverty in 2003 (using World Bank poverty line).........................................................................................................................53 Figure 4.45 : Topography-minority category shares in the incidence, depth and severity of consumption poverty in 2003 (using World Bank poverty line) ....................................................................................55 Figure 4.46 : Village-level concentration of poverty and the potential gains from village vs. household targeting for rural China, 2001 ................................................................................................................56 Figure 4.47 : Poverty concentration declines at lower levels of poverty across region-topography-ethnicity categories in rural China, 2003...............................................................................................................57 Figure 4.48 : The relative risk of poverty amongst the women and elderly in China, 2003 ............................62 Figure 4.49 : Relative probability of being disadvantaged and share amongst the urban disadvantaged for adults (age 16 and above) by employment category, 2003 ....................................................................74 Figure 5.50 : Completion of plans and initiation of investments in designated poor villages..........................81 Figure 5.51 : Percent of villages in each income per capita quintile designated as poor in 2001 ..................81 Figure 5.52: Potential gains from improved targeting under village and county targeting..............................83 Figure 5.53 : Fraction of the income poor living in designated poor villages..................................................83 Figure 5.54 : Potential errors in designating poor villages .............................................................................85 Figure 5.55 : Rising tide of rural migrants, 2001-2007 ...................................................................................94 Figure 5.56 : Sources and destinations of migrant workers from rural areas, 2003.......................................96 Figure 5.57 : Destinations by source of migrant workers from rural areas in 2003 ........................................97 Figure 5.58 : The 30 largest inter-provincial migration flows, 1995-2000.......................................................98 Figure 5.59 : Two-thirds of the rural migrants went to medium and large cities in 2004 ................................98 Figure 5.60 : The majority of rural migrants worked in manufacturing and construction in 2004 ...................98 Figure 5.61 : Male and female migration rates by age, 2003.........................................................................99 Figure 5.62 : Male and female migration rates by education level, 2003.....................................................100 Figure 5.63 : Migration reduced the probability of poverty for rural households, 2003.................................101 Figure 5.64 : The relationship between migration probability and log income per capita at the household level, 2003 ............................................................................................................................................102 Figure 6.65 : An egalitarian distribution of land in China, 2004....................................................................110 Figure 6.66 : An unchanging concentration of land over 20 years...............................................................111 Figure 6.67 : Farm yields vary significantly across counties ........................................................................111 Figure 6.68 : Predicted effects of a 10% increase in farm yields on poverty, 2003......................................112 Figure 6.69 : Declining incidence of major land readjustments in designated poor and non-poor villages ..114 Figure 6.70 : Rural Di Bao thresholds vary enormously across villages, 2004 ............................................122 Figure 6.71 : Social assistance benefits depend on village revenues, 2004................................................123 Figure 6.72 : Rising share of private and unregistered businesses in urban employment ...........................128 Figure 6.73 : China's declining urban labor force participation rate: 1996 to 2006 ......................................132 Figure 6.74 : Lower coverage of social insurance for the poor and the bottom income quintiles in Chinese cities, 2005 ...........................................................................................................................................139 Figure 6.75 : Urban social insurance coverage under informal and formal employment in 2002.................140 Figure 6.76 : Higher income levels are associated with higher per-capita contributions and benefits of urban BMI and pension programs across provinces, 2005.............................................................................141 Figure 6.77 : Targeting performance of Urban Di Bao, 2003 .......................................................................148 Figure 6.78 : Does Di Bao reach the poorest population in urban areas? ...................................................150 Figure 6.79 : Horizontal equity of Di Bao program.......................................................................................151 Figure 6.80 : Incentive effects of the urban Di Bao program........................................................................153 Figure 7.81 : Labor productivity differentials between sectors and rural and urban areas ...........................158 Figure 7.82 : Differences in human capital endowments between rural and urban areas in 2003...............159 Figure 7.83 : Differences in per-capita incomes between rural and urban areas in 2003, controlling for human capital endowments..................................................................................................................159 Figure 7.84 : The cost of schooling is high for many rural households ........................................................161 Figure 7.85 : The rising skill premium in urban China: 1988 to 2003...........................................................163 Figure 7.86 : Regional urban wage differences in 2003...............................................................................164 Figure 7.87 : There are signs of growing integration of urban labor markets...............................................164 Figure 7.88 : Health and education outcomes at the provincial level are more closely tied to incomes than they used to be .....................................................................................................................................166 Figure 7.89 : Senior secondary enrolments are less than half of enrolments at the junior secondary level.169 Figure 7.90 : Large disparities in local government revenues (provincial government revenue, yuan per capita, 2006).........................................................................................................................................171 Figure 7.91 : The provincial distribution of expenditures in 2004 favored the richer provinces....................172 Figure 7.92 : Poverty-related expenditures in 2004 were slightly more pro-poor than total expenditures....173 Figure 7.93 : The village-level distribution of expenditures in 2004 was even less pro-poor........................174 Figure 7.94 : The contribution of rural-urban income differences to overall inequality varies by region.......175 Figure 7.95 : Child sex ratio amongst the poor and non-poor in China, 2003.............................................175 Figure 7.96 : Housing conditions of local residents and migrants................................................................184 Figure 7.97 : Social insurance coverage of local residents and migrants ....................................................184 Figure 8.98 : Combined central and local government revenues have been growing very rapidly in recent years.....................................................................................................................................................202 Preface and acknowledgements This report brings together findings from a multi-year analytical activity undertaken by the World Bank on a policy-oriented assessment of poverty in the People's Republic of China. The report was prepared by a poverty assessment team led by Gaurav Datt and Shubham Chaudhuri from the East Asia Poverty Reduction and Economic Management Department of the World Bank. The other members of the core team who contributed to the analytical work undertaken for the report and its drafting included Albert Park (Oxford University), Sangui Wang (School of Agricultural and Rural Development, Renmin University), Yaohui Zhao (China Center for Economic Research, Peking University), and Fang Cai (Institute of Population and Labor Economics, Chinese Academy of Social Sciences). The larger team also included Tao Kong (Australian National University), Dewen Wang and Du Yang (Institute of Population and Labor Economics, Chinese Academy of Social Sciences), Liu Xuejun (Beijing Normal University). The poverty assessment study was conducted and completed by the World Bank, and reflects the Bank's latest analysis, judgments and views on poverty and inequality issues in China. The State Council's Leading Group Office of Poverty Alleviation and Development (LGOPAD) provided a great deal of support in the process of the study and the revisions of the report, as well as coordinated the process of seeking comments from other Chinese agencies and experts involved in the poverty reduction program. In particular, we are grateful to Zhang Lei, Wu Zhong, Huang Chengwei, Tian Weiping and Cao Hongmin for their positive contribution throughout the process of the poverty assessment study. The primary counterparts on the technical side were the Rural and Urban Household Survey Organizations of the National Bureau of Statistics with whom collaborative work arrangements were established that enabled detailed analyses of data from their rural and urban household surveys. The poverty assessment team is grateful to the NBS for not only giving access to the household survey data, but also for providing technical advice and feedback at different stages of the work, and actually participating in the analytical work in the course of four data analysis workshops organized by the poverty assessment team during 2004-6 in Washington DC, Beijing and Sydney. In particular, we would like to acknowledge our gratitude to Sheng Laiyun, Yang Junxiong, Yan Fang, Chen Xiaolong and Wang Youjuan for their generous support. We are also grateful to the Rural Survey Organization of NBS for conducting the 2004 NBS-World Bank Special Purpose Village-Level Survey covering 3037 villages throughout China for the purposes of this poverty assessment. We are also grateful to the Institute of Population and Labor Economics, CASS, for organizing and conducting the China Urban Labor Survey in 5 large and 5 small cities in 2005 for this poverty assessment, and to the Institute of Sociology, CASS, for conducting a qualitative study of poverty in 12 villages across six provinces in China. Our thanks also go to John Taylor for coordinating this qualitative study. On November 3, 2005, a draft of the report was released at a workshop jointly hosted by the World Bank and LGOPAD. Suggestions of various participants at the workshop were very helpful in the preparation of this report. On February 27, 2008, a group of Chinese experts were invited by the World Bank and LGOPAD for a discussion of the draft of the report. The twelve Chinese experts gave a thorough evaluation of the report and put forth many useful suggestions for revising the report. From September to November 2008, with the support of LGOPAD, we also asked for suggestions from the National Reform and Development Commission, the Ministry of Agriculture, the Ministry of Human Resources and Social Security, the Ministry of Civil Affairs, the Ministry of Health, the Ministry of Education, the National Bureau of Statistics P r e f a c e a n d a c k n o w l e d g e m e n t s i and other government counterparts, and received a lot of positive feedback. This has played a significant role in improving the report. The support from DFID for this work from the Trust Fund under the World Bank-DFID China AAA Partnership is gratefully acknowledged. Interaction and exchange of ideas with Christopher Athayde, Arjan De Haan, Rahul Malhotra, and Jillian Popkins from DFID was also very helpful in the development of the work for this report. The report benefited from the comments of peer reviewers Emmanuel Jimenez (Sector Director, EASHD), Ravi Kanbur (Cornell University), Justin Yifu Lin (Chief Economist, World Bank), Shahid Yusuf (Economic Adviser, DECRG), and Christine Wong (University of Washington). The report was prepared under the overall guidance of Vikram Nehru (Sector Director, EASPR). Strategic guidance was also provided by David Dollar (Country Director, China), and Bert Hofman (Country Director, Philippines, and ex-Lead Economist for China, EASPR). The report also benefited from the guidance of Homi Kharas and Indermit Gill during their tenures as Sector Director and Sector Manger at EASPR. We are also grateful to Ardo Hansson (Lead Economist for China, EASPR) for support and guidance during the final preparation and review of the report. Jianqing Chen and Zijing Niu provided invaluable project management throughout this activity for which we are extremely grateful. Louisa Huang helped us with the management of the DFID Trust Fund. A number of people helped in the preparation of this report by providing useful feedback, comments and various other forms of input at different stages of the work undertaken for the report. From within the World Bank, we are grateful for this support to Deepak Bhattasali, Shaohua Chen, Klaus Deininger, Achim Fock, John Giles, Li Guo, Louis Kuijs, Magnus Lindelow, Xiaofan Liu, Philip O'Keefe, Alan Piazza, Martin Ravallion, Adam Wagstaff, Kin Bing Wu, and Xiaoqing Yu. Amongst those from outside the World Bank, we are thankful to Sara Cook, Jikun Huang, Scott Roselle, and Pinping Wang. P r e f a c e a n d a c k n o w l e d g e m e n t s i i Executive summary China's record of poverty reduction and growth is enviable China's record over the China's progress in poverty reduction over the last 25 years is enviable. One cannot fail last quarter century in to be impressed by what this vast nation of 1.3 billion people has achieved in so little time. reducing extreme poverty In terms of a wide range of indicators, the progress has been remarkable. Poverty in has been remarkable terms of Income and consumption has been dramatically reduced. Progress has also been substantial in terms of human development indicators. Most of the Millennium Development Goals have either already been achieved or the country is well on the way to achieving them. As a result of this progress, the country is now at a very different stage of development than it was at the dawn of the economic reforms at the beginning of the 1980s. Between 1981 and 2004, By China's official poverty standard, the poverty rate (headcount ratio) in rural China fell the fraction of the from 18.5% in 1981 to 2.8% in 2004 and the number of rural poor declined from 152 population consuming million to 26 million. Measured in terms of the World Bank poverty standard of (of 888 less than a dollar-a-day yuan per person per year at 2003 rural prices), China's poverty reduction performance has fell from 65% to 10%, and more than half a billion been even more striking. Between 1981 and 2004, the fraction of the population people were lifted out of consuming below this poverty line fell from 65% to 10%, and the absolute number of poor poverty fell from 652 million to 135 million, a decline of over half a billion people (Figure 0.1). A fall in the number of poor of this magnitude over such a short period is without historical precedent. To put this in perspective, the absolute number of poor in the developing world as a whole declined from 1.5 to 1.0 billion over the same period (World Bank, 2007); in other words, but for China there would have been no decline in the numbers of poor in the developing world over the last two decades of the 20th century. Measured by the new international poverty standard of $1.25 per person per day (using 2005 Purchasing Power Parity for China), the levels of poverty are higher, but the decline since 1981 is no less impressive (from 85% in 1981 to 27% in 2004). Figure 0.1 : China's record of poverty reduction and growth over the last quarter century has been remarkable 70 3200 6th Plan 7th Plan 8th Plan 9th Plan 10th Plan 60 2800 )%( World Bank consumption HCR Per capita GDP 2400 50 ces)ipr HCR)(yt 2000 1978 40 at pover ofeatr 1600 uanY( 30 Income HCR 1200 GDPat 20 800 capi Headcount Per 10 Official rural HCR 400 0 0 1981 1982 1983 1984 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 Sources and notes: See main report. Between 2001 and 2004 The decline in the incidence of poverty was, however, uneven over the two and a half the pace of poverty decades since 1981. Five broad phases can be distinguished, broadly coinciding with the reduction accelerated, five Five-year Plan periods. The most rapid declines in poverty, in both the poverty rate and there are indications and the number of poor, occurred during the 6th, 8th, and 10th Plans. During the 7th Plan E x e c u t i v e s u m m a r y i i i that poverty has period the number of poor actually rose, while in the 9th Plan period, the poverty rate continued to decline declined only marginally. But the pace of poverty reduction resumed between 2001 and rapidly up to 2007 2004 and there are indications that during the first couple of years of the 11th Plan poverty has continued to decline rapidly. The percentage of China's population consuming less than the World Bank poverty line is estimated to have fallen from 10.3% in 2004 to 4.0% in 2007 (Figure 0.2). Figure 0.2 : Progress in poverty reduction during 2005-2007 (World Bank poverty line: 888 yuan per person per year at 2003 rural prices) Headcount Poverty gap Squared poverty gap 9.4 10 8 6.9 6.7 5.4 6 4.9 4.0 4 1.6 2 1.20.3 0.80.2 0.60.2 0.5 1.10.3 0.80.2 0 2005 2006 2007 2005 2006 2007 National Rural Source and notes: See main report. Growth has been central The rapid growth that has resulted from the series of economic reforms that China to China's poverty adopted as it transitioned from a planned to a market-oriented economy has been central reduction performance... to China's poverty reduction performance. During the period 1981-2005 as a whole, real per capita GDP grew at the trend rate of 8.3% per year, while the headcount index of poverty fell at the rate of 7.1% per year. Thus, every 10% increase in per capita GDP was associated with a 9% fall in the incidence of poverty. Key reforms have included the introduction of the Household Responsibility System in agriculture in the early 1980s, creation of a conducive environment for the rise of Township and Village Enterprises in rural areas, which up to the mid-1990s grew rapidly to absorb a large share of the rural labor force, restructuring of the state industrial sector in the late 1990s, the opening up of the economy to global trade and investment and the resultant rapid growth of the urban economy which has been absorbing an ever-growing mass of migrant workers from rural areas. Policies that have directly supported rural incomes, such as the increase in agricultural procurement prices between 1993 and 1995, the elimination of agricultural taxes and fees between 2003 and 2006, the developmental poverty reduction programs and the recent expansion of social assistance have been important as well. But the task of poverty reduction continues and in some respects has become harder Extreme poverty, in the The most recent official estimate of rural poverty in China for 2007 puts the number of sense of not being able to poor at 14.79 million, or less than 2% of the rural population (NBS, 2008). While there is meet the most elementary no official urban poverty line, estimates by others have found poverty levels in urban areas food and clothing needs, to be negligible using an urban poverty line that is comparable to the official poverty line has almost been eliminated in China for rural areas. These estimates thus suggest that only about 1% of China's population is currently in extreme poverty. In other words, extreme poverty, in the sense of not being able to meet the most elementary food and clothing needs, has almost been eliminated in China. But the task of poverty Notwithstanding this tremendous success, the central thesis of this report is that the task reduction continues and of poverty reduction in many ways continues and in some respects has become more in some respects has demanding. China's successes have bred their own challenges. The many factors-- become harder policies as well as processes--that have contributed to the past success have also brought about structural changes, profoundly transforming the country's economic and social landscape, and in the process have produced new challenges. E x e c u t i v e s u m m a r y i v Thus, the task of poverty reduction continues and in some respects has become harder because: Measured by · Measured by international standards for identifying and counting the poor, the number international standards, the number of poor in of poor in China remains high. Because of the sheer size of China's population, using China remains high an international poverty standard, China still has the second largest number of consumption poor in the world after India. As of 2005, the latest year for which direct survey-based estimates are available, China still had 254 million people consuming less than $1.25 per day in 2005 PPP dollars. If this assessment of the number of remaining poor in China appears to be at odds with the official estimates of only 15 million rural poor, it is because the official poverty line (set at 785 Yuan per person per year for 2007, or $0.57 per person per day in 2005 PPP dollars) is particularly stringent, and is the lowest amongst a sample of 75 countries (Figure 0.3). The official poverty line seems low not only compared to international standards, but also relative to the rapid rise in mean incomes and growing aspirations within China. Furthermore, recent calculations suggest that even by the objective standard that a poverty line aims to capture--the minimum expense necessary for subsistence and to meet basic living needs--the official poverty line may be too low to adequately cover basic food and nonfood needs. Figure 0.3 : China's poverty line in comparison with other countries 9 Poverty lines across countries 8 (dollars per person per day at 2005 PPPs) 7 6 5 4 3 2 1 0 Sources and notes: See main report. Vulnerability to poverty is · Vulnerability to poverty because of a variety of income shocks remains widespread. widespread, especially in rural China, and number For instance, using the World Bank poverty line, nearly a third of China's rural of those vulnerable to the population was consumption poor at least once between 2001 and 2004, about twice risk of falling into poverty the number who were poor in any one year (Figure 0.4). Nearly 70% of the severity is about twice as high as of income poverty and 40% of the severity of consumption poverty in rural China is the number of poor in a attributable to risk. Even for those persistently poor, exposure to risk adds to the given year severity of their poverty. And the exposure to risk and the resulting need for precautionary saving may be one reason why, in 2003, 43% of rural households with per-capita annual incomes below the poverty line were saving, and amongst households with per capita incomes between the poverty line and twice the poverty line, the median savings rate was 17.5%. If domestic consumption is to supplement investment and external trade as a central driver of China's economic growth, it will be important to reduce vulnerability to poverty. E x e c u t i v e s u m m a r y v Figure 0.4 : Almost a third of China's rural population was consumption poor in at least one year between 2001 and 2004 (% of the rural population that was dollar-a-day consumption poor in one or more years) Fraction (%) of rural population 100.0 80.0 60.0 69.1% 40.0 15.1% 30.9% 17.8% 20.0 6.9% 8.9% 0.0 lla 2 3 1 3 r ta e y in in in 1 in years the the poo any year oorP 3 Poor of years Poor of years in Not oorP least year verageA cidenc in povertfo The incidence of persistent and transient poverty in rural China in the three years: 2001, 2003 and 2004 Source and notes: See main report. · Across regions and topographic/ethnicity categories, the relative contribution of risk rises as poverty becomes less severe, which suggests that as China makes further progress in poverty reduction, the poverty challenge facing the country will increasingly take the form of tackling transient poverty. A relatively high share of transient poverty however does not mean that the problem of poverty has been rendered necessarily less serious or its solutions relatively easier. But, it does imply that social policies may have to focus on risk mitigation and risk management strategies more than was necessary when chronic poverty was the dominant form of poverty. As poverty rates have · Similarly, as the poverty rate has fallen, it has become harder to eliminate the fallen, the remaining poor are harder to reach as remaining poverty because the remaining poor are more dispersed. While the they are more dispersed incidence and severity of poverty is the highest in western provinces, nearly half the poor are dispersed in the rest of China, and likewise, while poverty is the most severe in mountainous and minority areas, more than half the poor are in non-mountainous non-minority areas. And as poverty levels are reduced, the village-level concentration of poverty tends to decline (Figure 0.5). This carries the important implication that as China makes further progress in alleviating poverty, the remaining poverty could be expected to be more dispersed thus eroding some of the potential benefits from area- based targeting relative to household-based targeting approaches. Figure 0.5 : Poverty concentration declines at lower levels of poverty across region- topography-ethnicity categories in rural China, 2003 1.0 0.9 0.8 index 0.7 0.6 iont 0.5 0.4 0.3 Concentra0.2 0.1 0.0 0 10 20 30 40 50 60 Poverty incidence (%) Sources and notes: See main report. E x e c u t i v e s u m m a r y v i While economic growth · Economic growth has been critical for poverty reduction. The only period in the last has been critical for poverty reduction, the quarter century when there was an increase in the poverty rate, albeit a relatively responsiveness of small one, was during the 7th Five Year Plan, between 1986 and 1990 when the poverty to economic growth rate fell to less than 4%. However, the task of poverty reduction has become growth has decreased harder because the poverty is now less responsive to economic growth. Whereas during the 6th Five Year Plan each percentage point of growth in real per capita GDP was associated with more than 2% reduction in poverty headcount rate, during the 10th Five Year Plan, the elasticity of poverty reduction with respect to growth had declined to about 1. Table 0.1 : Responsiveness of poverty to growth has declined, 1981-2005 Elasticity of headcount index with respect to real Annual percentage change in: per capita GDP: Per capita GDP Controlling Headcount (constant Gini Un- for index prices) index conditional inequality 6th Plan 1981-1985 -26.1 10.2 -0.8 -2.52 -2.40 7th Plan 1986-1990 7.1 5.8 2.5 0.95 -0.43 8th Plan 1990-1995 -10.2 11.2 2.9 -0.90 -1.36 9th Plan 1996-2000 -3.4 7.0 2.3 -0.48 -1.54 10th Plan 2001-2005 -9.0 8.9 1.4 -1.03 -1.55 1981-2005 -7.1 8.3 2.0 -0.87 -1.79 Sources and notes: See main report. Income inequality has · Not everyone has participated in the economic success equally. Income inequality in risen significantly because of a widening China has increased significantly since the start of economic reforms, and China is no rural-urban gap and longer the low-inequality country it was a quarter century ago. The (unadjusted) Gini increasing inequality index of income inequality rose from 30.9% in 1981 to 45.3% in 2003 (Figure 0.6). within both rural and Even adjusting for rural-urban cost of living differentials, the Gini index increased from urban areas 32.9% in 1990 to 44.3% in 2005, a level comparable to that in many other middle- income economies, including some in the East Asia region such as Thailand and Malaysia, though lower than that in many Latin American countries. Where China stands out is in the magnitude of the increase in inequality and the pace at which it has occurred. The rise in inequality is the result of both a widening income gap between the cities and the countryside, as well as growing inequality within rural and urban areas. Figure 0.6 : The evolution of income inequality in China since the start of reforms (Gini index of income inequality from 1981 to 2005) Adjusted for rural-urban cost of living differences 50 45.3 44.3 45 Without adjustments for 40 spatial cost of 41.1 35 living differences Adjusted for 30 spatial cost of living differences 25 1980 1985 1990 1995 2000 2005 Source and notes: See main report. E x e c u t i v e s u m m a r y v i i Disparities in other · China's successes have not only been in raising incomes. China's progress in human aspects of human development persist and development has been equally, if not more, impressive. In terms of various human mirror the disparities in development indicators, China compares favorably to levels achieved in middle- income... income countries. But as in the case of income growth and poverty reduction, the progress has been uneven, and disparities in many non-income aspects of human development have grown in recent years across rural and urban areas, provinces and households. ... because incomes · That disparities in human development mirror disparities in income is largely due to matter more now than they used to for access the fact that incomes matter more now than they used to in determining access and and outcomes in health outcomes in health and education, especially in rural areas (Figure 0.7). Incomes and education, and the matter more now than they used to because the change in the institutional burden of health and arrangements for the provision of health and education led to the marketization and education expenditures monetization of service delivery. The dismantling of the communes in rural areas and for the poor has the move to the Household Responsibility System was arguably the single most increased important reason for the rapid decline in poverty China experienced in the first half of the 1980s. However, a byproduct of this institutional transformation was the increasing monetization and marketization of public service delivery in rural areas. In place of the communes, local governments were charged with administering and financing these services, which in poorer areas, they were unable to fully do because of a lack of fiscal resources. Schools and health facilities have therefore had to increasingly rely on charging user fees in order to cover their costs. Figure 0.7 : Health and education outcomes at the provincial level are more closely tied to incomes than they used to be Infant survival rate (logit) High school enrollment rate (%) 5.5 100 5.0 2003 80 4.5 60 4.0 1981 2000 40 3.5 estimated 1990 1990 best-fit line 3.0 20 2.5 0 5.25 6.25 7.25 8.25 9.25 10.25 5.5 6.5 7.5 8.5 9.5 Provincial GDP per-capita (log) Provincial mean per-capita income (log) Sources and notes: See main report. · A direct consequence of this is that the burden of health and education expenditures has increased for rural households. For instance, it is estimated that the share of educational expenditure in the household budget increased from 1.0 to 8.3 percent between 1988 and 2003, while the budget share of health expenditures increased from 1.6 to 5.1percent over the same period. Increases in the education and health budget shares for the bottom deciles were especially burdensome. With large-scale · Beginning in the mid-1990s, China undertook aggressive restructuring of loss-making restructuring of state- owned enterprises, state-owned enterprises, leading to the layoffs of tens of millions of urban workers. unemployment has risen Between 1994, when it was at its peak, and 2006, employment in state-owned and labor force enterprises and urban collectives fell by 73 million, from 145 million workers to 72 participation has declined million. The restructuring of the urban labor market and the dismantling of the iron rice in urban areas bowl led to rising unemployment and declining labor force participation rates (Figure 0.8). Unemployment rates have fallen from their peak around five years ago, but are still higher than they were in mid-1990s. The urban labor force participation rate fell by 10 percentage points between 1999 and 2003 and remains much lower than in the mid-1990s. Because the decline was associated with restructuring, there has been concern that many of those leaving the labor force are discouraged workers who do not get counted among the ranks of officially unemployed. Relative poverty in urban areas is increasingly associated with the work status of household members. E x e c u t i v e s u m m a r y v i i i Figure 0.8 : Estimates of urban unemployment rates and labor force participation rate (%) 12.0 CULS-based 75 ) ) UER: urban (% (%)RE 10.0 residents rate (U Labor force 70 no 8.0 ratetne participation rate atipic rti myolp 6.0 pa Labor Force 65 me Survey-based force 4.0 UER Official Un registered UER Labor 2.0 60 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 Sources and notes: See main report. Growing informalization · A corollary of the downsizing of the state sector in urban areas has been the of the urban labor market raises concerns about diversification of the urban labor market, with non-state entities accounting for a the welfare of workers growing share of urban employment since the mid-1990s. The fastest growing segment has been the category of "other" unregistered workers--unreported workers in registered enterprises, workers in unregistered informal enterprises and undocumented rural migrants in urban areas. The large increase in the number of such workers suggests that informal urban employment has increased significantly in recent years, and may account for nearly half of all urban employment (Table 0.2). The growing informalization (and diversification away from the state sector) of the urban labor market, while it has undoubtedly helped create jobs and facilitated the allocation of labor according to market principles, has also exposed urban workers to new forms of insecurity because informal employment is usually not contract-based and by its nature is "hidden". That makes it harder to enforce protective regulations that ensure safe working environments and fair treatment of workers. It also poses a real challenge for developing and adequately financing sustainable social insurance systems that can offer adequate protection to the urban population. Table 0.2 : Share of informal employment for local residents in urban China in 2002 Share of Share of Share of informal informal informal employ- employ- employ- ment ment ment (%) (%) (%) All 45.3 By age: By education level: By sector: 16-24 57.2 Primary and below 78.4 State and collective 22.9 25-34 44.6 Middle school 62.6 Other sectors 56.3 35-44 42.2 High school 48.6 By sex: 45-54 42.2 Junior technical school 35.9 Male 42.2 55-64 55.6 Specialized college 24.6 Female 49.5 >65 83.9 College and above 14.8 Source and notes: See main report. The emergence of a · As an integral part of its growth process, China has witnessed internal migration from "floating population" poses continuing rural to urban areas on an unprecedented scale during the last decade a half, with the challenges for social numbers of migrants continuing to grow in recent years (Figure 0.9). Described as policy the largest peacetime movement of people in history, large numbers of rural workers have migrated to the cities to work, pulled by the jobs in China's fast-growing manufacturing and service industries, and pushed by the growing pool of surplus labor resulting from reforms in the agricultural sector. The result has been the emergence of a large (and still growing) "floating population"--estimated to be as E x e c u t i v e s u m m a r y i x many as 150 million people--who live and earn their livelihoods for increasingly longer periods each year in locations where, lacking a local (urban) resident status, they often have limited and highly variable access to basic social services, potentially suffer discrimination in access to economic opportunities, and are often treated as no better than "guest workers". This has raised concerns about the emergence of an urban underclass whose coverage under the existing social protection programs and policies remains a continuing challenge. Figure 0.9 : Rising tide of rural migrants, 2001-2007 (Number of rural migrants, million) 137.0 140 132.1 125.8 118.2 120 113.9 104.7 100 84.0 80 2001 2002 2003 2004 2005 2006 2007 Sources and notes: See main report. New systems of social · The dismantling of the collective commune system in rural areas and the large-scale protection are still evolving and many economic restructuring of the state and collective sector in urban areas completely challenges remain overturned the foundations of the earlier social welfare and social security system. While China is well on its way to establishing new systems of social protection in urban and rural areas, this is still work-in-progress and many challenges remain. Policy initiatives in response to these challenges suggest a broader poverty reduction agenda is evolving These developments and challenges suggest that the task of poverty reduction needs to be viewed in broader terms. Partly as a result of China's economic success, and partly as a result of wide-ranging economic reforms that have profoundly transformed the structure of the economy, the nature of the (poverty) problem to be tackled has changed in important ways. The changing nature of the problem suggests that the task can not be limited just to the established focus on poor-area development, but that there is a case for broadening the agenda not just in terms of the spatial coverage of poverty reduction efforts, but also in terms of the basic objectives of such efforts, and the range of instruments and approaches adopted: · First, the notion of income and consumption poverty itself may need to be broadened beyond the austere threshold of meeting "survival" needs of food and clothing to one that meets a more generous set of basic needs and capabilities for the country's population. · Second, policy and structural changes as well as the important contribution of risk to current poverty point to a large social protection agenda in both rural and urban areas. · Third, the experience with marketization of public services that eroded the near- universal access to basic health and education in the pre-reform period indicates the need for renewed focus on re-attaining this goal as part of the overall poverty reduction agenda. · Fourth, the challenge of making the process of growth more pro-poor suggests the need to focus on ways of increasing opportunities for the poor to participate more fully E x e c u t i v e s u m m a r y x in the growth process not just by increasing agricultural incomes but also through off- farm employment and realizing the potential of migration for poverty reduction. · Fifth, greater dispersion of the poor suggests there may be a case for rebalancing the focus of poverty reduction efforts from poor areas to poor people. · Sixth, the rapid increase in inequalities between as well as within rural and urban areas suggests that the task of fostering equity can not be delinked from the overall poverty reduction agenda. A poverty reduction Implicit in this call for a broadening of the poverty reduction agenda is the recognition that agenda consistent with China has had tremendous success in the last quarter century in eliminating the most the vision of a xiaokang extreme forms of chronic poverty. During this period, China has also become much society wealthier and more prosperous. Both developments suggest that China is now at the stage where it can take on the challenge of a broader poverty reduction agenda consistent with the vision of a xiaokang society that is better-off in an all-round way. The changes and trends described above have not gone unnoticed in government circles. On the contrary, a number of recent policy initiatives indicate an increasing commitment of the government to a broader poverty reduction, social protection, and human development agenda. Key policy responses to the emerging challenges have been evident in recent years through such initiatives as: · The launch, in 2000, of the Western Region development strategy, aimed at 'opening up the West', through construction of infrastructure, support for human capital formation, environmental protection, improvements in regional investment climate and local natural-resource-based industrial development. · Restructuring of poverty alleviation investments. Area-based poverty investment programs were reoriented in 2001 with a shift in focus from the 592 designated national poor counties to 148,000 poor villages. This was partly in recognition of the dispersed nature of poverty, and the fact that many of the poor did not live in the designated poor counties and the poor counties also had many non-poor. · Development of the urban social security system. While there is no official urban poverty line, the government has been rapidly developing an urban social security system following the economic restructuring of the public sector since the mid-1990s and the shattering of the "iron rice bowl". The new and evolving urban social security system has had three main elements: (i) assistance to laid-off workers (xiagang) from state-owned enterprises, which has now been integrated with unemployment insurance, (ii) a means-tested minimum income support (Di Bao) program, and (iii) social insurance programs including pensions, medical, unemployment, work injury and maternity insurance. The pace of efforts has quickened substantially since 2003: · Training program to support transfer of rural surplus labor. In 2004, the government launched a multi-ministry program ­ called the Sunshine program for Training and Transferring Rural Labor Force ­ which provides short-term training to farmers leading to their transfer to non-agricultural employment often in urban areas. The program aims to train and transfer up to 40 million farmers by 2010. · Elimination of agricultural taxes. The central government began to remove all local informal fee charges on farmers but raised the formal agricultural tax rate from 2-3% to 7% in 2002 to compensate partially for local revenue shortfalls. Local governments were subsequently asked to cut this tax rate by 1-2% per year, and by the end of 2006 all agricultural taxes and fees were fully phased out. · Supporting farm incomes. During the 11th Five Year Plan period, agricultural policies have moved from net taxation to net subsidization of agriculture. Direct subsidies to agriculture, including grain and input subsidies (for fuel and fertilizers), increased 3.5 times between 2004 and 2007. · A nationwide rural social assistance system. Perhaps, one of the most important developments has been the decision in March 2007 to set up a nationwide rural social assistance system--the Di Bao or minimum subsistence allowance program-- financed in part by the central budget. A similar program--the urban Di Bao--has been in place in urban China since the late 1990s and currently provides income E x e c u t i v e s u m m a r y x i support to about 22 million urban residents. In rural areas, until recently only some provinces and counties had such an income support program. However, under the new initiative, the rural Di Bao program has been rolled out on a nationwide basis, and by the end of 2007, the number of rural Di Bao beneficiaries had grown to nearly 35 million. · A rural health insurance scheme. Another major initiative has been the scaling up of the National Cooperative Medical Scheme (NCMS) in rural areas. The NCMS is a new voluntary rural health insurance scheme that is operated by counties and subsidized by the local and central governments. Introduced on a pilot basis in 2003, the NCMS has expanded rapidly to cover about 86% of all counties by the end of 2007. · Urban residents' basic medical insurance. Complementary to the existing basic medical insurance scheme for urban formal sector workers, a new urban residents' basic medical insurance scheme was introduced in 2007 (following pilots in 2005) to expand urban health insurance to urban unemployed residents, students and children. · A medical assistance scheme in rural and urban areas. A related initiative has been the rapid roll out of the Medical Assistance scheme in rural and urban areas to provide financial assistance to the poor and vulnerable groups in rural and urban areas with their health insurance contributions and copayments. Piloted in rural areas in 2003 and in urban areas in 2005, the scheme had expanded swiftly to cover all rural counties and 86% of cities by the end of 2007. · Compulsory education finance reform. A major new initiative in education has been the introduction of free compulsory education. Introduced in March 2006, the compulsory education finance reform aims to promote universal primary and junior secondary education by repealing tuition/miscellaneous fees and textbook fees for all students, and by providing boarding subsidies to poor students (the so-called "two exemptions and one subsidy" policy). The reform is financed through large-scale intergovernmental transfers. Substantial policy development and implementation is thus already evident along the lines of what could be considered a broader poverty reduction agenda. However, many challenges remain, and this report's assessment of both the evolving poverty situation and the key programs and policies suggests a number of priorities for further action. What the review suggests about the main priorities for poverty reduction and what is needed to implement them Adopting a broader conception of poverty and an adequate threshold for identifying and targeting the poor It may be time for China To begin with, it may be time for China to consider establishing a higher threshold to to consider adopting a measure poverty that could be consistently applied to both rural and urban areas. As higher threshold for suggested earlier, the current official poverty line of 785 Yuan per person per year at 2007 identifying and targeting rural prices (or about 71 cents a day in terms of the 1993 PPP dollars or about 57 cents a the poor day in terms of the new 2005 PPP dollars or about 31 cents a day using the current exchange rate) is arguably low ­ by international standards (in fact, it is one of the lowest in the developing world), relative to average incomes and growing aspirations within China, and even by the objective standard of the minimum expense currently needed to meet basic food and nonfood needs. The principle of a higher poverty threshold was endorsed by President Hu Jintao in his speech to the 17th Party Congress in October 2007. Adopting a higher poverty threshold is however only an initial step in the formulation of a forward-looking poverty reduction strategy. But it is an important first step as it can be critical for focusing policy attention on the right target group ­ one that is more appropriate for China's current stage of development. The new threshold should form the basis of not only the measurement and monitoring of poverty in the country, but more importantly, the development of a consistent operational approach to targeting a range of poverty alleviation and social protection programs to the poor in rural and urban areas. E x e c u t i v e s u m m a r y x i i It may be also worth considering measuring poverty in terms of consumption rather than income. While this report presents both income and consumption-based analysis, there are well-known reasons for preferring consumption over income, the main one being that consumption is a more stable measure of welfare. The emphasis on consumption is also in keeping with the current national goal of raising the share of consumption in GDP. Moreover, in the case of China, while many of the poor consume below their income levels, it is arguable that this is mostly in the nature of saving for a rainy day with a view to maintaining their consumption levels at a relatively stable level even as their incomes may fluctuate. This too suggests that consumption may be a more stable and reliable measure of welfare. But the conception of poverty could also be broadened in two other respects. First, beyond income and consumption, access to affordable education and healthcare also ought to be seen as part of the poverty reduction agenda. Higher income and consumption can be a means for improving this access; however, income and consumption can themselves be limited by the lack of such access. The evidence suggests that inability to cope with health shocks is often an important factor in households falling into poverty, and limited human capital is often the reason for many households being unable to make use of the opportunities provided by economic growth to move out of poverty. Second, since the contribution of (uninsured) risks to observed poverty is quite significant (estimated at about 40% for consumption poverty) and there are indications that this contribution tends to rise with further progress in poverty reduction, offering social protection to groups vulnerable to poverty also ought to be seen as an integral part of the poverty reduction agenda. Thus, the poverty reduction agenda is broader than may be conventionally recognized. There are indications that in practice the government is already viewing the poverty challenge in such wider terms, but as discussed further below much still remains to be done in implementing this broader agenda. Retaining rural poverty reduction as the top priority Reduction of rural Even with a broader conception of poverty, the evidence and analysis presented in this poverty should continue report suggests that the reduction of rural poverty should remain the top priority. Whether to be recognized as a top measured in terms of income or consumption, the poor in China are predominantly in or priority from rural areas. Even if migrant workers are not included in the rural population, as is commonly done in the national household surveys, rural poor account for about 90% of all those living below the poverty line. In contrast, depending upon the precise urban-rural cost of living differentials used to measure real incomes and consumption, urban residents (excluding migrants) account for 1-3% of the total number of poor in the country. Table 0.3 : Rural-urban distribution of poverty in China, 2003 World Bank poverty line Income Consumption Share of % who are % who are Share of population poor Share of poor poor poor Rural 72.5 9.5 99.2 17.9 99.4 Urban 27.5 0.2 0.8 0.3 0.6 Total 100.0 6.9 100.0 13.1 100.0 Source: See main report. The rural-urban income While income inequality within rural areas is somewhat higher than that within urban gap has grown in recent areas, the difference in rural and urban poverty rates largely stems from the widely-noted years... rural-urban gap in mean incomes, which has been steadily growing in recent years in both absolute and relative terms. Despite large-scale migration from rural areas, this gap is now at a historical high. The rural-urban income The rural-urban income gap is however only one manifestation of the great divide that gap is only one separates China's urban and rural areas. On almost every dimension of welfare, the rural E x e c u t i v e s u m m a r y x i i i manifestation of the great population lags behind the urban population. Literacy rates, enrollment rates (especially divide that separates past primary school), and educational attainment rates (especially past junior high school) China's urban and rural are all much lower in rural areas, and while there have been improvements in all these areas indicators in rural areas in absolute terms, the gap between rural and urban areas has continued to widen. Thus, rural poverty reduction must continue to be a top priority. What perhaps has made these rural-urban differences so salient and a focus of considerable policy attention is the fact that in contrast to the significantly poorer outcomes for the rural population, the country's highly decentralized fiscal system continues to harbor an urban bias. On a per capita basis, let alone a per poor person basis, urban areas have much greater fiscal resources and these are reflected in correspondingly higher levels of public spending. For instance, in terms of publicly financed per-capita expenditures on education and health, rural areas lag far behind urban areas. Spending on social assistance and social insurance, too, is disproportionately higher for the urban population. Several recent Recently, the government has taken several steps towards addressing the rural-urban government initiatives gap. The major initiatives include: (i) the elimination of agricultural taxes and fees, and an are beginning to address increase in direct agricultural subsidies, (ii) the waiving of tuition fees for compulsory the rural-urban imbalance education in rural areas, (iii) a major expansion of rural health insurance to about 86% of counties nationwide, and (iv) a nationwide roll-out of the rural Medical Assistance scheme and the rural Di Bao program of minimum income support. These are all important initiatives which reflect increasing commitment of the government to addressing the rural- urban imbalance, and they are already beginning to be reflected in noticeable reductions in rural and hence national poverty levels. While they are unlikely to make a big dent in the rural-urban gap as urban areas continue to grow faster than the countryside, It will be important to maintain the momentum on these initiatives in future while also addressing several outstanding challenges (as discussed below) that will help deepen their impact on rural poverty reduction. But it would be imprudent Also, while an emphasis on tackling rural poverty is entirely appropriate, it would however to ignore the urban be imprudent to neglect the situation and problems of the relatively disadvantaged groups disadvantaged groups in urban areas. For several reasons. First, rural and urban poverty issues are now increasingly linked due to the large and growing flow of labor from rural to urban areas. Second, there could a degree of underestimation of urban poverty due to the inadequate coverage of migrant households in national surveys. Third, the urban labor market itself has undergone a profound transformation with the economic restructuring of the state and collective sector since the mid-1990s that has exposed the urban working population to new vulnerabilities which if left unaddressed could lead to the emergence of an urban underclass or future urban poverty. With increasing urbanization, the development of a nationally integrated approach to both rural and urban poverty reduction and social protection is an increasingly important policy imperative. Promoting opportunity by raising the returns to labor In both rural and urban In both rural and urban areas, an inability to work is only a relatively small part of the areas, inability to work is problem. While it is true that households with someone who is unable to work ­ due to only a small part of the disability or old age ­ are more likely to be poor, nearly three-quarters of China's rural poverty problem poor and all but 2% of the urban disadvantaged are in households where there is no lack of work capacity. Raising the returns to What these households lack are adequate returns to their main asset, which is their labor, labor--the main asset of and what they need is greater and more rewarding employment. Raising the returns to the the most of the poor--is labor of the rural poor and the urban disadvantaged is, therefore, critical to promoting their therefore critical opportunities for increased incomes and improved livelihoods. This can and is being done in several ways. Realizing the potential of migration for poverty reduction The scale of migration is A promising avenue for rural poverty reduction has been through migration to urban areas. large and has been The scale of migration from rural areas has increased rapidly during the 1990s. While growing rapidly estimates differ depending upon data sources and definitions, the size of migrant E x e c u t i v e s u m m a r y x i v population could be as high as 150 million making this the largest peacetime movement of people in world history. An idea of the scale and significance of this phenomenon could be had from the following basic facts (for 2003): · Roughly one in every five rural workers is a migrant worker. · About 43% of the rural population lives in households that have one or more migrant workers. · The share of migrant income in total rural income is nearly 30%. ...and it has been poverty While it is difficult to precisely estimate the contribution of migration to poverty reduction, reducing there seems little doubt that internal migration and remittances have played an important role in poverty reduction in rural areas. For instance, evidence from the NBS Rural Household Surveys indicates that households with migrant workers are (30%) less likely to be poor than households without any migrant workers. It is estimated that for 2003 alone, an additional 11.8 million persons would have been income poor in the absence of migration, or an increment in the total number of rural poor by 13%. And this is likely to be an underestimate as it does not take into account the indirect benefits of migration through productive investments supported by remittances or through the positive effects of migration on rural wages. But many of the However, migration's effects on poverty are limited by the fact that many of the poor are remaining poor are less unable to migrate. The relationship between migration probability (defined as the able to migrate... probability of a household having a migrant worker) and per capita income has an inverted-U shape (Figure 0.10). Both the poorest and richest are less likely to migrate although for different reasons. The rich have fewer incentives to migrate, while many of the poor are not able to make use of migration opportunities due to their limiting circumstances and endowments. Some of the key factors limiting migration include: the level of educational attainment of rural labor, limited access to information and labor recruitment networks especially where the informal (family and friends-based) networks are not well-established as in the case of certain minority communities, the expenses of movement and other costs associated with finding a job in cities, higher costs of health and education for migrants' children in urban areas which have been especially limiting of family as opposed to individual migration, and a degree of occupational and wage discrimination against migrants relative to local urban residents in relatively segmented urban labor markets. Figure 0.10 : The relationship between migration probability and log income per capita at the household level, 2003 0.44 Twice WB line Thrice WB line WB line holde rker 0.40 wot hous y anrg 0.36 ilitb mi a Proba ash 0.32 0.28 5.5 6 6.5 7 7.5 8 8.5 9 9.5 Predicted log per-capita income of household (without migration) Source and notes: See main report. ... and migration is also While migration has been an important means of raising incomes of rural labor, it has also creating some problems had some adverse consequences in source areas, including negative effects on in source areas educational attainments amongst the rural youth, greater vulnerability of households to work injury or disability of migrant workers, the problems of the left-behind children and E x e c u t i v e s u m m a r y x v the elderly, and the erosion of village cohesion and community participation. While some effects such as those on community participation may be difficult to mitigate in villages that have experienced a large-scale labor exodus, most of the others are indicative of policy failures related to factors such as relatively limited human capital of migrant workers, limited social insurance for migrants and their families, discriminatory rules and practices restricting migrant workers' access to urban services and ultimately limiting the scope for family migration. In view of both the impediments to mobility that are particularly relevant to the poor, and some of the adverse effects in source areas, the potential of migration for poverty reduction remains under-realized. To better realize this potential, policy initiatives in several areas will be helpful. · Improving the education and skill base of potential migrants especially those from poor households who are more deficient in these respects. Ensuring junior high school level of education for all in rural areas is a necessary first step. And in this regard, the recent introduction of free compulsory education is a very useful initiative. But beyond that there is also a need to increase public support for senior/post- secondary and technical education for the rural population if migrants are to be able to access better-paid higher skill jobs in urban labor markets. · Providing assistance for reducing information and search costs associated with migration by supplementing informal social networks with more organized placement and intermediation services. A special effort is likely to be needed to reach some of the minority groups, and those in some of the poor areas where existing informal networks are underdeveloped. · Improving investment climate in cities in the main sending regions in Central and Western provinces. The pattern of migration in China in many ways mirrors the geographic concentration of economic activity. The sending areas are concentrated in the Center and the West and the receiving areas along the coastal belt in the East. Most migration is inter-provincial and over long distances. However, like elsewhere in the world, distance increases the costs of migration, and reduces its magnitude. At the same time, the industrial clusters in the coastal belt are becoming increasingly skill and capital-intensive. Hence, developing the relatively more labor-intensive segments of industry in the central and western provinces can play a useful role in promoting further absorption of rural labor. This in turn will require several steps to improve the investment climate in these areas. · Improving labor market conditions and the provision of public services and social security for migrants in the cities. Despite its large scale, the current system of migration could with some justification be described as a `guest worker system'. A range of discriminatory practices in destination areas abridge migrants' entitlements as workers or citizens. There is an increasing need to eliminate these practices which are limiting both the scale of migration, especially family migration, and the potential welfare benefits from migration. In particular, this will require better enforcement of the provisions of labor laws for migrant workers, access to basic services (education, health and housing) to be determined on the basis of place of residence rather than the place of hukou, and the extension of the coverage of social protection programs to migrant workers. Insofar as this promotes family migration, this will also help address the problems of the left-behind children. Government initiatives in some of these areas are already underway as, for instance, laid out in the Opinions of the State Council on Issues Concerning Rural Migrant Workers (SC Document No. 5, March 2006), and the MOLSS Circular on the Implementation of SC Document No. 5 (April, 2006) as well as the action plans of the inter-ministerial Joint Conference on Rural Migrants Issues established in 2006. The long list of policy guidelines for migrant workers cover many areas including worker rights and protection, public services and living conditions, social security, employment services and skills development. However, not much systematic information is available on the actual implementation of these wide-ranging guidelines. It will be important to monitor the E x e c u t i v e s u m m a r y x v i progress on these new initiatives not only to assess if they are being implemented as intended, but also to refine the different initiatives based on the implementation experience of what seems to be working, where and in what ways. Promoting agricultural and rural development in poor areas Promoting agricultural However, migration will not be the answer for all the rural poor. There is clearly a role for and rural development promoting agricultural and rural development within a poverty alleviation strategy, as also has a role within a suggested by the analysis of the taxonomy of rural poor presented in this report. And this poverty reduction has also been a key element of the government's approach to poverty alleviation which strategy from the outset has been development-oriented and targeted to poor areas. China's flagship poverty alleviation program, the Integrated Village Development Program (IVDP), is an area-based program that currently targets about one fifth of all Chinese villages (that are designated as poor villages) using a participatory approach to village investments. Other important developmental poverty alleviation programs include subsidized credit and development of agribusiness, the food-for-work program and a labor transfer program which in fact seeks to promote migration. But area-based programs The review of these programs in this report highlights several areas for potential are unable to target the improvement. In relation to the integrated village investment program (IVDP), the review poor effectively and the highlights two key issues: (i) the relatively poor targeting of the program (a large fraction of impact on incomes of the the poor simply do not live in designated poor villages), and (ii) the very limited impact of poor who are targeted is small village investments on incomes of the poor within designated villages (reflecting the relatively small scale of investments, the inability of many of the poor to make complementary investments, and a number of factors linked to limited participation of villagers, especially the poor, in the planning and implementation of projects). Besides the IVDP, the subsidized credit program, although accounting for about half of all poverty alleviation funds, is not set up in a way to deliver significant benefits to the poor as its main focus has been to support large agribusiness enterprises. And participation of the poor in the training programs for labor transfer has also been limited by the relatively high costs of participation. Figure 0.11 : Fraction of the income poor living in designated poor villages (Fraction of the income poor living in designated poor villages, by region and overall) % of income poor who live in designated poor villages 2001 2004 80 68.5 66.8 69.9 69.0 60 54.7 53.4 51.9 47.4 40 33.1 35.4 27.7 28.3 20 16.4 12.1 16.1 16.4 0 tal st ralt ts s r s l as ea hwe estw ie or ie po Co rth Cen rth Poo No ountc Overal Sout No on-N ountc Source and notes: See main report. To make them more effective instruments for rural poverty alleviation, further reform of these programs will be needed. For the IVDP, the following areas merit consideration: · Develop consistent, data-based criteria for selecting poor villages and a system for updating poor village designations; privilege poorest villages for earlier and larger investments. · Increase the size of village investments. This may require higher funding for the budgetary grant program, but the effective size of village investments could also be increased through improved coordination of fund disbursement for village plans by designing clear rules to guide the aligning of Food-for-Work investments and E x e c u t i v e s u m m a r y x v i i subsidized credit with village plans. There is also scope for releasing more funds per village by targeting fewer villages. · Develop institutional mechanisms to improve coordination of village investments with other line agencies. Since the larger part of investments in rural areas comes not through the poverty alleviation funds but through the budgets of other line agencies, the needs for better coordination of village-level investments goes well beyond the poverty alleviation funds themselves. The current institutional mechanism for inter- agency coordination through the local Poverty Alleviation and Development offices works only imperfectly, and needs considerable strengthening to be effective. · Place new emphasis on household-level targeting, as even with better identification of poor villages, it will be important to reach the poor within these villages. This can be done in several ways: by changing the mix of projects to include more household- oriented interventions, for instance, labor training to promote off-farm employment; by waiving the requirement for complementary investments by poor households or providing subsidized credit for this purpose; by developing new programs such as self-targeting public works programs or conditional cash transfers which could offer assistance to poor households conditional on other objectives, for instance, ensuring that their children do not drop out of school, or they themselves participate in labor training. · Conduct additional study of the village planning process to identify ways to increase participation of the poor, and experiment with new methods of implementing village plans. For the other developmental programs, too, there are several areas for potential reform that may be considered: · The subsidized credit program needs to be integrated with the village plans. New institutional arrangements will be necessary for this to happen; simply exhorting different agencies to coordinate better will not help. It also needs to be recognized that village-level coordination of funds and program management is unlikely to succeed without coordination at the county and provincial levels. There should thus be a single agency that is adequately empowered to ensure this coordination at the provincial and county levels. · There is a need to redirect subsidized credit from large agribusiness enterprises (which ought to be supported through the regular commercial banking networks) to poor farmers, in part through provision of cheap loans for complementary investments by them to better utilize the public goods created by village investments. · In relation to the labor transfer programs, boarding and other expenses should be waived for poor farmers to ensure their enhanced participation. Further gains from better Two further policy considerations about developmental poverty alleviation programs geographical targeting should also be noted. are limited ... hence the case for household- · Even with perfect village-level targeting, many of the poor will be left out (due to the oriented approaches and increasingly dispersed nature of poverty) and many of the benefits of investments in new targeting targeted villages will flow to the non-poor in these villages. There are thus limits to mechanisms such as further gains from area-based targeting and it will be important to supplement area proxy means targeting targeting with household targeting. But new mechanisms will need to be developed to target poor households, and in this regard approaches such as proxy-means targeting deserve to be actively explored. More household-oriented · If, as suggested above, more household-oriented interventions are to be adopted as interventions will raise the stakes of inter-agency part of these developmental poverty alleviation programs, then the issue of coordination of coordination with social assistance and social insurance programs, which also developmental and social provide critical support at the household-level, will become increasingly important. protection programs For instance, there will be a need for a common approach to identifying poor households by both the developmental and social protection programs, which raises the stakes of inter-agency coordination not just to the Poverty Alleviation Offices, Departments of Finance and Development Reform Commission and the ABC (or Rural Credit Cooperatives), but also the Departments of Civil Affairs and Labor and Social Security. The development of a coordinating mechanism for this related set of E x e c u t i v e s u m m a r y x v i i i poverty reduction interventions remains a big institutional challenge. Enhancing security by expanding and improving the coverage of the social protection system in rural and urban areas The evidence presented earlier in this report indicates that rural and urban households in China remain vulnerable to poverty because they are exposed to a variety of risks arising from natural disasters, ill-health, fluctuations in agricultural yields and prices, and job loss. In rural areas, for instance, the size of population vulnerable to poverty may be twice as large as the number of poor in any given year. Social protection systems ­ through social assistance and social insurance programs ­ help individuals cope with such risk and vulnerability, and hence the development of such systems is an integral part of the overall poverty alleviation strategy. China has come a long China has come a long way in reestablishing its rural and urban social protection systems way in reestablishing its after the collapse of the traditional commune and enterprise-based systems during the rural and urban social process of economic reforms and restructuring of the 1980s and 1990s. Major revamping protection systems, but of the urban system started in the mid-1990s while the rebuilding of the rural system has many policy challenges remain been the focus of more recent policy attention especially over the last five years. While the substantial development of the new social protection system over a relatively short period of time is impressive, this has happened in the context of massive structural change in the rural and urban economic systems, and there remain many policy challenges. Drawing upon the analysis presented in this report, the key challenges and potential areas for reform for the rural and urban systems are noted below. Improving rural social protection Over the last 3-4 years in particular, there has been significant progress in the expansion of several elements of the rural social protection system, which is beginning to address its longstanding underdeveloped state. The recent initiatives are consistent with the recognition that access to land alone, despite the relatively egalitarian distribution of use rights, cannot be relied upon to provide an effective rural safety net. The main thrust of these initiatives has been on expanded coverage, and as a result, the coverage of key programs, in particular the rural Di Bao, rural heath insurance and the Medical Assistance scheme, has rapidly improved. At the same time, however, the coverage of other programs such as rural pensions and disaster relief remains limited, and the benefit levels of most rural programs are quite low. This is in contrast to the urban sector where following the major restructuring of the state-owned enterprises during the 1990s, a much greater level of government effort and resources have been devoted to the development of a relatively modern non-enterprise-based social welfare system. The recent progress is starting to bridge the gap between the rural and urban systems, but much still remains to be done. Overall, there remain four major challenges for the rural social protection system. · The coverage (in terms of the number of beneficiaries as different to the number of implementing counties) of several programs needs to be expanded to more fully cover the rural poor and vulnerable population. · The benefit levels of most programs, which are only a fraction of the benefits under urban programs, will also need to rise if they are to offer effective protection to those covered by the programs. · There is a need for coordination and harmonization of different programs within the rural social protection system as well as between the rural and urban systems. A harmonized national system is important not only for bridging the rural-urban gap in the social protection, but also for ensuring adequate protection to migrant workers who are currently only imperfectly covered by the rural and urban systems. · All this will require further infusion of funds from the central and provincial governments. Such commitment of resources from higher levels of government is also needed to break the dependence of social protection spending on local fiscal capacity, and hence address the wide disparities in both the coverage and benefit levels of programs within rural areas due to the large differences in the revenue base E x e c u t i v e s u m m a r y x i x of local governments (Figure 0.12). Figure 0.12 : Social assistance benefits depend on village revenues, 2004 80 60 Across all villages atpiacr Within counties petif 40 neeB Between counties 20 0 0 500 1000 1500 2000 2500 Revenue per capita Source and notes: See main report. There are specific challenges for individual programs. · Rural Di Bao: The rural Di Bao has clearly emerged as the main focal assistance program for rural areas. Its rapid expansion since 2005 has been impressive with gross coverage rate now equivalent to about two-thirds of the number of rural poor. However, as it expands further, the program will need to be streamlined with a more consistent approach to coverage and targeting. Here, many of the challenges will be of a similar nature to those already being encountered with the more established urban Di Bao program (see further discussion below), namely, realizing the scope for better targeting and hence improved coverage of the poor, rationalization of Di Bao lines and eligibility criteria across different local governments so that program coverage is better aligned to needs rather than local fiscal capacity, and development of a benefit design framework that limits adverse work incentives. · Rural health insurance: The recent rapid expansion of rural health insurance is indicative of its recognition as an important priority for the government. With an estimated participation rate of 86% at the end of 2007, the coverage issue has been largely addressed. The next challenge will be to ensure the provision of an adequate level of benefits that can secure a sizeable reduction in the burden of private health costs. The targeted increase in the budget through a doubling of contributions in 2008 will help provide more resources for raising benefit levels, but further increments of funding from central and provincial governments will be needed for net reimbursements to cover a larger fraction of the out-of-pocket health costs. Coverage of outpatient services also remains a further challenge. Another major challenge is the coverage of migrant workers who seem to be left out by both the urban and rural health insurance systems, and this raises the broader challenge of the coordination and harmonization of two systems, which is all the more pertinent now as the NCMS is beginning to operate at a national scale. · Medical Assistance: The Medical Assistance program is a useful supplement to the rural health insurance scheme in extending the latter's coverage to the rural poor. E x e c u t i v e s u m m a r y x x However, despite its recent scaling up to the national level, its coverage in terms of the number of beneficiaries is still relatively limited; the gross coverage rates are equivalent to about half the number of rural poor. Moreover, its relationship with the Di Bao program will also need to be clarified. As both programs target the poor, there is a need for a unified approach to targeting assistance. The government may also want to consider using the MA scheme to introduce a means-tested safety net provision to health insurance whereby after reaching a certain threshold of per capita healthcare expenses in any year, the insured are entitled to higher (even 100%) reimbursements. · Disaster relief: Additional government resources will be needed to improve the coverage and benefit levels of disaster relief which remain limited relative to the scale of the problem. However, most of the government response is ex post in nature and there is a conspicuous absence of disaster insurance. Recently, China Life (China's largest life insurer) launched the county's first accident insurance covering six types of natural disasters. However, the private uptake of such insurance may be limited, and there is a strong case for the government to consider public provision of disaster insurance with subsidized premia for the low-income groups. · Rural pension system: The reach of rural pension program remains limited and uneven. A significant expansion of the program along the lines being considered by the government will entail a high degree of "subsidization' with a much larger government contribution (into individual pensions). Such a highly subsidized program however makes it more like a transfer or assistance program for the rural elderly. There is thus a need for a careful evaluation of different options for providing support to the elderly while recognizing that social insurance programs need to be considered together with social assistance programs. Improving urban social protection While the urban social protection system is substantially more developed than the rural system, there remain many areas of concern in relation to the ability of the core programs to serve as an effective safety net for the urban population. The review presented in this report identifies several challenges for both social insurance and social assistance programs in urban areas. Key policy challenges and areas for reform for urban social insurance programs include the following. · Limited coverage, especially of the poor and lower income groups and those in informal employment. Despite the expansion over the last decade, the coverage of urban social insurance programs still remains limited. Currently the key programs cover between 40-50% of urban workers which indicates that there is a lot of ground to be covered still. The system is also heavily focused on the formal sector, and coverage rates for informal workers are much lower still, often as low as one-fourth of those in the formal sector. Similarly, the coverage rates for the poor and low-income workers are a fraction of those at the upper end of the distribution. Further expansion of the system needs to pay particular attention to better coverage of these groups (Figure 0.13). E x e c u t i v e s u m m a r y x x i Figure 0.13 : Lower coverage of social insurance for the poor in Chinese cities, 2005 (% of workers with various forms of social insurance coverage) 5 Large cities 5 Small cities 79 80 76 80 69 56 60 47 50 46 37 35 40 33 24 17 20 9 0 Poor Non- Poor Non- Poor Non- Poor Non- poor poor poor poor Pensions Health insurance Unemployment Work injury insurance insurance Source and notes: See main report. · Limited coverage of migrant workers. Large-scale rural to urban migration continues to pose challenges for the urban social protection system. The coverage rates for pension, health and unemployment insurance for migrants are estimated to be about one-seventh of those for local urban residents. Since 2006, there has been a significant momentum on the extension of work injury and health insurance which, according to administrative data, covered about a third and a quarter of migrant workers respectively in 2007. However, much still remains to be done in these two areas. The inclusion of migrants in pension and unemployment insurance is more limited and poses a bigger challenge as the design of these programs is not well- suited to serving a mobile labor force that changes employers and jobs more frequently. · Low level of pooling of social insurance funds and very limited portability of benefits. The contributions into most social insurance funds are typically pooled only at the local government (municipality or city) level. Low level pooling is however inefficient from a risk management perspective and limits the insurance benefits of the programs. It also has the effect of creating inequities in the system by tying local social insurance expenditures to local revenues. Low level of pooling also goes hand in hand with very limited portability of benefits. This is not only a significant barrier to improved coverage of the programs, but is also a major impediment to labor mobility and the development of a flexible labor market. Progress towards higher level pooling and greater portability has remained limited. Higher levels of pooling can be unattractive to richer local governments who may view it as an implicit tax. The reform of the system is thus likely to be difficult, but remains an important area for future policy development. · Need to balance provision of greater protection with maintaining incentives and competitiveness. The urban social insurance system is costly with its financing by employers and employees accounting for about 43% of the wage bill. The employers' contribution for the five key programs alone comes to about 32% of the payroll. Some of this may be passed on to the workers in lower wages, but private incentives for non-compliance remain high. Incentives for non-compliance also exist for the poorer workers. These factors are already contributing to the informaiization of the urban labor market. The existing system also has important implications for maintaining competitiveness of Chinese business and industry. Greater effort at enforcing compliance, though desirable, is unlikely to be enough to tackle the dual problems of informalization and ensuring competitiveness. A part of the solution will almost certainly involve greater public funding of the social insurance system. To protect the interests of the poorer workers, the government may also consider reducing or waiving the employee contribution and even reducing the employer contribution for workers below a certain income threshold. E x e c u t i v e s u m m a r y x x i i Urban Di Bao. The main social assistance program for the urban disadvantaged is the urban Di Bao program. By international standards, the program is exceedingly well- targeted with 64% of the program beneficiaries in the bottom decile of the urban population who account for about 70% of program benefits. While such performance is exemplary, the review of the program in this report suggests that it still faces a number of policy challenges, the most important of which include: (a) limited coverage of the urban poor (bottom 5-10% of the urban population) despite overall excellent targeting in terms of concentration of benefits amongst the bottom end of the distribution, (b) non-coverage of the migrant population, (c) eligibility thresholds (Di Bao lines) and benefit levels varying with local fiscal capacity rather than need, and (d) potential work disincentive effects of the program design. In view of these challenges, several areas of reform are suggested; many of these also have relevance for the rural Di Bao program. · One area of potential reform is in relation to the rationalization of Di Bao lines. There is a case for eliminating inequities in the system by establishing a set of eligibility lines that refer to comparable standards of living across locations. If it is not possible to develop city-specific lines with the currently available data, consistently-estimated provincial eligibility lines could be a useful starting point. This approach will also delink Di Bao lines from considerations of local fiscal capacity. However, such a reform can only be effective if it is supported by transfers from higher levels of government to poorer municipalities. Central funding to provincial and local governments will thus need to be based on the estimated number of beneficiaries using the consistent set of eligibility lines. · A related issue is the rationalization of Di Bao benefit levels which too vary significantly across cities and municipalities in a manner that is related not to differences in the depth of poverty but to what the local governments can afford or are willing to fund. Recent regulations (in February 2008) to raise the benefit levels, for instance, also permit better-off areas to implement higher increases. A more nationally consistent and equitable approach is clearly needed. · Improving the coverage of the poor will need a further expansion of the program in areas where coverage has been limited by relatively low thresholds. But it will also require improved methods of targeting such as proxy-means targeting. Household incomes are difficult to observe and measure accurately, and some form of implicit proxy-means targeting is already being practiced by the neighborhood committees as evident in the use of various non-income criteria. While the implicit proxy-means mechanism appears to have worked reasonably well thus far, many of the urban disadvantaged remain excluded. Hence, there is a case for developing a more explicit proxy-means targeting system based on relatively easily-observed household characteristics that are correlated with income. The list of household characteristics could include indicators related to household demographics, labor force participation, any work disability, educational attainment of adult members, ownership of specific assets, and characteristics of dwellings, amongst others. There is now a good deal of international experience with proxy-means targeting that could be drawn upon in developing a system tailored to China. · Coverage of the migrant population needs special attention. Migrants include both individual migrant workers and migrant families. As a first step, extending Di Bao coverage to migrant families who have been resident in an urban area for some minimum period of time and meet the program's means test ought to be considered. Until such time that the urban and rural Di Bao programs can be coordinated, individual migrant workers may have to be covered under rural Di Bao through their families in rural areas. · There is also a case for reconsidering the top-up design formula for the determination of Di Bao benefits. Such a design implies a 100% marginal tax rate on incomes of those eligible for the program. The reason why the program has managed to avoid serious adverse effects on work incentives thus far is that the top-up design appears not to have been followed in practice. Thus, the actual implementation practices are already pointing to the need for an alternative benefit structure. One alternative here is uniform per capita benefits for those eligible. While this is attractive for its simplicity and transparency, uniform benefits may not be considered sufficiently progressive. E x e c u t i v e s u m m a r y x x i i i Another alternative could be to base benefits on the proxy-means scores of households, with lower benefits for those with higher scores but calibrated so as to maintain relatively low marginal tax rates. Harmonizing the rural and urban systems As already mentioned, the overall social protection system is still more focused on the urban formal sector, despite the recent expansion of the rural system. At the same time, with the growth in the rural system, the issue of the integration of the rural and urban social protection systems has become increasingly important, not only to close the gap between the two but also to promote a more flexible labor market environment and to facilitate the coverage of migrant population. The harmonization of the two systems may need to proceed in a phased manner. By capping urban benefits at current levels and gradually raising rural benefit levels over time, conditions can be created for the eventual integration and harmonization of the two systems. Fostering equity and reducing poverty by ensuring secondary school education and basic healthcare for all A large part of the current The analysis presented in this report has repeatedly found education to be of critical poverty and inequality in importance in determining better welfare outcomes for China's population. China can be traced to the underlying disparities · The educational attainment of a household's adult workers is the clearest observable in educational attainment predictor of poverty in China, in both rural and urban areas. While the poor differ from each other in other ways, low levels of educational attainment seem to be common to virtually all of them. A high school education for the adult working members virtually guarantees a household's exit out of poverty. Even with middle school education of adult workers, consistent with the goal of ensuring universal 9-year education, the probability of being poor declines to negligible levels. At the same time, over 90% of the poor in rural areas and nearly 60% of the disadvantaged in urban areas live in households where this target has not yet been attained. Investing in middle and high school education for all is thus the key to making a substantial dent in remaining poverty. · In terms of explaining income inequality too, differences in educational attainment, and the returns to education, turn out to be critical. The components of inequality that contributed most directly to the rise in inequality--the rural-urban gap and increasing inequality within rural and urban areas--can all be traced to inequalities in human capital (Figure 0.14). Figure 0.14 : Differences in human capital endowments and returns underlie much of income inequality between and within rural and urban areas, 2003 60% Rural 51.6% Urban 45.6% iont 40% 30.5% populafo 20.6% 23.6% 20% 16.1% 8.7% % 3.3% 0% Less than 6 Between 6 and 9 Between 9 and 12 More than 12 Average years of education of adult household members E x e c u t i v e s u m m a r y x x i v larur 9,000 8,290 8,457 Rural Urban in 6,007 6,000 4,963 4,850 meo se 4,560 3,040 inc a pric3,000 2,330 pitac-r 0 Pe Less than 6 Between 6 and 9 Between 9 and More than 12 12 Average years of education of adult household members Source and notes: See main report. · As already noted, at least a junior high school level of education is an important determinant of migration and off-farm employment for rural laborers, and higher level of education will be important for migrant workers to tap into higher-skill and better- paid jobs in the cities. · In urban areas, the rising inequality in labor incomes can in turn be traced to, on the one hand, the increased incidence of unemployment and declining labor force participation associated with the massive restructuring of state-owned enterprises, and on the other, to increasing dispersion in wages among those who are employed. Educational attainment has played a central role in both processes by determining who were most likely to retain their jobs in the face of widespread layoffs, who were most likely to be re-employed after an unemployment spell, and who were likely to be paid more when employed. · A large part of the gender income gap can be traced to differences in educational attainment of men and women. While private returns to education have increased with economic reforms and rapid growth of the Chinese economy, the persistence of underlying disparities in educational attainment suggests that the improved private incentives for greater investment in human capital are not enough especially for the poor. Since social returns to education are even higher than private returns, there is a strong case for greater subsidization of education. Unequal access to basic Another major area of policy challenge that has been opened up by China's wide-ranging healthcare also remains economic reforms during the 1980s and 1990s is in relation to provision of basic another major policy healthcare. challenge · Despite significant income poverty reduction over the last 25 years, China's progress in improving health outcomes (for instance, child mortality and communicable disease mortality) has slowed down. · Private costs of healthcare have risen dramatically (even for those who are insured) and are impinging on the utilization of healthcare services. · Health shocks and the burden of medical expenses and the loss of income they entail, are one of the most common reasons why households fall into poverty. · With only the well-off being able to afford reasonable quality healthcare, inequalities in income are transmitted to those in health outcomes. There are thus large disparities in health outcomes between urban and rural areas and between the rich and the poor within both urban and rural areas. Ensuring that every household has access to education and healthcare thus remains an important part of the current poverty reduction agenda and is also critical fostering greater equity. The government is already responding to this challenge in a number of ways, but a lot more remains to be accomplished. E x e c u t i v e s u m m a r y x x v Improving access to affordable education · Recently, good progress is being made in universalizing 9-year compulsory education. The waiving of school fees and the provision of boarding subsidies under the compulsory education finance reform, introduced in rural areas in 2006 and gradually extended to urban areas in 2007, is an important step in the right direction. The momentum on this initiative should be maintained so as to achieve the goal for 2008 of achieving free compulsory education nationwide. It will also be necessary to ensure that this policy can be implemented through China's decentralized fiscal system on a sustained basis. Poorer local governments (at the provincial and sub- provincial levels) will thus need to be adequately resourced, which in turn will be important for ensuring that free compulsory education does not come at the expense of a compromised quality of education. · There is also a need to review the formula for compulsory education transfers, which should be allocated more equitably based on needs rather than the pre-existing level of fees. · The impact of the policy of school consolidation and the closure of teaching points needs to reviewed. While the government has indicated the closures will be subject to prior community consultation, the implementation of this provision needs to be monitored and if the closures are likely to compromise school access in particular areas, the policy should be suitably amended. · A solution needs to be found for the problem of school debts perhaps by allowing a fraction of the transfer to go towards debt repayment. · China has achieved close to universal primary education. But significant inequalities in educational attainment beyond primary school ­ middle and high school, and eventually tertiary education ­ remain, and these are an increasingly important source of disadvantage for those who are not well-off. While middle school education is being covered by the compulsory education initiative, there is a need to turn attention to reducing the private costs of high school education especially for the poor. Means- tested subsidies for high school education could be an important future policy initiative, and these could be offered to poor households conditional on school attendance by their children, especially girls. · There is also a case for developing a credit market for education where students (or their parents) can borrow for investments in education against future earnings. And the subsidies could of course be combined with loans for education. There are several examples of the application of this approach internationally and drawing upon that experience China could develop a model suited to its own conditions. · The education of children from migrant families in urban areas remains an area of particular concern. While steps have been taken over the last few years (especially since the State Council Document No. 5 of March 2006 and the inter-ministerial Joint Conference on Rural Migrants Issues) to reduce effective educational discrimination against migrants, progress remains uneven across urban areas and the implementation of non-discriminatory practices needs to be further monitored and consolidated. · There also remain large differences in the quality of educational inputs across regions, especially across rural and urban areas. Further investments into Improving the quality of educational inputs in rural areas will help strengthen the demand and private incentives for education which in turn will be important for addressing the rural-urban education gap. Improving access to affordable healthcare Significant progress has been made in recent years in extending the coverage of basic healthcare to the Chinese population. Important initiatives in (i) rapid expansion of the coverage of rural health insurance to nearly 86% of counties in the country, (ii) expansion of the Medical Assistance scheme in rural and urban areas, (iii) the introduction of the Urban Resident Basic Medical Insurance in 2007 have all helped the country move forward towards the eventual goal of universal healthcare provision. Many of the E x e c u t i v e s u m m a r y x x v i remaining challenges have already been referred to above, and most importantly, include the following: · While the coverage of rural health insurance has now expanded to include the bulk of the rural population, the level of benefits (net reimbursement) needs to be increased to achieve a sizeable reduction in out of pocket expenses. The current government contribution of 40 yuan per person is clearly quite limited. In 2008, the government contribution was raised to 100 yuan per capita. However, this is probably still not enough. For instance, in 2006, rural households on average spent 192 yuan per person on healthcare and medical expenses, and even the bottom quintile spent 118 yuan. Thus, the standard should be gradually raised further. At the same time, the coverage of rural health insurance needs to be broadened to include outpatient services. · The recent rapid expansion of the Medical Assistance program is also laudable. There is however scope for further expansion of coverage beyond the 29 million rural and 4 million urban participants in 2007. As noted above, the relationship of MA to the Di Bao program also needs to be clarified with a view to establishing a unified approach to targeting assistance to the poor and the disadvantaged. The MA program could also be used for a means-tested safety net provision under health insurance for higher or even full reimbursement of healthcare expenses above a threshold. · In urban areas, the introduction of the new Urban Resident Basic Medical Insurance should help improve the inclusion of those in the growing informal sector whose participation in the largely formal-sector focused health insurance has been quite limited. The 41 million enrollees under this scheme by the end of 2007 is encouraging but still limited relative to the size of the informal sector. · The effective inclusion of migrant workers in health insurance remains an area of concern. The directions from MOLSS in May 2006 to set up a separate medical insurance pooling fund to mainly cover inpatient expenses for migrants, and to improve the settlement method for those in the urban system who return to their hometown for treatment or others in the rural system who seek treatment in urban areas, are indicative of the government's intent to address this issue. But the case of the migrant workers highlights the larger policy issue of the integration of rural and urban health insurance. This issue is all the more pressing with the recent expansion of both the rural and urban systems. There is also a risk of excessive fragmentation of heath insurance with different schemes with varying provisions covering different segments of the population and the lack of an effective mechanism for portability across schemes. Though a difficult issue, complicated further by disparities in fiscal capacity under the decentralized fiscal system, it is one that will need to be confronted in future policy development for universal healthcare. Supplementing area-based poverty reduction efforts with a household-oriented approach Supplementing area- Implementation of the broader poverty reduction agenda outlined above will require based poverty reduction supplementing the traditional area-based poverty reduction efforts with household- efforts with a household- oriented approaches and interventions. The case for a household-oriented approach has oriented approach makes already been mentioned in the context of the government's poor village-based IVDP sense... program. But a focus on household-oriented approaches has a broader cross-cutting relevance for the entire poverty reduction agenda. ...for several reasons There are several related reasons why such a focus makes sense. · The large and growing scale of labor mobility has diminished the promise of exclusively area-based approaches for poverty reduction. With greater prospects of migration, it has become relatively more important to invest in people (who are increasingly mobile) rather than the areas they come from. · The remaining poor are more dispersed throughout the villages of rural China, and as already noted this is one of the reasons why an area-based approach such as the one used in the IVDP program for designated poor villages is likely, even with a better identification of poor villages, to miss reaching many of the poor. E x e c u t i v e s u m m a r y x x v i i · The poor are also a heterogeneous group. They are poor for multiple and varying reasons related to their particular household endowments and circumstances, not all of which are reducible to where they live. A taxonomy of poverty suggests that several different clusters of poor can be identified depending upon different mix of factors underlying their poverty. Some of these factors are common to most, if not all, of the poor. Almost all the poor live in households where the adult workers are poorly educated. Few of the poor have access to employment off the farm. But the other reasons for poverty are more varied, and are relevant only for particular sections of the poor. In the case of some, it is that they have limited land to work with; others may be poor because of the burden of family members who are incapable of working; for still others their minority status may be a source of additional disadvantage. A household-oriented approach is needed to address this hetereogeneity. · Virtually the entire social protection agenda consists of interventions that are intrinsically household-oriented. Whether it is social assistance programs such as the rural and urban Di Bao, or social insurance programs such health or unemployment insurance, they are all directly targeted to households. · Finally, even within the context of area-based development-oriented programs, there is a case for supplementary targeted household-level assistance to enhance the poverty impact of these programs. The evidence indicates that without such direct assistance, the poorest households do not benefit. It is thus recommended that China's area-based poverty programs place new emphasis on household-level targeting, and incorporate a broader range of instruments for providing direct development-oriented assistance to poor households. These might include subsidies for complementary investments to make public goods more accessible to the poor, as well as development of new programs such as conditional cash transfers to increase rural school attendance, or self-targeted public works programs. Investing in the Greater focus on household-oriented approaches to poverty reduction also raises the development of a unified importance of developing suitable household targeting mechanisms to reach the poor. targeting system such as The common challenge faced by current programs, whether it is area-based interventions, proxy-means testing labor training for migrants, subsidized credit, or various social assistance or insurance could have large payoffs programs, is their ability to reach the poor and the disadvantaged. As discussed at several points in the report, there is significant scope for improvement of current targeting mechanisms, and there is a need for a unified approach to targeting the wide range of ongoing programs, many of which have expanded substantially in recent years. In particular, there is a case for considering mechanisms such as a proxy means targeting system at least on a pilot basis. Investing in the development of such a system could have large payoffs as the same underlying system could be employed to target multiple programs, perhaps with the use of different thresholds and ancillary criteria for individual programs. Even if a proxy means system is not directly used to determine program eligibilities, it could serve an important role in the verification of currently used forms of screening and means testing, and thus help improve the overall targeting effectiveness of different programs. Providing an adequate and equitable allocation of resources for local governments A more adequate and In China, as in most of the world, local governments are at the frontlines of service- equitable allocation of delivery, social protection, and poverty reduction. However, unlike in much of the world, resources for local within China's highly decentralized fiscal system, local governments at the county level governments is needed and below, as well as village collectives, also bear the primary responsibility for financing these efforts. But local governments in China vary widely in terms of their revenue-raising capacity, and the variation is only partially mitigated by transfers from higher levels of government. The resulting large disparities in fiscal resources across local jurisdictions translate into large disparities in levels of public spending. And that in turn leads to large differences in the quality and level of services provided, in the extent of social protection and unevenness in the pace of poverty reduction, thereby contributing to the persistence, and even magnification, of inequalities. Implementing the broader poverty reduction agenda and ensuring that local governments all across China have the resources to provide a basic nationally determined level and quality of services and social protection to those living within their jurisdictions will require a more equitable and adequate allocation of resources for local governments. E x e c u t i v e s u m m a r y x x v i i i This is a two-fold task. There is certainly a need to move towards a more equitable allocation of resources through further equalizing transfers. The evidence indicates that the pattern of public expenditures favors richer provinces, and within provinces, richer local governments. For instance, at the provincial level while transfers from the center have been equalizing, but because the disparity across provinces in revenue-raising capacity has grown noticeably during 1995-2004, provincial expenditures in 2004 still favored richer provinces as much as they did in 1995 (Figure 0.15). Figure 0.15 : The provincial distribution of expenditures in 2004 favored the richer provinces (Concentration curves of cumulative % of expenditures/revenues) 100 2004 expenditures 2004 revenues 1995 expenditures 1995 revenues 80 cial nues reve/ provin 60 of resu % 40 lativeu expendit Cum 20 public 0 0 20 40 60 80 100 Cumulative % of population ordered in terms of 2004 provincial per-capita income from poorest to richest Sources and notes: See main report. Disparities below the Spending disparities below the provincial level tend to be even more pronounced. Data provincial level tend to be from the Special Purpose Survey of 3036 villages, undertaken jointly by the NBS and the more pronounced, even World Bank for the purposes of this poverty assessment, indicate that the village-level for poverty-related distribution of social assistance and infrastructure investment expenditures is much less expenditures pro-poor than the provincial distribution would suggest. Much, therefore, still remains to be done in terms of making poverty-related public expenditures more equitable and pro-poor. Equalizing transfers from the center need to continue and be scaled up, but attention has to be paid as well to how equitable the allocation of resources is within provinces, and going one step further down, within counties. There is a need for But the challenge of financing the broader poverty reduction agenda is not simply one of additional resources, not redirecting resources from richer to poorer areas. A simple budget-neutral spatial re- just a more equitable allocation of existing resources, even if it were feasible or advisable, would not be allocation of resources adequate to improve the coverage and benefit levels of several programs. Therefore, there is a need as well to devote substantial additional resources to development-oriented poverty reduction, service delivery and social protection. And in doing so, because of existing disparities it will be critical to target all incremental resources to the localities where funding is currently inadequate. The additional resources What would these additional resources finance? The key areas include: could help finance critical expansion and · further expansion of the rural Di Bao, improvements in many · expansion of health insurance to achieve universal coverage of rural and urban key areas of the broader populations, poverty alleviation E x e c u t i v e s u m m a r y x x i x agenda · expanded coverage of rural and urban Medical Assistance schemes, · expansion of disaster relief and provision of disaster insurance, · targeted subsidies for high-school education, · higher benefit levels of rural social assistance and social insurance programs, · greater subsidization of urban social insurance programs to ensure better coverage of the informal sector, · scaling up of basic training and job placement services for migrant workers, and improved provision of education and housing services for their families, and · the development of a unified targeting system. The scaling up of the While it is difficult to estimate precisely the cost of the expanded initiatives mentioned poverty reduction above, with the total government revenue in 2007 of nearly 5 trillion RMB, the broader initiatives is affordable poverty reduction agenda is in principle affordable. This is especially so in light of the rapid growth in the combined central and local government revenues at about 20% per year in nominal terms since 2000, significantly faster than the growth in GDP. The broader poverty Very rough estimates of the incremental cost of the key additional initiatives mentioned reduction agenda is above suggest an added cost of around 155 billion RMB, which equivalent to 3.1% of the affordable combined tax revenues of central and local governments and 0.6% of the GDP in 2007 (Table 0.4). The scaling up of the poverty reduction efforts along these lines thus certainly seems affordable. Table 0.4 : A rough estimate of the additional cost of scaling up poverty reduction efforts Program Description of additional initiatives Cost (billion Yuan) Rural Di Bao Coverage of an additional 35 million beneficiaries and an average benefit 17.5 level of 400 yuan per capita for all beneficiaries Senior secondary education Targeted subsidy for 25 million students (equal to the total enrolment in 25.0 2006) to cover tuition and miscellaneous fees of 1000 yuan per student Rural health insurance 100% coverage of the population with rural hukou and per capita 71.2 government contribution raised to 100 yuan per capita Urban health insurance 100 yuan per capita subsidy to an additional 50 million beneficiaries 5.0 under the Urban Resient Basic Medical Insurance Medical assistance 200 yuan per capita subsidy to an additional 40 million rural and 18 11.6 million urban beneficiaries (to bring the total number of beneficiaries in line with expanded coverage of rural Di Bao and current coverage of urban Di Bao) Disaster insurance Per capita subsidy of 33 Yuan (about half the cost of private catastrophic 3.3 insurance) for 100 million insured Support for labor transfer 10 million rural workers at an average cost of 2000 yuan per worker 20.0 (including 1000 yuan for boarding costs) Unified targeting system Development of a unified targeting system with a registry of 100 million 1.0 households at the aveage cost of 10 yuan per household Total additional spending for poverty reduction 154.6 As % of total tax revenue (central and local governments in 2007) 3.1 As % of GDP in 2007 0.6 Sources and notes: See main report. Strengthening institutional arrangements to promote participation, enhance accountability and improve coordination But an even harder challenge remains, namely to improve governance--capacity, accountability and responsiveness--notably (but not only) at the local level. A concern about channeling larger volumes of central funds to local governments is the lack of adequate financial management systems and monitoring and oversight capacity to ensure that the funds are well-spent. E x e c u t i v e s u m m a r y x x x The government has Experience in other countries suggests that local communities can play an important role recognized that local in ensuring the effective implementation and targeting of poverty alleviation efforts, and to communities can help in making programs responsive to the needs of the poor. The Chinese government effective implementation recognized the importance of this principle when it mandated that the design of poor and targeting of poverty reduction efforts village investment plans be based on a participatory process. According to official training materials, projects are supposed to be selected by a plenary session of the entire village, with the views of poor households given added weight. But genuine local However, as discussed in this report, the actual practice rarely adheres to all of the participation remains guidelines, and all too often, there was little genuine participation by villagers. The limited, and more needs evaluation of the IVDP in this report also found that the better quality of village governance to be learnt about ways of (e.g., village committees with more members, more frequent meetings) enhanced program increasing participation by the poor benefits for both rich and poor in program villages. However, the poorest households did even worse relative to less poor households in villages with highly educated village leaders, a finding consistent with elite capture of program benefits. In light of these findings, high priority should be given to additional study of the village planning process to identify ways to increase participation of the poor, and to experiment with new methods of implementing village plans. There is now considerable international experience with community-based development models, which can provide some lessons for improving village planning and local accountability in China, but an appropriate model for China will also need to consider local institutional and power arrangements in Chinese villages. The need for inter-agency Another huge challenge is one of inter-agency coordination. A central theme of this report coordination for the has been that the China's current poverty reduction agenda is not limited to just the implementation of the developmental poverty alleviation programs, but also includes pro-poor interventions broader poverty across a wide range of areas covering education, health, agriculture, migration policies, reduction agenda cannot be overemphasized social assistance and social insurance. This broad spectrum of poverty reduction efforts raises the enormous challenge of institutional coordination across different agencies that are responsible for their implementation. However, as in the past, coordination among government agencies has proven difficult because each agency has its own priorities and is reluctant to relinquish control over resources. The lack of effective coordination limits the overall effectiveness of the diverse set of poverty reduction initiatives. The symptomatic effects of this are already visible in many forms. For instance: · The lack of coordination amongst different agencies responsible for developmental poverty alleviation programs has limited the size of village investments, and the diversion of funds to uses that are neither consistent with village plans nor are well- directed to supporting poor households. · There is a conspicuous absence of a unified targeting mechanism to reach the poor. The area-based programs follow one approach, the social protection programs follow another, not to mention the differences across individual social assistance and social insurance programs themselves. While the 5-year planning process does provide an overall coordinating framework, there is no effective institutional mechanism to ensure coordination at the implementation level. This gap was highlighted in the case of migration policies, for instance, where in recognition of the multi-sectoral nature of policies related to migrants, in 2006 the government established a Joint Conference on Rural Migrants with representation from Ministries of Labor and Social Security, Civil Affairs, Agriculture, Construction, Education, Health, Justice, Propaganda, Public Security, the National Development Reform Commission, the National Bureau of Statistics, All China Federation of Trade Unions, All China Women's and Youth Federations, the General Bureau of Safety Inspections, and the Population and Family Planning Commission. If this was deemed necessary for migration policies alone, the coordination requirements for the wide range of poverty reduction programs and policies (of which those related to migrants are but one part) are much greater still. And these requirements have become even larger with the 11th Five-Year Plan's renewed emphasis and intensification of a broad range of initiatives on social development and the reduction of disparities. Yet, E x e c u t i v e s u m m a r y x x x i there is absence of an effective body with a large enough mandate and practical authority over other government units to play such a coordinating role. While it is not clear whether any of the existing agencies can be groomed to play this role, establishing an institutional mechanism for inter-agency coordination is critically important to realize the full benefits of the ongoing and future poverty reduction efforts. Enhancing statistical monitoring and evaluation capacity China has a well- China has a well-developed statistical system that in many ways exceeds the capabilities developed statistical of other statistical systems around the world. One example of this capacity is its long- system with substantial standing tradition of conducting a wide array of annual, large-scale and nationally- capabilities representative household and enterprise surveys in the rural and urban areas of the country. It also produces a very large body of statistical information based on non-survey based reporting systems. Most of the analysis presented in this report would have been impossible in the absence of this rich data base. But there remain some However, the analysis undertaken for this report also highlighted some critical gaps in the critical gaps in the information base for the monitoring and policy analysis of poverty and poverty reduction information base for policies and programs. The implementation of the poverty reduction agenda as outlined monitoring and poverty above will, among other things, also require efforts at enhancing statistical capacity to fill analysis of poverty, including ... these gaps. The key gaps are in the following areas. · One important gap relates to information on migrant population in the national ... information on the living and working household surveys. China's statistical system has two concurrent annual household conditions of migrant surveys conducted independently in rural and urban areas of the country. Starting population 2001, NBS' Rural Household Surveys (RHS) began collecting information on individual migrant workers from rural areas who were considered part of their households in the source areas. However, for obvious reasons the RHS is not equipped to gather information on migrants who moved out of the villages with their families. But few of them are picked up by the Urban Household Surveys either whose sample frame is primarily confined to urban residents. Thus, very little information is available on full-family migrants in urban areas, and this alone could potentially lead to some underestimation of urban poverty. More generally, there is a lack of systematic and nationally-representative information on the working and living conditions of migrant population in destination areas that is important for developing future policies in relation to labor mobility. In view of both the large (and growing) scale of migration and its key role in achieving poverty reduction, it will be important to fill this gap. · Another gap relates to information on spatial cost of living differences. In order to ... information on spatial cost of living differentials have comparable measures of income and consumption across rural and urban areas and across regions and provinces, it is necessary to adjust nominal magnitudes by spatial cost of living indices. While different researchers have attempted to estimate these indices using available price information, and one of the most comprehensive such attempts (Brandt and Holz) has been used for the analysis in this report, the magnitudes of spatial cost of living differentials remains a subject of some debate. As this is critical for a reliable assessment of poverty, greater effort devoted to a regular production of such spatial price differences will be important for both policy analysis of poverty and the monitoring of future progress in poverty reduction. · A third important gap relates to information on program participation. As mentioned ... information on program participation at a number of points in the report, monitoring and evaluation are key to further and utilization of public reform and refinement of a wide range of initiatives, whether in education or health, services development poverty alleviation or social protection programs. While rural and urban household surveys conducted by the NBS have many strengths (e.g. large sample sizes, detailed information on household income and consumption), they have very little information on households' participation in various government programs and their utilization of various public services. This seriously limits the potential of these national surveys for program and policy evaluation. To enhance this potential, the inclusion of one or more additional modules on program participation and service E x e c u t i v e s u m m a r y x x x i i utilization ought to be given a priority in the future development of the main household surveys. · Finally, the idea of a unified targeting system and the potential use of proxy-means ... more detailed information on household targeting has been suggested at a number of places in the foregoing discussion of the characteristics that could current poverty reduction agenda. An implementation of the idea will require data on be used for proxy-means a range of household characteristics that can serve as proxies for income or targeting consumption as a measure of welfare. Some of these data will be available from the existing household surveys, but more information may need to be collected by augmenting data collection efforts in this regard. Once the means-proxies are identified, it will also be necessary to develop a unified household information registry to implement such a targeting system. E x e c u t i v e s u m m a r y x x x i i i PART I WHY REVIEW CHINA'S POVERTY REDUCTION AGENDA? 1 1.China's poverty reduction record over the last quarter century has been remarkable China has experienced unprecedented rates of economic growth over the last quarter century. The rapid growth has resulted from the series of market-oriented reforms that China adopted as it transitioned from a planned economy to a market-oriented economy. Key reforms have included the introduction of the Household Responsibility System in agriculture in the early 1980s, creation of a conducive environment for the rise of Township and Village Enterprises in rural areas, which up to the mid-1990s grew rapidly to absorb a large share of the rural labor force, restructuring of the state industrial sector in the late 1990s, and the opening up of the economy to global trade and investment. And this growth has been accompanied by, and arguably central to, a dramatic reduction in poverty over this period (Figure 1.16). Figure 1.16 : China's record of poverty reduction and growth over the last quarter century has been remarkable 70 3200 6th Plan 7th Plan 8th Plan 9th Plan 10th Plan 60 2800 )%( World Bank consumption HCR Per capita GDP 2400 50 ces)ipr HCR)(yt 2000 1978 40 at pover ofeatr 1600 uanY( 30 Income HCR 1200 GDPat 20 800 capi Headcount Per 10 Official rural HCR 400 0 0 1981 1982 1983 1984 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 Sources and notes: Estimates for consumption poverty are in terms of the World Bank poverty line (also equivalent to $1.08 per person per day in 1993 PPP dollars) using per capita consumption for all years (see Box 1.1 for further details) . Consumption poverty estimates prior to 2001 are based on World Bank calculations using grouped consumption distribution data for rural and urban China for select survey years; for 2001, 2003 and 2004, these are based on World Bank calculations from the NBS Rural and Urban Household Surveys. Estimates for income poverty up to 2001 are based on rural and urban poverty estimates from Ravallion and Chen (2007) and are in terms of the basic-needs line estimated there; for 2003 and 2004, income poverty estimates are based on World Bank calculations from the NBS Rural and Urban Household Surveys. The national income and consumption headcount rates are derived as population-weighted averages of rural and urban headcount rates, using rural and urban population shares consistent with the sampling frame for the NBS Household Surveys. For further details on the population numbers, also see Box 1.4. The official headcount rate is from NBS (2005a) and gives the proportion of rural population below the official poverty line. a. The incidence, depth and severity of poverty have declined dramatically both according to official poverty estimates and by international standards The decline in poverty in China over the last quarter century is evident across a number of alternative approaches to measuring poverty--whether in terms of the official poverty or by international standards, in terms of income or consumption, and whether in terms of the absolute number of poor or the incidence, depth and severity of poverty. (The choice of welfare measures and poverty lines is discussed in Box 1.1). C h i n a ' s p o v e r t y r e d u c t i o n r e c o r d 2 Between 1981 and 2004, the fraction of the population consuming less than a dollar-a- day fell from 65% to 10% China's official poverty line--the per-capita standard by which the poor in rural areas are counted, policies are targeted, and funding allocated--was 637 Yuan per year in 2003 or 71 cents a day in (1993) international prices. By this standard, which is meant to capture the minimum income required to meet basic subsistence needs such as food and clothing, and is applied using a mixed-income and consumption cutoff, the poverty rate in rural China fell from 18.5% in 1981 to 2.8% in 2004 (Figure 1.16). Measured in terms of the World Bank's poverty line of 888 Yuan per person per year in 2003 rural prices (also equivalent to the international poverty standard of a "dollar-a-day" per-capita consumption at comparable 1993 international prices; see Box 1.1 for details), China's poverty reduction performance has been even more striking. By this standard, China's poverty rate in 1981 was 65%, comparable to that of India or much of sub- Saharan Africa at the time. By 1990 the poverty rate had already been cut nearly in half to 33%. This rate was cut in half again in the 1990s, falling to 15% of the population in 2001, and by 2004 the incidence of consumption poverty had declined further to 10% (Figure 1.16). Box 1.1 : Welfare measures and poverty lines used in this report Two central issues in poverty measurement relate to (i) the choice of a measure of welfare, and (ii) the choice of a poverty line. Both issues have been a subject of some debate in academic and policy discussions of poverty in China. Measuring individual welfare The most commonly-used approach in China has been to utilize per capita income or per capita consumption or a combination of both. 1 Generally, income can be considered a measure of welfare opportunity and consumption a measure of welfare achievement. 2 There are strong conceptual and philosophical justifications for both. A key economic issue here is whether the poor save and what may be the motives for their savings. Consumption-based measures accord zero weight to household savings, while income-based measures treat them on a par with consumption. In this poverty assessment, both income and consumption-based poverty measures will often be reported. Which poverty lines? Even in terms of income or consumption, there is no single definition of poverty in China. Which poverty lines are used is not just a question of analysis but a matter of policy choice. A range of poverty lines have been used, notable amongst which are the following: · Official Poverty Line ­ that has traditionally been used by the government to monitor rural poverty in China. For 2003, the value of the official poverty is 637 yuan per person per year. · Low Income Line ­ introduced by the government in 2000 to monitor the situation of households above the official poverty line, but nevertheless below a low income threshold. For 2003, the value of the low income line is 882 yuan per person per year. · Ravallion-Chen Basic Needs Line ­ developed by Ravallion and Chen anchored to 2100 calories per person per day (with 75% of the calories from foodgrains) and an allowance for basic nonfood consumption. They estimate a rural and an urban line at 850 and 1200 yuan at 2002 rural and urban prices respectively. World Bank poverty line This poverty assessment uses a poverty line of 888 yuan per person per year in rural areas in 2003 prices. There are two independent justifications for this poverty line. _________________________________________________________ 1The available data do not allow an implementation of multi-dimensional poverty measures. However, the non-income/consumption dimensions of poverty are discussed below. 2Income itself is only an imperfect measure of opportunities; discussions of equality of opportunity have been at pains to point out that unequal opportunities can not often be reduced to ­ or measured as ­ unequal incomes (World Bank, 2005a). C h i n a ' s p o v e r t y r e d u c t i o n r e c o r d 3 First, this poverty line can be derived from the basic needs lines of Ravallion and Chen (heareafter, R-C) and the cost of living differentials estimated by Brandt and Holz (2006). Note that the urban and rural lines estimated by R-C imply a 41.2% urban-rural cost of living differential for 2002. This poverty assessment instead uses a 27.2% urban-to-rural cost of living differential for 2002 based on estimates by Brandt and Holz (2006). Thus, based on R-C's work, we have two alternative estimates of the rural line for 2002: 850 yuan per person per year which is a direct estimate of the rural line, and 943 yuan per person per year which is the rural equivalent of their urban line (i.e. 1200/1.272). The World Bank poverty line for rural areas is then taken to be a weighted average of the direct and indirect estimates of the rural line, where the former (which is based on NBS' Rural Household Survey) is weighted by the rural population and the latter (which is based on the Urban Household Survey) is weighted by the urban population in 2002. This comes to 874 yuan person per year for 2002. Adjusting for the 1.6% rural inflation between 2002 and 2003 (based on the official rural Consumer Price Index), the World Bank poverty line comes to 888 yuan per person per year for 2003 for rural areas. Second, the poverty line of 888 yuan per person per year can also be derived as yuan equivalent of the international "dollar-a-day" line using 1993 Purchasing Power Parities (PPP).3 Strictly, the dollar-a-day line corresponds to $1.08 per person per day or $32.74 per month at 1993 PPP US dollars. Using the 1993 PPP for China of 1.419 yuan per US dollar, this comes to 557 yuan (=32.74x12x1.419) per person per year at 1993 prices. Adjusting for the 59.3% rural inflation over the decade 1993-03 (based on the official rural Consumer Price Index), this also comes to 888 yuan per person per year. Thus, either approach yields a poverty line of 888 yuan per person per year for rural China for 2003. The corresponding urban poverty line for 2003 is then derived at 1124 yuan per person per year when the urban-to-rural price differential of 26.5% for 2003 (based on Brandt and Holz, 2006) is incorporated. Using the rural and urban Consumer Price Indices, the values of the rural and urban World Bank poverty lines for 2007 are 1018 and 1251 yuan per person per year respectively. The monetary value of these lines at 2003 prices is shown in Table 1.5. It is notable that the low- income line is very close to (99.3% of) the World Bank poverty line. Table 1.5 : Poverty lines for China Value in 2003 (yuan per- capita per year) POVERTY LINES Calculated by: Rural line Urban line Official poverty line National Bureau of 637 Statistics Low-income line National Bureau of 882 Statistics Ravallion-Chen basic needs line Ravallion and Chen (2007) 864 1211 World Bank poverty line World Bank 888 1124 Dollar-a-day line ($1.08/person World Bank 888 1124 /day at 1993 PPP dollars) Sources and notes: National Bureau of Statistics (2005), Ravallion and Chen (2007). Value of poverty lines is in all-China rural and urban prices for 2003. Each of these lines can be applied to either per capita income or consumption of households, which yields several possible combinations. However, the results presented in this assessment mainly focus on the 888 yuan line with per capita consumption or income as the measure of welfare. Occasionally, the report will also make a reference to (a) estimates using a poverty threshold that is twice this poverty line, and (b) estimates using the low-income line as well as the official poverty estimates which use a combination of per capita income and consumption and the official poverty line. 3Purchasing Power Parities (PPP) were developed as part of the International Comparisons Project (ICP) to facilitate comparisons of real income and consumption across countries. PPPs differ from exchange rates, and are typically derived using surveys of "average" market prices of a bundle of commodities for countries participating in the ICP. The 1993 PPP for China is 1.419 Yuan/USD, which is about 14% of the official exchange rate for that year. Recently, a new set of PPPs have been established for 2005 as part of the latest round of the ICP in which China participated for the first time. See further discussion below. C h i n a ' s p o v e r t y r e d u c t i o n r e c o r d 4 The progress in poverty reduction is equally impressive if based on the new international poverty standard of $1.25 per person per day using the recently-released Purchasing Power Parities for 2005 (Ravallion, Chen and Sangrula, 2008). The new PPPs are based on country-level price surveys for 2005 conducted by the International Comparisons Program (ICP) in which China participated for the first time. The new estimate of household consumption PPP for China is 4.09 Yuan to a dollar for 2005 as compared with Ravallion and Chen (2008). While this substantial upward revision of the PPP has entailed about a 40% reduction in the size of China's economy, and has also implied a significant upward revision of the poverty series using the new (2005) international standard of $1.25 per day, the trend decline in the poverty rates is similar to and no less impressive than that using the old (1993) standard of $1.08 per day (Figure 1.17). For instance, the incidence of consumption poverty declined from nearly 85% in 1981 to 27% in 2004 using the new standard as compared with a decline from 65% to 10% over the same period using the old standard. Figure 1.17 : China's poverty reduction using the new 2005 international poverty standard is no less impressive than using the old 1993 international standard 90 )%( 80 70 CR)H(yt 60 Consumption HCR ($1.25/day, 2005 PPP) pover 50 ofeatr 40 30 20 Consumption HCR ($1.08/day, Headcount 1993 PPP) 10 0 1981 1982 1983 1984 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 Sources and notes: Estimates of consumption poverty incidence for $1.08 per day at 1993 PPPs are the same as in Figure 1.16. The estimates for $1.25 per day at 2005 PPPs are based on rural and urban poverty estimates from Chen and Ravallion (2008) weighted by rural and urban population shares consistent with the sample frame used in NBS Household Surveys. The estimates using the 2005 PPPs also make an adjustment for rural prices being lower than urban prices. ...and half a billion people were lifted out of poverty China's progress in poverty reduction is also strongly reflected in the absolute numbers of people lifted out of poverty. The number of rural poor under the official poverty threshold using the government's mixed income and consumption criterion declined from 152 million in 1981 to 26 million in 2004. By the international standard, the absolute number of poor consuming less than $1.08 per day (1993 PPP dollars) fell in China from 652 million in 1981 to 135 million in 2004, a decline of over half a billion people. A decline in the number of poor of this magnitude over such a short period is without historical precedent. To put this in perspective, the absolute number of poor (using the same standard) in the developing world as a whole declined from 1.5 to 1.1 billion over the same period (World Bank, 2002); in other words, but for China there would have been no decline in the numbers of poor in the developing world over the last two decades of the 20th century. C h i n a ' s p o v e r t y r e d u c t i o n r e c o r d 5 Table 1.6 : Half a billion Chinese have been lifted out of poverty in the last quarter century (Number of poor (millions) under alternative poverty standards, 1981 and 2004) Poverty standard Number of poor Change between (millions) 1981 and 2004 1981 2004 (millions) Official poverty line: income & consumption 152.0 26.1 -125.9 Basic needs line: income 536.7 93.1 -443.6 $1.08/day (1993 PPP): consumption 652.4 134.9 -517.5 $1.25/day (2005 PPP): consumption 848.2 351.1 -497.1 Sources and notes: The official poverty estimates are from NBS (2005a) and relate to rural poor only. The sources for all the other estimates are the same as in Figure 1.16 and Figure 1.17. The depth and severity of poverty also declined considerably It is well-known that the headcount index of poverty is not sensitive to distribution of income or consumption below the poverty line, which raises the question of whether the fall in poverty is also reflected in measures of the depth and severity of poverty (which take into account not only whether a person is poor but also how poor he or she is). Table 1.7 presents the poverty gap and the squared poverty gap indices for the dollar-a-day line using consumption and for the basic needs line using income.4 The Table indicates that the trends are similar across all poverty measures indicating a sharp decline between 1981 and 2001. Table 1.7 : The incidence, depth and severity of poverty: 1981-2001 (percentage) Consumption Income Incidence Depth Severity Incidence Depth Severity (H) (PG) (SPG) (H) (PG) (SPG) 1981 65.2 23.9 11.1 53.9 16.5 7.0 1984 42.7 12.3 4.9 25.1 6.1 2.1 1987 30.2 8.7 3.5 17.6 4.3 1.5 1990 35.0 9.4 3.5 23.4 6.0 2.2 1993 30.3 7.8 2.9 21.3 6.1 2.4 1996 18.7 4.1 1.2 10.5 2.7 1.1 1999 19.9 4.7 1.5 8.5 2.0 0.7 2001 15.6 3.7 1.3 9.3 2.5 0.9 Sources and notes: HCR, PG and SPG refer to the headcount, the poverty gap and the squared poverty gap indices of poverty respectively. Consumption poverty estimates are for the World Bank poverty line; income poverty estimates are for the basic needs poverty line in Ravallion and Chen (2007). The sources are the same as in Figure 1.16 and Figure 1.17. b. Between 2001 and 2004 the pace of poverty reduction accelerated For the purposes of this poverty assessment, a detailed analysis was also carried out using unit-record (household-level) data from the NBS Rural and Urban Household Surveys for 2001, 2003 and 2004. This analysis indicates that poverty reduction accelerated during the 10th Five-Year Plan, especially in terms of consumption (Table 1.8). Both the proportion of the population in poverty and the total number of the poor declined significantly during 2001-2004. For instance, at the World Bank line using per capita income as a measure of welfare, the incidence of poverty declined from 7.7% to 6% of the national population, while the number of poor dropped from about 99 million to 78 million. _________________________________________________________ 4The headcount, poverty gap index and squared poverty gap indices respectively are also commonly referred to as measures of the incidence, depth and severity of poverty (Ravallion, 1994). C h i n a ' s p o v e r t y r e d u c t i o n r e c o r d 6 Table 1.8 : Recent changes in poverty, 2001-2004 Value of line in 2003 all-China rural prices % decline Poverty line: measure of welfare (Yuan/ person/ 2001 2003 2004 during year) 2001-04 World Bank estimates: national poverty % of national (rural+urban) population below the poverty line Official line: income 637 3.42 3.04 2.60 24.1 Official line: consumption 637 5.61 4.59 3.64 35.2 WB line: income 888 7.72 6.93 5.96 22.8 WB line: consumption 888 15.56 13.09 10.38 33.3 Twice the WB line: income 1776 31.19 26.89 23.66 24.1 Twice the WB line: consumption 1776 52.05 45.51 41.45 20.4 Number of poor nationwide (rural+urban) (million) WB line: income 888 98.59 89.51 77.54 21.4 WB line: consumption 888 198.53 169.20 134.90 32.1 World Bank estimates: rural poverty % of rural population below the poverty line WB line: income 888 10.46 9.47 8.17 21.8 WB line: consumption 888 21.04 17.93 14.31 32.0 Rural poverty gap index (%) WB line: income 888 3.08 2.95 2.46 20.2 WB line: consumption 888 4.99 4.27 3.32 33.5 Rural squared poverty gap index (%) WB line: income 888 1.43 1.44 1.21 15.3 WB line: consumption 888 1.75 1.50 1.14 34.6 Sources and notes: World Bank calculations based on NBS household survey data. Estimates for 2001 and 2004 are based on sub-samples of the NBS Rural and Urban Household Surveys. Estimates for 2003 are based on the national sample. The decline is faster for poverty measured in terms of per capita consumption. Consumption poverty incidence dropped from 15.6% to 10.4% while the number of poor fell from 199 to 135 million. The faster decline in consumption poverty relative to income poverty is not limited to just the World Bank line. Using the official poverty line (equivalent to $0.71/day in terms of 1993 PPP dollars), the incidence of consumption poverty declined by 35% compared to a 24% decline in income poverty incidence. Only for the much higher poverty threshold of twice the World Bank line, do we see more comparable rates of decline in income and consumption poverty (24% and 20% respectively). Thus, consumption at the lower end of the distribution has grown faster than income during this period. While the official NBS estimates of poverty, which are limited to rural areas only, differ from the World Bank estimates, they exhibit similar trends of a continued decline in both the proportion and the absolute number of poor during 2001-04 (Table 1.9). The key points of difference in the construction of these two sets of poverty estimates are explained in Box 1.2. C h i n a ' s p o v e r t y r e d u c t i o n r e c o r d 7 Table 1.9 : Official estimates of recent changes in poverty in rural areas, 2001-2004 Value of line in 2003 all-China rural prices % decline Poverty line: measure of welfare (Yuan/ person/ 2001 2003 2004 during year) 2001-04 Official NBS estimates % of rural population below the poverty line Official line: income & cons. 637 3.20 3.10 2.80 12.5 Low-income line: income & cons. 882 9.80 9.10 8.10 17.3 Number of rural poor (million) Official line: income & cons. 637 29.27 29.00 26.10 10.8 Low-income line: income & cons. 882 90.29 85.17 75.87 16.0 Sources and notes: NBS (2005a). Box 1.2 : Understanding the differences between official and World Bank poverty estimates The official NBS estimates of poverty are limited to poverty in rural areas only. These estimates differ from the World Bank poverty assessment (PA) estimates for a number of reasons. Apart from the obvious difference in the official and World bank poverty lines (637 versus 888 yuan per capita per year, as noted above), there are three additional differences in methodology: 1) The PA estimates use either only per capita income or only per capita consumption to estimate poverty measures. The official estimates, on the other hand, use a combination of per capita income and consumption. In particular, in the official methodology, a household is defined to be poor if it meets EITHER of the following conditions: i) household per capita income < the poverty line AND household per capita consumption < 1.2 times poverty line, OR ii) household per capita consumption < the poverty line AND household per capita income < 1.2 times official poverty line. 2) The PA estimates allow for cost of living differentials across provinces and across urban and rural areas using spatial price indices, while the official estimates do not make any provision for such cost of living differences. The spatial price indices used in the PA are based on the work of Brandt and Holz (2006). 3) For 2001 and 2004, the PA estimates are based on a sub-sample of the Rural Household Surveys (RHS) for those years. In particular, the sub-samples cover 386 (or about 45%) of the 857 counties included in the national sample; in particular, all of the 199 poor counties were included in the sub-sample and 187 (or about one-third) of the non-poor counties were included. The non-poor counties were randomly selected, and the sample was appropriately re-weighted to ensure unbiased estimates at the national level. The official estimates for all years are based on the full RHS samples. The last factor however does not appear to matter much. For the year 2003, it is possible to compare the PA estimates based on the full RHS sample and those based on the sub-sample. For rural areas, for instance, the full-sample consumption poverty estimates for the World Bank poverty line is 17.93% as against 18.19% using the sub-sample. The estimates for other poverty lines are similarly close. The decline in the incidence of poverty during 2001-2004 is also paralleled in measures of the depth and severity of poverty, shown in Table 1.8 for rural areas. Overall, these trends for 2001-4 thus not only indicate a continuation of China's longer-term strong performance in reducing poverty, but in particular, relative to the late-1990s, they are also are suggestive of a quickening of the pace of poverty reduction. C h i n a ' s p o v e r t y r e d u c t i o n r e c o r d 8 c. ...and there are indications that poverty has continued to decline up to 2007 Projections of poverty based on grouped consumption distribution data for 2005 and rural and urban per capita consumption growth since 2005 indicate that poverty reduction has continued up to 2007. For instance, the percentage of population with consumption below the World Bank poverty line is projected to have fallen from 6.9% in 2005 to 4.0% in 2007 (Figure 1.18). The projected fall in rural poverty incidence is of similar magnitude, and measures of the depth and severity of poverty, both national and rural, are also projected to have declined in parallel. Figure 1.18 : Progress in poverty reduction during 2005-2007 Headcount Poverty gap Squared poverty gap 9.4 10 8 6.9 6.7 5.4 6 4.9 4.0 4 1.6 2 1.20.3 0.80.2 0.60.2 0.5 1.10.3 0.80.2 0 2005 2006 2007 2005 2006 2007 National Rural Source and notes: The graph plots projections of consumption poverty for the World Bank poverty line using grouped consumption distribution data from the Rural and Urban Household Surveys for 2005 and estimates of rural and urban per capita consumption growth during 2005-07 as published in the NBS's annual Statistical Communiques. These projections allow for cost of living differences between urban and rural areas but not across provinces. All poverty measures are shown as percentages. Apart from continued economic growth, a factor that has contributed to the decline in poverty since 2001 is the reduction in rural taxes and fees. As discussed in Box 1.3, these taxes have been regressive in nature, and their continued reduction since 2001 is likely to have contributed to a more pro-poor incidence of growth over this period. The rapid rise in agricultural subsidies (especially subsidies for agricultural inputs) and the significant expansion of rural social assistance programs (especially rural Di Bao and medical assistance) since 2005 has further contributed to the decline in poverty. C h i n a ' s p o v e r t y r e d u c t i o n r e c o r d 9 Box 1.3 : The impact of reduction in rural taxes and fees on poverty and inequality in China Table 1.10: Rural taxes and fees have been reduced significantly since 2001 Agricultural Taxes and fees Net income (yuan Taxation subsidies Year (yuan per capita) per capita) rate (billion yuan) 2000 95.5 2253.4 0.041 2001 91.2 2366.4 0.037 2002 78.7 2475.6 0.031 2003 67.3 2622.2 0.025 2004 37.5 2936.4 0.013 14.5 2005 13.1 3254.9 0.004 17.4 2006 10.9 3587.0 0.003 30.9 2007 50.3 Sources and notes: China Rural Survey Yearbook, various years. The poor benefited the most in proportional terms from the reduction in taxes and fees Because many taxes and fees were assessed on a per capita basis, China's rural taxation system was highly regressive, with the poorest households paying a higher percentage of their income in taxes and fees. Thus, the poor stood to benefit relatively more from the tax reductions than the rich. This can be seen in Figure 1.19, which plots the mean rural taxation rates by income decile between 2002 and 2005, for households living in China's poor counties. Figure 1.19: The poor proportionally benefited the most from the reduction in taxes and fees Mean taxation rates, by income decile (1=poorest to 10=richest) Source and notes: Calculations by authors using data from the China Poverty Monitoring Report (various years). Calculations assume that the population in each income per capita interval has mean income and pays the mean tax amount. The mean taxation rate for the poorest decile fell from 13.1% in 2002 to 9.6% in 2003, 3.5% in 2005, and 0.6% in 2006. The mean taxation rate for the richest decile fell from 2.5% to 0.3% over the same period. Put another way, for the poorest decile the taxation rate in 2002 was 22 times the rate in 2006 while for the richest decile the 2002 taxation rate was 8 times the 2006 rate. The reduction in rural taxes and fees likely reduced poverty and inequality The 2003 NBS rural household survey data can be used to assess the likely impact that the reduction in rural taxes and fees had on rural poverty and inequality. Because the year 2003 predates the main tax and fee reductions it is possible to use the household-level data to simulate what the level of inequality and poverty would have been had there been no government taxes, fees, or direct subsidies. This exercise makes several simplifications. First, it assumes all taxes and fees are eliminated, even though actual taxation rates have not been fully eliminated. Second, it does not account for how household behavior would have changed with a lower tax rate (e.g. household investment and labor decisions). Third, by simulating the elimination of subsidies (which are only 0.07% of income compared to 2.55% for taxes and fees), the simulation slightly understates the influence of tax and fee reductions. Table 1.11 shows the basic findings from this simulation exercise. C h i n a ' s p o v e r t y r e d u c t i o n r e c o r d 1 0 Table 1.11: The reduction in taxes and fees likely reduced poverty and inequality in rural China Observed: Simulated: WITH taxes and WITHOUT taxes subsidies and subsidies Change % change Gini coefficient of inequality 0.3730 0.3670 0.0060 -1.6 Headcount index 0.0930 0.0832 0.0098 -10.5 Poverty gap index 0.0325 0.0286 0.0039 -12.0 Sources and notes: Calculations using 2003 NBS rural household survey data. The poverty measures are based on per capita income and refer to the "basic needs" poverty line estimated by Ravallion and Chen (2007) of 864 yuan per year (in 2003 prices). Table 1.11 shows that rural inequality and poverty are lower absent direct taxes and subsidies. The Gini coefficient of income inequality falls from 0.373 to 0.367 and the rural poverty headcount calculated using the dollar-per-day poverty line falls by nearly a full percentage point, from 9.30% to 8.32%, a reduction of over 10%. Similarly, the poverty gap falls by 12%. d. But the progress in poverty reduction has been temporally uneven... The progress in poverty The decline in the incidence of poverty was, however, uneven over the two and a half reduction was temporally decades since 1981. Five broad phases can be distinguished, broadly coinciding with the uneven... five Five-year Plan periods. The most rapid declines in poverty, in both the poverty rate and the number of poor, occurred during the 6th, 8th, and 10th Plans. During the 7th and 9th Plan periods, the poverty rate and the number of poor declined only marginally (Figure 1.20).5 The five-phase pattern described above for the incidence of poverty also appears to hold for other measures of poverty (Table 1.7 and Table 1.8). Figure 1.20 : Progress in poverty reduction was temporally uneven Starting poverty rate Ending poverty rate ) 50 70 illions 0 60 m( -50 50 ) (% poorfor -100 40 iontp te ra um -150 30 mbeun onsC verty -200 Change in 20 po in number of poor ngeahC-250 10 -300 0 6th Plan 7th Plan 8th Plan 9th Plan 10th Plan (1981-1985) (1986-1990) (1991-1995) (1996-2000) (2001-2004) Source and notes: Consumption poverty estimates are based on the World Bank poverty line. The source is the same as in Figure 1.16. e. ...and, at the household level, income growth has been uneven Over the period 1990-2004, the annualized mean real income growth rate for the population as a whole was 5.5%, lower than the rate of growth of per-capita GDP, but _________________________________________________________ 5The unevenness is less marked for the official (rural) poverty rates which are indicative of a relatively more steady decline; by the same token, the official poverty statistics do not suggest an acceleration in poverty decline since 2001 (NBS, 2005a) C h i n a ' s p o v e r t y r e d u c t i o n r e c o r d 1 1 impressive nonetheless. However, at the household-level, income growth was uneven. The unevenness of income growth can be assessed by estimating growth incidence curves, which display the annualized rate of growth between two points in time, for each percentile of the income distribution ordered from the poorest to the richest. These were estimated using grouped-data from the Rural and Urban Household Surveys of NBS. Box 1.4 provides details regarding the methodology. Box 1.4 : Constructing growth-incidence curves from grouped-data Growth incidence curves are ideally constructed from unit-record (i.e., household-level) data for multiple years. However, except for 2001 and 2003, nationally representative household data were not available for this analysis. In the absence of unit-record data, grouped-data from the Rural and Urban Household Surveys--which give the percentage of the population in various per-capita income ranges--were used to estimate the incidence of growth in China since 1990. Estimation of growth incidence curves using grouped data involves three steps: · Adjusting incomes for inflation and cost-of-living differences · Fitting a parameterized Lorenz curve to the grouped-data · Aggregating the rural and urban income distributions (Lorenz curves and means) into a single national distribution using population-share-based weights Each of the three steps involves choices. · In adjusting incomes, there are three choices: (i) adjust only for inflation using national rural and urban CPIs, (ii) adjust further for rural-urban cost-of-living differences at the national level, or (iii) use province-specific rural-urban price indices. Since the available grouped data did not provide distributions at the provincial level, only (i) and (ii) were applied. A 27% average cost-of-living differential between rural and urban areas for 2003 (estimated by Brandt and Holz, 2006) was used. · In fitting a parameterized Lorenz curve to the grouped-data, the choice has to do with the particular parametric distributional form that is fitted. A generalized quadratic parametric form was used based on the goodness-of-fit. · And in aggregating the rural and urban income distributions, the main decision is whether to use the rural population share given in Table 4-1 of the 2006 China Statistical Yearbook, which is based on the censuses/annual population surveys (and on residence at the time of the census/survey), or to use the rural population share given in Table 13- 3, which is based on household registration (hukou) and on the administrative classification of rural areas from which the sampling frame for the Rural Household Survey is derived. NBS, to be consistent with the sampling frame of the RHS, in its official estimates of poverty in rural China uses the latter definition of the rural population. In keeping with the procedure followed by the NBS, the growth-incidence curves shown in Figure 1.22 and 1.22 are estimated using rural-urban population-share weights based on the (hukou-based) rural population size given in Table 13-3 of the 2006 China Statistical Yearbook. C h i n a ' s p o v e r t y r e d u c t i o n r e c o r d 1 2 Figure 1.21 displays the growth incidence curve for China between 1990 and 2004. Figure 1.21 : Between 1990 and 2004 incomes grew throughout the distribution, but grew more at the upper end 8 7 Annualized % growth in income: 1990-2004 (by percentile of the income distribution) 6 5 4 3 2 1 0 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% Percentiles of the income distribution from bottom 1% to top 1% Sources and notes: World Bank estimates from grouped-data from the RHS and UHS. Per-capita incomes are adjusted for inflation using the rural and urban CPIs as well as a 27% average cost-of-living differential between rural and urban areas (as estimated by Brandt and Holz (2006). The rural and urban income distributions are aggregated using hukou-based population weights implied by the rural population numbers given in Table 13-3 of the 2006 China Statistical Yearbook. Real income growth rates over the period varied considerably for households at different points in the income distribution, ranging from 3% per year for the poorest percentile to over 8% for the richest percentile. The more modest but positive growth at the bottom of the distribution is consistent with both the overall decrease in the income poverty rate between 1990 and 2004, from 22% to 6%, as well as the remaining widespread vulnerability to poverty (evidence of which is presented in the next section), which suggests that a significant fraction of the population remains near poor. The record over the entire period masks considerable variation in the incidence of growth in various sub-periods (Figure 1.22). There were periods of significant pro-poor growth, notably from 1993 to 1996 and most recently, from 2003 to 2004. But there were also extended periods where income growth lagged (and even declined) at the bottom of the distribution--from 1990 to 1993 and more recently, from 1999 to 2003. In fact, the period as a whole is defined more by these variations than the long-term trend. Only between 1996 and 1999 did the incidence of growth match the long-term trend. C h i n a ' s p o v e r t y r e d u c t i o n r e c o r d 1 3 Figure 1.22 : Income growth incidence curves for China for various sub-periods between 1990 and 2004 Annualized % growth in income 1996-1999 1999-2003 20 16 2003-2004 12 1993-1996 8 1999-2003 4 1996-1999 1990-1993 0 -4 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% Percentiles of the income distribution from bottom 1% to top 1% Sources and notes: World Bank estimates from grouped-data from the RHS and UHS. Per-capita incomes are adjusted for inflation using the rural and urban CPIs as well as a 27% average cost-of-living differential between rural and urban areas (as estimated by Brandt and Holz (2006). The rural and urban income distributions are aggregated using hukou-based population weights implied by the rural population numbers given in Table 13-3 of the 2006 China Statistical Yearbook. Incomes at the bottom of Like the earlier episode between 1990 and 1993 when incomes at the bottom of the the distribution lagged distribution declined, the period between 1999 and 2003 was characterized by a crisis in between 1990 and 1993 farm incomes, in particular income from grain production. Grain prices were low--in part and between 1999 and because domestic prices moved closer to international prices, which were lower, with 2003, in both instances, because of a crisis in China's accession to the WTO--and production fell, as did grain output per worker. This farm incomes coincided with rising prices of inputs, such as fertilizers and diesel oil. For the poorest households, who are most dependent on incomes from grain production, this meant a decline in incomes. In both cases, however, these episodes were followed by periods of significantly pro-poor growth. Between 1993 and 1996, mean growth in per-capita household income was 9% and this growth rate was matched and even exceeded at the bottom of the distribution. Between 2003 and 2004, growth was even more markedly pro-poor with the growth rate exceeding 15% at the poorest percentile and declining monotonically thereafter. The dramatic shift in the incidence of growth was, in both instances, triggered at least in part by favorable government policies towards the agricultural sector. Between 1993 and 1996, agricultural procurement prices were raised, bringing them closer to market prices and thereby, lowering significantly the implicit tax on farmers. In 2004, the response by government took the form of speeding up the elimination of agricultural taxes and fees, efforts to decrease the monopoly of state-owned companies in grain markets, lowering of grain taxes, and provision of increased subsidies for grain production. The increased commitment of the government to supporting farm incomes is reflected in the jump in government budgetary expenditures on agriculture, including expenditures to support rural production, capital construction in rural areas and funds rural relief. Total government budgetary expenditures on agriculture rose from 175 billion RMB in 2003 to 234 billion RMB in 2004, an increase of 30%, and the share of these expenditures in total expenditures rose from 7.1% to 9.7%. The results from the grouped-data are confirmed for the two years, 2001 and 2003, for which unit-record household-level data are available. Between 2001 and 2003 mean income in the bottom quintile increased by 1.7% while the mean income in top quintile C h i n a ' s p o v e r t y r e d u c t i o n r e c o r d 1 4 rose by 15.6% (Figure 1.23). A similar contrast is also notable for consumption growth for the poorest and richest quintiles. Figure 1.23 : The unevenness of growth between 2001 and 2003 (% change in mean per-capita income and consumption between 2001 and 2003, by quintile) % change: income % change: consumption 15.6 16.0 14.2 14.0 11.8 11.9 12.0 11.1 9.5 10.0 7.9 8.0 7.3 6.0 4.6 4.0 1.7 2.0 0.0 2nd 3rd 4th 20% Poorest Per-capita income and consumption quintiles 20% Richest (from poorest to richest) Sources and notes: World Bank estimates from NBS Rural and Urban Household Surveys. The data from the RHS are from a nationally representative 40% sub-sample of the full national sample. The data from the UHS are for the full national sample. Incomes and consumptions are adjusted for inflation and for province-specific rural and urban cost-of-living differences. Four broad conclusions are suggested by these findings on the incidence of growth: · First, it is clear that a rise in inequality does not have to be the price of growth--the two periods with the most pro-poor growth and falling inequality were also the periods when mean income growth was the highest. · Second, the overall rate of growth by itself does not indicate how broad-based growth is--both periods when the incidence of growth was unfavorable for the poor were periods where the mean growth rate was quite respectable. · Third, efforts to bring about broad-based increases in rural incomes clearly can make a difference. · Fourth, while, in a narrow sense, the periods when incomes stagnated at the bottom of the distribution highlight the continued vulnerability of segments of the rural population to adverse grain price and yield shocks, in a broader sense, what these episodes reflect is the fact that a segment of the rural population has been unable to connect to the broader processes of growth by diversifying their income sources and moving beyond their traditional income-generating activities. Efforts to directly support their incomes from agriculture through subsidies and elimination of taxes, while useful for alleviating the short-run burden faced by these households, have to be supplemented with longer-term measures to address the structural causes of their poverty and enable them to move to "modern" higher-productivity activities that are connected to the dynamism of China's economy. And here, as the evidence presented later suggests, investments in human capital will be critical. C h i n a ' s p o v e r t y r e d u c t i o n r e c o r d 1 5 2.But the task of poverty reduction is not complete and in some respects has become harder The official estimate of The most recent official estimate of rural poverty in China for 2007 puts the number of 14.8 million poor in 2007 poor at 14.79 million, or less than 2% of the rural population (NBS, 2008). While there is suggests that China has no official urban poverty line, estimates by others have found poverty levels in urban areas practically solved its to be negligible using an urban poverty line that is comparable to the official poverty line poverty problem for rural areas (Ravallion and Chen, 2007). These estimates thus suggest that only a little more than 1% of China's population is currently poor, which would suggest that China has practically solved its poverty problem. In terms of meeting the most elementary food and clothing needs of most of its population, that is perhaps true. But the evidence that will be presented in this chapter suggests that the task of poverty reduction is not complete and, in some respects, has become harder. ...but the task of poverty The task is not complete because: reduction is not complete · By the international standard for identifying and counting the poor, the number of poor in China remains high, and the official poverty line seems low relative to both international standards and rising incomes and growing aspirations within China. · Vulnerability to poverty because of a variety of income shocks remains widespread. China's success has bred The task has become harder in part because China's success has bred its own its own challenges... challenges. As the poverty rate has fallen, it has become harder to eliminate the remaining poverty because the remaining poor are harder to reach and face more difficult challenges as economic growth is making less of an impact on poverty. ...and new challenges The task has also become harder because the process of structural transformation China have emerged as a result has undergone in the last quarter century as it has continued its transition to a market of China's structural economy has given rise to several trends that have emerged as new areas of concern and transformation imply new challenges for poverty reduction policy: · Income inequalityhas risen between and within rural and urban areas. · Disparities in other dimensions of human development remain and in some cases have grown. Because of the marketization of public service delivery, incomes matter more than they used to for access to health and education, and the burden of health and education expenditures has gone up for households. · The restructuring of state-owned enterprises and the dismantling of the "iron rice bowl" has led to rising unemployment and declining labor force participation... · ...while the increasing share of the private sector in employment has led to growing informalization of the urban labor market. · The growth of a large "floating population" of rural migrants in urban areas has raised concerns about the emergence of an urban underclass. a. Measured by international standards, the number of poor in China remains high China still has the second Though China's contribution has been central to poverty reduction regionally and globally, largest number of because of the sheer size of China's population, by international standards, China still consumption poor in the accounts for a large share of the world's poor. As of 2005, the latest year for which direct world survey-based estimates are available, China still had 254 million people consuming less than $1.25 per day in 2005 PPP dollars (90 million people consuming less than $1.08 per day in 1993 PPP dollars), the second largest concentration of extreme poor in the world after India. And so, though China was responsible for three-quarters of the poverty reduction in East Asia between 2001 and 2004, China also remains the home of nearly four-fifths of the poor in the region (World Bank, 2007b). If this assessment of the number of remaining poor in China appears to be at odds with the official estimates of only 15 million rural poor, it is because the official poverty line (set A b r o a d e r p o v e r t y r e d u c t i o n a g e n d a 1 6 at 785 Yuan per person per year for 2007) is particularly stringent. The official poverty line seems low both compared to international standards and relative to mean incomes and growing aspirations within China. Furthermore, some recent calculations suggest that even by the objective standard that a poverty line aims to capture--the minimum expense necessary for subsistence and to meet basic living needs--the poverty line may be too low. China's official poverty line is low by international standards China's official poverty China's official poverty line (of about 71 cents a day in international prices using 1993 line is low compared to Purchasing Power Parities) is relatively low when compared with national poverty lines international standards used by other countries around the world. Of the 63 countries for which such information is available, only eight use national poverty lines that are below a dollar-a-day (in 1993 PPP dollars). China is one of the eight, and of the eight, has the lowest ratio of the official poverty rate to the dollar-a-day poverty rate (Figure 2.24; top panel). The same is also true if national poverty lines are compared using the new sets of PPPs for 2005. China's official poverty line (equivalent to about $0.57 per person per day in 2005 PPP dollars) is the lowest amongst a sample of 75 countries (Figure 2.24; bottom panel). For most countries, the dollar-a-day poverty threshold itself has no special normative significance but helps to establish a standard for the purposes of international comparison. In the case of China, however, as discussed in chapter 1, the 1993 dollar-a-day threshold coincides with the estimate of a "basic needs" line using a well-accepted international methodology (Ravallion and Chen, 2005) and so it does have some relevance as an objective standard. But even if that were not the case, the evidence does suggest that China's national poverty line is low relative to standards commonly used by other countries--including many at lower stages of development than China--to monitor poverty. Figure 2.24 : China's poverty line in comparison with other countries Ratio of poverty rate under national poverty line to poverty rate under dollar-a-day line 12 10 8 6 4 NA 2 CHI 0 li s a nd ilaahT Braz esin DRP etnam bodia India donI Vi Lao ongolia M Philippine maC W h y t h e t a s k i s n o t c o m p l e t e 1 7 9 Poverty lines across countries 8 (dollars per person per day at 2005 PPPs) 7 6 5 4 3 2 1 0 Sources and notes: Top panel: For each of 63 countries for which data are available, the figure plots the ratio of the estimated poverty rate based on the national poverty line used in the country to that using the international (1993 dollar-a-day) line based on 1993 PPPs. The poverty estimates are for different years for different countries, but with one exception (Sierra Leone, 1989) they all relate to the 1990s or 2000-03. These estimates are taken from World Bank (2005b). Bottom panel: The figure plots national poverty lines for 75 countries where the poverty line for each country is converted into US dollars per person per day using the new set of PPPs for consumption for 2005 (Ravallion, Chen and Sangrula, 2008). China's official poverty line seems low relative to rising incomes and growing aspirations China's official poverty As countries develop, they tend to periodically raise the thresholds they use to monitor line is low relative to levels of national poverty. This is consistent with the notion that with a rise in the general mean incomes... standard of living what is commonly considered an acceptable minimum level of living also tends to increase. This explains, for instance, why the poverty lines used in developed countries are typically higher in real terms than those in relatively poorer countries. China's official poverty line was first developed in the mid-1980s and it has been used to monitor poverty levels for the period beginning the early 1980s (Park and Wang, 2001). However, since then China's real per capita GDP has grown more than five times, and reflecting this enormous growth in real incomes, the official poverty line, which was about 68% of the rural mean income in 1980, has fallen to 19% of the rural mean income by 2007 (Figure 2.25). With this growth in living standards, the country and its people's aspirations have grown and it is arguable that current poverty threshold now does not adequately reflect these aspirations. W h y t h e t a s k i s n o t c o m p l e t e 1 8 Figure 2.25 : The relative decline of China's poverty line, 1980-2007 (figure sub-title) Ratio of official poverty line to rural mean income 0.8 0.68 0.6 0.4 0.19 0.2 0 1980 1982 1984 1986 1988 1990 1992 1994 1996 1998 2000 2002 2004 2006 Source and notes: NBS ( 2005, 2007, 2008) Subjective assessments More direct evidence on the changing norms regarding the minimum acceptable level of of the minimum living is provided by the subjective poverty lines estimated using recent data from the acceptable level of living 2004 NBS Urban Short Form Survey. This survey included questions on the households' are several times higher own assessment of the adequacy of their income to meet their basic needs. Using this than the official poverty line information, the subjective poverty line for 2003 for urban areas is estimated at 4147 Yuan per capita per year for 2003 (see Box 2.5). The rural equivalent of this line is estimated at 2935 Yuan per capita per year, which is several times the official poverty line. Box 2.5 : Estimating subjective poverty lines in China More direct evidence of the changing norms of the minimum acceptable level of living is provided by recent data from the 2004 NBS Urban Short Form Household Survey which included the following question on the households' self-assessment of income adequacy: "Compared to your household's basic needs, your household income is: (i) not enough, (ii) just enough, (iii) more than enough, (iv) much more than enough." Using respondents' answers to this question, it is possible to estimate a subjective poverty line for urban households. This line has the following property: on average, those households whose per capita income is above this line consider their incomes to be enough or more than enough to meet their basic needs, while households whose income is below this line consider their incomes to be inadequate to meet their basic needs. In Figure 2.26, this subjective poverty line is depicted as z*. For 2003, the subjective poverty line is estimated at 4147 Yuan per capita per year at all-China 2003 urban prices.6 _________________________________________________________ 6 The subjective poverty line is determined by fitting an ordered probit model to the categorical variable corresponding to the income adequacy question, which is taken to be a function of log per capita income of the household. Following Pradhan and Ravallion (2000), the subjective poverty line is estimated as the fixed point of this function (i.e. where the latent minimum required income equals the actual income of the household). W h y t h e t a s k i s n o t c o m p l e t e 1 9 Figure 2.26 : The subjective poverty line Subjective minimum income Subjective poverty 45-degree line Actual line income z* Note: The figure plots the reported or estimated minimum income required on actual income of the household and the subjective poverty line is determined by the point (z*) where the minimum required income equals actual income. The Urban Short Form Household Survey however measures household income differently to how it is measured in NBS's annual Urban Household Survey. The former used a single question on household income, while the latter has a more thorough and arguably more accurate measure of different components of household incomes. Thus, in order to estimate the subjective poverty line for 2003, the Urban Short Form Survey was calibrated with the NBS' regular Urban Household Survey for 2003 as follows. The direct estimate of the subjectively poverty line based on the Urban Short Form Survey was 2986 Yuan per person per year, and the incidence of subjective poverty was about 17% of the urban population in 2003. The equivalent subjective poverty line using the NBS Urban Household Survey for 2003 was thus established as the line that also generated a poverty rate of 17%, which came to 4147 Yuan per capita per year. While similar data do not exist to carry out a parallel exercise for rural areas, the rural equivalent of the subjective poverty line could be estimated by deflating the urban line by the urban-rural cost of living differential for 2003. Two alternative estimates of the latter are 41.3% (Ravallion and Chen, 2007) and 26.5% for 2003 (Brandt and Holz, 2006). Even if one were to use the higher estimate of 41.3%, the equivalent rural line is estimated at 2935 Yuan per capita per year at all-China rural prices, or 4.6 times the official poverty line. Higher aspirations with economic development are also revealed by the 12-village qualitative poverty assessment. For instance, instead of the adobe houses typical of the 1980s, villagers are increasingly building brick houses, and young families are less willing to live in shared housing. Having a house is seen as a precondition for marriage and is one of the largest expenses undertaken by rural households who often go into considerable debt for this purpose. However, for farmers, house building is not intended to meet the needs of survival or for conspicuous consumption, but simply represents a necessary condition for family reproduction (CASS-IOS, 2006). Higher norms of a minimum level of income are also being reflected in some elements of government policy. For instance, the Di Bao minimum income support program administered by the Ministry of Civil Affairs in urban areas uses a set of city-specific Di Bao lines to target eligible households with annual per capita incomes below these lines (MOCA, 2003). The national average Di Bao line in 2003 was 1928 Yuan per person at national urban prices, or a rural equivalent of 1364-1524 Yuan depending upon the urban- rural cost of living differential used. The key point here is not that the subjective poverty line or the Di Bao line should now be used for future monitoring of poverty in China, but rather that people (and some parts of the government) in China themselves have significantly revised upwards their notions of what should be considered an adequate minimum level of income, and thus the official poverty line of just over 2 Yuan per person per day is now arguably out-of-step with the currently prevalent norms and expectations. W h y t h e t a s k i s n o t c o m p l e t e 2 0 China's poverty line may not cover essential food and non-food needs Recent calculations of food poverty lines in China indicate that the current official poverty line may be deficient in meeting the current cost of basic food and non-food needs. Two sets of estimates are pertinent here: first, by Ravallion and Chen (2007) and second by Meng, Gregory and Wang (2005). Both studies estimate food poverty lines separately for each province; Ravallion and Chen estimate both rural and urban lines for 2002 while Meng, Gregory and Wang estimate only urban lines for the period 1986-2000. Both follow a similar methodology where the food poverty lines are anchored to a nutritional requirement of 2100 calories per person per day (with 75% of the calories from foodgrains in Ravallion and Chen). Both studies use province-specific food bundles for a reference group of relatively poor households. 7 The average food poverty line for 2002 for rural areas estimated by Ravallion and Chen is 620 Yuan per person per year while that for urban areas is 865 Yuan. The 2003 equivalents of these lines are 630Yuan and 873 Yuan for rural and urban areas respectively. Similarly, Meng, Gregory and Wang estimate the average urban food poverty line for 2000 at 1316 Yuan per person per year8, which is equivalent to 1324 Yuan for 2003.9 Applying a 39% urban-rural differential implied by the Ravallion-Chen food poverty lines, the rural equivalent of the Meng-Gregory-Wang line for 2003 could be roughly estimated at 955 Yuan. Relative to these food poverty lines, the official poverty line of 637 Yuan appears to be on the low side. Even limiting the comparison to the lower food poverty line of Ravallion and Chen, it appears that the official poverty line would have to be almost fully allocated to food in order to meet the nutritional requirement of 2100 calories, with no allowance for basic non-food needs. Or put differently, if some allowance is made for basic non-food needs, then the official poverty line seems to be inadequate for meeting the minimum caloric requirement. The government's poverty line was also found to be too low during the qualitative poverty assessment undertaken in 12 study villages: Box 2.6 : Views from China's villages about the poverty line In the study villages in Jiangxi province, villagers argued that there is no household whose per capita income is less than 625 Yuan in their village and provided evidence for this in two ways: one is that the local governmental support for Wu Bao households has reached 800 Yuan per person a year. It indicates that households could not survive if per capita annual income were 625 Yuan or below; the other is that by their calculation, at least 900 Yuan is needed for one's basic living expenditure in a year. At one of the research sites in Sichuan Province, at least 1500 Yuan is needed for local villagers to survive. In this case, households whose incomes are below the poverty line cannot continue their daily life and need social relief. A similar situation was observed in Jiangsu Province. Jiangsu Province identified 637 Yuan per capita per year as the absolute poverty line, but this amount cannot ensure survival in Z Village. Most farmland in Z Village has been encroached by industrialization and the remainder cannot be cultivated due to water pollution. Therefore, vegetables and food cannot be obtained from the farmland and must be purchased at the market, as must drinkable water, electricity and other daily items. The daily expenditure of the villagers is similar to that of urban residents. Therefore, with at least 6000 Yuan per year a family of three people can survive. The local subsidy for Wu Bao hu (households without work capacity or other means of support) has reached 3000 Yuan per person a year. The local government of WJ District where Z Village is located established the minimal living support (Di Bao) line of 2160 Yuan per person per year, which is much higher than the absolute _________________________________________________________ 7Although in case of Ravallion and Chen these are based on the actual consumption bundles of those between the poorest 15th and 25th percentile nationally, while Meng, Gregory and Wang base it on the consumption bundle of the poorest 20% in each province. 8This is a population-weighted average of provincial poverty lines reported in Meng, Gregory and Wang (2005). 9Similar or even higher poverty lines have also been estimated by other researchers; for instance; Hussain (2003) estimated an average urban food poverty line of 1392 Yuan per person per year for 1998 also using a norm of 2100 calories per person per day. W h y t h e t a s k i s n o t c o m p l e t e 2 1 poverty line as well as the Di Bao line in other provinces. However, with regard to the consumption level in southern Jiangsu Province, Di Bao should be regarded as the poverty line between the standards of the poorest and the basic living standard. However, the poverty line based on the villagers' perceptions is higher than this poverty line. According to the villagers, the poor households not only include Di Bao households, but also those who are not Di Bao according to their income, yet their income cannot balance expenditure and households have to be in debt and seek help from relatives and friends. Source: Fieldwork in Jiangxi, Sichuan and Jiangsu provinces. CASS-IOS (2006). Thus, to conclude, whether in relation to poverty standards used in other countries, or the rise in living standards within China, or in relation to recent valuations of the cost of meeting basic needs, China's current poverty threshold seems to be relatively low. While this threshold may have served its purpose well in the 1980s and 1990s (when the overall levels of poverty were relatively high), continuing reliance on it now as a guide to future poverty reduction policies runs the risk of demarcating the problem (and hence its attempted solutions) too narrowly. It may be time for China to consider adopting a higher poverty threshold. b. Vulnerability to poverty because of income shocks remains widespread Vulnerability to poverty If those who exited poverty did so permanently and no new poor emerged, the task of because of income eliminating poverty would be relatively simple. However, in practice, even amongst the shocks remains non-poor, some households may face the possibility of falling into poverty at some point in widespread... the future because they are unable to cope with income shocks against which they have limited or sometimes no insurance. Such shocks can be large-scale--the financial crisis of the late 1990s, the SARS epidemic, and the tsunami disaster of 2004 are leading examples of large-scale shocks that have affected East Asia--in which case they tend to generate the most public attention. But a number of recent studies indicate that even in the absence of such high-profile aggregate shocks, there is considerable `churning' in and out of poverty, reflecting the exposure of households to a myriad of more localized and idiosyncratic shocks, ranging from the ill health of the primary income-earner and job loss, to crop-yield variability, adverse commodity price movements and area-specific natural hazards. At any given point in time, therefore, besides the group of people who are currently observed to be poor, there may be many who face a significant risk of falling into poverty--i.e., are vulnerable to poverty. How large then is the challenge of poverty alleviation in China viewed from this (dynamic and stochastic) perspective? This section presents some recent findings based on analyses of panel data for 25987 households that were included in each of three Rural Household Surveys for 2001, 2003 and 2004 conducted by the National Bureau of Statistics. Since the same survey instruments were used in each year, the panel provides consistent measures of income and consumption for each household at three points in time.10 It is thus possible to track the poverty status of each household for each of the three years. Two major findings emerge from this analysis: · there is a much larger population at risk of poverty than is likely to be inferred from observed poverty rates at any given time, · for both the minority who are persistently poor and the much larger number who are vulnerable to transitory episodes of poverty, the severity of poverty is largely attributable exposure to uninsured risks _________________________________________________________ 10The longitudinal panel was specially constructed for the analysis undertaken for this poverty assessment as part of a collaborative effort with the National Bureau of Statistics. The panel is a sub-sample of the annual Rural Household Surveys (RHS), but it has national representation. It covers 386 of the 847 counties included in the RHS (excluding Tibet). In particular, the panel includes households for all the 199 designated poor counties and for 187 out of 648 non-poor counties included in the RHS sample. Random selection amongst the non-poor counties ensured that the panel is nationally representative. W h y t h e t a s k i s n o t c o m p l e t e 2 2 These findings are described in greater detail below. Nearly a third of China's rural population was poor at least once between 2001 and 2004 Typical poverty monitoring is based on cross-sectional surveys which present information on income/consumption distributions one date at a time. Poverty measures are estimated from these data one year at a time. However, since different sets of households could be poor in different years, a cross-sectional series of poverty measures does not shed light on the size of the population that faces a significant risk of poverty over a period of time. The evidence summarized in Table 2.12 and Table 2.13 shows that this population is relatively large in China. Table 2.12 : The incidence of persistent and transient income poverty in rural China: 2001-2004 Income poverty Share of rural population (%) Poor in Average Poor in 2 only 1 of Poor in at headcount Poor in of the 3 the 3 least one Not poor index Vulnerability-to- all 3 years years years year in any year 2001-04 poverty ratio (4)=(1)+(2)+ Region (1) (2) (3) (3) (5) (7) (8)=(4)/(7) Coastal 0.1 0.8 5.4 6.3 93.7 2.6 2.4 (1.6) (13.2) (85.1) (100) Northeast 0.5 4.3 15.8 20.6 79.4 9.0 2.3 (2.6) (20.6) (76.8) (100) Central 0.7 3.1 12.3 16.1 83.9 7.0 2.3 (4.3) (19.2) (76.4) (100) Southwest 5.3 7.1 14.3 26.7 73.3 15.2 1.8 (20.0) (26.7) (53.4) (100) Northwest 9.3 16.8 24.5 50.5 49.5 28.2 1.8 (18.4) (33.1) (48.5) (100) Area Non-mountainous, non-minority 0.5 2.4 9.4 12.3 87.7 5.2 2.4 (4.1) (19.3) (76.6) (100) Non-mountainous, minority 6.8 9.9 20.3 36.9 63.0 20.3 1.8 (18.3) (26.9) (54.8) (100) Mountainous, non-minority 4.8 8.5 16.0 29.3 70.7 15.8 1.9 (32.2) (57.0) (10.8) (100) Mountainous, minority 10.3 12.5 20.2 42.9 57.1 25.8 1.7 (23.9) (29.1) (47.0) (100) All rural 2.3 4.6 11.9 18.8 81.2 9.3 2.0 (12.2) (24.4) (63.4) (100) Sources and notes: World Bank estimates based on panel data from NBS' Rural Household Surveys for 2001, 2003 and 2004, using the poverty line of 888 Yuan per persons per year at 2003 rural prices. For a definition of regions and topographic/ethnicity categories, see Box 4.10. The numbers in parentheses report the shares of the poor in all 3 years, in 2 years and in one year respectively as percentages of the vulnerable population (poor in at least one year). Thus, for instance, using per capita income as the measure of welfare, during 2001-04, on average, 9.3% of the rural population in China was below the World Bank poverty line of 888 Yuan per person per year at 2003 prices. (This is an average of rural headcount indices of 10.3, 9.6 and 8.1% for 2001, 2003 and 2004 respectively.) This can be compared with the percentage of the rural population who were poor in only one year, two years or all three years; these are estimated at 11.9, 4.6 and 2.3% respectively. One can also calculate the proportion of population who were poor in at least one of the three years ­ who, in other words, had a one-third or higher probability of falling into poverty. If this segment were to be designated as "vulnerable", the proportion of this vulnerable group in rural population is estimated to be 18.8%. Thus, the population facing a significant risk of poverty is about twice as large as the population typically estimated to be in poverty. As shown in Table 2.13, the pattern is similar for consumption poverty; while the average proportion in poverty over the three years was 17.8% (4.2%), the proportion of poor in at least one of the three years was 30.9% (9.6%), implying a vulnerability-poverty ratio of 1.7 (2.3). W h y t h e t a s k i s n o t c o m p l e t e 2 3 Table 2.13 : The incidence of persistent and transient consumption poverty in rural China: 2001-2004 Consumption poverty Share of rural population (%) Poor in Average Poor in 2 only 1 of Poor in at headcount Poor in of the 3 the 3 least one Not poor index Vulnerability-to- all 3 years years years year in any year 2001-04 poverty ratio (4)=(1)+(2)+ Region (1) (2) (3) (3) (5) (7) (8)=(4)/(7) Coastal 2.0 4.4 10.7 17.1 82.9 8.7 2.0 (11.7) (25.9) (62.5) (100) Northeast 3.3 7.3 17.7 28.4 71.6 14.6 1.9 (11.7) (25.8) (62.5) (100) Central 5.0 9.0 17.2 31.2 68.8 17.0 1.8 (16.1) (28.8) (55.1) (100) Southwest 11.3 11.2 16.2 38.6 61.4 24.8 1.6 (29.3) (28.9) (41.8) (100) Northwest 21.8 20.1 18.5 60.4 39.6 41.0 1.5 (36.1) (33.3) (30.7) (100) Area Non-mountainous, non-minority 4.0 7.4 14.4 25.7 74.3 13.7 1.9 (15.4) (28.8) (55.8) (100) Non-mountainous, minority 16.1 12.4 21.9 50.4 49.6 32.4 1.6 (31.9) (24.6) (43.5) (100) Mountainous, non-minority 10.9 11.1 15.7 37.7 62.2 23.2 1.6 (28.9) (29.5) (41.7) (100) Mountainous, minority 19.1 15.6 17.2 51.8 48.2 36.2 1.4 (36.8) (30.1) (33.1) (100) All rural 6.9 8.9 15.1 30.9 69.1 17.8 1.7 (22.3) (28.9) (48.9) (100) Sources and notes: World Bank estimates based on panel data from NBS' Rural Household Surveys for 2001, 2003 and 2004, using the poverty line of 888 Yuan per persons per year at 2003 rural prices.. For a definition of regions and topographic/ethnicity categories, see Box 4.10. The numbers in parentheses report the shares of the poor in all 3 years, in 2 years and in one year respectively as percentage of the vulnerable population (poor in at least one year). Another useful ratio to look at is the share of those who are persistently poor (poor in all three years) to the vulnerable (poor in at least one year). For income poverty, this share is 12%, and it is 22% for consumption poverty. Thus, one in every eighth or fifth vulnerable person is persistently poor, while the rest are poor only periodically. This is indicative of a relatively high degree of temporary poverty. However, there are marked differences in the ratio of persistently poor (poor in all three years) to the vulnerable. Across regions, for instance, in the case of income poverty the ratio of persistently poor is less than 5% in coastal, northeast and central regions, while it is around 20% in the western region. A similar regional contrast is also notable for dollar- a-day consumption poverty, where the share of persistently poor amongst the vulnerable in western region is 2-3 times as that in other provinces. Similarly, remoteness and minority status are associated with significantly higher shares of persistently poor amongst the vulnerable; for consumption poverty, for instance, this share is 15% for non-mountainous, non-minority areas, while it is 37% for mountainous, minority areas. The increasing importance of vulnerability is also strongly indicated in the qualitative poverty assessment in the 12 study villages. Rural poverty is increasingly seen by the villagers not as a constant lack of adequate food and clothing (although that was seemed to be true for a few households) but mainly in terms of economic vulnerability, poverty circulation of households and inconsistencies with social development. The villagers alluded to the unstable balance of income and expenditure which (in the absence of adequate social relief) is easily broken by many events such as illness, disasters, accidents or business failure. Being in debt (in order to meet any of these contingencies) was widely reported to be an important indicator for identifying households in poverty. W h y t h e t a s k i s n o t c o m p l e t e 2 4 The villagers' accounts provide many examples of how new risks have emerged or existing risks have become more severe. Box 2.7 : New or aggravated risks of falling into poverty in study villages The villagers in the study villages identify economic vulnerability as a major issue for rural poverty. Their accounts provide many examples of the risks faced by households that make them vulnerable to falling into poverty: Ill health has become one of the most important factors leading to rural poverty in an environment of unaffordable medical care. According to the township hospital in J township of Jiangxi Province, the average expenditure to see a doctor is about 30 Yuan for one visit. Medical care has become a major part of daily living expenditure. According to fieldwork in Q Village of Inner Mongolia, one household paid more than 4000 Yuan for hospitalization due to a serious disease in 2005, but was only reimbursed 750 Yuan. According to CMS, households still need to bear the burden of a large part of the costs of hospitalization; hence the system cannot solve the problem that many households do not go to hospital when it is necessary. Moreover, some chronic diseases are the main factors leading to poverty, but expenditure on them cannot be reimbursed simply because the patients are not hospitalized. While farmers have gained from greater economic integration of agricultural markets, they are also more exposed to price shocks. For instance, most of the households in J Village in Sichuan had moved out of poverty through growing mushrooms on a large scale in the early 1990s. However since 1995, the price of mushrooms in the local market has plummeted, and as a result, households' incomes have declined rapidly, some households have become indebted, and many villagers have left for cities to seek jobs. Another major risk is that rural migrant workers cannot get their wages on schedule, and there are large wage arrears. Villagers in DH Village of Gansu were unable to afford their children's education due to delays in payment of their wages and could not migrate for work any more. Source: CASS-IOS (2006). A large part of the severity of poverty in rural China is attributable to risk With comparable data for the same households for several years, it is also possible to decompose observed average poverty over a period of time into what have been termed chronic and transient poverty components. Chronic poverty refers to poverty on account of low mean incomes over time. This is the poverty level that would result even if there was no variability in household incomes over time. The remainder ­ the difference between average observed poverty and chronic poverty ­ is the transient component of poverty and is attributable to variation in incomes over time.11 The share of transient poverty in aggregate poverty can thus be interpreted as the contribution of risk to overall poverty. There is an underlying concern that the nature of poverty may be changing over time in China. In particular, with the large-scale structural transformations in the country and China's success in combating extreme poverty, the issue is whether transient poverty is now becoming the dominant form of poverty in China. Ultimately, this issue is of concern because different sets of policies may be needed to address chronic and transient poverty. Table 2.14 presents the main results where the aggregate squared poverty gap indices for rural China for 2001-04 are decomposed into transient and chronic components.12 In interpreting these results, one should note that some of transient poverty could be on account of random errors in measuring household income or consumption consistently over time; the transient component is thus better interpreted as an upper bound on the extent of transient poverty. _________________________________________________________ 11For a formal derivation of the chronic-transient poverty decomposition, see Ravallion (1988). 12The squared poverty gap index is more appropriate for such decomposition because of its property that it is sensitive to variations in income or consumption below the poverty line. The commonly-used headcount index does not have this property. For instance, a household who is poor in all time periods, but has a highly variable income, will still have both its aggregate observed poverty and the chronic poverty measures equal to unity. The headcount index fails to capture transient poverty despite the variability in the household's income. See Ravallion (1988) for further discussion. W h y t h e t a s k i s n o t c o m p l e t e 2 5 Table 2.14 : The contribution of risk to income and consumption poverty in rural China (% of aggregate squared poverty gap index 2001-04 attributable to risk) Severity of poverty Income Consumption Overall % attributable Overall % attributable level to risk level to risk Region Coastal 0.49 94.8 0.51 60.8 Northeast 2.43 89.9 1.05 61.1 Central 0.72 86.6 0.95 53.8 Southwest 1.76 49.5 2.26 34.3 Northwest 4.54 62.1 4.61 32.6 Type of area Non-mountainous, non-minority 0.78 87.0 0.86 54.3 Non-mountainous, minority 3.38 62.4 3.41 31.0 Mountainous, non-minority 2.12 61.8 2.25 34.9 Mountainous, minority 3.36 48.2 3.74 31.6 All rural 1.32 69.7 1.44 42.0 Sources and notes: World Bank estimates based on panel data from NBS' Rural Household Surveys for 2001, 2003 and 2004 using the poverty line of 888 Yuan per persons per year at 2003 rural prices.. For a definition of regions and topographic/ethnicity categories, see Box 4.10. The severity of poverty refers to aggregate squared poverty gap indices which are expressed as percentages. A key finding from the decompositions is that the contribution of risk to aggregate poverty is high. It is almost 70% for income poverty. Using consumption as a measure of welfare for the same line, this declines to about 42% which is indicative of a degree of consumption smoothing by the poor in the face of income shocks.13 There are also large variations in the share of transient poverty across regions and topographical/ethnicity categories. The highest shares of transient poverty are found in the coastal, northeast and central regions, and in non-mountainous, non-minority areas, while chronic poverty is relatively more prevalent in the western region and especially in mountainous, minority areas. The relative contribution The numbers in Table 2.14 suggest that the relative contribution of risk rises as poverty of risk rises as poverty becomes less severe. For instance, in the coastal region, where income poverty is least becomes less severe severe, nearly 95% is attributable to risk; in contrast, risk accounts for less than two-thirds of the overall severity of poverty in the northwest region where poverty is most severe. Figure 2.27, which plots the share of the aggregate squared poverty gap attributable to risk against the squared poverty gap indices across 20 region-topography-ethnicity categories, confirms this pattern at a more disaggregated level. Thus, as poverty levels decline the problem of risk-induced poverty tends to become relatively more important. _________________________________________________________ 13Other studies have found comparable results. For instance, Jalan and Ravallion (1998, 2000) found a 49% share of transient poverty for the rural areas of four provinces in southern China for 1985-90 using consumption as a measure of welfare. Yue, Li and Wang (2005) on the other hand report a variety of estimates for different poverty lines and welfare measures for the designated poor counties in China for the years 1997-2001. For a poverty line close to the dollar-a-day poverty line, they report a 77% share of transient poverty using per capita income as a measure of welfare and a 42% share using consumption per capita. Similarly, for the same poverty line, Whalley and Yue (2006) report transient poverty shares of 53 and 41% respectively of income and consumption-based welfare measures using longitudinal data for rural China for the years 1998-2002. W h y t h e t a s k i s n o t c o m p l e t e 2 6 Figure 2.27 : The relative contribution of risk rises as poverty becomes less severe (% of the aggregate squared consumption poverty gap attributable to risk across 20 region- topography-ethnicity segments of the rural population) to 100 risk 80 of (%) 60 n 40 All rural poverty 20 ontributioC 0 0 1 2 3 4 5 6 7 Aggregate squared poverty gap (%) Source and notes: World Bank estimates based on panel data from NBS' Rural Household Surveys for 2001, 2003 and 2004. For a definition of regions and topographic/ethnicity categories, see Box 4.10. Each point refers to one of the 20 segments of the rural population formed as combinations of five regions, two topographical categories (mountainous and non-mountainous areas) and two ethnicity categories (minority and non-minority areas). The squared poverty gap is in terms of consumption poverty relative to the World Bank poverty line of 888 Yuan per person per year at 2003 rural prices. These findings have two implications. First, although we do not have the requisite data to conduct this analysis for urban areas, the transient component of urban poverty is likely to be high in view of the comparatively low levels of poverty in urban areas. Second, it also suggests that as China makes further progress in poverty reduction, the poverty challenge facing the country will increasingly take the form of tackling transient poverty. One should however note that a relatively high share of transient poverty does not mean that the problem of poverty has been rendered necessarily less serious or its solutions relatively easier. But, it does imply that social policies may have to focus on risk mitigation and risk management strategies more than was necessary when chronic poverty was the dominant form of poverty. But risk is an element of At the same time, it should be noted that even for the persistent poor, exposure to risk poverty even for the adds to the severity of poverty. For instance, as shown in Figure 2.28, though chronic persistently poor poverty associated with low mean levels of consumption is responsible for a greater fraction of the severity of poverty for rural residents of the northwest, in absolute terms, the component of the aggregate squared poverty gap attributable to risk is large, larger even than for households in more prosperous areas--e.g., the coastal region--where exposure to risk is the primary cause of poverty. What this suggests is that even for the persistent poor, policies are needed not just to build up asset bases and long-term income-generating capacity, but also to mitigate exposure to risk. In fact, research elsewhere in the developing world indicates that because exposure to risk can keep households from adopting high-risk but high-return income-enhancing strategies, reducing this exposure can itself enable the poor to increase their long-term income prospects. W h y t h e t a s k i s n o t c o m p l e t e 2 7 Figure 2.28 : Even for the persistently poor, risk adds to the severity of poverty (Aggregate squared poverty gap(2001-2004) and its components, by segment ) Component attributable to low-mean Component attributable to risk 7 6 5 4 3 2 1 0 n n n n n -mi -mi -mi -mi -mi Rural nsia nsia nsia nsia nsia Plains-oth Pl Mount-oth Mount-min Plains-oth Pl Mount-oth Mount-min Plains-oth Pl Mount-oth Mount-min Plains-oth Pl Mount-oth Mount-min Plains-oth Pl Mount-oth Mount-min Coastal Northeast Central Southwest Northwest All Source and notes: World Bank estimates based on panel data from NBS' Rural Household Surveys for 2001, 2003 and 2004. For a definition of regions and topographic/ethnicity categories, see Box 4.10. Each point refers to one of the 20 segments of the rural population formed as combinations of five regions, two topographical categories (mountainous and non-mountainous areas) and two ethnicity categories (minority and non-minority areas). The squared poverty gap is in terms of consumption poverty relative to the World Bank poverty line of 888 Yuan per person per year at 2003 rural prices. Exposure to risk may be one reason why China's poor have unusually high savings rates Exposure to risk and the Exposure to risk can also depress consumption levels as households, even the poor, resulting need for engage in precautionary savings to protect themselves from future income shocks. The precautionary saving may considerable risk that rural households are exposed to may be one reason why a be one reason why so remarkably high fraction of even China's poor and low-income households save, and why many of China's poor and low-income households household savings rates are so high (Table 2.15). For instance, in rural areas, 43% of the China tend to save... households with per-capita annual incomes below the World Bank poverty line of 888 Yuan per person per year save, and the median savings rate amongst low-income households earning between the poverty line and twice that line is over 17%. W h y t h e t a s k i s n o t c o m p l e t e 2 8 Table 2.15 : Even the poor in China tend to save and household saving rates are high (Median savings rates and % of households saving in different income ranges) Rural Urban Median Median % of house- savings rate % of house- savings rate Annual per-capita household holds (as % of holds (as % of income range saving income) saving income) Less than 888 yuan (WB poverty line) 43.1 -5.7 23.5 -13.2 Between 888 and 1776 yuan 70.6 17.5 48.5 -0.8 Between 1776 and 2664 yuan 83.0 30.4 60.3 4.3 Between 2664 and 3552 yuan 87.7 38.6 70.7 10.5 Between 3552 and 4440 yuan 89.5 43.7 72.5 12.8 Between 4440 and 5328 yuan 90.3 46.8 77.3 16.8 Between 5328 and 6216 yuan 90.8 48.6 80.3 19.6 Greater than 6216 yuan 92.1 55.0 86.5 29.2 All households 79.7 30.4 81.8 23.2 Sources and notes: World Bank estimates from national samples of NBS' 2003 Rural and Urban Household Surveys. Savings are defined as the difference between net income and living expenditures. The savings rate is calculated as a percentage of income. ...and why the number of Because of the high savings rates amongst low-income households in China, many of the consumption poor in near poor in terms of income are consumption poor (Figure 2.29). The result is that the China is roughly double number of consumption poor is almost double the number of income poor. the number of income poor Figure 2.29 : Many households are not income-poor but are consumption-poor (% of households that are consume less than the poverty line within various income ranges ) World Bank taht poverty line 100% an th ay 80% holdse essl 60% e m 40% hous 20% of sun lar-a-dlod a % co 0% 4 8 2 6 0 4 8 2 6 0 4 8 2 6 0 4 8 2 44 88 133 177 222 266 310 355 399 444 488 532 577 621 666 710 754 799 Annual per-capita income range (yuan) Source and notes: World Bank estimates from national samples of NBS' 2003 Rural and Urban Household Surveys. The World Bank poverty line is equivalent to 888 Yuan per-capita annually in 2003 China prices. c. Eliminating the remaining poverty has become harder as the poverty rate has declined China's success at poverty reduction has bred its own challenges. As the poverty rate has declined, it has become harder to eliminate the remaining poverty. ...because the remaining poor are harder to reach Through the 1980s and up to the mid-1990s, broad-based growth in rural incomes led to sharply falling poverty rates in the densely populated rural areas of the central and coastal regions. Poverty rates did fall in the more sparsely populated West and in more mountainous areas but by less. The result was a pronounced increase in the share of the poor living in the Western region of the country. There remained many poor people in the coastal and central regions but they constituted a smaller fraction of the poor nationwide. W h y t h e t a s k i s n o t c o m p l e t e 2 9 Since the mid-1990s, poverty rates in the Western region have fallen as well except in especially remote mountainous and minority areas, where because of geography and agro-climatic conditions, the challenges of ensuring sustainable livelihoods are particularly severe. A consequence of this success, however, is that the share of the poor who are dispersed through the non-mountainous areas outside of the Western region has risen (Figure 2.30). Figure 2.30 : The Western region's share of China's rural poor has fallen to about half in the last decade 80 Western region's share of rural 60 poor (%) 40 Official rural 20 poverty rate(%) 0 1988 1991 1996 2001 2003 Source and notes: The official poverty rates are from NBS (2005a). The share of the Western region in rural poor is based on World Bank (2001) for all years before 2001, and on World Bank estimates from NBS Rural Household Surveys for 2001 and 2003 using the official poverty line. ...because the remaining The result is that the remaining poor in China today are roughly equally divided between poor are harder to reach remote upland communities, primarily in the western region, and households dispersed across villages throughout China's non-mountainous areas, including in the more prosperous coastal and central regions. And both groups, for distinct reasons, are hard to reach. ...and because the responsiveness of poverty reduction to economic growth has fallen As noted earlier, over the two-and-a-half decades since the beginning of economic reforms in the late-1970s, China's performance in alleviating absolute poverty has been highly impressive. For instance, by 2004 the proportion in absolute poverty (those with per capita income below the basic needs line) fell to less than 12% of its level in 1981, while the number of poor declined to about 15% of the level in 1981. A major factor underlying this stellar performance has been China's strong economic growth over this period. During 1981-2004, the trend rate of growth in China's real per capita GDP has been about 7.3% per annum.14 At this rate, per capita income doubles every ten years, and over the period 1981-2004 China's per capita GDP increased by a factor of about 5.8. As Figure 2.31 shows, the impressive fall in poverty rates largely mirrors the impressive growth in average incomes. Growth has been pro-poor in China. _________________________________________________________ 14Real GDP here is defined as GDP at current prices deflated by the national Consumer Price Index (CPI), and uses the revised GDP series recently released by the government. W h y t h e t a s k i s n o t c o m p l e t e 3 0 Figure 2.31 : The poverty-growth scissors: 1981-2005 ) %( 60 3500 ytrevop 1978 50 Per capita GDP 3000 at 40 2500 na of Headcount index of x 2000 (yu 30 poverty ces) de 1500 in pri 20 GDP 1000 tai ountcdaeH 10 500 0 0 capreP 1981 1983 1985 1987 1989 1991 1993 1995 1997 1999 2001 2003 2005 Source and notes: Estimates of the headcount indices refer to income poverty at the national level. The poverty series for 1981-2001 is based on estimates Ravallion and Chen (2007) and is in terms of the basic needs poverty line estimated there. The estimates for 2001-2005 are based on calculations from NBS Rural and Urban Household Surveys using World Bank poverty line, and are spliced with the 1981-2001 series. Per capita GDP is from NBS China Statistical Yearbook: 2007; it is deflated by the national CPI also from the same source. While poverty has clearly been responsive to growth, concerns have however been expressed if this responsiveness has been declining over time, and whether this is beginning to impose some limits to the prospects for future growth-mediated reduction of poverty. This section reviews recent evidence on poverty and growth from this perspective. A simple measure of the responsiveness of poverty to economic growth is the elasticity of poverty measures with respect to real per capita income. This elasticity for the national headcount index is reported in Table 2.16.15 Table 2.16 : Responsiveness of poverty to growth has declined, 1981-2005 Elasticity of headcount index with respect to real Annual percentage change in: per capita GDP: Per capita GDP Controlling Headcount (constant Gini Un- for index prices) index conditional inequality 6th Plan 1981-1985 -26.1 10.2 -0.8 -2.52 -2.40 7th Plan 1986-1990 7.1 5.8 2.5 0.95 -0.43 8th Plan 1990-1995 -10.2 11.2 2.9 -0.90 -1.36 9th Plan 1996-2000 -3.4 7.0 2.3 -0.48 -1.54 10th Plan 2001-2005 -9.0 8.9 1.4 -1.03 -1.55 1981-2005 -7.1 8.3 2.0 -0.87 -1.79 Sources and notes: Same as for Figure 2.31. The elasticities are estimated from regressions of log poverty measures on log per capita income and interactions with dummy variables for each plan period. The elasticities controlling for inequality also include log Gini indices interacted with plan dummy variables as additional regressors. For the period 1981-2005 as a whole, the proportion of population below the poverty line fell at a trend rate of about 7.1% annually. Over the same period, per capita GDP grew at the rate of 8.3% per year. Thus, for the period as a whole, the (absolute) elasticity of the headcount index with respect to per capita survey income was 0.9, implying that a 10% increase in per capita GDP was associated with a 9% fall in the proportion of population _________________________________________________________ 15The results are similar for the poverty gap and the squared poverty gap measures. W h y t h e t a s k i s n o t c o m p l e t e 3 1 below the poverty line. However, there was also considerable variation in trends across the different Plan periods. What can be learnt from this experience regarding the responsiveness of poverty to growth? Setting aside the somewhat exceptional experience of the 7th Plan period (1986-90), the main thrust of the evidence is that while growth continues to be pro-poor it is significantly less so now than it used to be. For instance, on comparing the early 1980s with the most recent period 2001-05, the absolute elasticity of the headcount index with respect to per capita GDP fell from 2.5 to 1. Thus, poverty is now about 40% as responsive to growth as it used to be in the early 1980s. A principal factor underlying the declining responsiveness of poverty to growth is the rise in inequality over this period. The Gini index of inequality of per capita incomes increased at a trend rate of 2% per annum over the whole period. This rise in inequality has served to dampen the poverty impact of growth. This is evident from the results presented in the last two columns of Table 2.16., which show that once changes in inequality are controlled for, not only is the absolute growth elasticity of poverty about twice as high as the unconditional elasticity for the 25-year period as a whole, but also the declining trend in the absolute growth elasticities of poverty is substantially attenuated. This evidence thus suggests that while growth has been a major driver of poverty reduction, the efficiency of growth in reducing poverty has been declining, and unless the rise in inequalities can be halted or reversed, more growth will now be needed to deliver the same rate of poverty reduction. d. Income inequality has increased significantly China is no longer the low-inequality country it was a quarter century ago Income inequality has There has been much concern and public debate about the rise in income inequality in risen significantly since China. Income inequality in China has indeed increased significantly since the start of the beginning of reforms economic reforms a quarter century ago. Estimates from the national samples of the NBS Rural and Urban Household Surveys indicate that the Gini index of income inequality has risen from 30.9% in 1981 to 45.3% in 2003 (Figure 2.32). Figure 2.32 : The evolution of income inequality in China since the start of reforms (Gini index of income inequality from 1981 to 2005) Adjusted for rural-urban cost of living differences 50 45.3 44.3 45 Without adjustments for 40 spatial cost of 41.1 35 living differences Adjusted for 30 spatial cost of living differences 25 1980 1985 1990 1995 2000 2005 Source and notes: Chen and Ravallion (2007) for all years up to 2001; for later years, World Bank estimates from NBS Rural and Urban Household Surveys. Adjusting for spatial cost Adjusting for spatial ­ inter-provincial and rural-urban ­ cost of living differences dampens of living differences the level of inequality, but does not alter the increasing trend. Adjusting for only rural- dampens the level of urban cost of living differences also confirms the increasing trend in Gini indices of income inequality, but does not inequality between 1990 and 2005. And other measures of inequality, such as the Theil alter the increasing trend. index and the ratio of mean incomes in the richest and poorest deciles of the population, confirm the overall trend reflected by the Gini. A more disaggregated picture of the distribution of income in 2003--the latest year for which household-level data from a large-scale nationally representative sample is W h y t h e t a s k i s n o t c o m p l e t e 3 2 available--is provided in Figure 2.33, which shows the shares of overall income accruing to deciles of the population ordered from the poorest 10% to the richest 10%. The poorest 10% of the population received a 1.8% share of overall income while the share of the richest 10% was 29.9%. Figure 2.33 : The distribution of income in China in 2003 (Share of overall income, by decile, from the poorest 10% to the richest 10%) 35% Ratio of mean income in richest decile to that 29.9% 30% in poorest decile : 16.3 25% Ratio of 90th percentile to 10th percentile : 7.3 20% 17.0% 15% 12.8% 10.1% 10% 6.8% 8.2% 4.5% 5.6% 5% 1.8% 3.3% 0% Poorest 10% Richest 10% Source and notes: World Bank estimates from 2003 NBS Rural and Urban Household Surveys. Incomes adjusted for spatial cost-of-living differences using the spatial price indices calculated by Brandt and Holz (2006). From a global perspective, China, today, is no longer the low-income-inequality country it was at the start of the reforms a quarter century ago. But it is worth bearing in mind that contrary to what is sometimes implied, neither can it be considered a high-income- inequality nation. The measured level of inequality in China is, in fact, comparable to that in many other middle-income economies, including some in the East Asia region such as Thailand and Malaysia and is still lower than that in many Latin American countries (Figure 2.34).16 Figure 2.34 : Income inequality in China from a cross-country perspective (Gini indices of inequality for a number of countries) Brazil-2001 (inc.) 59.0 Mexico-2002 (inc.) 49.0 China-2003 (con.) 47.4 Malaysia-2002 (con.) 46.1 China-2003 (inc.) 45.3 Thailand-2002 (con.) 44.2 U.S.A.-2000 (inc.) 38.0 Vietnam-2002 (con.) 35.0 Indonesia-2000 (con.) 34.0 India-2000 (con.) 33.0 Russia-2002 (con.) 32.0 Korea-1998 (inc.) 32.0 China-1981 (inc.) 31.0 Germany-2000 (inc.) 28.0 Sweden-2000 (inc.) 25.0 0.0 10.0 20.0 30.0 40.0 50.0 60.0 Source and notes: World Bank (2005a) World Development Report, 2006 for countries other than China; Ravallion and Chen (2007) for China pre-2003; World Bank estimates from NBS Rural and Urban Household Surveys for China 2003. Where China stands out is in the magnitude of the increase in inequality and the pace at which it has occurred (Figure 2.35). _________________________________________________________ 16However, China's method of using self-recorded diaries to measure income are likely to lead to lower inequality estimates than traditional (one-time) survey methods employed in most countries, so in comparative perspective the estimates for China are likely to be lower bounds. W h y t h e t a s k i s n o t c o m p l e t e 3 3 Figure 2.35 : China stands out globally in the magnitude and pace with which incomes and income inequality have increased (Per-capita income and Gini index of income inequality trajectories for the last two decades) BRA SAF 60 of CHL NGA 50 MEX ARG PHL MYS HKG 40 CHN THA SGP GBR USA coefficient equalityni VNM IDN 30 KOR RUS Gini IND FRA POL ESP 20 JPN 6.5 7.5 8.5 9.5 10.5 PPP per-capita GDP (constant 2000 PPP$): log scale Source and notes: World Bank (2005a) except estimates for China for 2003. Each line segment depicts the trajectory of per-capita GDP and income inequality over the last two decades for a particular country. The beginning and end-years vary from country to country because of data availability, but in general span the period from the early-to-mid 1980s to the late 1990s and early years of the twenty-first century. The rural-urban income gap has grown... The rural-urban income Media reports in recent years have made frequent mention of the fact that the difference gap has grown in recent between rural and urban incomes in China has reached historically high levels. That is years... certainly true in terms of the ratio of nominal mean urban incomes to mean rural incomes, which in 2007 has reached 3.3. When adjustments are made for inflation and for cost-of- living differences between rural and urban areas, the trend is less strong, but rural-urban inequality still increased significantly since the mid-1990s (Figure 2.36), while there are reasons to believe that the measured disparity could still be an underestimate (see Box 2.8). Moreover, because mean household incomes have grown at such a rapid rate, even in the absence of any long-term trend in the ratio of urban to rural incomes, the absolute gap has widened tremendously. Figure 2.36 : The rural-urban income gap has grown in recent years (Ratio of mean urban incomes to mean rural incomes from 1980 to 2007) 3.3 3.5 3.0 2.7 Nominal 2.5 2.0 Real cost of living adjusted 1.5 1980 1982 1984 1986 1988 1990 1992 1994 1996 1998 2000 2002 2004 2006 Sources and notes: World Bank estimates based on published data from China Statistical Yearbooks (NBS). W h y t h e t a s k i s n o t c o m p l e t e 3 4 Box 2.8 : The rural-urban income gap as measured through household surveys could be underestimated The rural-urban income gaps measured by the NBS urban and rural household surveys are subject to a number of sources of potential bias. First, the urban sample excluded migrants living in urban areas until 2001, and afterwards migrants comprised less than 2% of the urban sample population even though the 2000 census found that 14.6% of the population living in cities were migrants. However, studies that include migrants do not find that including migrants in urban income calculations reduces rural-urban income gaps appreciably (Sicular et al, 2007; Park and Wang 2007). A related problem is that the NBS rural household survey includes as household members individuals who live away from home for more than 6 months but whose economic life remains closely tied to the household (e.g., spouses, unmarried children). Many such individuals actually reside in urban areas which means their inclusion as rural household members could create upward bias in measured rural income per capita. Another potential source of bias is that some categories of income are excluded from NBS's definition of income. According to Sicular and others (2007), including the rental value of housing as income increases the urban-rural income ratio by 6% in 2002. Urban incomes also do not include non-wage benefits such as housing subsidies, health care, pension, and unemployment insurance benefits. One study that attempts to estimate the value of such benefits finds that they increase the urban-rural income ratio by as much as one third (Li and Luo 2006). Finally, if the more rapid increase in the cost of living in urban areas compared to rural areas reflects increases in the quality of goods consumed rather than an increase in the prices of goods of the same quality, the comparison of real incomes in rural and urban areas may underestimate the increase in the rural-urban gap. These factors tend to suggest that the rural-urban income gap as measured through the sueveys could well be understated. ...and inequality within both rural and urban areas has risen Inequality within both The widening rural-urban income gap has received the most attention in the popular press rural and urban areas has in discussions of the overall rise in income inequality. Equally important, however, for the risen overall rise in inequality, has been the increase in income inequality within both rural and urban areas (Figure 2.37). Figure 2.37 : Income inequality within rural and urban areas has increased (Gini index of income inequality within rural and urban areas from 1981 to 2005) e m 40 conifot 35 Rural y lit 30 en ci qua 25 Urban effi ine20 coin 15 Gi 1981 1983 1985 1987 1989 1991 1993 1995 1997 1999 2001 2003 2005 Sources and notes: Chen and Ravallion (2007) for all years except 2003; for 2003, World Bank estimates from NBS Rural and Urban Household Surveys. ...while the contribution of inter-provincial income differences to overall inequality has narrowed ...while the contribution On the other hand, regional and inter-provincial differences that have figured prominently of inter-provincial in past discussions of income inequality in China, appear to have narrowed in the last differences to inequality decade, particularly in urban areas. This can be seen in Figure 2.38, which, for each of has narrowed three years--1987, 1995 and 2003--presents a decomposition of the Theil index of W h y t h e t a s k i s n o t c o m p l e t e 3 5 income inequality into five components: · differences in mean incomes between rural and urban areas nationally · differences in mean incomes between rural areas of different provinces · inequality within rural areas within each province · differences in mean incomes between urban areas of different provinces · inequality within urban areas within each province Figure 2.38 : Inequality within rural and urban areas has become an increasingly important component of overall inequality (Theil decomposition of the levels and changes in income inequality between 1987 and 2003) 40.0 Within urban, y within provinces litau 30.0 Within urban, eq between in provinces meocnifo Within rural, 20.0 within provinces Within rural, xed between provinces in ilehT 10.0 Between rural and urban areas 0.0 1987 1995 2003 Level Change 1987 1995 2003 1987-1995 1995-2003 Overall inequality (Theil index) 16.6 25.6 36.1 9.0 10.5 Components of overall inequality: Share(%) of overall Contribution (%) to inequality change in inequality ...between rural and urban areas 31.9 39.6 40.2 46.6 40.9 ... between rural areas of different provinces 11.9 13.5 7.9 14.1 -3.4 ...within rural areas within provinces 41.6 22.9 24.8 12.4 32.5 ...between urban areas of different provinces 3.0 9.9 5.0 12.6 -8.3 ...within urban areas within provinces 11.6 14.2 22.2 14.3 38.4 Source and notes: World Bank (1997) for 1987 and 1995. World Bank estimates from NBS Rural and Urban Household Surveys for 2003. All estimates based on large nationally representative samples without adjustments for spatial price differences. Between 1995 and 2003 inter-provincial differences in both rural and urban incomes narrowed to the point where their contribution to the overall change in inequality was negative. And as a share of inequality in 2003, the component attributable to such differences had fallen to 12.9%. On the other hand, rural-urban income differences and income inequality within rural and urban areas within provinces continued to grow in importance, both in terms of their share of overall inequality in 2003, and their contribution to the rise in inequality over time. Adjustments for spatial price differences further magnify the importance of growing inequality within rural and urban areas of provinces, though the rural-urban gap remains important. W h y t h e t a s k i s n o t c o m p l e t e 3 6 e. Disparities in other aspects of human development remain and in some cases have grown Disparities in other China's successes have not only been in raising incomes. China's progress in human aspects of human development has been equally, if not more, impressive. In terms of various human development mirror the development indicators, China compares favorably to levels achieved in middle-income disparities in income... countries. But as in the case of income growth and poverty reduction, the progress has been uneven. And like income, disparities in many non-income aspects of human development have grown in recent years. For instance, using provincial-level aggregates, for rural and urban areas from published Census reports, Zhang and Kanbur (2005) document rising inequality in illiteracy and infant-mortality rates between 1981 and 2000. Because of the marketization of public services, incomes now matter more for access to health and education than they used to ...because incomes That disparities in human development mirror disparities in income is largely due to the matter more now than fact that incomes matter more now than they used to in determining access to health and they used to for access toeducation. This is evident both at the household-level and across provinces. For instance, health and education at the provincial level, the differences in per-capita GDP and income are associated, in recent years, with much greater differences in infant survival rates and high school enrollment rates than they used to in 1990 (see Figure 7.88 in chapter 7). In 1990, a 10% higher provincial per-capita mean income was, on average associated with a 0.9 percentage point higher high school enrollment rate; in 2003, a similar income difference translated into a 2.5 percentage point difference in high school enrollment rates. Income differences seem to matter even more for disparities in infant mortality rates. ...because of the Incomes matter more now than they used to because the change in the institutional marketization of public arrangements for the provision of health and education has led to a marketization of service delivery service delivery. The dismantling of the communes in rural areas and the move to the Household Responsibility System was arguably the single most important reason for the rapid decline in poverty China experienced in the first half of the 1980s. However, a byproduct of this institutional transformation has been the increasing marketization of public service delivery in rural areas. In place of the communes, local governments were charged with administering and financing these services. In poorer areas, they were unable to fully do so because of a lack of fiscal resources. Schools and health facilities have therefore had to increasingly rely on charging user fees to rural households in order to cover their costs. ...and the burden of health and education expenditures has increased for households ...and that has increased A direct consequence of this is that the burden of health and education expenditures has the burden of health and increased for rural households. For instance, it is estimated that the share of educational education expenditures expenditure in the household budget increased from 1.0 to 8.3 percent between 1988 and for the poor 2003, while the budget share of health expenditures increased from 1.6 to 5.1percent over the same period (Figure 2.39). By one estimate, in the 10 years up to 2004, out-of-pocket medical expenditures of households have increased 6-fold, and by forty times since 1980. And more than 800 million Chinese remain uncovered by health insurance (World Bank, 2006). Similarly, it is estimated that between 1997 and 2000, about 50% of the operational fees for basic education was paid by the farmers. In the survey of 3037 villages undertaken in connection with this report, primary school fees per year for a fifth grade student averaged 260 Yuan in 2004 while fees per student per year for middle school averaged 442 Yuan, or about 40% and 70%, respectively of the official poverty line. W h y t h e t a s k i s n o t c o m p l e t e 3 7 Figure 2.39 : The burden of health and education expenditures has increased for rural households (Mean budget shares of education and health expenditures, by per-capita income decile) Share(%) of education expenditures Share (%) of health expenditures in household budget in household budget 10.0 10.0 8.0 2003 8.0 6.0 6.0 2003 4.0 4.0 2.0 1988 2.0 1988 0.0 0.0 l l 10%ts 2nd 3rd 4th 5th 6th 7th 8th 9th Al 10% 10%ts 2nd 3rd 4th 5th 6th 7th 8th 9th Al 10% chest chest Poore Per-capita income decile Ri Poore Per-capita income decile Ri Sources: The estimates for 1988 are from Gustafsson and Li (2004). For 2003, the estimates are World Bank estimates from the national sample of the NBS 2003 Rural Household Survey. Deciles are defined in terms of per-capita incomes adjusted for spatial price differences. f. Restructuring of the urban labor market and the dismantling of the iron rice bowl has led to new challenges Unemployment has risen in urban areas while labor force participation has decreased Rising unemployment in Beginning in the mid-1990s, China undertook aggressive restructuring of loss-making urban areas state-owned enterprises, leading to the layoffs of tens of millions of urban workers. Between 1994, when it was at its peak, and 2006, employment in state-owned enterprises and urban collectives fell by 73 million, from 145 million workers to 72 million. For the first time since the founding of the PRC, China's urban areas experienced high rates of unemployment, causing substantial dislocation and hardship for many urban families. The officially reported unemployment rate, calculated as the number of administratively registered unemployed workers divided by the total number of employed and registered unemployed workers stayed low, increasing from 3.1% in 1996 to 4.3% in 2003, falling slightly to 4.1% in 2006. However, other survey-based estimates suggest much higher unemployment rates (Figure 2.40). Figure 2.40 : Estimates of urban unemployment rates and labor force participation rate (%) 12.0 CULS-based 75 ) ) UER: urban (% (%)RE 10.0 residents rate (U Labor force 70 no 8.0 ratetne participation rate atipic rti myolp 6.0 pa Labor Force 65 me Survey-based force 4.0 UER Official Un registered UER Labor 2.0 60 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 Sources and notes: 1.China Labor Statistical Yearbook (2005); 2.Cai, Park, and Zhao (2006); 3.Giles, Park, and Zhang (2005); 4. World Bank calculations based on 2003 NBS Urban Household Survey. W h y t h e t a s k i s n o t c o m p l e t e 3 8 Estimates based on national labor force survey aggregate statistics suggest that the unemployment rate has been in the 5.6% to 7.6% range since 1998. Census-based extrapolations form China Urban Labor Survey data in 5 large cities using an internationally standard definition of unemployment finds the unemployment rate of local residents (excluding migrants) increased form 6.8% in 1996 to 11.1% in 2002 then declined to 6.7% in 2005. Analysis of the NBS urban survey data (which includes very few migrants) produced an unemployment rate of local residents of 8.8% in 2003. ...and declining labor In addition to higher unemployment, China's recent period of economic restructuring also force participation was characterized by a steadily declining labor force participation rate (Figure 2.40). The main decline occurs after 1999, with the labor force participation rate falling by 10 percentage points by 2003. Because the decline was associated with restructuring, there should be concern that many of those leaving the labor force were discouraged workers. Growing informalization of the urban labor market has raised concerns about the welfare of urban workers Growing informalization A corollary of the downsizing of the state sector in urban areas has been the of the urban labor market diversification of the urban labor market, with non-state entities accounting for a growing has raised concerns share of urban employment since the mid-1990s. The fastest growing segment has been about the welfare of the category of "other" unregistered workers, their share in urban employment increasing urban workers from under 10% in 1995 to 35% in 2006 (see Figure 6.72 in chapter 6). These workers can be described as "missing" workers, because their number reflects the difference between the total urban employment estimated from labor force surveys, and the number of workers reported through administrative channels by all registered entities. In principle, therefore, these missing workers include unreported urban workers, unregistered informal employment, including undocumented work by migrants in urban areas. The large increase in the number of such workers thus suggests that informal urban employment may have increased significantly in recent years. The informalization (and diversification away from the state sector) of the Chinese labor market, while it has undoubtedly helped create jobs and facilitated the allocation of labor according to market principles, has also raised concerns because informal employment is usually not contract-based and by its nature is "hidden". That makes its harder to enforce protective regulations for workers that ensure safe working environments and fair treatment of workers. It also makes it difficult to develop and adequately finance sustainable social insurance systems that cover most of the population. g. A large "floating" population of rural migrants has emerged in urban areas The emergence of a China has witnessed internal migration from rural to urban areas on an unprecedented "floating population" of scale during the last decade a half. This migration has been described as the largest migrant workers from peacetime movement of people in history. Large numbers of rural workers have migrated rural areas to the cities to work, pulled by the jobs in China's fast-growing manufacturing and service industries, and pushed by the growing pool of surplus labor resulting from reforms in the agricultural sector. Arguably, migration has played a foundational role in China's growth over the last two decades. In urban areas, the abundant supply of migrant workers has been an important element of China's global competitive advantage. This movement has occurred despite longstanding policy impediments to labor mobility, notably, the "hukou" or household registration system. An individual's "hukou" status is assigned at birth and gives certain guarantees--an allocation of arable land in rural areas and rights to social services provided by local government in urban areas. But, except for college graduates, and except in certain provinces and some smaller towns and cities, switching one's hukou from rural to urban and from one location to another is difficult. The result has been the emergence of a large "floating population" who live and earn their livelihoods--for many months of each year--in locations where they do not have hukou and thus have only limited and highly variable access to basic social services and potentially suffer discrimination in access to economic opportunities. This has raised concerns about the emergence of an urban underclass. W h y t h e t a s k i s n o t c o m p l e t e 3 9 The exact scale of internal migration in China remains a subject of considerable debate. Estimates differ depending upon the source of data and how migrants are defined in terms of geographic scope, the duration of stay outside the source area and whether the movement was accompanied by a change in hukou status. Rough estimates of the size of the floating population--those who live for more than six months outside the township where they have their hukou--indicate that it constituted nearly 150 million people in recent years, nearly ten times the number in the late 1980s (see chapter 5 for further discussion). Of these, nearly two-thirds were migrant workers from rural areas. Counting migrants as those migrating for at least one month during the year, surveys conducted by the Ministry of Agriculture (MOA) and the Ministry of Labor and Social Security (MOLSS) suggest that the number of migrants increased more than 2.5 times between 1997 and 2004 (Figure 2.41). Of these, in 2004 about two-fifths migrated to another province.17 Figure 2.41 : Recent trends in the number of rural migrant workers 120 Total migration Inter-provincial migration 98.3 102.6 rkers 100 wo 84.0 78.5 antrgi ) 80 61.3 m 49.4 52.0 illions 60 43.0 ral 38.9 36.8 39.0 40.3 rufo m( 40 28.2 21.2 14.8 18.6 erb 20 m Nu 0 1997 1998 1999 2000 2001 2002 2003 2004 Sources and notes: Zhang Hongyu, Research Report on Rural Labor Migration and Employment, in Proceeding of International Conference on China's Transitional Labor Market, 2001; Liu Jianjing, Rural Labor Employment and Transition, in China Employment Report 2003-2004, China Labor and Social Security Press, 2004; MOA Production Policy and Rules Department, Analysis of Rural Labor Migration Employment Situation 2003. Recent estimates from the National Bureau of Statistics put the number of rural migrants (including migrant families) at 137 million in 2007, an increase by over 60% since 2001 (see section 5e for further discussion). The nature of migration is changing and integrating migrant workers and their families into urban areas poses new challenges But the nature of But the nature of migration is changing. Migrants increasingly are resettling permanently migration is changing... in urban areas, often with their families, and more and more migrants are sending their children to school in urban areas despite the difficulties of educational access. According to the CULS in 2005, nearly half of the migrant population have lived in their current city of residence for more than 5 years, and 20% have lived there for more than 10 years. _________________________________________________________ 17Other sources put the share of inter-provincial migration even higher; see further discussion in section 5e. W h y t h e t a s k i s n o t c o m p l e t e 4 0 Figure 2.42 : Migrants are living in the cities for long periods of time % of migrants 28.5 30.0 25.0 18.4 18.4 20.0 15.9 17.2 15.0 10.0 5.0 1.7 0.0 <=1 year 1-3 years 3-5 years 5-10 years 10-20 years >20 years Years in city in which currently living Sources and notes: China Urban Labor Survey 2005 of 12 cities. ...and migrants fare less When migrants do settle in the cities, they fare considerably worse than local residents in well in non-income nearly all non-income measures of welfare, including the quality and costs of housing, measures of welfare access to education, and access to social insurance and social assistance programs. For example, according to the 2005 CULS, compared to local residents migrants live in housing with much less space (10.7 sq. m. versus 18.4 sq. m.), and less access to drinking water (78.1% versus 98.8%), toilets (45.5% versus 88.2%) and less heating (31.8% versus 61.7%). Migration may also be a cause of rising social inequities in urban areas. Where migrant workers do bring their families to the cities, they are often denied access to government- funded health care facilities and priced out of local government schools. These migrant workers Migrants remain a vulnerable segment of the urban labor market. Analysis undertaken for constitute a particularly this report reveals that, even controlling for occupation and skills, migrant workers earn vulnerable segment of less than registered urban workers and they are more likely to experience wage arrears the urban labor market... and work without labor contracts. Few migrant workers participate in the main social insurance programs in urban China. For instance, participation rates are 8.3% for pensions, 6.8% for medical insurance, far lower than for registered urban workers.18 These developments suggest that in the future migration may become more permanent rather than selective, creating greater and more diverse needs for government policy. _________________________________________________________ 18Further evidence in relation to migrants is discussed in chapters 5 and 7. W h y t h e t a s k i s n o t c o m p l e t e 4 1 3.A poverty reduction agenda consistent with the vision of a xiaokang society a. The challenges that remain as well as those that have emerged suggest a case for reviewing and broadening China's poverty reduction agenda China's official poverty reduction program has been development-oriented and targeted to poor rural areas. The official programs have focused on enhancing the income-generating capacity of those living in poor areas by improving rural infrastructure and agricultural practices, developing agribusiness, and promoting labor migration. In addition, government social assistance programs aim to meet the basic food and clothing needs of the extreme poor in rural areas. The developments and challenges described above suggest that the task of poverty reduction needs to be viewed in broader terms. Partly as a result of China's economic success, and partly as a result of wide-ranging economic reforms that have profoundly transformed the structure of the economy, the nature of the (poverty) problem to be tackled has changed in important ways. The changing nature of the problem suggests that the task can not be limited just to the established focus on poor-area development, but that there is a case for broadening the agenda not just in terms of the spatial coverage of poverty reduction efforts, but also in terms of the basic objectives of such efforts, and the range of instruments and approaches adopted: · First, the notion of income and consumption poverty itself may need to be broadened beyond the austere threshold of meeting "survival" needs of food and clothing to one that meets a more generous set of basic needs and capabilities for the country's population. · Second, policy and structural changes as well as the important contribution of risk to current poverty point to a large social protection agenda in both rural and urban areas. · Third, the experience with marketization of public services that eroded the near- universal access to basic health and education in the pre-reform period indicates the need for renewed focus on re-attaining this goal as part of the overall poverty reduction agenda. · Fourth, the challenge of making the process of growth more pro-poor suggests the need to focus on ways of increasing opportunities for the poor to participate more fully in the growth process not just by increasing agricultural incomes but also through off- farm employment and realizing the potential of migration for poverty reduction. · Fifth, greater dispersion of the poor suggests there may be a case for rebalancing the focus of poverty reduction efforts from poor areas to poor people. · Sixth, the rapid increase in inequalities between rural and urban areas as well as within rural and urban areas suggests that the task of fostering equity can not be delinked from the overall poverty reduction agenda. A poverty reduction Implicit in this call for a broadening of the poverty reduction agenda is the recognition that agenda consistent with China has had tremendous success in the last quarter century in eliminating the most the vision of a xiaokang extreme forms of chronic poverty. During this period, China has also become much society wealthier and more prosperous. Both developments suggest that China is now at the stage where it can take on the challenge of a broader poverty reduction agenda consistent with the vision of a xiaokang society that is better-off in an all-round way. b. Recent policy initiatives suggest that a broader poverty reduction agenda is indeed evolving The changes and trends described above have not gone unnoticed in government circles. On the contrary, a number of recent policy initiatives indicate an increasing commitment of the government to a broader poverty reduction, social protection, and human development A b r o a d e r p o v e r t y r e d u c t i o n a g e n d a 4 2 agenda. Key policy responses to the emerging challenges have been evident in recent years through such initiatives as: · The launch, in 2000, of the Western Region development strategy, aimed at 'opening up the West', through construction of infrastructure, support for human capital formation, environmental protection, improvements in regional investment climate and local natural-resource-based industrial development. · Restructuring of poverty alleviation investments. Area-based poverty investment programs were reoriented in 2001 with a shift in focus from the 592 designated national poor counties to 148,000 poor villages. This was partly in recognition of the dispersed nature of poverty, and the fact that many of the poor did not live in the designated poor counties and the poor counties also had many non-poor. · Development of the urban social security system. While there is no official urban poverty line, the government has been rapidly developing an urban social security system following the economic restructuring of the public sector since the mid-1990s and the shattering of the "iron rice bowl". The new and evolving urban social security system has had three main elements: (i) assistance to laid-off workers (xiagang) from state-owned enterprises, which has now been integrated with unemployment insurance, (ii) a means-tested minimum income support (Di Bao) program, and (iii) social insurance programs including pensions, medical, unemployment, work injury and maternity insurance. The pace of efforts has quickened substantially since 2003: · Training program to support transfer of rural surplus labor. In 2004, the government launched a multi-ministry program ­ called the Sunshine program for Training and Transferring Rural Labor Force ­ which provides short-term training to farmers leading to their transfer to non-agricultural employment often in urban areas. The program aims to train and transfer up to 40 million farmers by 2010. · Elimination of agricultural taxes. The central government began to remove all local informal fee charges on farmers but raised the formal agricultural tax rate from 2-3% to 7% in 2002 to compensate partially for local revenue shortfalls. Local governments were subsequently asked to cut this tax rate by 1-2% per year, and by the end of 2006 all agricultural taxes and fees were fully phased out. · Supporting farm incomes. During the 11th Five Year Plan period, agricultural policies have moved from net taxation to net subsidization of agriculture. Direct subsidies to agriculture, including grain and input subsidies (for fuel and fertilizers), increased 3.5 times between 2004 and 2007. · A nationwide rural social assistance system. Perhaps, one of the most important developments has been the decision, announced by Premier Wen Jiabao at the National Peoples' Congress in March 2007, to set up a nationwide rural social assistance system--the Di Bao or minimum subsistence allowance program-- financed in part by the central budget. A similar scheme--the urban Di Bao--has been in place in urban China since the late 1990s and currently provides income support to about 22 million urban residents. In rural areas, until recently only some provinces and counties had such a program. However, under the new initiative, the rural Di Bao program has been rolled out on a nationwide basis, and by the end of 2007, the number of rural Di Bao beneficiaries had grown to nearly 35 million. · A rural health insurance scheme. Another important initiative has been the scaling up of the National Cooperative Medical Scheme (NCMS) in rural areas. The NCMS is a new voluntary rural health insurance scheme that is operated by counties and subsidized by the local and central governments. Introduced on a pilot basis in 2003, the NCMS has expanded rapidly to cover about 86% of all counties by the end of 2007. · Urban residents' basic medical insurance. Complementary to the existing basic medical insurance scheme for urban formal sector workers, a new urban residents' basic medical insurance scheme was introduced in 2007 (following pilots in 2005) to expand urban health insurance to urban unemployed residents, students and children. · A medical assistance scheme in rural and urban areas. A related initiative has been the rapid roll out of the Medical Assistance scheme in rural and urban areas to A b r o a d e r p o v e r t y r e d u c t i o n a g e n d a 4 3 provide financial assistance to the poor and vulnerable groups in rural and urban areas with their health insurance contributions and copayments. Piloted in rural areas in 2003 and in urban areas in 2005, the scheme had expanded swiftly to cover all rural counties and 86% of cities by the end of 2007. · Compulsory education finance reform. A major new initiative in education has been the introduction of free compulsory education. Introduced in March 2006, the compulsory education finance reform aims to promote universal primary and junior secondary education by repealing tuition/miscellaneous fees and textbook fees for all students, and by providing boarding subsidies to poor students (the so-called "two exemptions and one subsidy" policy). The reform is financed through large-scale intergovernmental transfers. Many of these new initiatives were reiterated and reinforced in President Hu Jintao's report to the 17th Party Congress in October 2007 (see Box 3.9). Box 3.9 : "Accelerating Social Development with the Focus on Improving People's Livelihood" ­ Extract of Hu Jintao's report at the 17th Party Congress Social development is closely related to the people's well-being. More importance must therefore be attached to social development on the basis of economic growth to ensure and improve people's livelihood, carry out social restructuring, expand public services, improve social management and promote social equity and justice. We must do our best to ensure that all our people enjoy their rights to education, employment, medical and old-age care, and housing, so as to build a harmonious society. 1. Give priority to education and turn China into a country rich in human resources. Education is the cornerstone of national rejuvenation, and equal access to education provides an important underpinning for social equity. ...We will optimize the educational structure, promote balanced development of compulsory education, move faster toward universal access to senior secondary education, vigorously develop vocational education, and improve the quality of higher education. ...We will continue to run education on a nonprofit basis, increase government spending on education, regulate the collection of education-related fees, support the development of education in poverty-stricken and ethnic autonomous areas, improve the system of financial aid to students, and ensure that children from poor families and of rural migrant workers in cities enjoy equal access to compulsory education as other children. 2. Implement a development strategy that promotes job creation and encourage entrepreneurship to create more employment opportunities. ...We will improve vocational education and training for the labor force and intensify pre-employment training for surplus labor transferred from rural areas. We will establish a unified, standardized labor market and a mechanism that ensures equal employment opportunities for both urban and rural residents. We will improve employment assistance to the needy and make it a priority to help zero-employment families to have job opportunities. ...We will regulate and coordinate labor relations, improve and implement government policies concerning rural migrant workers in cities, and protect the rights and interests of every worker in accordance with the law. 3. Deepen reform of the income distribution system and increase the income of urban and rural residents. ...A proper balance will be struck between efficiency and equity in both primary distribution and redistribution, with particular emphasis on equity in redistribution. We will gradually increase the share of personal income in the distribution of national income, and raise that of work remuneration in primary distribution. Vigorous efforts will be made to raise the income of low- income groups, gradually increase poverty-alleviation aid and the minimum wage, and set up a mechanism of regular pay increases for enterprise employees and a mechanism for guaranteeing payment of their salaries. ...We will increase transfer payments, intensify the regulation of incomes through taxation, break business monopolies, create equal opportunities, and overhaul income distribution practices with a view to gradually reversing the growing income disparity. 4. Accelerate the establishment of a social security system covering both urban and rural residents and guarantee their basic living conditions. ...We will step up the building of a sound social security system that is based on social insurance, assistance and welfare, with basic old-age pension, basic medical care and subsistence allowances as its backbone, and supplemented by charity and commercial insurance. We will promote reform of basic old-age insurance systems in enterprises, Party and government organs and public institutions and explore ways to set up an old- age insurance system in rural areas. We will promote the development of basic medical insurance systems for urban workers and residents and a new type of cooperative medical care system in rural areas. We will improve the system of subsistence allowances for both urban and rural residents and gradually increase such allowances. We will work continuously to improve the A b r o a d e r p o v e r t y r e d u c t i o n a g e n d a 4 4 unemployment insurance, workers' compensation and maternity insurance. We will upgrade management of social security funds to higher-level authorities and work out unified methods for transferring social security accounts nationwide. We will provide more social security funds through various means and strengthen their oversight and management to maintain and increase their value. We will improve social assistance. We will do a good job providing special assistance to families of martyrs and servicemen and helping demobilized military personnel return to civilian life. We will promote programs for the disabled in a humanitarian spirit. We will develop more programs for the elderly. We will intensify our efforts in disaster prevention and reduction. We will improve the low-rent housing system and speed up resolution of the housing difficulties of low-income families in urban areas. 5. Establish a basic medical and health care system and improve the health of the whole nation. ...We will separate government administration from medical institutions, management from operation, medical care from pharmaceuticals, and for-profit from nonprofit operations. We will increase government responsibilities and spending, improve the national health policy, and encourage greater participation of nongovernmental sectors so as to develop systems of public health services, medical services, medical security and medicine supply to provide both urban and rural residents with safe, effective, convenient and affordable medical and health services. 6. Improve social management and safeguard social stability and unity. Social stability is the common aspiration of the people and an important prerequisite for reform and development. We need to improve the structure of social management comprising Party committee leadership, government responsibility, nongovernmental support and public participation, and improve the system of social management at the primary level. Source: President Hu Jintao's Report to the Seventeenth National Congress of the Communist Party of China on October 15, 2007. Clearly then, substantial policy development and implementation is already evident along the lines of what could be considered a broader poverty reduction agenda. This report hopes to contribute to this ongoing process by undertaking analyses that hope to establish the basic facts regarding the current poverty situation, examine the above policy initiatives and the main challenges in some depth, and thereby elicit the key priorities for the implementation of the evolving poverty reduction agenda. The results of this attempt are presented in Parts II and III of the report. A b r o a d e r p o v e r t y r e d u c t i o n a g e n d a 4 5 PART II REVIEWING CHINA'S POVERTY REDUCTION AGENDA 4 6 4.Establishing the basic facts about poverty and the poor in China If the poverty problem in China is not yet solved as suggested above, what is the nature of the problem to be solved? Conventional wisdom regarding poverty in China is characterized by several competing or complementary claims about the nature of income or consumption poverty in the country: · that it is exclusively a rural phenomenon, · that it is predominantly a problem of the western region, · that it is primarily a problem of remote, mountainous or minority areas · that it tends to be highly concentrated in clusters of poor villages; that it is more of a village-level rather than household-level phenomenon, · that it tends to be more of a problem for girls, women and the elderly, · that it mainly involves those who are unable to work. This chapter examines this conventional wisdom based on a detailed analysis of micro data from NBS' Rural and Urban Household Surveys. As the analysis discussed below hopes to show, while there are elements of truth in several of these claims, poverty in the country is differentiated in character and will require an equally differentiated policy response. a. The poor in China are predominantly from rural areas Virtually all of the China's poor are in or from rural areas The estimates of poverty for the year 2003 presented in Table 4.17 (based on national samples of the Rural and Urban Household Surveys conducted by NBS using the World Bank poverty line) suggest that absolute poverty in China is almost exclusively rural. According to these estimates: · Virtually all (99%) of the poor are in rural areas. This is true for both income and consumption poverty. · The incidence of poverty (i.e. the proportion of the population who are poor) within rural areas is much higher (at 9.5 or 17.9% for two poverty measures) than the national average (of 6.9 or 13.1%). At the same time, urban poverty incidence is negligibly low (0.2-0.3%). Table 4.17 : Rural-urban distribution of poverty in China, 2003 World Bank poverty line Income Consumption Share of % who are % who are Share of population poor Share of poor poor poor Rural 72.5 9.5 99.2 17.9 99.4 Urban 27.5 0.2 0.8 0.3 0.6 Total 100.0 6.9 100.0 13.1 100.0 Source: World Bank estimates from national samples of NBS' 2003 Rural and Urban Household Surveys using the poverty lines of 888 (1124) Yuan per person per year at 2003 rural (urban) prices. The very high share of rural poverty is not just limited to the year 2003 nor are they limited to the headcount measure of poverty. As Table 4.18 shows, similarly high rural shares characterize poverty in China right through the period 1981-2001, and high rural shares are equally notable for measures of depth and severity of poverty. T h e b a s i c f a c t s a b o u t p o v e r t y a n d t h e p o o r 4 7 Table 4.18 : Share of rural poverty in China, 1981-2001 Average share of rural poverty (%) Year Headcount rate Poverty gap Squared poverty gap 1981-1985 98.5 99.1 99.1 1986-1990 95.9 96.0 94.6 1991-1995 98.3 98.3 97.3 1996-2001 97.3 96.5 93.1 Source and notes: Based on estimates of poverty for the basic needs poverty line (using per capita income as the measure of welfare) reported in Ravallion and Chen (2007). H, PG and SPG refer to the headcount, the poverty gap and the squared poverty gap indices of poverty respectively. However, these inferences about the almost exclusively rural nature of poverty in China need to be qualified in two important respects. The first has to do with the underlying definition of rural and urban populations which is closely tied to the issue of how the migrant or the so-called "floating" population is treated in these estimates. The second has to do with the cost of living differentials between urban and rural areas implicit in these estimates. The numbers of non-hukou migrants from rural areas have grown dramatically ... NBS' Rural Household A lesser-known fact about the NBS Rural Household Surveys is that the definition of the Surveys include permanent residents of rural household includes individual migrant workers.19 On the individual migrant other hand, the Urban Household Surveys are predominantly confined to urban workers who are mostly permanent residents. For instance, in the 2003 UHS, the share of urban population with excluded from the Urban Household Surveys non-local or agricultural hukou (registration) was only 2.8%. Thus, at best the UHS appears to cover only a small fraction of migrants, possibly some of those who migrated with their families. But family migrants are a relatively small proportion of all the migrants most of whom are individual workers whose families continue to live in rural areas (see below). The size of the migrant For 2003 (and later surveys) it is possible to identify rural migrant workers in urban areas population is large and from the RHS data. 20 Using questions on work outside the village, the number of increasing individual rural migrant workers is estimated to be 125.8 million when defined as those who have worked outside for any length of time and to be 98.9 million when defined as those who worked outside for six months or more. In comparison the MOA/MOLSS survey reported in Figure 2.41 finds that in the same year, there were 98.3 million rural workers who migrated for one month or more. In addition to the individual migrants, recent NBS research on rural labor (NBS, 2005b) indicates that the population of whole-family migrants was 24.3 million in 2003 (24.7 million in 2004). Thus, the total stock of migrants for 2003 could be estimated at about 150 million (migrants for any length of time + family migrants) or 123 million (migrants for 6 months or longer + family migrants). One simple way to estimate the size of the rural-urban migrant population is as the difference between the number of those with rural hukou minus the official estimates of the rural resident population, based on the annual 0.1% National Population Sample Survey (NPSS). The difference was 169 million for 2003 and 185.5 million for 2004. The trend in this estimate of migrant population is shown in Figure 4.43. _________________________________________________________ 19In particular, permanent residents "refer to all members in a household who have a permanent residence registered in that household and usually stay at the household for over 6 months and are economically linked with the household. It also covers those who are engaged in their work away from home for over 6 months and bring their wages back to the household and are economically linked with the household." (NBS, 2005 China Yearbook: Rural Household Survey). 20The RHS however only includes individual migrant workers. Those who migrated as a whole family for longer than a year are not included in the survey. T h e b a s i c f a c t s a b o u t p o v e r t y a n d t h e p o o r 4 8 Figure 4.43 : The growing difference between rural registered and rural resident population in China, 1978-2005 (Population in millions) 1000 Rural population 950 based on hukou 900 850 800 Rural population 750 based on resisdence 700 1978 1985 1990 1992 1994 1996 1998 2000 2002 2004 Source: China Statistical Year Book 2006 (NBS). However, this is likely to overestimate the number of migrants, because some urban areas contain populations with rural hukou, so that some of the migrants measured in this way are persons with agricultural hukou who live in urban areas without migrating. Also, estimated changes in migration using this method are influenced by re-categorization of rural regions as urban.21 Another direct estimate of the floating population is the number of individuals whose usual place of residence is different to the place where they are registered, based on data from the 2000 Census or the subsequent NPSS: according to the former, the size of the floating population was 144 million while the 2005 NPSS put it at 147 million for 2005 (NBS, 2006). Unlike the previous estimates of rural migration, this estimate includes rural-rural, urban-rural, and urban-urban migrants. Excluding migrant workers from the rural population, 90% of poverty is still rural An important implication of the particular treatment of migrant workers in NBS' Rural Household Surveys (RHS) is that the RHS-based rural poverty estimates, such as those presented above in Table 4.17 and Table 4.18, also include migrant workers who are actually not part of the resident rural population; the average duration of migrant work outside the village is upwards of 8 months (NBS, 2005b). Being able to identify individual migrant workers in the Rural Household Survey also opens up the possibility of presenting estimates of poverty separating out such migrant workers from rural residents (Table 4.19). However, since data on individual income or consumption are not available, in constructing these estimates, migrant workers are assigned the per capita income or consumption of the households to which they belong. This procedure assumes perfect pooling of income or consumption within the household. While migrant workers send relatively large amounts of remittances back home, this may nonetheless understate the income and consumption of migrant workers and overstate the income and consumption of other resident household members. In fact, if migrants pooled little of their income with other household members, the NBS rural survey data could understate the magnitude of rural poverty, even if migrants were not counted among the rural poor. On the other hand, these calculations do not make an allowance for the higher cost of living in urban areas and to that extent overstate the income/consumption of migrant workers, somewhat counteracting the bias from assuming perfect pooling. The net effect is difficult to predict and hence the following results should be interpreted as mainly indicative in nature. Later, in section 7e, some evidence is presented suggesting that including family migrants in _________________________________________________________ 21The hukou-based estimate of rural population refers to persons residing and registered in rural areas. This estimate is calculated on the basis of the town classification standards of 1964 which includes rural population in what are now established towns (NBS, 2006). T h e b a s i c f a c t s a b o u t p o v e r t y a n d t h e p o o r 4 9 urban poverty calculations may not significantly increase measured urban relative poverty rates. Table 4.19 : Poverty estimates for rural residents, migrants and urban residents in China, 2003 Income poverty Consumption poverty Mean % Share Mean % Share Share in PCY who No. of of PCE who No. of of Population population (Yuan/ are poor poor (Yuan/ are poor poor (million) (%) year) poor (million) (%) year) poor (million) (%) Migrants defined as those working outside the village for any length of time Rural residents 811.6 62.8 2786 9.9 80.6 90.0 2000 18.3 148.2 87.6 Migrants 125.8 9.7 2850 6.5 8.2 9.2 1985 15.8 19.8 11.7 Urban residents 354.9 27.5 6488 0.2 0.7 0.8 4982 0.3 1.1 0.6 Total 1292.3 100.0 3809 6.9 89.5 100.0 2818 13.1 169.1 100.0 Migrants defined as those working outside the village for 6 months or more Rural residents 838.5 64.9 2775 10.0 83.5 93.3 1993 18.3 153.8 90.9 Migrants 98.9 7.7 2964 5.4 5.3 5.9 2047 14.4 14.3 8.4 Urban residents 354.9 27.5 6488 0.2 0.7 0.8 4982 0.3 1.1 0.6 Total 1292.3 100.0 3809 6.9 89.5 100.0 2818 13.1 169.1 100.0 Source and notes: World Bank estimates from national samples of NBS' 2003 Rural and Urban Household Surveys. PCY is per capita income at 2003 all-China rural prices; PCE is per capita living expenditure also at 2003 all-China rural prices. Income and consumption poverty is estimated using the World Bank poverty line of 888 (1124) Yuan per person per year at 2003 rural (urban) prices. If the definition of migrants were restricted to those having worked outside the village for more than 6 months, the implied share of urban poverty including migrant workers would be between 7 to 9% in 2003. Using the World Bank poverty line, the income poverty rate of migrants is 5.4%, which is much higher than that of urban residents (0.2%) but lower than that of rural residents (10%). If one were to define migrant workers as those working outside the village for any length of time, migrants would account for 9% (income) or 12% (consumption) of the poor. In this case, the share of urban poverty would be between 10 to 12% in contrast to the less than 1% share if migrants were counted as part of the rural population.22 Thus, to summarize, while poverty remains predominantly rural, if one assumes full pooling of incomes and counts migrant workers as part of the urban sector, where they reside for the bulk of the year, the share of urban poor (in the total number of poor) increases ten times to about 10%. However, if incomes are not fully pooled, it is unclear how counting migrants as part of the urban sector affects the relative importance of urban versus rural poverty; it is even possible that doing so could reduce the share of the urban poor. Only with a 50% or higher urban-rural cost of living differential is there any noticeable poverty amongst urban residents Accounting for urban-rural cost of living differentials (COL) also has an important bearing on the estimates of poverty and its sectoral distribution. The poverty estimates presented above utilize the spatial price indices developed by Brandt and Holz (2006) who price a common bundle of goods constructed for the year 1990 across rural and urban areas of provinces over the period 1990-2001. Making adjustments to 2003 price indices using provincial urban and rural price indices, it is calculated that the average urban-rural cost of living differential for 2003 was 26.5%; it is this differential that underlies the poverty _________________________________________________________ 22Counting the number of poor migrants in urban areas does not capture many important questions about how migration affects overall poverty including impacts on rural poverty. This issue is further discussed in section 5d. Nor does this discussion cover other important issues of how migration is circumscribed by the residence permit (hukou) system and other policy issues related to employment and social protection of migrant population. These issues are further discussed elsewhere in the report (sections 5d and 7e as well as sections 6d and 6e). T h e b a s i c f a c t s a b o u t p o v e r t y a n d t h e p o o r 5 0 calculations presented above. However, the true cost of living differentials across urban and rural areas remains a subject of considerable debate. For instance, an alternative estimate of this differential can be based on the rural and urban poverty lines for 2002 estimated by Ravallion and Chen (2007). Their rural and urban poverty lines of 850 and 1200 Yuan per person per year imply a COL differential of 41.2% for 2002, or 41.3% for 2003 when updated with the urban and rural CPIs. Even this could be considered an underestimate as is suggested by the relatively high urban poverty lines estimated by Meng, Gregory and Wang (2005). The 2003 equivalent of their average urban poverty line, for instance, is 1789 Yuan per person per year compared to 1211 Yuan calculated by Ravallion and Chen when both estimates are anchored to the same caloric threshold and use comparable methodologies.23 A higher urban-rural cost of living differential will necessarily imply higher levels of urban poverty and hence higher shares of the urban sector in total poverty. To assess the broad magnitude these effects, Table 4.20 presents poverty estimates assuming a 50% differential in urban-rural cost of living.24 As the table shows, the number of poor amongst urban residents and migrant workers increases (relative to the estimates in Table 4.19) as they now have lower levels of real income and consumption, and the share of urban poverty (including migrant workers) rises to about 17-19%. Table 4.20 : Poverty estimates for rural residents, migrants and urban residents in China, 2003, assuming a higher (50%) urban-rural cost of living differential World Bank poverty line Income Consumption No. of poor Share of No. of poor Share of (millions) poor (millions) poor Rural residents 62.8 79.2 83.2 145.8 80.9 Migrants 9.7 13.7 14.3 29.7 16.5 Urban residents 27.5 2.4 2.5 4.6 2.6 Total 100.0 95.2 100.0 180.1 100.0 Sources and notes: World Bank estimates from national samples of NBS' 2003 Rural and Urban Household Surveys. Migrants defined as those working outside their home village and township for any length of time. To sum up, poverty in China appears to be entirely a rural phenomenon if no account is taken of the fact that migrant workers are included amongst rural households in the national surveys. If one assumes perfect income pooling between migrants and their families, taking account of migrant workers (and giving priority to their residence rather than registration status) reduces the share of rural poverty to around 90%. If a further allowance is made for potentially higher differentials in the cost of living between urban and rural areas, the share of rural poverty declines to around 81-83% which could be treated as a lower bound on this share. A further caveat on the estimates of urban poverty is worth noting. While the Rural Household Surveys include individual migrant workers, migrant workers who moved to cities with families are not covered. And relatively few of them are covered by the Urban Household Surveys whose sample frame is largely limited to the local urban residents.25 Since many of these migrant families may be poorer than urban residents, this could lead to some underestimation of urban poverty. _________________________________________________________ 23One significant difference between the two relates to the choice of the reference group for calculating the food poverty line. For Ravallion and Chen, this is the group between the 15th and the 25th percentile nationally, while for Meng et al it is the bottom 20% of each province. Since the bottom 20% of richer provinces could be considerably richer than the bottom 20% of poor provinces, Meng et al's methodology could lead to relatively higher food (and hence non-food) poverty lines for richer provinces. 24These estimates are based on elasticities of poverty measures with respect to the poverty line for urban residents and migrant workers. For migrant workers, the estimates assume that the burden of a higher cost of living is entirely borne by them rather than rural resident members. 25For instance, the share of urban population with non-local or agricultural hukou in the 2003 Urban Household Survey is 2.8% or about 10 million people which is only about 40% of the estimated number of full-family migrants according to independent estimates in Sheng and Peng (2005). T h e b a s i c f a c t s a b o u t p o v e r t y a n d t h e p o o r 5 1 The overall evidence, therefore, suggests that while poverty in China may not be entirely rural, it is clearly predominantly so. b. Geography and ethnicity matter but are not the only determinants of poverty The common folklore on poverty in China often characterizes it in particular geographic and ethnic terms. For instance, it is believed that poverty in the country is predominantly a problem of the Western provinces, or of remote and mountainous areas, or one that mainly afflicts the minority communities. While there has been some scattered information in support of such characterizations, available household survey data have typically not been analyzed in a manner to provide more definite evidence on this issue. Using nationally-representative survey data from the 2003 Rural and Urban Household Surveys conducted by the National Bureau of Statistics, this section presents key findings on the geographic and ethnic profile of poverty in China. (The definitions of geographic and ethnic categories used in the analysis are presented in Box 4.10) Box 4.10 : Definitions of regional and topographical/ethnicity categories used in the analysis Five main regions are distinguished in the analysis, as below: Coastal: Beijing, Tianjin, Hebei, Shanghai, Jiangsu, Zhejiang, Fujian, Shandong, Guangdong, Hainan Northeast: Liaoning, Jilin, Heilongjiang Central: Shanxi, Inner Mongolia, Anhui, Jiangxi, Henan, Hubei, Hunan Southwest: Guangxi, Chongqing, Sichuan, Guizhou, Yunnan, Tibet Northwest: Shaanxi, Gansu, Qinghai, Ningxia, Xinjiang Mountainous/minority categories are identified at the village level. The Rural Household Survey instrument distinguishes three different categories for topographical information in the village information form: (i) plain, (ii) hilly, and (iii) mountainous. Plain refers to an extensive level area of land that is basically free of mountains and hills and is relatively low in elevation. Plain includes level plain, flood plain, swamp, lake area and meadow of grazing land. Hilly area refers to an area of land with rolling small mountains, including semi-mountainous area and low hills. Mountainous area refers to an area of land where there are a number of mountains, including mountains covered with grazing grassland. Minority areas refer to areas where non-Han ethnic populations are concentrated, following the standards of the State Ethnic Affairs Commission (SEAC), PRC. According to SEAC standards, minority villages are designated by local county governments. No national specification explicitly defines minority villages. Based on local regulations, the designation of minority villages, in principal, should satisfy the criterion that minority (non-Han) population is at least 30% of the total village population. If such criterion is satisfied, the village committee can apply for a minority village status at the local county government, and the county government makes the decision upon such application. The incidence and severity of poverty is highest in the western region but nearly half of the poor are elsewhere Table 4.21 shows the incidence of poverty and the distribution of the poor across the five main regions of China for 2003. Poverty estimates are presented using per capita income and per capita consumption as two alternative measures of household welfare. T h e b a s i c f a c t s a b o u t p o v e r t y a n d t h e p o o r 5 2 Table 4.21 : Is poverty in China a Western province phenomenon? Evidence for 2003 World Bank poverty line in terms of: Share of Income Consumption Region population % who are Share of % who are Share of (%) poor poor (%) poor poor (%) Coastal 34.6 2.3 11.3 6.3 16.8 Northeast 8.4 5.6 6.8 6.6 4.2 Central 28.3 5.8 23.7 13.0 28.0 Southwest 19.6 11.8 33.3 20.7 31.0 Northwest 9.1 18.9 24.9 28.6 20.0 Total 100.0 6.9 100.0 13.1 100.0 Sources and notes: World Bank estimates from national samples of NBS 2003 Rural and Urban Household Surveys. Poverty incidence in The evidence does suggest that poverty rates ­ both by the income and consumption western provinces is criteria ­ decline as we move from the west to the east. In terms of consumption, the about 3-4 times higher highest incidence of poverty is observed in the northwestern provinces followed by than in coastal and southwestern provinces ­ about 29% and 21% respectively. Central provinces have the northeastern provinces next highest poverty incidence of about 13%, while poverty rates are relatively low ­ about 7% and 6% respectively ­ in the Northeast and Coastal provinces. A similar profile also emerges for income poverty; the only notable difference is that the central and the northeastern regions have a similar incidence of poverty (about 6% each). Income poverty rates are the lowest in coastal areas. About half of China's The western provinces thus account for about half of China's poor which is well above poor are in western their 29% share in the country's population. However, these provinces are even poorer in provinces terms of the depth and severity of poverty (as measured by the poverty gap and the squared poverty gap indices). Thus, their share in the aggregate depth and severity of poverty are even higher, about 60 and 66 percent respectively (Figure 4.44). In other words, if one were to assess poverty by a measure that is sensitive to how poor the poor are, then about two-thirds of China's poverty is to be found in the north- and south- western provinces. Figure 4.44 : Regional shares in the incidence, depth and severity of consumption poverty in 2003 (using World Bank poverty line) (%) Southwest Northwest Central Northeast Coastal Incidence of poverty 31.0 20.0 28.0 16.8 4.2 Depth of poverty 34.7 24.9 23.0 13.8 3.7 Severity of poverty 37.3 28.4 19.3 11.7 3.4 Source: World Bank estimates from national samples of NBS 2003 Rural and Urban Household Surveys. But about 45% of the It is nonetheless instructive to note that about 31% of the aggregate severity of poverty (by poor are still in central the consumption criterion) is still located in central and coastal provinces which together and coastal provinces also account for about 45% of the poor. By the count in 2003, that is still almost 76 million people. T h e b a s i c f a c t s a b o u t p o v e r t y a n d t h e p o o r 5 3 Poverty is most severe in mountainous and minority areas, but more than half the poor are in non-mountainous non-minority areas About 19% of China's population (or about 26% of the rural population) lives in mountainous areas, and about 8% of the population comprises of non-Han minority ethnic groups.26 The minority population itself is concentrated in the mountainous areas; as compared with about a fifth of the overall population, nearly two-thirds of them live in such locations. While it is widely believed that the ethnic minorities and others living in these locations are relatively disadvantages, do they account for most for the observed poverty in China? Table 4.22 sheds some light on this question by presenting poverty estimates where the rural population is disaggregated by mountainous/non-mountainous and minority/non- minority categories while the urban population is disaggregated by type of city. Three types of cities are distinguished in order of size: the four provincial cities of Beijng, Tianjin, Shanghai and Chongqing; the provincial capitals; and prefectural cities. As noted before, the urban population mainly refers to urban residents, while the rural population includes migrant workers. Table 4.22 : Poverty in China by topography-minority area categories and by type of city, 2003 World Bank poverty line Share of Income Consumption population % who are Share of % who are Share of Region (%) poor poor (%) poor poor (%) Minority 7.7 24.1 26.9 36.6 21.6 Non-minority 64.8 7.7 72.3 15.7 77.7 Mountainous 18.5 18.8 50.2 27.9 39.4 Non-mountainous 54.0 6.3 49.0 14.5 59.9 Non-mountain, Non-minority 51.3 5.7 42.1 13.7 53.7 Non-mountain, Minority 2.7 17.6 6.9 30.0 6.2 Mountain, Non-minority 13.5 15.5 30.2 23.3 24.1 Mountain, Minority 5.0 27.6 20.0 40.2 15.4 4 Provincial cities 3.0 0.0 0.0 0.0 0.0 Provincial capitals 5.7 0.1 0.1 0.1 0.0 Prefectural cities 18.8 0.3 0.7 0.4 0.6 Total 100.0 6.9 100.0 13.1 100.0 Sources and notes: World Bank estimates from national samples of NBS 2003 Rural and Urban Household Surveys. Nearly all urban poverty With regards to the urban population, the Table confirms what was noted earlier, viz., the is a small (prefectural) rates of poverty amongst urban residents are very low, and they account for only about city phenomenon 1% of the total number of poor. (As discussed above, this mostly excludes migrant workers who are instead included in the rural population.) Table 4.22 also indicates that whatever poverty exists amongst urban residents, it is mostly at the level of prefectural cities rather than provincial capitals or provincial cities. Poverty incidence With respect to the rural population, the estimates indicate that minority communities are amongst minority significantly poorer than Han majority communities; for the dollar-a-day line their communities is 2-3 times consumption poverty rate is more than twice as high and their income poverty rate is three higher than for the Han times as high as that of Han communities. Similarly, consumption and income poverty majority ... and it is also 2-3 times higher in incidence in mountainous areas is also about twice and three times as high, respectively, as that in non-mountainous areas. It is also the case that within the minority category, _________________________________________________________ 26There are 55 registered ethnic minorities in China. The 8% population share of these 55 minority groups is based on the 2000 census (NBS, 2005, China Statistical Yearbook). This is very close to the estimate in Table 4.22 which is based on the NBS Rural Household Survey (RHS) for 2003. Strictly speaking, however, the RHS estimate refers to population in minority villages (rather than minority population) where a minority village is defined as one with at least one-third of its population comprising of minority groups. T h e b a s i c f a c t s a b o u t p o v e r t y a n d t h e p o o r 5 4 mountainous than in non- those in mountainous locations have still higher poverty rates, and conversely, amongst mountainous areas those in mountainous locations the minority groups are poorer still. Those with the double disadvantage of minority status and mountainous location thus have the highest poverty rate of 40% (28%) in the case of consumption (income) poverty. But non-minority and However, living in a mountainous terrain or a minority area does not fully determine the non-mountainous poor. The non-minority and non-mountainous population still accounts for about 42% of population still accounts the poor by the income criterion and 54% by the consumption criterion. Their shares in for about half the poor measures of the depth and severity of poverty are lower; nonetheless, the non-minority and non-mountainous population still accounts for about two-fifths of the aggregate severity of poverty in the country (Figure 4.45). Figure 4.45 : Topography-minority category shares in the incidence, depth and severity of consumption poverty in 2003 (using World Bank poverty line) (%) Mountain, Non-minority Mountain, Minority Non-mountain, Minority Non-mountain, Non-minority Incidence of poverty 24.1 15.4 6.2 53.7 Depth of poverty 27.0 18.0 7.4 47.1 Severity of poverty 29.4 19.6 8.4 42.2 Source: World Bank estimates from national samples of NBS 2003 Rural and Urban Household Surveys. Thus, in summary, while poverty is certainly concentrated in western provinces, and in mountainous and minority areas, it is not confined to them. While these provinces and areas deserve greater attention in poverty alleviation efforts, an exclusive focus on them would be tantamount to addressing only about half the problem. c. The poor are dispersed throughout China's villages Consider an area with ten equal-sized villages where twenty percent of the area's total population is poor (a proportion not too different to the consumption poverty incidence for rural China for 2003 using the dollar-a-day line). Consider one scenario where all the poor are concentrated in two villages while everyone in the other eight villages is non-poor. Consider another scenario where 20% of the population in each village is poor. Clearly, the two scenarios have very different implications for policies aiming to reduce poverty in the area: while a narrowly-focused village-based line of attack would make sense in the first scenario, an area-wide household-based approach would be better suited to the second. This rather artificial example illustrates an important principle: how dispersed poverty is in any given area can matter a great deal to the appropriate design of poverty reduction policies. Measuring the extent to which poverty is concentrated within villages How dispersed or concentrated is rural poverty across China's villages? To assess this, a simple measure of village-level concentration of poverty is used (see Box 4.11).27 The measure takes the minimum value of zero if poverty is perfectly dispersed across villages _________________________________________________________ 27Village is chosen as the unit of observation for its policy relevance in view of the government main poverty alleviation program which targets designated poor villages (see chapter 5). But the concentration measure itself can be used to assess concentration for any other unit of observation. T h e b a s i c f a c t s a b o u t p o v e r t y a n d t h e p o o r 5 5 in the sense that all villages have the same poverty rate, and it takes the maximum value of one if poverty is perfectly concentrated across villages in the sense that the universe consists of only two types of villages where either everyone or no one is poor. Between these extremes, higher values of the measure indicate higher degree of village-level concentration of poverty. (The measure can also be interpreted as the variance of village headcount indices normalized by the maximum variance possible so that measure lies between zero and one.) Box 4.11 : Village-level concentration of poverty and implications for targeting The degree of village-level poverty concentration is indicative of the relative gains from household- based targeting versus village-level targeting. This is illustrated in Figure 1 which shows for rural China in 2001 the share of poor covered by the bottom x% of the population under two alternative conditions: (i) when villages are ranked by descending poverty incidence (curve OBA), and (ii) when households are ranked by ascending per capita (curve OCA). The former thus shows the coverage rate of the poor under perfect village-level targeting while the latter shows the coverage rate under perfect household-level targeting. Figure 4.46 : Village-level concentration of poverty and the potential gains from village vs. household targeting for rural China, 2001 1 C Area H B A .8 Area V oropfo .6 arehs.4 m.uC .2 0 O 0 .2 .4 .6 .8 1 Cum. Share of population Sources and notes: World Bank estimates based on data from NBS' Rural Household Survey for 2001. Note that the dotted line indicates the incidence of rural poverty for 2001 using the World Bank poverty line. The area between the village-targeting curve and the diagonal (marked as area V) measures the potential gains from village-level targeting. Similarly, the area between the household-targeting curve and the diagonal shows the maximum gains from household-level targeting. The area marked H indicates the additional gains in moving from village to household-level targeting. If poverty were more dispersed (concentrated) across villages, the village-targeting curve would be closer (farther away) from the diagonal, and gains from village-targeting (area V) would shrink while the additional gains from household targeting (area H) would expand. The concentration of poverty within villages is highest in the southwestern region and least in the coastal region Table 4.23 reports the values of these concentration indices for the five main regions in China using data from 2003 Rural Household Survey. T h e b a s i c f a c t s a b o u t p o v e r t y a n d t h e p o o r 5 6 Table 4.23 : Village-level concentration of rural poverty in China, 2003 Consumption poverty Income poverty Concentration Poverty Concentration Poverty Region index incidence (%) index incidence (%) Coastal 0.33 8.8 0.22 3.1 Northeast 0.25 12.5 0.25 10.5 Central 0.31 16.8 0.27 7.6 Southwest 0.46 25.7 0.44 14.5 Northwest 0.36 42.6 0.37 28.1 All rural 0.35 17.9 0.29 9.5 Sources and notes: World Bank estimates based on data from NBS' Rural Household Survey for 2003. Note that the poverty estimates are for the World Bank poverty line. The overall village concentration index for consumption poverty (using the dollar-a-day poverty line) in rural China is 0.35. The concentration of income poverty is lower (with an index of 0.29) which reflects the greater variability of income both within and across villages. A value of 0.35 is suggestive of a moderate level of concentration. For instance, in the 10-village example above this is similar to the case where no one is poor in six villages and half the population is poor in the remaining four; this case yields a concentration index of 0.38. As seen in Table 4.23, village-level concentration of both income and consumption poverty is the highest in the southwestern region. It is the lowest in coastal provinces for income poverty and the northeastern provinces for consumption poverty. The concentration of poverty tends to decline as the incidence of poverty itself declines As poverty declines, it In general, Table 4.23 suggests that poorer regions tend to have greater concentration of tends to be more poverty. This relationship is further confirmed in Figure 4.47 which plots village poverty dispersed concentration indices against the poverty rates (for the dollar-a-day line) across 20 region- topography-ethnicity categories. 28 As seen in the graph, there is a clear positive relationship between the level of poverty and its concentration. The concentration of poverty tends to decline as the incidence of poverty itself declines. Figure 4.47 : Poverty concentration declines at lower levels of poverty across region- topography-ethnicity categories in rural China, 2003 1.0 0.9 0.8 index 0.7 0.6 iont 0.5 0.4 0.3 Concentra0.2 0.1 0.0 0 10 20 30 40 50 60 Poverty incidence (%) Sources and notes: World Bank estimates based on data from NBS' Rural Household Survey for 2003. Each point refers to one of the 20 segments of the rural population formed as combinations of five regions, two topographical categories (mountainous and non-mountainous areas) and two ethnicity categories (minority and non-minority areas). Poverty rates are calculated in terms of per capita consumption using the World Bank poverty line. _________________________________________________________ 28The 20 categories are combinations of five regions, two topographical categories (mountainous and non-mountainous areas) and two ethnicity categories (minority and non-minority areas). T h e b a s i c f a c t s a b o u t p o v e r t y a n d t h e p o o r 5 7 This has an important implication. As China makes further progress in reducing poverty, poverty could be expected to become more dispersed thus eroding some of the potential benefits from area-based targeting. This suggests that household-based targeting approaches are likely to become increasingly relevant in future poverty reduction initiatives. d. Most of China's poor are able to work In 2003, according to official poverty statistics, China's absolute poor numbered 29.0 million persons. This poverty count was based on analysis of the 2003 National Bureau of Statistics (NBS) Rural Household Survey data and China's official poverty line of 637 Yuan per capita per year (NBS, 2005b). In April of 2003, the Ministry of Civil Affairs (MOCA), which is in charge of China's major social assistance programs, conducted an independent nationwide investigation to evaluate the social assistance needs of China's rural population. They estimated that 25.2 million individuals had not met their basic needs for food and clothing and were unable to work, and thus required public assistance. The closeness of these two numbers has sometimes been seen as evidence for a commonly-expressed view that most of the remaining poor in China live in households without productive labor and thus cannot benefit from the income-generation projects which have been the focus of China's official poverty reduction programs. Rather, the government should concentrate its efforts on accurately providing targeted consumption subsidies for the remaining poor. Do China's poor lack the capacity to work? Given the policy significance of this question, it is important to take a critical look at the evidence from survey data. Nearly three-quarters of China's rural poor live in households where no-one lacks work capacity and 97% live in households with two or more productive members Using the Rural Household Survey data for 2003, the same source as that used for the official poverty count, Table 4.24 presents tabulations of poverty incidence and the number of poor disaggregated by the number of laborers in the household. Laborers are defined as those who are above the age of 16 years, are not full-time students and do not lack the capacity to work. The number of laborers includes the elderly who are able to work as well as migrants who may be away for work for most of the year but whose economic life is closely tied to the household; it excludes all adults who are incapable of work. Poverty incidence is calculated using income net of any public transfers (including payments from government programs and social organizations) to avoid underestimation of the poverty rate for households without productive labor insofar as at least some of the public transfers could themselves be linked to the household's lack of work capacity. Table 4.24 : Poverty by the number of laborers in the household, based on household per capita income before receiving public transfers, in rural China, 2003 Official poverty line World Bank poverty line Number of Percent of Number of Number of laborers in rural % who poor %of all % who poor % of all household population are poor (millions) poor are poor (millions) poor 0 0.2 7.4 0.2 0.4 16.6 0.3 0.4 1 1.6 5.8 0.9 2.2 11.6 1.7 1.9 2 41.3 3.9 15.2 38.6 8.9 34.4 38.7 3+ 56.9 4.3 23.1 58.8 9.8 52.3 58.9 All 100.0 4.2 39.3 100.0 9.5 88.8 100.0 Sources and notes: World Bank estimates from national samples of NBS 2003 Rural and Urban Household Surveys. Per capita income is used as the measure of welfare. Households without any Table 4.24 shows that while persons living in households with no labor have a higher productive labor account poverty incidence (7.4% as against 4.2% for the rural population as a whole using the for 0.4% of the poor official poverty line and 16.6% as against 9.5% using the World Bank poverty line), they account for only 0.4% of the rural poor. In fact, nearly 98% of the poor are in households with two or more laborers. No-labor households altogether account for only 0.2% of the T h e b a s i c f a c t s a b o u t p o v e r t y a n d t h e p o o r 5 8 rural population. Even if all of them were poor, they would account for only 1.94 million poor. A household that has no productive labor at all is however a limiting case of the lack of work capacity. More generally, those who are unable to work are part of households with productive workers. Table 4.25 thus presents two further sets of poverty estimates: (i) amongst households that have at least one adult member who is unable to work relative to households that do not have any work disability, (ii) amongst individuals who lack work capacity relative to the rest of the population. Table 4.25 : Poverty and lack of work capacity within households in rural China, 2003 Official poverty line World Bank poverty line Number Number % of rural % who of poor % of all % who of poor % of all population are poor (million) poor are poor (million) poor Does the households have at least one adult unable ot work? Yes 19.9 5.8 10.8 27.7 13.1 24.5 27.7 No 80.1 3.8 28.2 72.3 8.6 64.2 72.3 Is the individual an adult who lacks work capacity? Yes 5.3 5.5 2.7 6.9 12.4 6.1 6.9 No 94.7 4.1 36..4 93.1 9.3 82.7 93.1 All rural 100.00 4.2 39.1 100.0 9.5 88.8 100.0 Sources and notes: World Bank estimates from national samples of NBS 2003 Rural Household Survey. Per capita income is used as the measure of welfare. Nearly three-fourths of A much larger fraction of the rural population (about 20%) lives in households with at least the rural poor live in one adult who is unable to work. This is not surprising as households often have an households where all elderly person who has lost the ability to work. The poverty rate amongst such adults are able to work households is higher than for households where all adults can work (5.8% versus 3.8% using the official poverty line, and 13.1% versus 8.6% using t the World Bank poverty line). However, they account for about 28% of the rural poor. Nearly three-quarters of the rural poor thus do not have even a single adult in their household who is unable to work. Only 7% of poor Focusing on individuals within households who lack work capacity, the survey data individuals lack the indicate that they constitute only 5.25% of the rural population, or about 49 million capacity to work persons.29 They are somewhat poorer than the rest of the rural population, with a headcount index of 5.5% (12.4%) as against 4.1% (9.3%) for the rest of the population using the official (World Bank) poverty line. Only 2.7 million of those who lack work capacity are poor using the official poverty line. Using either line, only 6.9% of the rural poor lack the capacity to work. In summary, therefore, the evidence from the NBS survey data clearly shows that nearly all of the rural poor in fact live in households that do have productive labor, and while adults who lack work capacity face a higher risk of poverty, they account for only about 7% of the rural poor. It is possible that the NBS survey undersamples those who are poor and lack work capacity, if for example, such individuals are illiterate or handicapped in a way that makes it impossible for them to keep self-recorded diaries. However, the evidence points to the vast majority of the poor being working poor. While those who lack the capacity to work need support in view of their relatively higher poverty rate, a much bigger part of the overall poverty challenge is the task of improving the productivity of those who are able to work. _________________________________________________________ 29The China Disabled Person's Federation (CDPF), which is responsible for maintaining statistics on disability in China, reports that China has 60 million disabled, of which 21 million have hearing disabilities, 12 million are mentally retarded, 9 million have physical disabilities, 9 million have visual disabilities, 2 million have mental illnesses, and 8 million have multiple disabilities (see www.cdpf.org.cn). Eighty percent of the disabled live in rural areas (Information Office of the State Council, 2001). This estimate of the disabled in rural areas is very similar to the estimate of 49 million rural adults who are unable to work. However, there are important differences between the measures of disability and the lack of work capacity. The estimate of disabled persons also includes children who are disabled. On the other hand, those without work capacity include many who are not disabled (for instance, those who are too old to work). T h e b a s i c f a c t s a b o u t p o v e r t y a n d t h e p o o r 5 9 What the poor lack is human capital If the vast majority of the poor in China are working poor, the issue of improving their productivity becomes centrally important. Education and human capital development have a key role to play here in adding value to the key resource of the poor, viz., their labor power. The importance of human capital development in reducing poverty is widely recognized. China is no exception. The evidence from the NBS survey data in Table 4.26 readily confirms this. The table presents poverty rates by the average years of schooling of labor members in the household. The latter are defined as those who are 16 years or older, are not students, and have the ability to work. Table 4.26 : Income poverty in China by level of education, 2003 Twice the World World Bank line Bank line Average years of Share of Mean income schooling of adult population (Yuan/ person/ % who Share of % who Share of household workers (%) year) are poor poor (%) are poor poor (%) Less than 6 16.1 2330 16.3 38.0 48.8 29.2 Between 6 and 9 51.6 3040 7.2 53.6 31.1 59.6 Between 9 and 12 23.6 4850 2.5 8.4 12.5 11.0 Greater than 12 8.7 8290 0.1 0.1 0.6 0.2 Total 100 3808 6.9 100 26.9 100 Sources and notes: World Bank estimates from national samples of NBS' 2003 Rural and Urban Household Surveys. Dollar-a-day income There is a clear negative relationship between educational attainment of the working poverty is virtually population and the incidence of poverty. For instance, for the World Bank poverty line, eliminated when the the incidence of poverty is 16% for up to 6 years of schooling. It declines to less than half average worker has (about 7%) for three additional years of schooling. Crossing the 9-year threshold is higher secondary education associated with large gains in poverty reduction. The incidence of poverty drops to 2.5% for 9-12 years of schooling, and when the average laborer has more than 12 years of education poverty is virtually eliminated. Such a fall in poverty rates by level of education is not just limited to the World Bank poverty line; the decline is equally steep for a poverty line that is twice as high. Despite high returns to The improvement is also reflected in mean incomes. As the average worker's education education, 90% of the level rises from up to 6 years, to 6-9, to 9-12, and above 12, the index of mean incomes poor still have 9 or fewer rises from 100, to 130, to 208, to 356. Clearly, there are large returns to education. The years of schooling social returns may be even higher which begs the question whether China and China's amongst their adult household members poor in particular are investing enough in education.30 Ninety percent of the poor still have 9 or fewer years of schooling amongst their adult household members, and 38% have six or less. e. More than the elderly, children, and in particular, girls, are likely to be poor How are the poor in China distributed amongst the children, adults and the elderly and amongst men and women? A demographic profile of the poor is a useful device to focus attention on the composition of the poor from an age and gender perspective. Table 4.27 presents the basic data for those who are poor in terms of their per capita consumption using the dollar-a-day line. _________________________________________________________ 30This issue is discussed in greater detail in chapter 7. T h e b a s i c f a c t s a b o u t p o v e r t y a n d t h e p o o r 6 0 Table 4.27 : The incidence of consumption poverty amongst women and the elderly in China, 2003 Share of Number of population Number % who are poor Share of (%) (million) poor (million) poor (%) Children (less than 16 years old) Boys 10.6 136.4 16.3 22.3 13.2 Girls 9.2 119.1 17.3 20.6 12.2 Working age adults (between 16 and 60 years old) Men 36.0 465.7 12.2 56.9 33.6 Women 35.7 460.8 12.0 55.3 32.7 Elderly (more than 60 years old) Men 4.2 54.3 12.8 6.9 4.1 Women 4.3 55.9 13.0 7.2 4.3 Total 100 1292.3 13.1 169.2 100 Sources and notes: World Bank estimates from national samples of the NBS 2003 Rural and Urban Household Surveys. Poverty rates are based on per-capita consumption relative to the World Bank poverty line. The data indicate that of the 169 million persons living below the World Bank poverty line in 2003, about 14 million (about 8%) were elderly above age 60, 112 million (66%) were adults in age group 16-60, and 43 million (about 25%) were children below 16. The data also indicate poverty incidence for different demographic groups. In particular, the poverty incidence for male and female adults and the elderly population is very similar ­ between 12 and 13 percent. The incidence amongst children below 16 is however higher: about 16% for boys and 17% for girls.31 The higher poverty rate for children is not an unusual finding, since in many countries larger households tend to have more children and also be poorer.32 If there are economies of scale in household expenditure, then the lower income or expenditure per capita in larger households may overstate actual differences in welfare. The lower income per capita of larger households with children also may be related to life cycles effects, since parents with children tend to be less productive than older adults. The higher poverty rate of children thus does not necessarily reflect a high prevalence in China of orphans, broken homes, or neglect within families. Nonetheless, the particular challenges faced by poor children and their families merit policy attention, because the effects of being poor when young can have long-lasting consequences. Table 4.27 also reveals some interesting information on gender bias in China. While much has been written on the adverse sex ratio in China (mostly based on census data), it is notable that the 2003 household survey data (on which the analysis of Table 4.27 is based) also reveal an adverse ratio of female to male children below 16; female children account for 9.2% of the national population as against 10.6% for male children, implying a ratio of 873 girls to 1000 boys. The data also reveal a further aspect of gender bias. The numbers in Table 4.27 can be used to calculate the relative risk of poverty for different demographic groups defined as the ratio of poverty incidence for a particular group to the overall national poverty incidence. The ratio indicates whether relative to the general population a particular group faces a higher risk of poverty. Figure 4.48 shows this relative poverty risk for different groups. _________________________________________________________ 31It should be noted that this estimate of child poverty is not based on information on the allocation of consumption within the household. It strictly refers to children living in households whose per capita consumption is below the poverty line. 32The negative relationship between poverty and household size is sensitive to the use of measures of welfare expressed in per capita terms. The per capita normalization implies that there are no economies of household size, that is, the needs of a household increase proportionately with the total number of members in the household. The reason for using this normalization is not that such economies of scale do not exist, but that there is no very satisfactory way to estimate them empirically (Lanjouw and Ravallion, 1995; Deaton, 1997). T h e b a s i c f a c t s a b o u t p o v e r t y a n d t h e p o o r 6 1 Figure 4.48 : The relative risk of poverty amongst the women and elderly in China, 2003 Age Male Female Greater than 60 0.99 0.97 Between 16 and 60 0.92 0.93 Less than 16 1.32 1.25 0.90 0.95 1.00 1.05 1.10 1.15 1.20 1.25 1.30 1.35 Sources and notes: World Bank estimates from national samples of the NBS 2003 Rural and Urban Household Surveys. Note that the poverty rates are based on per-capita consumption relative to the World Bank poverty line. Consistent with the higher poverty incidence amongst children noted above, the relative risk of poverty is above one only for those below age 16. Thus, relative to the general population, children are more likely to be poor or, strictly speaking, living in poor households. However, a key finding highlighted by the figure is that girls below 16 face a significantly higher risk of poverty (1.32) relative to boys below 16 (1.25). Thus, while children of both sexes are over-represented amongst the poor relative to their share in the population, girls are over-represented more than boys. Another way to describe this is that female-to-male sex ratio is in fact higher amongst the poor than non-poor. As shown later in Figure 7.95, while the overall sex ratio for those below 16 is 873, the ratio for the non-poor is 864 while in contrast the ratio for the poor is 924. Thus, poorer households on average have 60 more girls per thousand boys than non-poor households. The evidence is suggestive of underlying discrimination against female children. f. The poor have some characteristics in common, but differ in other ways While all the poor have their poverty status in common, the underlying causes of their individual poverty often vary. Some may be poor due to their deficient endowments, such as their low levels of human capital or limited land to work with, some others may be poor as a result of living in geographically disadvantaged locations such as remote mountainous areas, yet others may be poor because of the burden of family members who are incapable of working. A taxonomy of the rural poor is developed in this section to highlight the different underlying sources of different types of rural poverty in China. Specifically, the poor are classified into a number of groups based on their characteristics. This analysis departs from the standard poverty profile exercise which typically involves a comparison of the characteristics of the poor with that of the non-poor. By classifying the poor into different categories, this section intends to provide a basis for the formulation of various types of policies that may be required to reach them. Constructing a taxonomy of the rural poor in China The taxonomy is developed using cluster analysis. While cluster analysis has been used widely as a data exploration technique, this analytical approach is well-suited to identifying natural clusters of observations, in this case, poor households.33 The households are _________________________________________________________ 33In particular, the analysis is based on non-hierarchical k-means clustering (MacQueen, 1967, Wishart 1982, 2001) which is an iterative procedure that partitions data into k groups or clusters. It begins with k group centers, and subsequently involves assigning observations to the closest cluster centroid and calculating the mean of the observations of each cluster. This process is repeated until all observations remain in the same group from the previous iteration. Further discussion of different methods of cluster analysis can also be found in the STATA reference manual. T h e b a s i c f a c t s a b o u t p o v e r t y a n d t h e p o o r 6 2 classified into different groups or clusters such that households within a cluster are most similar to each other in terms of a set of chosen characteristics. Using data from the Rural Household Survey 2003, ten household characteristics are selected for this analysis. These characteristics which are indicative of various causes or correlates of poverty are described in Table 4.28. The Table also shows how frequently a particular characteristic is observed amongst the rural poor and total rural population respectively in the last two columns.34 Table 4.28 : Household characteristics used in constructing a taxonomy of rural poor in China Mean Mean value Characteristic Description value for for total rural rural poor population High dependency Dependency ratio is defined as the share of non labor members in the 0.53 0.39 ratio household. A share greater than one-third indicates high dependency ratio. Low work capacity A household has at least one adult member without work capacity. 0.28 0.20 Low education of Maximum education acquired by any labor member of a household is junior 0.80 0.71 labor high school level. Limited land A household is considered as having limited land if its land per labor ratio falls 0.21 0.25 in the bottom quartile. Low productivity of Agricultural productivity index of the county where the household lives falls in 0.56 0.25 land the bottom quartile. Minority area A household resides in a minority village. 0.27 0.11 Mountainous area A household resides in a mountainous area. 0.51 0.26 Far from county The distance from the household's village to the nearest county township is 0.49 0.39 town further than 10 kilometers. No migrant worker There is no migrant worker in a household. 0.66 0.57 No TVE worker There is no TVE worker in a household. 0.98 0.92 Sources and notes: World Bank estimates based on NBS Rural Household Survey data for 2003. The poor are defined as those with per-capita income below the World Bank poverty line. It is evident from Table 4.28 that with the exception of limited land per labor, for all the characteristics, the poor population has on average a higher, sometimes much higher, incidence of the specific challenges indicated by these characteristics. For example, while 39% of the total rural population are from households with high (greater than one-third) dependency ratios, amongst the rural poor this proportion is 53%. In general, relative to the rural population as a whole, the rural poor not only have higher dependency ratios, they also have lower work capacity, lower level of education, more limited land per unit of productive labor, are more likely to live in minority and mountainous areas, are more likely to be relatively far away from county towns, and are less likely to have opportunities for migration or TVE employment. While the data in Table 4.28 help establish a poverty profile for China, these characteristics can, as discussed above, be used to construct clusters of poor facing similar conditions. Without prior knowledge of how many clusters should be formed, the poor are grouped into 5, 7 or 10 clusters respectively, and the statistically most satisfactory classification is selected to represent the taxonomy.35 The rural poor can be usefully grouped into seven clusters, each distinguished by a particular disadvantage... Table 4.29 shows this taxonomy for the rural poor in 2003 where the poor are defined as those with per capita incomes below the World Bank poverty line. _________________________________________________________ 34The characteristics are represented as binary variables where value one indicates the presence of a particular characteristic and zero its absence. Hence, the mean values indicate the share of rural poor or rural population with that attribute. 35The choice of partitions is based on the comparison of the value of Calinski & Harabasz pseudo-F values associated with each run of the k-means cluster analysis. T h e b a s i c f a c t s a b o u t p o v e r t y a n d t h e p o o r 6 3 As seen in Table 4.29, seven clusters of rural poor are identified. Different clusters can be described in terms of the specific characteristics that are common to a majority of households within that cluster. For example, all of the rural poor in cluster 3 have limited land to work with, none have TVE workers in the household, over three-quarters have a relatively low level of education and over half have no migrant workers in the family. Table 4.29 also provides information on the size of the seven clusters in terms of the shares of rural poor as well as the number of poor included in each cluster. It is notable that clusters 5, 6, 4 and 1 ­ the four biggest clusters, each accounting for more than 15% of the rural poor ­ together make up more than 75% of the total rural poor, or more than 66 million people. Table 4.29 : A taxonomy of the rural poor in China, 2003 Percentage of households in the cluster with the listed characteristic Clusters (types) High Low Low Far of the poor, by depend- work Low prod- Mount- from No mig- Number shared ency capa- educ- Limited uctivity Minor- ainous county rant No TVE % of all of poor characteristics ratio city ation land of land ity area area town worker worker poor (million) 1 91.0 100.0 84.2 7.7 84.3 38.0 72.9 54.1 67.9 99.9 17.3 15.5 2 14.4 17.7 46.3 0.4 7.0 0.0 17.6 78.7 59.0 100.0 8.6 7.7 3 31.2 9.8 75.3 100.0 37.8 0.0 49.3 45.6 55.9 100.0 13.9 12.4 4 69.0 26.0 88.7 1.6 0.0 0.0 7.5 20.1 71.0 100.0 18.6 16.6 5 40.9 10.8 86.3 21.6 74.3 100.0 72.8 55.6 71.1 99.7 20.6 18.4 6 47.6 0.0 82.0 0.0 100.0 0.0 66.7 55.9 66.0 100.0 19.6 17.5 7 41.8 28.8 77.4 37.7 23.9 1.2 16.8 38.3 66.0 0.0 1.6 1.4 All rural poor 52.5 27.7 80.5 20.5 55.6 27.2 50.6 49.1 66.3 98.4 100.0 89.5 Sources and notes: World Bank estimates based on NBS Rural Household Survey data for 2003. The poor are defined as those with per-capita income below the World Bank poverty line. A number of interesting aspects of the taxonomy of the rural poor are revealed in Table 4.29. On the one hand, some common characteristics are shared amongst most, if not all, clusters, such as low education level of labor and no migrant labor or TVE workers in the household. On the other hand, in addition to these common factors, some clusters demonstrate distinctive attributes. For instance, cluster 1 stands out as being particularly afflicted by adult household members without work capability as well as by the relatively low productivity of land, high dependency ratio and the difficulties associated with living in mountainous terrain. The main distinguishing problem for cluster 2's poor seems to be distance from county town. While cluster 3 suffers from limited land to work with, cluster 4 has relatively high dependency ratio. In contrast, in cluster 5 the majority of the population live in minority and mountainous areas, and they also have less productive land. Cluster 6's main distinguishing feature is the low productivity of land. Cluster 7 ­ the smallest cluster with only 1.6% of the rural poor ­ is the only cluster that is poor despite the fact that the majority of its population lives in households with at least one TVE worker. Their main problem seems to be that the education level of labor is relatively low, and they are unable to avail themselves of significant migrant work opportunities. The regional distribution of the clusters varies, as does the importance of individual clusters within each region Table 4.30 offers a regional perspective on the taxonomy of rural poverty by showing how different types of poverty are distributed across the main regions. T h e b a s i c f a c t s a b o u t p o v e r t y a n d t h e p o o r 6 4 Table 4.30 : Regional shares of different categories of rural poor in China, 2003 (% of each category of poor) Type of Distinguishing North- South- North- poor characteristic Coastal east Central west west All 1 Low work capacity 2.8 6.4 10.6 45.7 34.6 100 2 Far from town 24.0 11.3 48.0 8.1 8.6 100 3 Land scarce 19.8 3.4 30.0 31.2 15.6 100 4 High dependency 23.5 3.2 58.5 5.0 9.7 100 5 Minority area 2.4 3.9 9.7 66.0 18.0 100 6 Low productivity 1.5 14.0 13.0 29.5 41.9 100 7 Other 40.3 8.3 22.7 6.9 21.8 100 All rural poor 11.1 6.8 25.9 33.3 22.9 100 Regional shares (%) of different categories of the poor % of all rural poor Coastal Northeast Central Southwest Northwest 17.3 Low work capacity 8.6 Far from town 13.9 Land scarce 18.6 High dependency 20.6 Minority area 19.6 Low productivity 1.6 Other 100.0 All rural poor 0.0 20.0 40.0 60.0 80.0 100.0 Sources and notes: World Bank estimates from the national samples of the NBS 2003 Rural Household Survey. The poor are defined as those with per-capita income below the World Bank poverty line. Table 4.30 shows a highly varied pattern, suggesting that different categories of the poor are concentrated in different regions. The following points are notable. · More than 70% of the poor in clusters 1, 5 and 6 are concentrated in Southwest and Northwest regions. Besides low education, lack of migration and TVE employment opportunities (which are shared by most of the rural poor), these clusters also have in common the following additional distinctive features of rural poverty in the West: low productivity of land and remoteness in terms of mountainous terrain and distance from county towns. · In particular, cluster 5, which has the largest share of rural poor, is mainly concentrated in the Southwest region (66% share of cluster 5, which translates into a poor population of approximately 12 million), which further highlights the role of minority status as a special feature of poverty in the Southwest. · The type of poverty represented by clusters 2 and 4 is mainly a Coastal and Central region phenomenon. These two regions account for 72% of the poor in cluster 2 and 82% of the poor in cluster 4; the Central region alone accounts for 48% and 59% of the poor in these clusters respectively. This highlights the importance of distance from county town and high dependency burden ­ the distinctive features of these clusters ­ as contributory factors to poverty in the Coastal and Central regions. · Cluster 3 is more evenly distributed across regions, but the Central and Southwest regions have relatively high shares of about 30% each. Limited land being a distinctive feature of this cluster, it points to the additional importance of this factor for the poor in Central and Southwest regions. · Cluster 7, which is the smallest group accounting for 1.6 per cent of the total rural poor, has a large concentration in the Coastal region, and to a more limited extent in the Central and Northwest regions. This group of the poor seems to be less afflicted by some of the problems faced by other groups such as remoteness, limited or poor T h e b a s i c f a c t s a b o u t p o v e r t y a n d t h e p o o r 6 5 quality land, or lack of TVE employment or minority status. Its main problems seem to be low levels of education and lack of work opportunities through migration. Complementary to the above discussion, Table 4.31 presents information on the composition of the poor in each region by type of poverty. While Table 4.30 showed where different types of poor are located, Table 4.31 shows for each location which types of poor are more important. The three main types of poverty in each region are highlighted. It is instructive to note that the pattern of poverty in the Southwest and the Northwest is similar with types 1, 5 and 6 dominating. Similarly, the Coastal and the Central regions also share a similar pattern with types 2, 3 and 4 dominating. The pattern in the Northeast is however different where types 1, 2 and 6 dominate. The distinctive pattern for each region suggests that the underlying causes and correlates of poverty vary from one region to another. Table 4.31 : Types of rural poor in different regions, China 2003--a region-specific taxonomy (% of each category of poor) Type of Distinguishing North- South- North- Share of all poor characteristic Coastal east Central west west rural poor 1 Low work capacity 4.4 16.1 7.1 23.7 26.1 17.3 2 Far from town 18.5 14.3 15.9 2.1 3.2 8.6 3 Land scarce 24.8 6.8 16.1 13.0 9.4 13.9 4 High dependency 39.4 8.8 42.0 2.8 7.9 18.6 5 Minority area 4.5 11.8 7.7 40.8 16.2 20.6 6 Low productivity 2.7 40.2 9.9 17.3 35.7 19.6 7 Other 5.7 1.9 1.4 0.3 1.5 1.6 All 100.0 100.0 100.0 100.0 100.0 100.0 Low work capacity Far from town Land scarce High dependency Minority area Low productivity Other 100% 80% 60% 40% 20% 0% ralt rurall Coastal Cen hwest poor Al Northeast Sout Northwest % of rural 11.1 6.8 25.9 33.3 22.9 100.0 poor Sources and notes: World Bank estimates from the national samples of the NBS 2003 Rural Household Survey. The poor are defined as those with per-capita income below the World Bank poverty line. In summary, the taxonomy of rural poverty offers useful information on the characteristics of various types of poverty as well as the regional distribution associated with them. The different characteristics of the poor across clusters and the regional aspects of such differences indicate the degree of heterogeneity in terms of the sources of poverty. Hence, policy responses may need to be tailored to address the varying nature of impediments faced by the poor in different locations. While some impediments seem to be common to all the poor, viz., low levels of education and limited opportunities for non- agricultural employment through migration or work in TVEs, other impediments are more relevant in specific contexts. There is a role for both household and area-based interventions, and the two may often have to go hand-in-hand. In particular, Western regions may need to focus more on alleviating low productivity of land and problems T h e b a s i c f a c t s a b o u t p o v e r t y a n d t h e p o o r 6 6 associated with limited work capacity and remoteness in terms of mountainous terrain as well as special difficulties faced by minority groups. In the Coastal and Central regions, greater attention may be warranted to address the problems of limited land per unit of labor, high dependency burden of households and remoteness (distance from county towns). In the Northeast, policies will need to address the challenges of low work capacity (problems of disability or old age), remoteness and low productivity of land. g. A profile of the disadvantaged in urban areas As noted above, absolute poverty remains very low amongst urban residents in China. For instance, using the World Bank poverty line of 1124 Yuan per person per year at 2003 urban prices, income poverty incidence is estimated at 0.2% for 2003 while consumption poverty incidence is estimated at 0.3%. The urban poor account for less than 1% of the total number of poor in terms of either income or consumption. Of course, the urban population canvassed by the NBS urban household surveys refers almost entirely to urban residents. It only includes a small fraction of long-term family migrants. But the vast numbers of individual migrant workers are not covered by the urban household surveys that are used for monitoring urban poverty.36 Yet, it would be a rare policy discussion of urban poverty that implicitly or explicitly does not include migrant population within its scope. Whether or not they included in the household surveys typically used for urban poverty monitoring, the scope of urban social policy can hardly be determined by the particular forms in which rural and urban household survey data are gathered. Later in this report (chapter 7), the problems related to the integration of migrant workers and their families in urban areas are discussed at length. However, independent of the issue of the treatment of migrant population in urban poverty calculations, there is also considerable debate about the urban poverty lines themselves. There is no single threshold for identifying the urban "poor" The Chinese government does not set any official poverty lines for the urban population and does not endorse any official estimates of urban poverty. Not surprisingly, there is considerable uncertainty about which urban poverty lines are most appropriate. Table 4.32 shows a range of urban poverty lines that have been proposed in different studies. The table also shows the 2003 equivalents of these poverty lines. It is evident that these lines span a very wide range: from 1211 Yuan per person per year (Ravallion-Chen) to a line that is almost twice as high line of 2310 Yuan (Hussain). _________________________________________________________ 36As discussed earlier, if individual migrants ­ who are currently covered in the rural household surveys ­ were included in the urban population, the share of urban poor increases to 10-12% of all the poor (see section 4a). T h e b a s i c f a c t s a b o u t p o v e r t y a n d t h e p o o r 6 7 Table 4.32 : Urban poverty lines in China (in Yuan per person per year) Urban-rural cost-of-living differential (%): urban line relative to the 2003 World 1995 1997 1998 2000 2002 2003 Bank rural line (888 Yuan) Urban CPI (1995=100) 100 112 111 111 111 112 Poverty lines World Bank line 1124 27 Ravallion-Chen 1200 1211 36 Meng-Gregory-Wang: lower line 1779 1789 102 Khan-Riskin 1604 1790 102 National Bureau of Statistics 1875 1886 112 Average dibao line 1928 117 Twice the World Bank 2247 153 line Meng-Gregory-Wang: upper line 2256 2269 156 Hussain 2310 2312 160 Sources and notes: For World Bank lines, see discussion in chapter 1 (Box 1.1); the other lines are from Ravallion and Chen (2007), Meng, Gregory and Wang (2005), the "extreme" poverty line from Khan and Riskin (2001). The NBS poverty line is reported in ADB (2004), and the average di-bao line is a population-weighted average of provincial Di Bao lines reported in MoCA (2003); prior to taking the average, the provincial Di Bao lines were expressed in national urban prices using spatial (inter- provincial) prices indices from Brandt and Holz (2006). How should one interpret some of the higher urban poverty lines such as those estimated by Meng, Gregory and Wang, NBS, Khan and Riskin, and Hussain? It is arguable that these higher poverty lines should be interpreted in relative rather than absolute terms. The main argument is that if interpreted in absolute terms these lines suggest urban-rural price differentials that seem too high to be plausible. Using the Brandt-Holz estimate of 2003 urban-to-rural cost of living differential of 26.5%, which is based on common bundles of goods defined in 1990, the rural World Bank line of 888 Yuan per person per year in 2003 is equivalent to 1124 Yuan per person per year in urban areas. However, if one uses the World Bank line in rural areas in combination with the other urban poverty lines in Table 4.32, the implied urban-to-rural COL differential is higher. It is only moderately high (36%) for the Ravallion-Chen line. But for the other lines, the implied COL differentials are much higher, from 102% to 160% (see last column of Table 4.32). Such COL differentials are high enough to wipe out most of the difference between real mean incomes in urban and rural areas; for instance; with a 160% differential, urban mean per capita income in 2003 would be only 13% higher than the rural mean income. This seems implausible. Upon further examination, the high urban lines seem to be the product of their underlying high food poverty lines (Table 4.33), though in some cases (Khan-Riskin, Hussain, Meng- Gregory-Wang-upper line), this is further accentuated by a higher allowance for the non- food component of the poverty line (as seen in the relatively low shares of food poverty lines in total poverty lines in Table 4.33). Since all food poverty lines conform to the norm of 2100 calories per person per day, it is possible to compute the price per calorie implied by these lines. These are shown in Table 4.33 along with the average price per calorie in rural areas implied by the rural food poverty line estimated by Ravallion and Chen. T h e b a s i c f a c t s a b o u t p o v e r t y a n d t h e p o o r 6 8 Table 4.33 : Food poverty lines and implied calorie prices in urban and rural areas (Yuan values in 2003 prices) Food poverty Share of food Urban calorie Calorie price line pov. line in price as a ratio (Yuan/1000 (Yuan/ person/ total pov. line of rural calorie cal.) year) (%) price Urban food poverty lines: Ravallion -Chen 873 72.1 1.14 1.39 Meng-Gregory-Wang: lower line 1316 73.5 1.72 2.09 Khan-Riskin 1406 78.6 1.83 2.23 Hussain 1393 60.3 1.82 2.21 Meng-Gregory-Wang: upper line 1316 58.0 1.72 2.09 Rural food poverty line: Ravallion -Chen 630 73.0 0.82 1.00 Sources and notes: See sources and notes for Table 4.32 As seen in Table 4.33, like the urban poverty lines themselves, the calorie prices corresponding to these lines also span a wide range from 1.14 to 1.83 Yuan per 1000 calories. In particular, with the exception of the Ravallion-Chen line, all the other lines imply that calorie prices in urban areas are more than twice as high as those in rural areas. As an indicator of underlying urban-rural food price differentials, this too seems implausibly high. It seems more likely that such differentials reflect in part food quality differences. One could argue that the problem is not that these urban calorie prices are too high, but that the rural calorie prices are too low. On looking further into the estimation of the average price per calorie, there is one significant difference between the Ravallion-Chen methodology and that used in studies by Meng-Gregory-Wang and Hussain. While both estimate the average price per calorie for a reference group of relatively poor households, their choice of the reference group is different. For Ravallion and Chen, this is the group of households between the 15th and the 25th percentile nationally, while for Meng et al and Hussain it is the bottom 20% of each province. However, since the bottom 20% of richer provinces are considerably richer than the bottom 20% of poor provinces, and the rich tend to consume more expensive forms of calories (higher quality, more expensive foods), the Meng et al-Hussain methodology is likely to yield higher calorie prices (and hence higher food and total poverty lines) for richer provinces. In other words, poverty lines for richer provinces would be higher not simply because the cost of living is higher in those areas, but because these poverty lines also correspond to a higher standard of living. In this sense, the higher poverty lines in Table 4.32 contain a relative poverty element. Thus, to summarize, in terms of absolute income or consumption poverty, the urban poverty rates are underestimated mainly because migrant workers are not included as part of the urban sector. But some of the higher estimates of poverty amongst (non- migrant) urban residents are difficult to interpret in absolute terms. There could be some under-estimation of the urban-to-rural cost of living differential, but even a high differential of the order of about 50% does not yield an incidence of consumption poverty for the World Bank line among urban residents that is upwards of 1% (see section 4a). Thus, from the perspective of a consistent national picture of poverty in China, it is difficult to claim a significant level of poverty amongst urban residents. Nearly 20 million people in urban areas consume less than twice the World Bank poverty line and may be considered the "urban disadvantaged" Twice the World Bank That said, there however do remain concerns about the emergence of an urban line can be used as a underclass (Oxford Analytica, 2006). For instance, this is the group the Ministry of Civil (relative poverty) Affairs has sought to reach through its urban Di Bao (minimum income support) program. threshold to identify the In order to focus attention on this disadvantaged group in urban areas and to understand urban disadvantaged the nature of emerging issues and challenges faced by them, urban poverty is better conceptualized in relative rather than absolute terms. In the following discussion T h e b a s i c f a c t s a b o u t p o v e r t y a n d t h e p o o r 6 9 therefore, an urban poverty line equivalent to twice the World Bank line is used which is interpreted as a relative poverty line for urban areas. It is a little higher than the average Di Bao line used by MOCA to determine eligibility and benefit levels for its program, but it is within the range of lines proposed in several studies of urban poverty. An additional relative poverty line of three times the World Bank line is also used for comparative purposes to assess the sensitivity of findings to the choice of the threshold for identifying the urban disadvantaged. There are up to 20 million Table 4.34 presents summary measures of relative poverty in urban China using these urban disadvantaged thresholds; the measures are presented using both per capita income and per capita using the $2/day consumption as measures of welfare. Using twice the World Bank line, the estimates for threshold 2003 indicate that there are between 10 and 20 million relatively poor or disadvantaged persons in urban areas (about 3-6 percent of the urban population) depending on whether one looks at income or consumption. With thrice the World Bank line, the urban disadvantaged population is estimated to be much higher ­ approximately 34 to 72 million (or about 10 to 20 percent of the urban population). Table 4.34 : The urban disadvantaged in China, 2003 Using per capita Using per capita income consumption Poverty line Number of Number of Poverty line (yuan/person/ Headcount poor Headcount poor year) ratio (%) (million) ratio (%) (million) Twice the World Bank line 2247 2.7 9.5 5.7 20.1 Thrice the World Bank line 3371 9.7 34.4 20.2 71.7 Sources and notes: World Bank staff estimates from the national sample of NBS' 2003 Urban Household Survey. The poverty lines in Yuan per person per year are at 2003 all-China urban prices. The term "poor" in this and following table is used synonymously with "disadvantage" in the relative poverty sense as noted in the text. Given these overall numbers, how can this relatively deprived section of the urban population be characterized? This rest of this section presents a brief profile of relative poverty in urban areas. The above numbers and the following profile are based on NBS' Urban Household Survey which, as noted before, does not cover many individual migrant workers and includes only a small faction of long-term migrant families living in urban areas.37 Hence, the profile below is essentially one of the relatively poor urban permanent resident population. Five aspects of the profile are discussed: (i) by region and city-type, (ii) age-sex distribution, (iii) by education level, (iv) by hukou registration and minority status, and (v) by employment status. The urban disadvantaged are relatively evenly spread across regions but heavily concentrated in smaller cities Five main regions and three types of cities are identified to describe where the urban disadvantaged are located. As for the national profile, the five regions are: coastal, Northeast, Central, Southwest and Northwest. The city-types include: provincial-level cities (Beijing, Shanghai, Tianjin and Chongqing), provincial capitals, and prefecture-level and other cities. Table 4.35 presents the main results. As seen in the table, the profiles for the two relative poverty lines are quite similar. Two features of the profile are notable. _________________________________________________________ 37For instance, the share of urban population with non-local or agricultural hukou as estimated from the 2003 Urban Household Survey is only 2.9%, which is at odds with the 120-150 million migrants estimated from other sources. See sections 4a for further discussion of the coverage of migrant workers in NBS' Rural and Urban Household Surveys. T h e b a s i c f a c t s a b o u t p o v e r t y a n d t h e p o o r 7 0 Table 4.35 : Urban disadvantaged in China by region and city-type, 2003 Twice the WB line Thrice the WB line Region/ City-type % Share of % who are Share of % who are Share of population poor poor (%) poor poor (%) Coastal 36.4 4.4 28.2 15.7 28.3 Northeast 14.6 6.3 16.4 22.7 16.4 Central 26.8 6.2 29.2 23.5 31.2 Southwest 14.1 6.6 16.5 20.5 14.3 Northwest 8.1 6.8 9.7 24.3 9.7 Four Provincial cities 10.9 1.0 1.8 6.3 3.4 Provincial capitals 20.6 3.2 11.5 13.3 13.6 Prefectural cities 68.5 7.2 86.7 24.5 83.0 Total 100.0 5.7 100.0 20.2 100.0 Sources and notes: World Bank staff estimates from the national sample of NBS' 2003 Urban Household Survey using per capita consumption as the measure of welfare.. Regions are defined as described in Box 4.10. First, there is no large concentration of the urban disadvantaged population in a particular region. The incidence of urban relative poverty is similar across regions, and correspondingly, the shares of different regions in the total disadvantaged population are comparable with their shares in the total urban population. The only notable exception is the coastal region where the incidence is lower, and the region thus accounts for a smaller share of the disadvantaged relative to its population share. Second, for both relative poverty thresholds, the incidence of urban disadvantage is inversely associated with city size. Provincial mega-cities have the lowest incidence, while at the other end prefecture-level cities have several times higher incidence. Urban disadvantage is thus primarily a small-city phenomenon: more than 80% of the urban disadvantaged population lives in prefecture- and lower-level cities. The elderly in urban areas are no more likely to be disadvantaged than other age groups Demographically, more than 70% of the urban disadvantaged population are adults between 16 to 60 years of age split evenly amongst males and females (Table 4.36). The elderly (those above 60 years) and children (below 16 years) constitute around 10% and 20% of the poor, respectively. Amongst the three age groups, children have the highest incidence of disadvantage (their share amongst the disadvantaged is higher than their share in the population). Just as for rural households, this likely reflects the fact that larger households typically have more children and they also tend to be poorer with adults who are relatively young. In particular, the data do not provide any supportive evidence for the view that the urban disadvantage or relative poverty is mainly a problem of the elderly. T h e b a s i c f a c t s a b o u t p o v e r t y a n d t h e p o o r 7 1 Table 4.36 : Urban disadvantaged in China by age and sex, 2003 Twice the WB line Thrice the WB line Share of Number of Number of population Number % who are poor Share of % who are poor Share of (%) (million) poor (million) poor (%) poor (million) poor (%) Children (less than 16 years old) Boys 7.5 26.6 7.3 1.9 9.8 24.7 6.6 9.2 Girls 6.9 24.5 7.1 1.7 8.7 23.5 5.8 8.0 Working-age adults (between 16 and 60 years old) Men 36.2 128.5 5.5 7.1 35.6 20.0 25.7 35.8 Women 38.0 134.9 5.5 7.4 37.3 19.5 26.3 36.7 Elderly (more than 60 years old) Men 5.8 20.6 4.0 0.8 4.1 16.3 3.4 4.7 Women 5.7 20.2 5.6 1.1 5.7 20.1 4.1 5.7 Total 100 354.9 5.6 19.9 100.0 20.2 71.7 100.0 Sources and notes: World Bank staff estimates from the national sample of NBS' 2003 Urban Household Survey using per capita consumption as the measure of welfare.. Households without high-school educated workers constitute most of the disadvantaged in urban areas It is not surprising that there is a strong negative relationship between the level of education and the incidence of relative poverty in urban areas (Table 4.37). There are significant returns to education and the completion of senior high school by the average labor member of the household virtually assures an exit out of poverty for the household (at twice the World Bank poverty threshold). However, relatively low levels of education are still an important attribute of the urban disadvantaged, more than half of whom have average education levels of their labor members below the junior high school stage. Table 4.37 : Education and urban disadvantaged in China, 2003 Twice Thrice the WB line the WB line Average years of Share of Mean income schooling of adult population (Yuan/ person/ % who Share of % who Share of household workers (%) year) are poor poor (%) are poor poor (%) Less than 6 3.3 5770 9.8 12.0 27.2 9.2 Between 6 and 9 20.6 6280 6.1 46.5 19.8 42.1 Between 9 and 12 45.6 7600 2.3 39.1 9.3 43.9 Greater than 12 30.5 10700 0.2 2.4 1.5 4.7 Total 100 8209 2.7 100 9.7 100 Sources and notes: World Bank staff estimates from the national sample of NBS' 2003 Urban Household Survey using per capita income as the measure of welfare.. Ethnic minorities and those without local urban hukou are more likely to be disadvantaged but are relatively few in number Despite some recent relaxations of the hukou registration system, the possession of a local hukou can matter a great deal to the welfare of those living in urban areas. However, as noted above, the NBS Urban Household Survey is not well-suited to examining this as its sample frame largely excludes migrant workers for whom this issue is most relevant. Amongst the urban population included in the Urban Household Survey, those without local non-agricultural hukou only account for less than 3% of the population (Table 4.38). While they have significantly higher incidence of relative poverty, given their low share in the urban population, they account for only about 5% of the urban disadvantaged. Minority status is also a relatively unimportant attribute of the urban disadvantaged population. While the incidence of relative poverty is higher for households with members belonging to ethnic minorities, the latter only account for about 6.5% of the urban T h e b a s i c f a c t s a b o u t p o v e r t y a n d t h e p o o r 7 2 population. Thus, only a small fraction (about 8-11%) of the urban disadvantaged comprises of ethnic minorities. Table 4.38 : Hukou and minority status and urban disadvantaged in China, 2003 Twice Thrice the WB line the WB line Share of population % who Share of % who Share of (%) are poor poor (%) are poor poor (%) Hukou status Local and non-agricultural 97.1 2.6 94.5 9.5 95.1 Non-local or agricultural 2.9 5.1 5.5 16.7 4.9 Minority status Household without minority members 93.5 2.6 89.1 9.5 91.7 Household with minority members 6.5 4.5 10.9 12.4 8.3 Sources and notes: World Bank staff estimates from the national sample of NBS' 2003 Urban Household Survey using per capita income as the measure of welfare.. More than half of the urban disadvantaged live in households with unemployed workers With the progress in economic reforms especially since the 1990s, China's urban labor market has become increasingly market-oriented with a rapid decline in the size of the state-owned and collective sector and an increasing degree of informalization of employment. This period has also witnessed the development of an urban social security system that is increasingly untied to employment in the public sector, though its coverage remains limited in many respects (see further discussion in chapter 6). This economic restructuring has induced major adjustments on the part of urban households as they find their feet in the new labor market environment. The potentially diverse impact of such labor adjustments suggests that the profile of relative poverty in urban China may now be more closely related to difference in the employment status of urban households. Table 4.39 presents some evidence on whether this is the case. Table 4.39 : Employment status and relative poverty in urban areas, 2003 % with per-capita income below: Twice the WB line Thrice the WB line Those living in households with/without... ...with ...without ...with ...without ...an adult not working 4.3 1.0 13.8 5.5 ...who is out of the labor force 3.3 2.3 11.2 8.6 ...and unable to work 11.9 2.6 28.1 9.6 ...who is unemployed 9.5 1.5 26.8 6.6 Share (%) of those with per- capita incomes below: Those living in households Share in urban Twice Thrice with ... population (%) the WB line the WB line ...an adult not working 50.6 81.3 72.0 ...who is out of labour force 42.3 51.3 49.0 ...and unable to work 0.5 2.4 1.6 ...who is unemployed 15.3 53.8 42.1 Sources and notes: World Bank staff estimates from the national sample of NBS' 2003 Urban Household Survey. A clear link between employment status and urban disadvantage is evident from Table 4.39. The following points are notable. · Households with a non-working adult member are much more likely to be poor (for both twice and thrice the World Bank poverty thresholds) than those where all adults work. · The "non- working" status is on account of either being out of the labor force or being unemployed. On looking at either of these two conditions separately, the data indicate that households with an adult being out of the labor force are also more likely T h e b a s i c f a c t s a b o u t p o v e r t y a n d t h e p o o r 7 3 to be poor than those without any such members, but not by much (3.3% versus 2.3% incidence for twice the World Bank line). · On the other hand, the difference in relative poverty incidence is much larger between households with and without an unemployed adult member (9.5% versus 1.5% for twice the World Bank line). · A subset of those out of the labor force is the group who are unable to work. Comparable with the case of the unemployed, there is also a large difference in poverty incidence between households with and without an adult who is unable to work: 11.9% versus 2.6% incidence for twice the World Bank line. · In sum, unemployment or lack of work capacity significantly increases the risk of a household being relatively poor in urban areas, while labor force participation by itself seems to matter much less. · However, in terms of the shares amongst the total urban disadvantaged population, it is the unemployment status that weighs much more heavily than the inability to work. For instance, households with an adult who is unable to work account for only 0.5% of the urban population and 2.4% of the urban disadvantaged (using twice the World Bank poverty line), but households with an unemployed member account for about 15% of the urban population and 54% of the disadvantaged. Households that have an unemployed or non-working adult member may nevertheless have others who work. Figure 4.49 takes a further look at data at the individual level, and shows for differing employment status of urban adults, their probability of being poor relative to the urban population as a whole, as well as their shares in total urban disadvantaged population. Figure 4.49 : Relative probability of being disadvantaged and share amongst the urban disadvantaged for adults (age 16 and above) by employment category, 2003 Relative risk of income poverty, by employment category Twice the WB line Thrice the WB line Unemployed 2.79 3.69 Out of labor force 2.48 3.07 Self-employed & informal 1.76 1.68 Students 1.17 1.23 Retired with pension 0.49 0.39 Formal: public & private 0.54 0.41 0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0 Shares of the income poor, by employment category Twice the WB line Thrice the WB line Unemployed 20.2 26.7 Out of labor force 11.113.7 Self-employed & informal 22.5 21.4 Students 10.6 11.1 Retired with pension 4.9 3.9 Formal: public & private 30.7 23.1 0 5 10 15 20 25 30 35 Sources and notes: World Bank estimates from national sample of the NBS 2003 Urban Household Surveys. Employment categories in the figure are mutually exclusive. Thus, out-of-labor-force category excludes students and those who are retired with pension. T h e b a s i c f a c t s a b o u t p o v e r t y a n d t h e p o o r 7 4 The figure shows that adults who are employed in the formal public or private sector or who are retired with pensions are less likely (about half as likely) to be poor than the average urban adult. On the other hand, those who are unemployed, out of the labor force, self-employed or in informal jobs are significantly more likely to be disadvantaged. The unemployed are the most vulnerable being about 3.7 times as likely to be poor (as the average urban adult); those out of the labor force38 are about 3.1 times as likely to be poor; and the self- or informally employed are about 1.7 times as likely to be poor. The latter three groups account for about 61% of all urban disadvantaged adults (using twice the World Bank poverty line). Unlike Table 4.39, those out of the labor force exclude retired workers and students, which shows that disaggregating the out-of-labor force group is important for identifying vulnerable groups within this category of individuals. Of the remaining, those in formal public or private employment still account for about a quarter, mainly because this group is such a large share of the total population. Together with the results mentioned above, these findings reinforce the importance of how individuals and households participate in the labor market as a determinant of economic welfare in urban areas. _________________________________________________________ 38Excluding adult students and those retired with pension. T h e b a s i c f a c t s a b o u t p o v e r t y a n d t h e p o o r 7 5 5.Promoting opportunity: helping the poor move out of poverty This chapter evaluates China's experience of two important pathways to poverty reduction: poor area development and migration. The policy debate has sometimes been cast in terms of "moving jobs to people" or "moving people to jobs". China has now accumulated a great deal of experience with both approaches. While the policy options need not be cast in terms of an all-or-nothing choice between these two alternatives, it is important to carefully evaluate the experience to understand how either or both could be harnessed better in meeting the challenge of future poverty reduction. a. China's poverty reduction efforts have been development-oriented China's poverty reduction policies have focused on providing opportunities by building assets for the poor and transferring assets to the poor, largely avoiding pure "handouts". Poverty reduction programs have sought to raise income generating capacity of poor rural households. In rural areas, they have provided important investments in infrastructure, agriculture and enterprises, and support to human capital development as for instance through construction of schools. China's poverty reduction policies can roughly be classified into four phases since 1978. These include the rural reform of 1978-1985, the National Poverty Reduction and Development Programs of 1986-1993, the 8-7 Poverty Reduction Plan of 1994-2000, and the New Century Rural Poverty Alleviation Plan for 2001-2010. First established in 1986, China's official poverty alleviation programs emphasize increasing the productive capacity of the rural poor through regionally-targeted investment programs. At each level of government (national, provincial, and county), an inter- ministerial Leading Group for Poverty Alleviation and Development coordinates the design and implementation of poverty alleviation programs through its Leading Group Office for Poverty Alleviation and Development (LGOPAD). Rural Reforms and the introduction of the Household Responsibility System were central to poverty reduction during the early 1980s The most pressing problem during the early years of the post-1978 reform period was to boost rural economic growth through rural reforms. The national economy was stagnating and poverty widespread all over the country, exacerbated by the shocks of the ten-year long Cultural Revolution and the long-standing restrictions on economic activity. To boost the rural economy, the Government of China launched institutional reforms of the rural production, distribution systems, and procurement prices. The fundamental institutional change was the land reform characterized by the implementation of the household responsibility system in rural China, whereby farm households became residual claimants to output by receiving land use rights in return for delivery of a certain quota of grain to the village and meeting their tax obligations. The early rural reforms delivered remarkable results in terms of poverty reduction as well as agricultural production and rural industries. A strong growth in grain yields (5.7% per year) and in rural industries, accompanied by sharp increases in agricultural procurement prices, raised rural incomes at 15% a year in real terms. The growth in agricultural production and farmers' income was spectacular in some extremely poverty-stricken regions, such as the Huanghuaihai region and Eastern Fujian (Wang, 1994). As poverty was widespread across China's rural areas in the early years of the post-1978 reforms, rural income growth delivered nearly universal poverty reduction. Between 1981 and 1984, income poverty at the basic needs (approximately $1/day) line fell from 53 to 24 percent, and the number of rural poor declined from 250 million in 1978 to 125 million in 1985 measured at the official poverty line (Ravaliion and Chen, 2007; NBS, 2005a). P r o m o t i n g o p p o r t u n i t y 7 6 Area-based efforts began with the National Poverty Reduction and Development Programs during 1986-1993 Once the poverty effect of the early rural reforms was realized, poverty reduction during 1986-1993 slowed down. By the mid-1980s, rural economic growth and living conditions appeared to lag again, particularly in the revolutionary, minority, and border regions. As a result, these areas received special attention in the Seventh Five Year Development Plan (1986-1990). This marks the beginning of China's targeted rural poverty reduction and development programs. The rural poverty reduction programs launched in mid-1980s involved a wide variety of actors, initiatives and funding channels. The State Council's Leading Group for Poverty Reduction and Development (LGPRD) was established in 1986 to provide coherence to the large number of poverty reduction initiatives and, in particular, to expedite economic development in poor areas. The government designated poor counties as the basic unit for poverty targeting. For the officially designated "poor" counties, the central government created special funds to support a subsidized loan program, food for work (FFW) program, and poor-area development budgetary grants. Subsidized loans covered both households and enterprises in industry and agriculture, food-for-work program utilized surplus farm labor mainly to develop infrastructure, and government budgetary grants supported investment in poor areas across all the sectors (World Bank, 2001). At the local level, most poor provinces, prefectures, and counties established Leading Group offices, and local governments were required to provide counterpart funds. Different agencies became responsible for different poverty reductions projects and activities (rural roads constructed under the FFW program, for example, have been implemented by local staff of the Transport Bureau). Some preferential taxation treatment was offered to poor regions (Office of the Leading Group for Poverty Reduction and Development, 1989). But the overall progress in poverty reduction in China was slower and even reversed in some years, coinciding with stagnation in the rural economy. Country-wide, the rapid growth of agricultural productivity halted in mid-1980s, and China's agriculture and rural economic development stagnated. Comparing the period 1986-1993 with 1978-1985, growth rate in the per-capita value added from agriculture dropped from 9.7% to 3.4%, and the growth rate of net income per capita fell from 12.2% to 2%. The decline on both fronts was closely associated with the relative decrease of agricultural prices and deteriorating rural-urban terms of trade in late 1980s and early 1990s. Farmer's annual net per capita income even declined in some years, and the income gap between urban and rural residents widened. By some estimates, poverty reduction programs on average contributed more than 1 percentage point to farmers' income and consumption growth (Jalan and Ravallion, 1998, Park, Wang and Wu, 2002). By the end of 1993, poverty rate was at 20% at the basic needs (approximately $1/day) line or, measured at the official poverty line, 80 million of rural population remained poor, lower by 45 million compared to 1985 (Ravaliion and Chen, 2007; NBS, 2005a). The 8-7 Poverty Reduction Plan continued and intensified these efforts during 1994- 2000 In 1994, the Government introduced the "8-7 Poverty Reduction Plan", aspiring to lift the majority of the remaining 80 million poor above the government's poverty line during the seven year period 1994-2000. The diminishing pace and increasing difficulties in poverty reduction revealed that greater determination and efforts were needed to solve the food and clothing problem of the remaining 80 million of the rural poor population. The 8-7 Plan was introduced to intensify the poverty reduction policies that had commenced in the second half of 1980s. It outlined several objectives. It aimed at 1) assisting poor households with land improvement, increased cash crop, tree crop and livestock production, and improved access to off-farm employment opportunities, 2) providing most townships with road access and electricity, and improving access to drinking water for most poor villages, 3) accomplishing universal primary education and basic preventive and curative health care, 4) graduating better-off counties in the coastal provinces from the newly-established list of nationally-designated poor counties, 5) managing well available funding, with attention to the appraisal and financial viability of poverty reduction investment activities, recovery of loan funds and leakage of poverty reduction funding to P r o m o t i n g o p p o r t u n i t y 7 7 alternative activities, and 6) enlisting involvement and support from all government ministries and agencies, the coastal provinces and major municipalities, and other domestic and international organizations (LGPRD, 1994). In implementing the 8-7 Plan, the Government refined its selection of "poor" counties and emphasized the responsibility of local government leaders for the effectiveness of poverty reduction work in their jurisdictions. In 1997, annual funding for poverty reduction jumped by about 50% in real terms, which reversed a decade of decline in real funding for poverty reduction in China. Over the next four years, allocations for poverty reduction continued to increase, almost doubling the 1997 level by 2001. The 8-7 Plan maintained the three channels of interventions launched in 1986: subsidized loans program, food-for-work program and government budgetary grants. Subsidized loans program accounted for the greatest increase in funding. Empirical analysis indicates that the 8-7 Plan had a positive impact on the designated "poor" counties. The allocation of poverty funds across the officially designated "poor" counties under the 8-7 Plan was correlated with poverty incidence, i.e., poorer counties received more funding (Li and Wang, 2004). Analysis also suggests that although the efficiency of poverty investments has been declining, designated poverty reduction funds under the 8-7 Plan achieved a rate of return over 10% (Park, Wang and Wu, 2002). Poverty reduction accelerated over the 8-7 Plan period, particularly during 1994-1996. Using the official poverty line, the number of rural poor declined from 80 million to 32 million between 1993 and 2000, or at an annual rate of about 12%, which was 3 percentage points higher than the average poverty reduction rate over the entire post- 1978 reform period. Measured by the basic needs line (approximately $1/day), the total number of poor dropped from 237 million (20% of the total population) in 1993 to 108 million (8.5% of the total population) in 2000 at an annual rate of 11% (Ravaliion and Chen, 2007; NBS, 2005a). Several factors contributed to faster poverty reduction during the 8-7 Plan period including more rapid economic growth, agricultural price increases (related to the 1994-1996 government procurement reform), rural-urban migration and poverty reduction efforts. The greatest effect on poverty reduction under the 8-7 Plan period was realized during 1994-1996, and appears to have been primarily associated with the procurement reforms which raised agricultural prices and boosted rural growth. With county as the basic targeting unit up to 2000, ineffective targeting of the poor seemed to be a major problem of China's poverty investments. Though poverty investments benefited poor areas and helped increase household incomes on average, the evidence indicated that poor households benefited less than better-off households (Wang, Li and Li, 2004). Low recovery of subsidized loans was another serious problem, which undermined the effectiveness of the overall program, and also distorted the rural financial market in poor areas. Further refinements were made in the New Century Poverty Reduction Strategy for 2001-2010 Learning from the experience of the "8-7 Plan", the Chinese Government launched a New Century Rural Poverty Alleviation Plan for 2001-2010. During the 1990s, China's rural poor population shrunk in scale. Based on the official rural poverty line, rural poverty incidence dropped to 3.0% in 2002 and under 9% even in poor counties. The poor population also was further concentrated in the western regions. However, within the western regions, poor households were relatively scattered across villages rather than concentrated in poor counties, which paved the way for a further refinement of the unit of targeting. China's Rural Poverty Reduction and Development Compendium (2001-2010) was issued in 2001. A number of important new policies were issued later on. One of the main policies was the shift from county targeting to village targeting although national key poor counties are still designated. A total of about 148,000 poor villages both within and outside the key poor counties were designated for receiving poverty assistance. The list of 592 poor counties was updated to focus more on the western regions and these key poor counties are favorably treated in tax reform and poverty fund allocation. The new P r o m o t i n g o p p o r t u n i t y 7 8 plan also emphasizes more on human capital and social development in poor localities and promotes participatory poverty reduction approaches. Integrated community and village development based on participatory village planning has emerged as the key strategy for rural poverty reduction (LGPRD, 2001; Gao, 2001). Participatory village planning was conducted in the 148,000 designated poor villages, of which about 53,000 villages had completed participatory village planning by 2004. In the village plans, detailed budgets and funding sources for specific projects are developed using processes that include participatory approaches. In principle, infrastructure projects are financed by the Food-for-Work program, social services including primary education and health, training and technology extension are financed by budgetary development funds and agricultural production is financed by the subsidized loan program. The funding for poverty alleviation programs comes from multiple sources Three different government agencies provide funding for official poverty alleviation programs: · the Ministry of Finance provides budgetary grants to support public investments as well as training and resettlement projects; · the National Development and Reform Commission (NDRC) funds Food-for-Work infrastructure projects, especially for roads and drinking water; · subsidized loans to enterprises and households are provided through the Agricultural Bank of China (ABC), and more recently thorough Rural Credit Cooperatives. Since 2001, on average about 28 billion Yuan has been provided annually by the central government through these three agencies (Table 5.40). This has been equivalent to more than 5% of the national budget. These totals, however, do not take into account matching funds and voluntary spending by local governments. About half of the total funding has been in the form of subsidized loans, although their share (as well as the absolute amount) has declined since 2005. Table 5.40 : Central government funding for poverty alleviation programs, 2001 to 2007 (billion Yuan) Year Subsidized loans Food for work Budgetary funds Total 2001 18.5 6.0 5.0 29.5 2002 18.5 6.0 5.6 30.1 2003 18.5 6.5 6.4 31.4 2004 18.5 6.4 8.2 33.1 2005 7.0 6.8 9.0 22.8 2006 6.9 4.6 9.7 21.2 2007 12.4 4.5 10.4 27.3 Average 2001-7 14.3 5.8 7.8 27.9 Sources and notes: LGOPAD and Ministry of Finance. Though the funding channels are the same as the 8-7 plan, the focus of the investments and the ways of using these funds changed dramatically. The LGOPAD is now promoting three major approaches for rural poverty reduction and poor areas development: · integrated village development program (IVDP) based on participatory village investment planning, begun in 2001 · labor training program for workers from poor counties , begun in 2003 · agri-business development for poverty reduction, begun in 2003, and is linked with the subsidized credit provided by the ABC. The three programs aim to improve income-generating opportunities via rural production, migration, and local enterprise development, respectively. The IVDP program is discussed in the following section, while the other developmental poverty reduction programs are discussed in section 5c. P r o m o t i n g o p p o r t u n i t y 7 9 b. Raising Incomes through poor-area development: the Integrated Village Development Program China's poor village investment program began in 2001, partly in response to concern that the previous system of county-level targeting had failed to reach many of China's poor (World Bank, 2001; Park, Wang, and Wu, 2002). The decision to base investments on participatory village plans reflected concerns that investment projects were not sufficiently focused on the needs of China's poor. A reasonably good effort was made to designate poor villages according to specified criteria In 2001, 148,131 villages, or 20.9% of all villages in China, were designated to participate in the IVDP program. According to official guidelines, poor villages were selected according to a weighted poverty index based on eight indicators: grain production per capita, cash income per capita, percentage of low quality houses, percentage of households with poor access to potable water, percentage of natural villages with reliable access to electricity, percentage of natural villages with all-weather road access to the county seat, percentage of women with long-term health problems, and percentage of eligible children not attending school. However, local governments could alter the weights or the indicators based on local circumstances, and they were often constrained by the limited administrative data available for each village in their jurisdiction. Generally, county governments were responsible for identifying poor villages, while provincial governments determined the number of village designations and approved the final list of poor villages.39 The regional breakdown of designated poor villages is presented in Table 5.41. Nearly half of all poor villages are in western China which is consistent with the share of the western region in total number of the poor (chapter 4). The percentage of villages designated as poor is 41% in the northwest, 32% in the southwest, 26% in the northeast, 22% in the central region, and 8% in coastal areas. Econometric analysis of the determinants of poor village designations suggests that a reasonably good effort was made to designate poor villages according to the formula-based approach. Low village per capita income, low agricultural productivity, limited access of village groups to roads and clean water as well as village being in a minority or old revolutionary base area all contributed significantly to the probability of its being designated as poor (Wang et al., 2006). Table 5.41 : Number of designated poor villages, by region Total no. of No. of designated % of villages Share of poor Region villages poor villages designated poor villages (%) Coastal 249,723 20,698 8.3 14.0 Northeast 35,540 9,182 25.8 6.2 Central 225,964 48,950 21.7 33.0 Southwest 132,879 42,647 32.1 28.8 Northwest 65,151 26,654 40.9 18.0 Total 709,257 148,131 20.9 100.0 Sources and notes: LGOPAD and National Agricultural Census Office. According to the World Bank-NBS Special Purpose Village-Level Survey, by the end of 2004, 55% of designated poor villages had completed village plans, and 37% of villages had begun investments based on the plans (Figure 5.50). An important reason why most villages had yet to begin planned investments three years after the program had begun is that county governments generally concentrated annual program allocations in a subset of villages. The decision to fund village plans sequentially rather than simultaneously was due to practical concerns such as economies of scale in investments and the fixed costs _________________________________________________________ 39In general, administrative villages are selected. However, in a few provinces such as Jiangxi, IVDP projects are implemented at the level of natural villages. P r o m o t i n g o p p o r t u n i t y 8 0 associated with supervising the design and implementation of plans in each village. The village data confirm that that increase in "treated" villages over time was due to a gradual expansion of investments in new villages within counties rather than expansion across counties. Under village planning, the top six investments were roads and bridges (34%), electricity (16%), irrigation (10%), schools (9%), afforestation (8%), and communication (8%). Figure 5.50 : Completion of plans and initiation of investments in designated poor villages (Cumulative % of designated poor villages) 60 50 40 30 20 10 0 t t t t letedp n n letedp n n letedp n n letedp n n m pla Begu pla pla pla Co vestmenni m Begu Co vestmenni m Begu Co vestmenni m Begu Co vestmenni 2001 2002 2003 2004 Source: World Bank calculations based on NBS 2001 Rural Household Survey and the 2004 World Bank-NBS Special Purpose Survey of 3037 villages. But more than half the poor live outside the designated villages But how effective was the program in targeting the poorest villages? Using NBS household survey data and the World Bank-NBS Special-Purpose Village-Level Survey, it is possible to examine whether poor village designations in fact went to the poorest villages measured by mean household income per capita (based on NBS survey of about 10 households per village). Figure 5.51 shows the percentage of villages in each income per capita quintile (where villages are ranked by their real mean per capita incomes) that are designated as poor villages. Even though villages with lower per capita incomes have a higher chance to be designated as poor villages, nearly 45% of villages in the poorest quintile are not designated as poor while at the same time, numerous richer villages, even some in the top income quintile, are designated as poor villages. Figure 5.51 : Percent of villages in each income per capita quintile designated as poor in 2001 (Percentage of villages in quintile designated as poor) 60 54.9 50 40 30 20.5 20 13.5 12.5 10 5.3 5.8 0 Poorest 2nd 3rd 4th Richest All Village mean per-capita income quintile Source: World Bank calculations based on NBS 2001 Rural Household Survey and the 2004 World Bank-NBS Special Purpose Survey of 3037 villages. The village income per capita data used in these calculations includes measurement error and overlooks non-income factors that could have influenced designations. Analysis of P r o m o t i n g o p p o r t u n i t y 8 1 the determinants of poor village designation in fact finds that village characteristics such as poor infrastructure and remote location increases the probability of being designated as a poor village even after controlling for village income level (Wang et al., 2006). Nonetheless, the substantial mistargeting suggested by Figure 5.51 merits greater scrutiny. The targeting error in poor village designations, defined as the percent of poor designations given to villages that would not be designated poor under perfect targeting based on village income per capita, is equal to 48%. Put differently, 48% of the villages had mean per capita incomes in 2001 that were above the official poverty line. Table 5.42 (top panel) presents further evidence on the extent of coverage and potential leakage of the village program, measured respectively by the percentage of the poor who live in poor villages and the percentage of the population living in poor villages who are non-poor. In 2001, about 60% of the poor by the official poverty line lived in officially designated poor villages, and 52% of those with incomes below the World Bank poverty line lived in designated villages. After three years (in 2004), these coverage rates had fallen considerably to 51% and 47% respectively. Potential leakage was equal to 85% in 2001 and 90% in 2004 using the official poverty line, and 72% and 79% using the World Bank poverty line. Table 5.42 : Targeting outcomes under poor village and poor county designations Targeting outcomes under poor village designation Coverage Potential Leakage (% of poor who live in (% of population in designated designated poor villages) poor villages that is not poor) 2001 2004 2001 2004 Official poverty line 59.5 50.8 84.5 89.7 World Bank poverty line 51.9 47.4 72.0 79.3 Targeting outcomes under poor county designation Coverage Potential Leakage (% of poor who live in (% of population in designated designated poor counties) poor counties that is not poor) 2001 2004 2001 2004 Official poverty line 70.8 62.0 83.9 88.9 World Bank poverty line 66.6 59.0 68.7 77.3 Sources and notes: World Bank calculations based on NBS 2001 Rural Household Survey and the 2004 World Bank-NBS Special Purpose Survey of 3037 villages. Per capita income is used as the measure of welfare in defining the poor. How does this compare to poor county targeting? It turns out that coverage is actually better under poor county targeting than under poor village targeting. As seen in the bottom panel of Table 3.3, coverage of the poor was 71% in 2001 and 62% in 2004 using the official poverty line, and 67% in 2001 and 59% in 2004 using the dollar-a-day line. Potential leakage was also slightly lower under county targeting. These results warrant concern because in principle, switching to lower-level targeting should increase potential targeting precision. In fact, if one compares the current coverage of the poor (about 60% using the official poverty line) with the maximum number of poor that could be targeted with the same overall percentage of the population covered (or budgeted) by the program, one finds that under perfect village targeting, it is possible to cover nearly 80% of the poor, while under perfect county-level targeting, one can cover at best 71% of the poor (see Figure 5.52). P r o m o t i n g o p p o r t u n i t y 8 2 Figure 5.52: Potential gains from improved targeting under village and county targeting Share of poor covered as program size increases under perfect targeting based on village vs. county per capita income Perfect village targeting Perfect county targeting 1 1 .8 .8 Potential gain Potential gain .6 .6 Actual coverage of the poor by the current program .4 .4 .2 .2 0 0 0 .2 .4 .6 .8 1 0 .2 .4 .6 .8 1 Fraction of population covered by program Fraction of population covered by program Source and notes: World Bank calculations based on NBS 2001 Rural Household Survey and the 2004 World Bank-NBS Special Purpose Survey of 3037 villages. The vertical axis shows the cumulative share of poor in villages (or counties) when villages (or counties) are ranked by declining poverty incidence using the official poverty line and per capita income as the measure of welfare. One reason for higher coverage of the poor by poor counties is greater population coverage of poor counties (24.7% of the rural population) compared to poor villages (20.7%). However, this is not a sufficient explanation. The accuracy of targeting was also greater under county targeting. In fact, targeting error rate for poor county designations in 2001 was 25%, compared to 48% for poor villages. Greater targeting error at the village level could be due to a number of factors: use of a broader set of non-income criteria in village designations, much more limited data on village versus county outcomes, much noisier village data, and greater susceptibility to political influence at lower levels of government. Was there spatial variation in targeting outcomes? Figure 5.53 plots the program coverage rates of those living below the World Bank poverty line in different regions and in poor and nonpoor counties, for both 2001 and 2004. First, the western regions have much higher coverage rates than non-Western regions. Second, coverage in poor counties is much more complete than coverage in non-poor counties. Third, there is much faster deterioration in coverage over time in the coastal and northeast regions compared to other regions, perhaps because poverty in richer areas is more transitory. Figure 5.53 : Fraction of the income poor living in designated poor villages (Fraction of the income poor living in designated poor villages, by region and overall) % of income poor who live in designated poor villages 2001 2004 80 68.5 66.8 69.9 69.0 60 54.7 53.4 51.9 47.4 40 33.1 35.4 27.7 28.3 20 16.4 12.1 16.1 16.4 0 tal st ralt ts s r s l as ea hwe estw ie or ie po Co rth Cen rth Poo No ountc Overal Sout No on-N ountc Source and notes: World Bank calculations based on NBS 2001 Rural Household Survey and the 2004 World Bank-NBS Special Purpose Survey of 3037 villages. Income poverty is relative to the World Bank poverty line. P r o m o t i n g o p p o r t u n i t y 8 3 How can one account for these regional differences in coverage rates? As noted earlier, regional targeting becomes more difficult when poverty is more dispersed, or less concentrated. Earlier section 4c introduced the idea of a poverty concentration index that takes a minimum value of zero if poverty is perfectly dispersed across villages, or all villages have the same poverty rate, and takes a maximum value of one if poverty is perfectly concentrated across villages so that there are only two types of villages where either everyone or no one is poor. The measure can also be interpreted as the normalized variance of village headcount indices. As noted earlier, concentration tends to fall with lower poverty lines. For instance, the village poverty concentration index for rural China is 0.319 using the official poverty line, 0.368 for the World Bank poverty line, and 0.388 for twice the World Bank poverty line. Concentration indices also tend to be higher in areas of greater poverty. As seen in Table 5.43, poverty concentration is highest in the western provinces, and the index is much higher for villages in poor counties (0452) than that for villages in nonpoor counties (0.267). These rankings correspond closely to the differences in program coverage displayed in Figure 5.53. Table 5.43 : Coverage rates adjusted for concentration of poverty and scale of program (% of income poor covered in 2001, using the World Bank poverty line) Poor Poverty Population Maximum Adjusted coverage concentration coverage per coverage coverage rate (%) index (%) poor person rate (%) rate (%) Nation 51.9 36.8 1.9 77.1 67.3 Coast 33.1 34.2 3.2 66.8 49.5 Northeast 35.4 26.6 1.8 61.6 57.4 Central 27.7 27.9 2.6 65.8 42.1 Southwest 68.5 52.8 1.6 86.2 79.5 Northwest 54.7 37.2 1.2 73.9 74.1 Nonpoor Counties 16.1 26.7 1.9 50.7 31.7 Poor Counties 69.9 45.2 1.9 92.2 75.8 Sources and notes: World Bank calculations based on NBS 2001 Rural Household Survey and the 2004 World Bank-NBS Special Purpose Survey of 3037 villages. Adjusted coverage equals poor coverage rate/maximum coverage. A second factor that affects coverage rates of the poor is the population coverage of the program, which can be considered a budget constraint. To explore the role of this factor, the population covered by the program is normalized by the total number of poor in the region. As seen in Table 5.43, population coverage per poor person turns out to be the highest in the Coast, followed by the Central region, Northeast, Southwest, and Northwest; so this does not help explain the observed patterns in coverage of the poor. Given a region's population coverage per poor person, it is straightforward to calculate the maximum coverage rate of the poor given the sample income distribution when villages are perfectly targeted based on village income per capita. This "maximum" coverage serves as a benchmark for evaluating to what extent observed coverage outcomes can be explained by fundamentals (poverty concentration, budget) versus targeting effectiveness. Adjusted coverage rates, which divide actual coverage rates by "maximum" coverage rates, are presented in the last column of Table 5.43. While the maximum coverage rates are higher in western provinces and poor counties, so are the adjusted coverage rates. Thus, even after controlling for differences in fundamentals, western provinces and poor counties do a substantially better job in reaching the poor. This intuition can be verified by calculating village targeting error rates by region (percent of designated villages with mean per capita incomes above the World bank poverty line), which are displayed in Figure 5.54. Similar to the adjusted coverage rates reported in Table 5.43, targeting errors are found to be greatest for the Coast, followed by the Northeast, Central, Northwest, and Southwest. Again, targeting errors are much smaller in the western regions and in poor counties. Possible reasons for better targeting performance in the western regions and in poor counties include the following: greater concentration of poverty in areas with have high poverty rates, stronger institutions P r o m o t i n g o p p o r t u n i t y 8 4 (county LGOPADs), greater experience in targeting poor villages (e.g. Yunnan began targeting villages before the rest of the country). Figure 5.54 : Potential errors in designating poor villages (% of villages designated poor where village mean-per-capita income exceeds the World Bank poverty line) 100 80 60 40 73.6 85.9 64.6 58.7 20 48.2 35.0 42.0 33.9 0 inahC aste tse Coast hw estw oor seit or seit llA rth Central rth No onpN Po Sout No ounC ounC Sources and notes: World Bank calculations based on NBS 2001 Rural Household Survey and the 2004 World Bank-NBS Special Purpose Survey of 3037 villages. Income poverty is relative to the World Bank poverty line. The program raised incomes in designated villages but the poorest households have failed to benefit How has poor-village investment program affected the welfare and behavior of rural households? A simple comparison of mean changes in income per capita from 2001 to 2004 presented in Table 3.6 finds that mean income growth was faster in poor villages (6.7% per year) than in non-poor villages (4.8% per year), and faster in poor villages with investments based on village planning (7.8% per year) that in poor villages without plan investments (6.1% per year). Even if one narrows the comparison to poor and non-poor villages within poor counties, mean income growth is faster in poor villages (7.2% per year) than in non-poor villages (6.1% per year). Table 5.44 : Comparing per-capita income growth in designated poor villages (in 2003 Yuan) Mean per-capita income Mean annual income 2001 2004 growth (%): 2001-2004 All villages 2454 2910 5.04 Designated poor villages 1637 2008 6.72 Other villages 2573 3043 4.80 Designated poor villages in poor counties 1322 1619 7.20 Other villages in poor counties 1585 1914 6.13 Designated poor villages with plan investments 1733 2211 7.77 Designated poor villages without plan investments 1580 1888 6.10 Sources and notes: World Bank calculations based on NBS 2001 Rural Household Survey and the 2004 World Bank-NBS Special Purpose Survey of 3037 villages. However, these differences in mean income growth are not necessarily attributable to the IVDP, since faster income growth of poorer households and villages is commonly observed due to mean reversion. For example, villages hit by temporary disasters in 2001 could have been both more likely to be designated as poor and have enjoyed faster subsequent income growth, or other factors not related to IVDP could explain the differences. Nonetheless, the sizable real rural income increases of the early 2000s did P r o m o t i n g o p p o r t u n i t y 8 5 reduce poverty. Using the official poverty line, we find that from 2001 to 2004, the poverty headcount rate fell from 15.6% to 10.3% in designated poor villages and fell from 2.8 % to 2.5% in nonpoor villages. Using the low income line, over the same period the poverty headcount rate fell from 28.0% to 20.7% in poor villages, and fell from 6.8% to 5.8% in non-poor villages. But what were the specific impacts of IVDP on total investment in the village and on rural households, especially the poor? To examine these questions systematically, using data from NBS rural household surveys before and after the implementation of the program (2001 and 2004) and the World Bank-NBS Special Purpose Village-Level Survey, program impacts were estimated by comparing changes (in impact variables) in designated poor villages that had already begun plan investments with changes in designated poor villages with similar characteristics that had yet to begin plan investments (Park and Wang, 2006).40 In villages beginning plan investments, total annual investment41 per capita increased from 60.7 Yuan before planned investments to 106.3 Yuan after the program started, compared to 44.6 Yuan before plan implementation and 62.7 Yuan afterwards in villages yet to begin planned investments. Overall, the plans increased total investments in the village significantly. Both village and government investments increased significantly, suggesting that government financing was a complement not a substitute for village financing in practice. Investments increased more in non-western regions where village matching investments increased along with government investments. In contrast, in the West, village investments did not increase in plan villages, perhaps because matching investments were not required or villages were unable to finance village investments (Park and Wang, 2006). The main findings on household impacts are described in Table 5.45. For the full sample of households, there are no statistically significant effects of the program on mean income growth, consumption growth or change in the propensity to migrate. An important finding is that the poorest households have not benefited from the IVDP program, while the program has increased incomes by 6.6% and consumption by 8.8% for the relatively better-off households in poor villages over the three years. Similarly, while the program has had no effect on the migration propensity of poorer households, for better-off households it has reduced the share of labor that migrates by 5.2%. Table 5.45 : Household-level impacts of the Integrated Village Development Program (% change between 2001 and 2004) Income per Consumption per Share of labor with capita capita migrants All households 3.0 1.0 -2.5 Poorer households -3.9 0.1 -0.5 Better-off households *6.6 **8.8 ***-5.2 Sources and notes: World Bank calculations based on NBS 2001 Rural Household Survey and the 2004 World Bank-NBS Special Purpose Survey of 3037 villages. Details of the estimation procedure are in Park and Wang (2006). Poorer households are those whose income per capita is below the median income of households in the matched sample of villages used for estimation, the better-off are those with incomes above the median. Statistical significance at the 90% confidence level is indicated by *, at the 95% level, by **, and at the 99% level, by ***. _________________________________________________________ 40This is better than comparing changes in designated poor and nonpoor villages because the highly visible designations are frequently used as the basis for targeting for other projects and policies initiated by local governments, social organizations, or international donors. Thus, a simple comparison with nonpoor villages could confound the IVDP impacts with the impacts of these other projects and policies. 41Total investment in the village includes monetary investment by the government and by the village itself, and it also includes the value of corvee labor days based on the mean daily male labor wage rate in 2004 as reported in the village questionnaire. P r o m o t i n g o p p o r t u n i t y 8 6 Program impacts hampered by difficulties in inter-agency coordination, inadequate resources and limited participation of poor households The poor-village investment program was intended to integrate and coordinate investments under the three official poverty investment programs. However, as in the past, coordination among government agencies has proven difficult because each agency is reluctant to relinquish control over resources and the LGOPAD has limited practical authority over other government units. In practice, the allocation of subsidized loans is rarely coordinated with village plans, and Food-for-Work projects are sometimes coordinated with village plans. In the end, village plans are mainly financed from budgetary grants (Wang, 2006). The lack of inter-agency coordination is one reason why village investment plans often have budgets that far exceed the amount of actual investments that are financed (Wang, 2006). According to official training materials (Gao, Wang, and Huang, 2001), the following principles are expected to guide the village planning process: projects helping the poor should be favored; participation of households and different groups (e.g., women) should be emphasized; plans should integrate resources from different sources; plans should be for a 3-5 year time horizon and should reflect local conditions and causes of poverty; and plans should be developed following standardized procedures set by the county government. Specific procedures are recommended for completing village investment plans. First, an analysis of the causes of poverty in the village should be conducted and possible solutions identified, based on analysis of village-level data and participatory workshops with 10 to 20 villagers. Next, with the support of technical experts, a SWOT (strength, weakness, opportunity, and threat) analysis and feasibility analysis should be completed for potential projects. Projects are selected by a plenary session of the entire village, with the views of poor households given added weight. Once projects are chosen, a more detailed assessment is completed describing project beneficiaries, project requirements, project implementation (schedule, budget, labor allocations), and a plan for monitoring and evaluation. The final product is the village planning document, which serves as a detailed blueprint for the investment projects to be undertaken. Field research in different regions revealed that actual practice rarely adheres to all of the guidelines. Plans are often designed by a group including village committee members, small group (hamlet) leaders, Communist Party representatives, and household representatives, most often under the supervision of township government officials who receive training from county LGOPAD staff. Because the program was scaled up so quickly, there is only limited evidence on the true extent of participatory planning that occurred. Field research on the implementation of village investment projects undertaken as part of the qualitative poverty assessment in 12 study villages found several factors that inhibited the success of the projects: small size of investments, inappropriate choice of activities, insufficient resources for maintenance of projects (Box 5.12 ). However, a key underlying cause seems to be little participation by villagers in the village planning process. Rather, local government officials played a prominent role in plan development. The research team frequently asked the villagers how planning was carried out and how the list of poor households was selected, but in most cases the answer was "don't know". Participatory planning seldom appears to have been successfully implemented in practice (CASS-IOS, 2006). Box 5.12 : Some factors inhibiting impact of village investment projects: evidence from qualitative field research Limited investments. While infrastructure improved in several villages through village investment projects, the investments have often been relatively small. The scale of investment is based on amount of money available from poverty alleviation agencies rather than in accordance with the needs of villagers. For instance, one villagers' group was left without a cement road in S village in Jiangxi Province, and as a result the villagers have to carry goods. Similarly, the problem of irrigation had not been solved even after projects were finished in A and Y villages in Yunnan Province. Inappropriate activities. Many income-generating activities end up being unsuccessful as they are pushed by the government rather than being based on market demand. Sometimes farmers P r o m o t i n g o p p o r t u n i t y 8 7 have lost a great deal of money and have gradually become disinterested in these activities. For instance, the goat breeding project in R village in Inner Mongolia failed. The main reason was that the price was very high at 800 Yuan a goat when the households bought the goats, but in less than a year it plummeted to 200-300 Yuan, sometimes even around 100 Yuan. Similarly, in DD village of Gansu Province, the sheep (from Qinghai province) purchased by local officials together with the village committee were not accustomed to local ecological environment. They had to be penned and could not be grazed in pastures, and hence costs increased. Moreover the sheep frequently caught many diseases. Lack of sustainable management and maintenance. Projects have often neglected to build a sustainable system of management and maintenance of infrastructure. As a result some of the infrastructure has been ruined and could not be used soon after it was built. For instance, in D village of Sichuan Province, more than 17,000 Yuan had been provided by government to build the water system and some pools to collect water. The villagers rejected the village leader's idea of establishing water rates to employ one person to be in charge of the water system and clean and disinfect the pools. As the village has no collective funds, it has been unable to deal with this issue. Source: Cass-IOS (2006). The failure of the poorest households to benefit from public infrastructure investments could be due to several factors: they may lack capacity to take advantage of public infrastructure, they may have failed to participate actively in the planning process, or they may have been actively excluded by village elites from influencing the planning process. Field interviews suggest that poor households are often unable to pay for matching investments (often exceeding half of total costs) required to access new public infrastructure. For instance, for a water project, rural households had to contribute 3.3 and 3.6 times the amount of government funding for a water cabinet and a biomass pool project respectively (CDRF, 2007). A more detailed look at the variation of IVDP impacts further revealed that benefits were greater for both the rich and the poor in villages with higher quality village committees (e.g., that have more members, or conduct more frequent meetings). However, it was also found that the poorest households did even worse relative to less poor households in villages with highly educated village leaders, a finding consistent with elite capture of program benefits (Park and Wang, 2006). These findings are also consistent with studies of the impact of community-based development programs in other countries which find mixed results on the ability of such programs to successfully target and benefit the poor. Much depends on village governance and the willingness and ability of villagers, especially the poor, to participate fully in community decision-making processes (Box 5.13). Box 5.13 : Social funds and community-based development: international experience Community-based development is "an umbrella term for projects that actively include beneficiaries in their design and management" (Mansuri and Rao, 2004) that has emerged as a popular model for development assistance amidst growing recognition that sound governance and local accountability are critical for the success of public projects. For instance, using a broad definition, World Bank lending to community-based development projects increased from $2 billion in 1996 to $7 billion in 2003 (Mansuri and Rao, 2004). Today, government agencies, international donors, and NGOs support a wide array of programs throughout the world to promote community-based development using many different models. Many share elements in common with the Social Fund projects supported by the World Bank and other international organizations since the 1980s. Initially intended to alleviate the social consequences of structural adjustment programs, today social funds are multi-sectoral programs that provide financing (usually grants) for small-scale public investments and transfers targeted at meeting the needs of the poor and vulnerable communities, and at contributing to social capital and development at the local level. They serve as innovators and demonstrators of new methods of decentralized participatory decision-making, management, and accountability that may be adopted for broader application by public sector organizations (World Bank, 2002). Typically such projects award funds on a competitive basis after communities submit spending plans developed themselves through participatory processes (e.g., voting by community members). Only the best plans are supported, with funds transferred directly to the community and bypassing higher levels of government. The plans and processes should be publicly announced and implemented in a way that is highly transparent in order to maximize legitimacy and accountability. Evaluations of social funds and other community-based development projects thus far have found P r o m o t i n g o p p o r t u n i t y 8 8 that the programs have the potential to improve targeting of the poor and to increase the effectiveness of fund use by allocating funds to projects more valued by the community and by creating incentives for better project implementation and maintenance (for a review, see Mansuri and Rao, 2004). However, success is by no means guaranteed, with one evaluation study finding that overall the Bank's social fund projects have not been particularly successful in targeting the poor (World Bank, 2002). Successful design requires tapping into local needs, understanding and building on the strengths of existing institutions, and defining the changes needed in intermediary implementing agencies to support community action. The impact of decentralization as embodied in community-based programs depends critically on the susceptibility to capture by elites or other interest groups, which is particularly likely to be a problem in developing country settings where information and monitoring systems as well as local democracy or other forms of local accountability are weak (Platteau, 2004; Bardhan and Mookherjee, 2005). Inequality and other forms of social heterogeneity also may impact the dynamics of community decision-making. Thus, one of the lessons from the experience of Social Fund projects is that without adequate monitoring and careful attention to the local political, historical, and social contexts, community- based programs may not by very successful. Implementation is particularly difficult initially, with the most successful programs improving over time based on rigorous evaluations of earlier experiences. Early piloting of experimental approaches and gradual scaling up with great emphasis placed on training of facilitators are highly recommended. c. Raising Incomes through other developmental poverty reduction programs: training for labor transfer, subsidized credit and agribusiness Training for labor transfer Rural-urban migration is increasingly being recognized as a critical avenue for poverty reduction and new poverty alleviation initiative are aiming to make it easier for rural inhabitants to benefit from new job opportunities arising in China's towns and cities. The labor training program formally started in 2004 when the Leading Group Office issued a document on "Strengthening labor training for migrants in poor areas". Thirty-two Model Training Bases were established by the Leading Group Office by mid-2005 all over China, and data from 14 provinces and municipalities in central and western China indicate that by 2005 about 1500 training bases existed to support labor transfer at the provincial, prefecture and county levels. Training bases are publicly or privately run vocational schools that have already engaged in vocational training for some time and have close connections with employers in coastal regions and cities. The targeted trainees are laborers in key poor counties who are interested in working as migrants in other areas. The county Leading Group Offices are responsible for identifying laborers eligible for training with the help from township and village leaders. It is stipulated that laborers from poor households and poor villages are to be preferred, but in practice anyone from poor counties who is willing and able to participate in the training is accepted. The trainings usually last for 1-6 months. Each participant receives 600-1200 yuan subsidy according to the length of the training to cover their tuition and in most cases he/she needs to take care of his/her own boarding expenses in the school, which is about 1000 yuan per person. The training schools first need to sign a contract with the employers to guarantee the employment of trainees after the completion of training in order to apply for tuition subsides. Most of the subsidies come from the budgetary development fund. Training courses are organized according to market demand. Domestic service, manufacture of clothing, secretarial/typing services, waitressing and hospitality, security services, electronics assembly, machinery manufacturing and construction are identified by the Leading Group Office as occupations in great demand. According to the statistics of the Leading Group Office, 2.4 million laborers had been trained by 2004 in 14 central and western provinces, of which 65% were employed within the provinces and 28% outside the provinces. P r o m o t i n g o p p o r t u n i t y 8 9 Besides LGOPAD, a range of labor training programs have also been organized by the Ministries of Agriculture, Education, Science and Technology, Human Resources and Social Security, Housing and Urban Construction. However, a key challenge faced by the training programs is that laborers from poor households often have a lesser chance of participation because they are less likely to be able to afford the boarding or other expenses for the training. More subsidies for laborers from poor households should be considered, but accurately identifying poor laborers is not an easy task in part because of favoritism and rent-seeking on the part of local officials. Subsidized credit and agribusiness About half of China's poverty funds are subsidized loans which until recently have been managed by the ABC. To make use of subsidized loans, the LGOPAD has been promoting specialized agricultural industrialization through the development of dragon- head agricultural enterprises in poor areas. Key policies for the development of agribusiness include the provision of subsidized loans to dragon head agricultural enterprises, provision of budgetary grants to these enterprises for technical innovation and new brand promotion, and the exemption of income tax for a certain years and priority in land use. In 2005, 260 Poverty Reduction Dragon-head enterprises were identified by the Leading Group Office of the State Council and the ABC to enjoy the above favorable policy. Hundreds of local Poverty Reduction Dragon-head enterprises were also nominated by the provincial Leading Group Offices. The majority of these enterprises are privately owned. Statistics from 11 provinces42 show that a total amount of 1.7 billion yuan were spent for the development of agribusiness in 2004, of which 62% are from subsidized loans and 19% from budgetary development funds. Some provinces use most of their subsidized loan for agribusiness, for example, 8 billion yuan of cumulated subsidized loans have been allocated to agribusiness in Hubei, accounting for 70% of the total subsidized loans in the province. The development of agribusiness in some regions has successfully promoted agricultural production and increased household incomes in the production base areas. For instance, Shunchang Company, a poverty reduction dragon-head enterprise in Jiangxi province, has established 20 production bases involving 120,000 households, and 15,000 poor households are estimated to have benefited from agricultural production and processing related to this company. The development of agribusiness has also helped increase fiscal revenue of the local governments, and the ABC has successfully reduced its non- performing subsidized loans by diverting its lending to relatively big agricultural enterprises. Agricultural industrialization projects now absorb the biggest share of interest-subsidized loans accounting for about a third of all subsidized lending by the ABC in 2005. However, there is evidence that local branches of the ABC are often reluctant to make subsidized loans to individual households in poor areas, undermining that part of the program (Box 5.14 ). With the ABC's drive for greater commercialization since 2001, its lending to rural households has suffered with the latter comprising only about 17% of the total subsidized loans in 2005. Box 5.14 : China's subsidized loan program for poverty alleviation International experience indicates that subsidized loan programs similar to the Chinese program generally fail to reach the poor because of rent-seeking activity by nonpoor groups. In China, local government leaders and bank officials have strong incentives to direct subsidized loans to tax- generating industrial projects or to borrowers with high repayment probability or low transaction costs. Since the start of the program, this has led to significant conflicts between local branches of the Agricultural Bank of China and local LGOPADs. Repayment rates on subsidized loans have been very low, indicating that the loans are often viewed as social grants rather than credit, _________________________________________________________ 42These eleven provinces include Heinongjiang, Liaoning, Hebei, Shanxi, Shandong, Jiangxi, Hubei, Hunan, Guangxi, Gansu and Xinjiang. P r o m o t i n g o p p o r t u n i t y 9 0 undermining the financial sustainability of ABC branches. Reported timely repayment rate in the mid-1990s was about 50%. In the late 1990s, following the example of successful microfinance pilot programs in China, the Chinese government initiated microfinance programs on a larger scale, featuring Grameen-style borrower groups. However, the government-run programs were not well-targeted to the poor, and failed to achieve high repayment rates (Park and Ren, 2001). According to the World Bank-NBS Special Purpose Village-Level Survey, in 2004, 1.4% of rural households in China had received poverty loans, including 4% of households in poor counties and 4.7% of households in poor villages. Thus, subsidized loans are not particularly well-targeted to poor villages, consistent with field observations that subsidized loans are generally not coordinated with the IVDP. The average loan size for households receiving loans was 1898 Yuan (in 2003 prices). For the most part, subsidized poverty loans are allocated to agro-industrial enterprises rather than households. The ABC, the agency that has been in charge of subsided loans, has generally ignored the IVDP and has not made any loans based on the IVDP village plans. While the shift of lending from rural households to economic entities and enterprises has helped improve the overall quality of the ABC's subsidized credit portfolio (the proportion of non-performing loans has declined) and has enhanced revenues of local governments, its impact on the poor has been limited. The main problem of this approach to poverty reduction is that the link between the development of agribusiness and the poor households is weak. There is no mechanism to ensure that poor households are the preferred ones to benefit from the agribusiness development. On the contrary, in most cases, rich households with large-scale specialized production capacity are contracted by the agribusiness companies in order to save management costs and to ensure a large supply of agricultural products. At the same time, the proportion of non-performing loans remains the highest (at upwards of 70%) for loans to rural households in ABC's portfolio, and there is a real tension between the efficiency of its operation as a commercial bank and aspirations of poverty reduction as the key goal of subsidized credit. d. How can developmental poverty reduction programs better help the poor? To sum up, the IVDP program has been important in introducing a participatory approach to planning of developmental projects, in increasing investments at the village level, and in adopting integrated development by combining projects for production, infrastructure, social services, training, and cultural enhancement. Even as poverty is getting more dispersed and village-level concentration of poverty is declining (chapter 2), there is a role for area-based investment programs in alleviating rural poverty (as suggested by the taxonomy of the poor in chapter 4). However, the above review highlights two key issues in relation to the effectiveness of village investment programs in alleviating rural poverty: their poor targeting performance (a large fraction of the poor simply do not live in designated poor villages), and limited impact of investments on incomes of the poor within designated villages (reflecting the relatively small scale of investments and a number of factors linked to limited participation of villagers, especially the poor, in the planning and implementation of projects). A number of steps could be taken to improve the poverty reduction impacts of IVDP, including the following: · Develop consistent, data-based criteria for selecting poor villages and a system for updating poor village designations; privilege poorest villages for earlier and larger investments. · Increase the size of village investments. This may require higher funding for the budgetary grant program, but the effective size of village investments could also be increased through improved coordination of fund disbursement for village plans by designing clear rules to guide the aligning of Food-for-Work investments and subsidized credit with village plans. There is also scope for releasing more funds per village by targeting fewer villages. · Develop institutional mechanisms to improve coordination of village investments with other line agencies. Since the larger part of investments in rural areas comes not through the poverty alleviation funds but through the budgets of other line agencies, the needs for better coordination of village-level investments goes well beyond the P r o m o t i n g o p p o r t u n i t y 9 1 poverty alleviation funds themselves. The current institutional mechanism for inter- agency coordination through the local Poverty Alleviation and Development offices works only imperfectly, and needs considerable strengthening to be effective. · Place new emphasis on household-level targeting, as even with better identification of poor villages, it will be important to reach the poor within these villages. This can be done in several ways: by changing the mix of projects to include more household- oriented interventions, for instance, labor training to promote off-farm employment; by waiving the requirement for complementary investments by poor households or providing subsidized credit for this purpose; by developing new programs such as self-targeted public works programs or conditional cash transfers which could offer assistance to poor households conditional on other objectives, for instance, ensuring that their children do not drop out of school. · Conduct additional study of the village planning process to identify ways to increase participation of the poor, experiment with new methods of implementing village plans. Apart from the IVDP, the subsidized credit program, though it accounts for about half of all poverty alleviation funds, is not set up in a way to deliver significant benefits to the poor as its main focus has been to support large agribusiness enterprises. And participation of the poor in the training programs for labor transfer has also been limited by the relatively high costs of participation. To make them more effective instruments for rural poverty alleviation, further reform of these programs will be needed, for which the following areas merit consideration: · The subsidized credit program needs to be integrated with the village plans. New institutional arrangements will be necessary for this to happen; simply exhorting different agencies to coordinate better will not help. It also needs to be recognized that village-level coordination of funds and program management is unlikely to succeed without coordination at the provincial and county levels. There should thus be a single agency empowered to ensure this coordination at the provincial and county level. · There is a need to redirect subsidized credit from large agribusiness enterprises (which ought to be supported through the commercial banking networks) to poor farmers, in part through provision of cheap loans for complementary investments by them to better utilize the public goods created by village investments. · In relation to the labor transfer programs, boarding and other expenses should be waived for poor farmers to ensure their enhanced participation. Two more general points about developmental poverty alleviation programs also should be noted. · Even with perfect village-level targeting, many of the poor will be left out (due to the increasingly dispersed nature of poverty) and many of the benefits of investments in targeted villages will flow to the non-poor in these villages. There are thus limits to further gains from area-based targeting and it will be important to supplement area targeting with household targeting. But new mechanisms will need to be developed to target poor households, and in this regard approaches such as proxy-means targeting deserve to be actively explored. · If, as suggested above, more household-oriented interventions are to be adopted as part of these developmental poverty alleviation programs, then the issue of coordination with social assistance and social insurance programs, which also provide critical support at the household-level, will become increasingly important. For instance, there will thus be a need for a common approach to identifying poor households by both the developmental and social protection programs, which raises the stakes of inter-agency coordination not just to the Poverty Alleviation Offices, Departments of Finance and Development Reform Commission and the ABC (or Rural Credit Cooperatives), but also the Departments of Civil Affairs and Labor and Social Security. The development of a coordinating mechanism for this related set of poverty reduction interventions remains a big institutional challenge. P r o m o t i n g o p p o r t u n i t y 9 2 e. Rural-to-urban migration as a path out of poverty One of the most significant developments associated with China's robust economic growth over the last two decades has been the large-scale movement of people to the cities and towns in response to the rapid rise in labor demand in urban areas where this growth has been concentrated. While the exact size of migrant population in urban areas remains debatable, recent estimates suggest the number could be as high as 150 million (see section 4a and further discussion below) making this the largest peacetime movement of people in world history. The process of urbanization has allowed China to benefit from the economies of agglomeration that have been an important source of productivity gains.43 While it has been claimed that China's cities are generally undersized in being below the levels that would maximize productivity per worker (Au and Henderson, 2006), the scale and pattern of migration has had profound consequences for both destination and source areas. This section focuses on implications for source areas. In particular, rural-to-urban migration has been hailed as an important pathway for rural poverty reduction. In the following, an attempt is made to assess this contribution of migration focusing in particular on whether some of the poorest amongst the rural population are able to exploit this pathway out of poverty. Large and growing scale of migration though full-family migration remains limited The exact scale of internal migration in China remains a subject of considerable debate. Estimates differ depending upon the source of data and how migrants are defined in terms of geographic scope, the duration of stay outside the source area and whether the movement was accompanied by a change in hukou status. Another point of distinction is between the stock of migrants at a point in time and the flow of migrants over a period of time. For instance, according to the 2000 Census, the size (stock) of the floating population ­ defined as those who have been away from their place of permanent household registration (hukou) for more than 6 months ­ was 144 million. 44 More recently, the Second National Agricultural Census puts the number of rural migrant workers at 131.8 million at the end of 2006.45 On the other hand, the 5-year flow measure of migrants from the census ­ defined as those who moved to their current township or urban district during 1995-2000 and have lived there for at least 6 months over the preceding year ­ is about 131 million, of which 85 million (about 65%) were non-hukou migrants i.e. those without full hukou in their current place of residence (Cai, Park and Zhao, 2006). Of the non-hukou migrants during this period, almost 42 million or about half were rural-to-urban migrants. As noted earlier in section 2g, the scale of migration from rural areas has also increased rapidly during the 1990s. Estimates from surveys of rural labor conducted by the Ministry of Agriculture (MOA) and the Ministry of Labor and Social Security (MOLSS) are indicative of this increasing trend. Counting migrants as those migrating for at least one month during the year, these surveys suggest that the number of migrants from rural areas increased more than 2.5 times between 1997 and 2004 (Figure 2.41). Estimates from the National Bureau of Statistics confirm the continuing rise in the number of rural migrants. The most recent estimate of the number of migrants (including family migrants) puts the number at 137 million in 2007, an increase by more than 60% over the six years since 2001 (Figure 5.55). _________________________________________________________ 43Between 1985 and 2005, China's urban population rose from 251 to 562 million or from about 24% to 43% of the total population (NBS, 2006). 44For 2005, the stock of floating population is estimated at 147 million according to the 1% National Population Sample Survey (NBS, 2006). 45Rural migrant workers are defined as employed members of rural households who engaged in work for one month or longer outside the subdivision or district of local town or township under the administrative jurisdiction in 2006. P r o m o t i n g o p p o r t u n i t y 9 3 Figure 5.55 : Rising tide of rural migrants, 2001-2007 (Number of rural migrants, million) 137.0 140 132.1 125.8 118.2 120 113.9 104.7 100 84.0 80 2001 2002 2003 2004 2005 2006 2007 Sources and notes: National Bureau of Statistics, China Yearbook of Rural Household Survey (various years), China Statistics Press. Perhaps the most important source of information on migrants from rural areas is the NBS' recent Rural Household Surveys (RHS) which started collecting information on migrants since 2002. The following discussion makes extensive use of the RHS data on migrants. Using this source, Table 5.46 presents some basic indicators on the scale of migration of rural workers for 2003. Table 5.46 : Basic information on migration from rural areas in China, 2003 Basic facts about the number of migrants from rural areas Number % of rural % of rural (million) labor population A Individual migrant workers who worked outside their home township for more than 6 months 98.9 15.3 10.5 B Individual migrant workers who worked outside their home township for any length of time 125.8 19.5 13.4 C Family migrants from rural areas 24.3 2.6 D = 0.5 x C Family migrant workers from rural areas 12.2 1.9 E = B + C Total number of migrants from rural areas 150.1 16.0 F = B + D Total number of migrant workers from rural areas 138.0 21.4 Basic facts about income from migrant work G Average monthly income of a migrant worker (yuan) 702 H Average number of months of migrant work (months) 8.1 I = G x H Average annual income of a migrant worker (yuan) 5686 J Average annual rural income per person (yuan) 2622 K = I / J Ratio of migrant worker income to per capita rural income (ratio) 2.17 L = B x (I/J) Share of migrant income in total rural income (%) 29.1 Sources and notes: World Bank estimates based on data from NBS' 2003 Rural Household Survey and Sheng and Peng (2005). Rural labor is defined as those aged 16 and above who are not students and who have work capacity. By this definition, for 2003 the rural labor force is about 646 million out of a total rural population of 937.5 million. (The latter measure of rural population is consistent with how rural households are defined in the NBS Rural Household Survey.) Note that this measure of the rural labor force is similar to the 640 million estimate of able-bodied and semi-able bodied laborers implied by the numbers reported in the 2005 China Yearbook: Rural Household Survey (NBS, 2005a). It is assumed that only half of family migrants are workers. As shown in Table 5.46, for 2003 the total number or individual migrant workers from rural areas who worked outside the township for at least 6 months is about 99 million, while those who worked outside the township for any length of time is about 126 million. These figures however refer to individual migrant workers only and do not include those where the entire family migrated outside the township. The number of the latter is estimated at 24.3 million for 2003 (Sheng and Peng, 2005). The total number of migrants from rural P r o m o t i n g o p p o r t u n i t y 9 4 areas is thus estimated at about 150 million which amounts to about 16% of the rural population. Assuming that half of the family migrants are workers46, the total number of migrant workers is estimated at about 138 million or about 21% of the rural work force. Using the stricter 6-month criterion, migrants account for 13 and 17 percent respectively of the rural population and work force. Thus, the scale of migration is large: roughly one in every five rural workers is a migrant worker. About 43% of the rural population lives in households that have one or more migrant workers. Going by the data presented in Table 5.46, it appears that the share of workers who migrate with families in the total number of migrant workers is relatively small ­ less than 10%. Further information on the extent of family out-migration is available from the 2004 World Bank-NBS Special Purpose Village-level Survey, which indicates that while more than three-quarters of villages report some family out-migration during the five-year period 2000-2004, only about 5% of the households were involved in such migration, or an annual rate of just over 1% of households (Table 5.47). The vast majority of migrant workers migrate as individual workers leaving their families behind. The limited incidence of family as opposed to individual migration was also noted in the qualitative poverty assessment in 12 study villages. For instance, fieldwork in Jiangxi province indicated that family migration happens only when migrants have secure jobs or can run their own businesses (CASS-IOS, 2006). Table 5.47 : The extent of family out-migration from rural areas, 2000-2004 % of villages reporting family out-migration between 2000 and 2004 76.5 In villages reporting family out-migration: Current total number of households in the village 355.8 Number of family migration households 20.5 Share of family migration houeholds in total number of households (%) 5.4 Number of family migration households whose land was taken away 6.2 Share of family migration households whose land was taken away (%) 30.2 Sources and notes: World Bank estimates based on the 2004 World Bank-NBS Special Purpose Survey of 3037 villages. Another measure of the overall significance of migration is given by the estimated share of income from migrant work in total rural income. As seen in Table 5.46, the mean annual income of a migrant worker is more than twice the mean rural per capita income, even taking account of the fact that, on average, a migrant worker works outside the township for about 8 months in a year. This large income premium translates into a high share of migrant income in total rural income; this share is estimated to be nearly 30% for 2003 (Table 5.46). Most migration is inter-provincial, from central and southwest to coastal areas, and to medium and large cities Data from the 2003 RHS and the Second National Agricultural Census for 2006 indicate that about half of all migrant workers work outside their home province. About another one-third work outside their county but within the province and the remaining one-sixth work outside their township but within the same county (Table 5.48). Thus, migration over long distances seems to be the norm. _________________________________________________________ 46The proportion of half is motivated by the fact that the share of able-bodied workers in total rural population for 2003 is about 48% (NBS, 2005a China Yearbook: Rural Household Survey). P r o m o t i n g o p p o r t u n i t y 9 5 Table 5.48 : Main working place of rural workers, by region, 2003 and 2006 2003 2006 North- South- North- China Coastal east Central west west China % of rural labor force migrating for work outside township 16.8 15.0 7.3 19.7 19.4 12.5 24.8 Of those migrating for work: % migrating out of township but within county 16.7 27.3 24.1 8.7 12.9 26.4 19.2 % migrating out of county but within province 31.6 54.5 56.9 17.3 20.8 35.7 31.5 % migrating out of province 51.8 18.3 18.9 74.1 66.2 37.9 49.3 Sources and notes: World Bank estimates from national sample of the NBS 2003 Rural Household Survey. The 2003 numbers in the table are derived from answer to the survey question "Where was you main place of work?" with possible answers: (1) within township, (2) out of township but within county, (3) out of county but within province, (4) out of province but within country, and (5) out of the country or in Hong Kong, Macao and Taiwan. The last category, which is extremely small, is omitted and responses (2), (3) and (4) are treated as migration. The numbers for 2006 are from NBS (2008) Communique on Major Data of the Second National Census of China (no. 5). The definition of migrant workers includes those working outside the township for one month or longer. However, there are regional variations. The two regions that had the highest migration rates, central and southwest, also had the highest proportions of out-of-province migrants (74 and 66 percent respectively). By contrast, in the northeastern region, which has the lowest migration rate, the proportion of migrants moving across provinces was also amongst the lowest at about 19%. The share of out-of-province migration is equally low for the coastal region (about 18%) despite a moderately high overall rate of migration, which mainly reflects proximity to the main centers of labor demand along the coastal belt. In both the coastal and northeastern regions, most migrants moved across counties but within the same province. In the northwest with a relatively low migration rate, migrants are evenly distributed between across county and across province movers. Amongst the regions, the largest share of migrants (about 37%) comes from central China provinces, while the coastal and the southwest regions each account for about a quarter (Figure 5.56). The northwest and the northeast account for relatively low shares (8% and 4% respectively). On the other hand, the chosen destination of nearly two-thirds of all migrants (including those moving within the same province) is the coastal region, with relatively small shares going to other regions. Figure 5.56 : Sources and destinations of migrant workers from rural areas, 2003 (% of all migrant workers from rural areas who worked for any period of time outside their home township) Where do migrants from rural areas come from? Where do migrants from rural areas go? 7.2% 7.7% from 10.6% to NW to NW 23.5% 28.2% SW from from south- coastal 12.6% to west central 65.5% to 3.6% from 36.9% coastal NE from 4.1% central to NE Sources and notes: World Bank estimates from national sample of the NBS 2003 Rural Household Survey. NE is the northeast region, and NW is the northwest region. The lists of provinces in each of the regions are given in Box 4.10. P r o m o t i n g o p p o r t u n i t y 9 6 The variation in destinations by source regions is quite revealing. Nearly all the migrants from the coastal region (even if they move across provinces) go to the coastal region itself (Figure 5.57 and Table 5.49). Nearly two-thirds of migrants from the central region and half of those from the southwest also move to the coastal provinces. For the northeast and the northwest, the coastal region is not an important destination; in their case, most of the movement is within the region itself. Figure 5.57 : Destinations by source of migrant workers from rural areas in 2003 NW to NW: F rom C O 6.0% to CO: F rom SW 27.0% to SW: Fro m 9.9% NW: 7.7% Fro m Fro m CO Fro m SW SW: 28.2% to CO: 23.5% 12.5% F rom C E: F rom N E 36.9% Fro m to NE: 3.2% CE to CE: 11.2% From CE to CO: 24.3% Sources and notes: World Bank estimates from national sample of the NBS 2003 Rural Household Survey. CO is the coastal region, NE, the northeast region, CE, the central region, SW, the southwest region, and NW, the northwest region. For lists of provinces in each of the regions, see Box 4.10. Table 5.49 : Destinations of migrants by source region, 2003 Total migration (both within and across provinces) To: From: Coastal Northeast Central Southwest Northwest Total Coastal 95.9 1.2 1.9 0.5 0.5 100.0 Northeast 9.1 88.6 1.7 0.2 0.4 100.0 Central 66.0 1.1 30.4 1.2 1.4 100.0 Southwest 53.1 0.4 2.0 42.3 2.3 100.0 Northwest 16.9 0.3 4.1 1.2 77.6 100.0 All China 65.5 4.1 12.6 10.6 7.2 100.0 Inter-provincial migration To: From: Coastal Northeast Central Southwest Northwest Total Coastal 79.3 5.8 9.8 2.4 2.7 100.0 Northeast 48.7 40.0 8.7 0.7 1.8 100.0 Central 90.8 2.0 4.0 1.6 1.7 100.0 Southwest 84.5 0.5 3.2 8.6 3.3 100.0 Northwest 48.1 0.9 12.3 4.2 34.5 100.0 All China 84.7 2.5 4.9 3.8 4.1 100.0 Sources and notes: World Bank estimates from national sample of the NBS 2003 Rural Household Survey. For lists of provinces in each of the regions, see Box 4.10. Overall, the coastal areas are clearly the dominant destination for migrant workers as is also borne out by the major inter-provincial migration flows depicted in Figure 5.58. It is estimated that more than a quarter of all migrant labor (about 26 million workers) works in Guandong alone, and both the absolute number and the proportion of migrant labor working in the eastern coastal region are still increasing (Sheng and Peng, 2005). P r o m o t i n g o p p o r t u n i t y 9 7 Figure 5.58 : The 30 largest inter-provincial migration flows, 1995-2000 Sources and notes: Chan (2006). More than 60% of rural migrant laborers work in medium-large cities at prefecture level or above (Figure 5.60), and this share has been increasing in recent years. During 2001-04, the share of medium-large cities increased from 57.2% to 62.4% (Sheng and Peng, 2005). Figure 5.59 : Two-thirds of the rural migrants went to medium and large cities in 2004 Others Townships City provinces 5.7% 11.4% 9.6% County level Prefecture level Provincial cities cities capitals 20.5% 34.3% 18.5% 0% 20% 40% 60% 80% 100% Sources and notes: Sheng and Peng (2005). The majority of migrant workers are employed in the manufacturing (30%) and construction (23%) sectors; the remaining are mostly employed in various kinds of services (Figure 5.60). Figure 5.60 : The majority of rural migrants worked in manufacturing and construction in 2004 Transport 3.4% y litaitps Other 7% Construction Manufacturing 6.services: 22.9% 30.3% Ho 10.4% Mining Wholesale 1.8% & retail 4.6% 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% Sources and notes: Sheng and Peng (2005). P r o m o t i n g o p p o r t u n i t y 9 8 Male young adults with junior high school education are more likely to migrate A little over one-third of the migrant workers are women (Table 5.50). There is a large gap in the migration rates for male and female rural laborers which are about 25 and 14 percent respectively. Male laborers are thus about 1.8 more likely to migrate than female laborers. Table 5.50 : Gender, age and education level of migrant workers, 2003 and 2006 2003 2006 Non- Female Male All migrants Migrants migrants migrants Migrants Share of rural workers or migrants (%) 100.0 80.6 19.4 33.8 66.2 24.8 Fraction female (%) 50.8 33.8 100.0 0.0 36.0 Age (years) 38.7 41.0 29.1 25.4 31.0 Average schooling (years) 7.9 7.6 8.9 8.6 9.0 Fraction illiterate (%) 7.0 8.3 1.6 2.4 1.2 1.2 Fraction with primary school (%) 29.8 33.0 16.2 18.8 14.9 18.7 Fraction with junior high school (%) 50.8 46.9 67.3 67.1 67.4 70.1 Fraction with senior high school (%) 11.9 11.3 14.0 10.9 15.6 8.7 Fraction with college and above (%) 0.6 0.5 0.9 0.8 0.9 1.3 Sources and notes: World Bank estimates from national sample of the NBS 2003 Rural Household Survey. The numbers for 2006 are from NBS (2008) Communique on Major Data of the Second National Census of China (no. 5). The definition of migrant workers includes those working outside the township for one month or longer. About 47% of migrant workers are young adults between 16 and 25 years of age. The average age of migrant workers is 29 years, 12 years less than the average age of non- migrants, and female migrant workers are on average six years younger than their male counterparts. Migration rates decline steadily with age for both male and female workers, although the decline is much steeper for women than for men. As seen in Figure 5.61, most of the difference between male and female migration rates is on account of much fewer women migrating within the age groups 25-35 and 35-44 years. Figure 5.61 : Male and female migration rates by age, 2003 ) %( 50 46.0 e 40 39.3 Male rat 42.2 oni 30 24.8 grati 20 17.4 23.3 m 13.7 10 10.2 Female 5.8 1.6 4.0 Labor 0 0.5 16 - 25 25 - 35 35 - 45 45 - 55 55 & above Age (in years) Sources and notes: World Bank estimates from national sample of the NBS 2003 Rural Household Survey. The dotted horizontal lines indicate the average migration rates for male and female rural labor. Migrant workers on average have more years of schooling (8.9 years) relative to rural workers who do not migrate (7.6 years). About 80% of all migrant workers have junior high school or higher level of education (Table 5.50). About 70% are in the junior high school category alone. The proportion is similar for both male and female migrant workers, and reflects employer preferences to hire those with some minimum level of education even as most of the migrant workers are engaged in relatively low-skill tasks. Nearly 72% of migrant workers have not received any kind of skill training, although three- quarters of those without training expressed willingness to participate in training (Sheng and Peng, 2005). Of the few who did receive training, only 10% participated in training organized through government organizations; the rest relied either on training organized by enterprises (30%) or more commonly they participated in training through their own P r o m o t i n g o p p o r t u n i t y 9 9 initiative (about 60%). There is evidence that migration rates rise with the level of education for both men and women with the rates for women rising faster (Figure 5.61). Figure 5.62 : Male and female migration rates by education level, 2003 30 29.2 29.7 (%) 24.0 e 28.7 24.8 rat 20 oni Male 15.9 20.8 20.7 at 13.7 gr mi 10 10.8 7.0 Female 2.8 Labor 0 Illiterate Primary Junior high Senior high Some college school school school and above Sources and notes: World Bank estimates from national sample of the NBS 2003 Rural Household Survey. The dotted horizontal lines indicate the average migration rates for male and female rural labor. Migration reduces poverty While migration to urban areas has been widely regarded as a path out of poverty for rural households, it is difficult to empirically estimate the contribution of migration to poverty reduction. There are two main difficulties in establishing this contribution. First, migration affects poverty through multiple channels. There are, firstly, the direct gains to the migrant workers themselves through the possibility of earning significantly higher incomes. These income gains are also shared with other family members back in rural areas through remittances. Beyond these direct gains, there are also potential knock-on effects insofar as higher incomes generated by migration further induce productive investments in the rural economy including further investment in human capital (in part by relieving the credit constraints faced by rural households). Migration is also likely to affect wage rates in rural (and urban) labor markets. The second difficulty arises from the fact that the ability to migrate may itself depend on the poverty status of households. As already hinted by the characteristics of migrant workers (and as further discussed below), many of the poor may not be able to make use of migration opportunities due to their limiting circumstances and endowments. Notwithstanding these difficulties, there are, however, several pieces of evidence that give an indication of how much migration may be contributing to poverty reduction. First, evidence from the NBS Rural Household Surveys indicates that households with migrant workers are less likely to be poor than households without any migrant workers. For instance, for the World Bank poverty line, the income poverty rate for households with migrants is 7.5% in contrast to 10.9% for households without migrants (Table 5.51). Nearly two-thirds of the rural poor live in households without a migrant worker, much larger than their 57% share in rural population. Table 5.51 : Rural households with migrant workers are less likely to be poor, 2003 Income poverty using the Share of WB poverty line rural % who are Share of population poor rural poor Rural households with no migrant workers 57.4 10.9 66.2 Rural households with migrant worker(s) 42.6 7.5 33.8 Total 100.0 6.9 100.0 Sources and notes: World Bank estimates from national sample of the NBS 2003 Rural Household Survey using the World Bank poverty line. However, households with migrants differ from households without migrants in many respects. On controlling for a number of household and area characteristics, it is P r o m o t i n g o p p o r t u n i t y 1 0 0 estimated that having a migrant worker in the household increases the household's per capita income and consumption by 8.7 and 6.1 percent respectively.47 The effects of migration on the probability of a rural household being poor are depicted in Figure 5.63. As seen in the Figure, migration reduces the probability of being poor. For instance, a household with a migrant worker that has a 50% probability of poverty would have faced a 57% probability of being poor in the absence of migration. The effects are of similar order of magnitude for income and consumption poverty. Figure 5.63 : Migration reduced the probability of poverty for rural households, 2003 Likelihood of income poverty(%) Likelihood of consumption poverty(%) Without migrant With migrant Without migrant With migrant 93% 93% re 85% re 84% ytrevop 76% 75% work 67% 90% ytre work 66% 90% nta 57% 80% nta 57% 80% 70% pov 70% of gri 47% of gri 46% m 37% 60% m 36% 60% ihoodle 26% 50% 25% 50% hout 14% 40% lihoode hout 40% 13% Lik 30% 30% h/wit Lik h/wit 20% wit10% 20% wit10% 10% 20% 30% 40% 50% 60% 70% 80% 90% 10% 20% 30% 40% 50% 60% 70% 80% 90% Likelihood of poverty with migrant worker Likelihood of poverty with migrant worker Sources and notes: World Bank estimates from national sample of the NBS 2003 Rural Household Survey. The estimates are based on household-level regressions of an indicator variable for whether a household is poor on whether it had a migrant worker during the previous year, while controlling for number of household, village and area characteristics. Poverty in measured with reference to the World bank poverty line. The results can also be used to obtain a rough estimate of the overall contribution of migration to rural poverty reduction. As noted in Table 5.51, rural households with migrants have a 7.5% probability of being income poor. In the absence of migration, this probability would rise to 10.5%. This translates into an additional 11.8 million persons being poor in 2003, or an increment in the total number of rural poor by 13%. This measure, however, is likely to be an underestimate as it does not take into account any of the positive indirect effects of migration through productive investments supported by remittances or through induced effects on rural wages. But many of the poor are unable to migrate While migration is an important channel for reducing rural poverty, its impact on poverty is limited by the fact that many of the poor do not migrate. As already noted in Table 5.51, two-thirds of the poor live in households that do not have any migrant workers. Figure 5.64 shows the relationship between the probability of the household having a migrant worker and the household's predicted log per capita income in the absence of migration. As evident from the Figure, the migration probability-income relationship has an inverted-U shape.48 Both the poorest and richest are less likely to migrate although for different reasons. Migration probability is around one-third for households below the World Bank _________________________________________________________ 47These estimates are similar to those in by Du, Park and Wang (2005) who report an effect on per capita incomes of between 8.5 and 13.1percent using longitudinal data from the China Poverty Monitoring Survey for 1997-2001. 48A similar inverted-U shaped relationship was also estimated by Du, Park and Wang (2005) for 1997 and 2001. P r o m o t i n g o p p o r t u n i t y 1 0 1 poverty line , it reaches a peak of around 0.41 at about three times the World Bank poverty line and declines thereafter. Figure 5.64 : The relationship between migration probability and log income per capita at the household level, 2003 0.44 Twice WB line Thrice WB line WB line holde rker 0.40 wot hous y anrg 0.36 ilitb mi a Proba ash 0.32 0.28 5.5 6 6.5 7 7.5 8 8.5 9 9.5 Predicted log per-capita income of household (without migration) Source and notes: World Bank estimates from national sample of the NBS 2003 Rural Household Survey. The graph shows the non-parametric relationship between the probability of a household having a migrant worker and the estimated log per capita income of the household in the absence of migration (with a 1% trip of the sample at either end). The latter is estimated from a regression model of log per capita income as a function of whether the household had a migrant worker in the previous year and a number of household, village and area characteristics. The relationship between migration probability and the actual log per capita income of the household is very similar. The different endowments and opportunities of the poor and the non-poor households in rural areas are also reflected in their occupational choices. Table 5.52 presents statistics on work activity of labor from poor and non-poor rural households distinguishing three main categories of work: farming, local non-farm work and migrant work. Here, as well as the subsequent analysis of migration and poverty, poverty is defined as consumption less than the World Bank poverty line. Two alternative measures are used. The first measure is by participation ­ any work outside the township for one month or more is considered migration; among non-migrants, any local non-farm work of one month or more is considered local non-farm participation. The second measure is by main occupation defined as the occupation a worker spent the most months in. By either measure, the proportion of labor from poor households engaging in non-farm activities, be it migration or local non-farm work, is lower than that for non-poor households. For instance, the vast majority (78%) of labor from poor households make a living by farming as their main occupation. Only 7% and 15% of them have local non-farm and migrant work as their main activity. By contrast, farm work is the main activity for 64% of labor from non-poor households while non-farm work and migration accounted for 18% each. Clearly, non- farm avenues of employment seem to be significantly more limited for the rural poor. P r o m o t i n g o p p o r t u n i t y 1 0 2 Table 5.52 : Distribution of rural workers among farm, local non-farm and migration activities, by poverty status, 2003 Poor Non-poor % of rural workers... households households ...who participate in... ...only farmwork 53.0 44.8 ...some local non-farm work but no migrant work 29.0 35.5 ...some migrant work 18.1 19.7 ...whose main activity is... Farm-work 78.1 64.1 Local non-farm work 7.2 18.0 Migrant work 14.7 17.9 % of rural migrant workers... ...who migrate within the province 41.8 51.1 ...who migrate outside the province 58.2 48.9 Sources and notes: World Bank estimates from national sample of the NBS 2003 Rural Household Survey. Poor households are defined as those with per capita consumption less than the World Bank poverty line . Table 5.52 also shows that the share of non-farm work by the main activity measure is much smaller than by the participation measure, especially for poor households. This is suggestive of the temporary or seasonal nature of available non-farm work that particularly affects labor from poor households. The scarcity of proximate off-farm work opportunities is also reflected in a higher proportion of migrants (58%) from poor households migrating to another province relative to those from non-poor households (49%). Limited local off-farm employment opportunities may also explain differing patterns of work amongst laborers from poor households in different regions. As seen in Table 5.53, in the central and western regions where local off-farm opportunities are relatively less abundant relative to the coastal region, the poor are many times more likely to migrate than to engage in local non-farm as their main occupation. Table 5.53 : Work activity of rural workers from poor households in different regions, 2003 % of rural workers from poor households... North- South- North- Coastal east Central west west ...who participate in... ...only farmwork 55.5 70.0 53.0 49.4 51.9 ...some local non-farm work but no migrant work 29.4 21.8 25.1 33.9 28.8 ...some migrant work 15.1 8.1 21.9 16.7 19.4 ...whose main activity is... Farm-work 75.7 84.8 74.3 81.5 79.4 Local non-farm work 12.0 8.9 7.2 4.2 7.2 Migrant work 12.2 6.3 18.5 14.3 13.5 Sources and notes: World Bank estimates from national sample of the NBS 2003 Rural Household Survey. Poor households are defined as those with per capita consumption less than the World Bank poverty line . One of the factors limiting migration possibilities by the poor is their relatively limited level of education. As already noted, more than 80% of all migrant workers have junior high school of higher level of education relative to 59% for non-migrant workers. However, the education levels of workers from poor households without migrants are lower still. Amongst them, only 50% have junior high school or higher education (Table 5.54). Similarly, the average years of schooling for migrant workers, all non-migrant workers and workers from poor non-migrant households are 8.9, 7.6 and 6.9 years respectively. The limiting effects of lower human capital on migration prospects are further discussed below. P r o m o t i n g o p p o r t u n i t y 1 0 3 Table 5.54 : Non- migrant workers from poor households without migrants, 2003 Workers from poor Migrant Non- migrant households without workers workers migrants Mean age 29.1 41.0 39.2 Fraction male (%) 66.2 48.4 48.5 Mean years of schooling 8.9 7.6 6.9 Fraction illiterate (%) 1.6 8.3 14.0 Fraction completed primary school (%) 16.2 33.0 36.4 Fraction completed junior high school (%) 67.3 46.8 42.6 Fraction completed senior high school (%) 14.0 11.3 6.8 Fraction with some college and above (%) 0.9 0.5 0.2 Sources and notes: World Bank estimates from national sample of the NBS 2003 Rural Household Survey. Poor households are defined as those with per capita consumption less than the World Bank poverty line . There remain several impediments to migration Further analysis of the data from the NBS 2003 Rural Household Survey sheds more light on some of the incentive for and impediments to migration from rural areas. This analysis focuses on the determinants of the probability of a household having a migrant worker. The analysis explores the effects of a large number of household, village, county and provincial characteristics. Table 5.55 presents the key results for some select variables only. Table 5.55 : Select factors influencing migration from rural areas, 2003 Marginal effects on the probability of a household having a migrant worker Household characteristics Ratio of labor members with junior high education 0.041** Ratio of labor members with senior high/college college education -0.016 Land area per labor member -0.010** Ratio of children age 0-5 years -0.218** Ratio of children age 6-12 years -0.118** Village, county and provincial characteristics Ratio of TVE workers in the village -0.007** If village in mountainous areas 0.064** If minority village -0.169** Agricultural productivity in the county 0.304** Urban-rural provincial per capita income ratio weighted by distance 0.078** Standardized index of urban labor market segmentation weighted by distance -0.042** Sources and notes: World Bank estimates from national sample of the NBS 2003 Rural Household Survey. The results are from a linear probability model estimated at the household level using the full rural sample for more than 65,000 households. A number of different versions of the model were estimated: with village-fixed effects, with county fixed effects, with provincial fixed effects and without fixed effects. The above results are from the model without fixed effects. The estimated effects for household-level variables were robust across different models. ** highly significant (at 1% level); * significant (at 5% level). The education effects are measured relative to primary or lower level of education. The last two provincial variables are weighted by the inverse of the distance between the provincial capitals of destination and source provinces. The role of education in determining migration probability is confirmed by the results in Table 5.55. A higher proportion of labor members with junior high education has a significant positive impact on the likelihood of migration. On the other hand, the results also indicate that senior high school education has no significant effect on migration probability. This needs to be interpreted carefully. As already noted, a very large fraction (about two-thirds) of migrant workers have junior high school education, which it appears is all that is needed for the low-skill jobs that migrant workers are typically engaged in. High-skill or white-collar jobs in urban areas are generally out of reach of rural migrants. The pathway to those job opportunities is mostly through city-based tertiary education. Much of the value of senior high school education comes from its serving as an input into tertiary education. Indeed, tertiary education in urban areas remains an important P r o m o t i n g o p p o r t u n i t y 1 0 4 mechanism for changing a rural hukou into an urban one. Thus, senior high school education does not necessarily promote higher incidence of migrant workers, but it remains valuable for rural students going on to pursue tertiary education in urban areas. Unfortunately, however, only a few amongst the rural population actually go on to pursue college or higher education. The welfare of children is clearly an important consideration. A higher ratio of children (under the age of 14) has a significant negative effect on the likelihood of migration. This reflects the significantly higher costs of raising children in urban areas which seriously limits the prospects for full-family migration. As noted above already, full-family migration remains a small component of overall migration from rural areas. The qualitative poverty assessment in the 12 study villages also highlights that provision for children's education is often the most important issue in family migration (CASS-IOS, 2006). That a lack of local non-farm employment opportunities is a reason for migration is supported by the finding that a higher proportion of TVE workers in the village has a negative effect on the migration probability. Similarly, households in relatively remote mountainous villages with limited local potential for productive employment are also more likely have migrant workers. Controlling for household endowments (including those related to educational attainment), households in minority villages have lower migration probabilities. This suggests minority communities face special difficulties in being able to make use of migration opportunities. This may have to do with employment discrimination in destination areas, and minority communities' limited access to information and labor recruitment networks. The evidence from NBS' 2004 Rural Household Survey indicates that the main channel for migration (accounting for two-thirds of all migrant workers) is through contacts or introduction by people of the same origin, families and friends (Sheng and Peng, 2005). These informal channels are less established and accessible for some categories of potential migrants (such as those from minority communities) than others. The research undertaken for the 12-village qualitative poverty assessment also clearly showed that households from minority villages are much less successful in migration compared to households from non- minority areas (CASS-IOS, 2006). As may be expected, higher labor-to-land ratio for the household tends to increase the probability of migration. On the other hand, counties with higher agricultural productivity per unit of land tend to have higher migration probabilities. This is consistent with the idea that poor, who are over-represented in areas with lower agricultural productivity, are less likely to migrate. This is also likely to reflect that within-county migration (which accounts for about 17% of all migrant workers) is likely to higher for better-off counties. Economic incentives in the form of relatively higher returns to labor in destination areas are an important driver of migration. This is indicated by the significant positive effect on the likelihood of migration of a higher differential between destination and source area income levels. Since this variable is weighted by the inverse of the distance between destination and source of migration, the results also underline the importance of distance as a factor in determining migration. Greater distance which is proxy for the monetary and other costs of movement tends to inhibit migration. Costs of migration can be substantial. For instance, fieldwork in Yunnan province indicated that the average cost for a villager to find a job in cities is about 1000 Yuan including traveling expense and the cost of living while looking for a job, and many often borrow money from relatives or friends (CASS- IOS, 2006). Finally, there is evidence that migrant workers do not appear to have the same work opportunities as local residents in urban labor markets, and this acts as a disincentive for migration.49 Using the 2000 census data, Park, Zhao, and Huang (2006) construct indices of labor market segmentation in different provinces by estimating the probability that migrants and local residents work in different occupations or sectors after controlling for _________________________________________________________ 49See section 7e for further discussion of this issue. P r o m o t i n g o p p o r t u n i t y 1 0 5 differences in individual characteristics. As seen in Table 5.55, greater segregation as measured by these indices has a significant negative impact on the likelihood of migration. What do potential migrants say? Direct responses from potential migrants in four designated poor counties on the most important reason for not migrating provide complementary information to the above quantitative analysis. As seen in Table 5.56, while agricultural labor requirements are cited as the most important reason for lack of migration by 31% of men and 26% of women, most of the other significant responses have to do with factors limiting the individual's ability to migrate in one way or another. The most important factors include: inadequate education or skills (for 21% of men, 17% of women), lack of information or social networks (17% of men, 9% of women), and inability to meet the costs the financial costs associated with migration (9% of men, 10% of women). Unwillingness to live away from children is a particularly important consideration for women (31% of women, 11% of men) which highlights the limited prospects for entire family migration. The hukou system, which can be restrictive of migrants' access to a variety of public services on equal terms with local residents in urban areas, itself is not considered a major impediment by potential migrants. But if a major effect of the hukou system is to restrict family as opposed to individual migration, then the responses in relation to separation from children also need to be taken into account. Table 5.56 What do potential migrants in poor areas say? % of rural non-migrants who said that reason for not migrating was... Men Women Farm labor demand 31.3 25.9 Education 20.8 16.5 Information 17.1 8.7 Take care of children 10.9 31.5 Financing constraints 9.0 9.7 High local income opportunities 2.8 1.7 Support from urban governments 2.1 5.2 Urban lifestyle 1.9 0.2 Support from local government 1.6 0.0 Hukou system 1.4 0.0 High urban cost of living 1.2 0.7 Sources and notes: Du, Park and Wang (2005) based on 2000 China Rural Poverty Survey in 4 designated poor counties. The downside of migration in source areas While migration is an important avenue for rural poverty reduction, it does not come without some negative effects on the migrants themselves and for those in source areas. Some of these are discussed below. First, migration opportunities appear to be creating a disincentive for continued increases in educational attainment amongst the rural youth. According to a recent estimate, one percent increase in migration from a village tends to reduce the probability that a junior high school graduate will attend senior high school in the following year by 0.18-0.23 percent (De Brauw and Giles, 2006). As the decision not to attend high school is often irreversible for youth in rural areas, this can permanently affect the lifetime earning ability of young migrants. Rapidly increasing private costs of tertiary education (for which post- secondary education is a necessary input) contribute further to a decline in the perceived net returns to senior high school education. The occupations migrant workers are concentrated in exposes them to greater risks of work-related injury and illness, especially in an environment of inadequate occupational safety standards. Yet, few of the migrants are covered by any social insurance programs. Thus, in the event of an injury or serious illness, the hardship imposed on the migrant and their family members can be immense. The qualitative poverty assessment in 12 study- villages conducted by the Institute of Sociology (IOS), China Academy of Social Sciences (CASS-IOS, 2006) indicates that in cases where migrants have died or returned home P r o m o t i n g o p p o r t u n i t y 1 0 6 with injury or disability, their households suffer an extremely difficult situation and often rapidly descend into poverty. Some illustrative examples are recounted in Box 5.12 . Box 5.12 : Hazards of migrant work: injury and disability without compensation LSL case from X village, Jiangxi province: In 2004, when he worked in the building site of a hospital in J city, his head was smashed by bricks. He stayed in hospital for three days. He left hospital because the boss told him that he could have a rest at home. The boss hid when he had to pay several hundred Yuan for LSL's medical treatment. When he went home, his wound worsened. He suffered paralysis because of serious cerebral concussion. He cannot take care of himself now. As a result, his wife left home. He lives with older parents and a younger child who is going to school now. Two cases in Q Village, Inner Mongolia: In 1990, Yang went to Liaoning and worked in a brickyard. His legs were broken by the machinery. Then he returned to the village. He did not marry until recently. He is very poor. Another person suffered semi-paralysis because of hard work. His household is included in the Di Bao hu now. Source: CASS-IOS (2006). Compared to other adult family members, the children left behind by migrants are likely to suffer disproportionately from the absence of one, and sometimes both, of their parents. Using the data of the 5th National Population Census, it is estimated that there are 19.8 million migrants' children below 14 years of age who are left behind in the home villages (Duan and Zhou, 2005). Studies conducted by the Women's Federation of PRC (2006) indicate that 60% of women migrants have left their children behind in the villages, and approximately 90% of them see their children once or twice a year. Many of the left- behind children thus experience extended periods of time without maternal or paternal care during some of the important formative years of their development. A recent study for 10 villages in Shaanxi, Ningxia, Hebei and Beijing documented that 88% of the left- behind children only contact their parents through (mostly short) phone calls, 65% have personal contact with their parents once a week or less frequently, and 9% are not in contact with their parents at all (Ye and Murry's, 2005). The study also suggests that lack of parental care has contributed not only to malnutrition and poor health of some children, but also to personality and psychological problems. Moreover, due to a relative shortage of labor in the left-behind households, school-aged children often share additional responsibilities for farming and housework. As a result of both this extra burden and the lack of parental care, many of them do not perform very well at school and some are discouraged to continue their education. Using China Household Income Project (CHIPs) data for 2002, Meng and Luo (2006) find that migrants' left-behind children systematically perform less satisfactorily than children whose parents are non-migrants. In the long-run, the under-investment in migrant children's education as well as health and nutrition, if not adequately addressed, is likely to cause systematic disadvantage for future generations of the rural population. The large scale of rural-urban migration has also significantly altered the population structure in many rural sending communities (CASS-IOS, 2006). In many villages, most youth and middle-aged people have left for employment outside, and only the elderly, a small number of adults, and pre-school children remain in the home village. This has substantially increased the burden on the elderly who themselves receive less care and often have to take on additional responsibility of household chores, farming as well as tending to grandchildren. Some of the elderly referred to this heavy burden by describing themselves as "tired as dogs" (12-village qualitative poverty assessment; CASS-IOS, 2006). From the perspective of rural communities, this hollowing out of the productive core of the village population has also dampened rural community participation and community building (CASS-IOS, 2006). Some villagers are concerned that community cohesion is gradually eroding, and the actions of village leaders cannot be controlled because of the migration of the youth and the active middle-aged population. Many village leaders, on the other hand, report that public affairs cannot be conducted properly because the elderly P r o m o t i n g o p p o r t u n i t y 1 0 7 at home cannot make decisions. Also as a consequence of out-migration of able-bodied labor, mutual-help is less prevalent in the village and public activities and communication between villagers has also declined. A village head from Sichuan province described that few villagers participate in traditional cultural activities, and villagers are generally not interested in participating in village congress and other public activities unless they receive a payment (CASS-IOS, 2006). Several of the adverse effects in source areas, to a large extent, are not caused by migration per se. Rather, they reflect the shortcomings of the prevailing `guest worker' system where (a) discriminatory rules and practices restrict migrants' access to urban services and ultimately limit the scope for migration of families as opposed to individual workers, (b) virtual absence of a social insurance system for migrant workers and at best limited enforcement of their worker rights exposes them and their families to risks against which they have little protection, and (c) the relatively limited human capital of migrant workers (reflecting an overall underinvestment in human capital in rural areas) results in a de facto segregation of migrant workers into certain low-skill occupations. The policy response to this situation ought not be to limit the flow of migration for that will serve to slow the pace of rural poverty reduction, but to address some of these underlying shortcomings. Policy implications In sum, therefore, the quantitative and qualitative evidence taken together suggests policy initiatives in the several directions that are likely to be helpful in facilitating labor mobility from rural areas and realizing its potential for poverty reduction: · Improving the education and skill base of potential migrants especially those from poor households who are more deficient in these respects. Ensuring junior high school level of education for all in rural areas is a necessary first step. But beyond that there is a need to increase public support for post-secondary and technical education in rural areas if migrants are to be able to access better-paid higher skill jobs in urban labor markets. · Assistance for reducing information and search costs associated with migration by supplementing informal social networks with more organized placement and intermediation services. A special effort may be needed to reach some of the minority groups, and those in some of the poor areas where existing informal networks are underdeveloped. · Improving investment climate in cities in the main sending regions in Central and Western provinces. The pattern of migration in China in many ways mirrors the geographic concentration of economic activity. The sending areas are concentrated in the Center and the West and the receiving areas along the coastal belt in the East. Most migration is inter-provincial and over long distances. However, like elsewhere in the world, distance increases the costs of migration, and reduces its magnitude. Hence, there is a case, especially as the industrial clusters in the coastal belt are becoming increasingly skill and capital-intensive, that some of relatively more labor- intensive segments of industry can be further developed in the central and western provinces. · Improving labor market conditions and public service and social security provision for migrants in the cities. Despite its large scale, the current system of migration could with some justification be described as a `guest worker system'. A range of discriminatory practices in destination areas abridge migrants' entitlements as workers or citizens. There is an increasing need to eliminate these practices which are limiting both the scale of migration, especially family migration, and the potential welfare benefits from migration. In particular, this will require better enforcement of the provisions of labor laws for migrant workers, access to basic services (education, health and housing) to be determined on the basis of place of residence rather than the place of hukou, and the extension of the coverage of social protection programs to migrant workers. Government initiatives in some of these areas are already underway as, for instance, laid out in the Opinions of the State Council on Issues Concerning Rural Migrant Workers (SC P r o m o t i n g o p p o r t u n i t y 1 0 8 Document No. 5, March 2006), and the MOLSS Circular on the Implementation of SC Document No. 5 (April, 2006) as well as the action plans of the inter-ministerial Joint Conference on Rural Migrants Issues established in 2006. The long list of policy guidelines for migrant workers cover many areas including worker rights and protection, public services and living conditions, social security, employment services and skills development (Box 5.15). Box 5.15 : Recent policy guidelines in relation to migrant workers With the release of State Council Document No. 5, "Opinions of the State Council on Issues Concerning Rural Migrant Workers", and the setting up of the inter-ministerial Joint Conference on Rural Migrant Issues in 2006, wide-ranging policy guidelines have been issued in relation to migrant workers. Covering a number of areas, the key provisions of these guidelines include: In the area of social security: · extending work injury insurance to almost all rural migrant workers employed in high-risk industries (e.g. coal mining and construction) · setting up separate medical insurance pooling fund to mainly cover inpatient medical expenses of migrants · improving settlement method for those who return to their hometown for treatment or those NCMS members who seek treatment in urban areas · exploring appropriate pension insurance plans for migrant workers In the area of public services and living conditions: · extending disease prevention and immunization for migrant children · government at the importing end to undertake responsibility of compulsory education for any accompanied children of migrant workers, by integrating their compulsory education into the local education development plans, and to ensure that migrant children receive the same treatment as local students in relat6ion to charges, fees and management · strengthening supervision by related departments to ensure that housing for rural migrant workers meets the basic hygiene and safety standards, and integrating the housing needs of rural migrant workers who live and work in the city on a long term basis into the urban development plan for housing construction In the area of employment services, training and transfer of rural labor: · at the exporting end, setting up county and township employment service networks to provide fee policy consultation and information · at the receiving end, opening public employment service agencies to the migrants to provide them with free job information, job introduction and consultation · strengthening skills development through the implementation of Rural Labor Skills Development and Employment Plan, the Sunshine Project for labor transfer, and the piloting of Poverty Reduction through Skills Development Project · moving related industries to mid-Western regions to create more local job opportunities, and developing small towns in an active and steady manner, and building their capacities to attract industries Source: Opinions of the State Council on Issues Concerning Rural Migrant Workers (SC Document No. 5, March 2006), and the MOLSS Circular on the Implementation of SC Document No. 5 (April, 2006). However, not much systematic information is available on the actual implementation of these wide-ranging policy guidelines. It will be important to monitor the progress on these new initiatives not only to assess if they are being implemented as intended, but also to refine the different initiatives based on the implementation experience of what seems to be working, where and in what ways. P r o m o t i n g o p p o r t u n i t y 1 0 9 6.Enhancing security: protecting households from falling into poverty a. Land allocations alone are an inadequate basis for the rural safety net Traditionally, land has been considered an important safety net for the rural population in China. The view has sometimes been starkly expressed as: the rural population does not need safety nets because they have land. China is indeed relatively unique amongst developing countries in having one of the most egalitarian distributions of land. For instance, in 2004, while net per capita income for the top quintile in rural areas was nearly seven times that for the bottom quintile, per capita cultivated land was not very different from the overall average of 2 mu across the quintiles (Figure 6.65). Figure 6.65 : An egalitarian distribution of land in China, 2004 (Distribution of cultivated land by quintiles of rural population ranked by net per capita income, 2004) Net income per capita (yuan) 0 1000 2000 3000 4000 5000 6000 7000 8000 6931 Richest 20% 2.18 3608 4th 2.24 2578 3rd 1.99 1842 2nd 1.88 1007 Poorest 20% 1.80 0 0.5 1 1.5 2 2.5 Land per capita (mu) Source: Computations based on NBS (2005): 2005 China Yearbook: Rural Household Survey. The egalitarian land distribution is a legacy of the Household Responsibility System introduced in 1978 when land use rights were granted to rural households on a more or less uniform per capita basis. Largely, the same system of land allocation has been maintained with changes in household population over time accommodated through periodic administrative reallocations carried out by village communes. Thus, the concentration of land has hardly changed at all over the last 20 years (Figure 6.66). E n h a n c i n g s e c u r i t y 1 1 0 Figure 6.66 : An unchanging concentration of land over 20 years nd 100 la deta 80 ivltuc 1985 60 of % 40 evit 2004 20 mulauC 0 0 20 40 60 80 100 Cum. % of population ranked by net per capita income Sources and notes: Computations based on NBS (2005a): 2005 China Yearbook: Rural Household Survey. The graphs for 1985 and 2004 show the concentration curves for cultivated land when rural population is ranked by net per capita income. Each marked point corresponds to the corresponding point on the grouped distribution available from published data. But is a per capita allocation of around 2 mu adequate? There are three reasons why this allocation, despite its egalitarian distribution, may be considered inadequate as a rural safety net. First, not all land is the same and there are large variations in the quality--and hence, the productivity--of land. Second, farm income is an increasingly small proportion of rural incomes. Third, there remain issues with the security of land tenure. These issues are further discussed below. There are large variations in the productivity of land Unsurprisingly, there are significant variations in the quality and productivity of land across the vast expanse of the Chinese hinterland. Figure 6.67 gives an indication of this variation across the 863 counties covered by the National Bureau of Statistics' Rural Household Survey. For 2003, it shows the distribution of county grain yields (value of grain output per mu of sown area with grains) as predicted by basic agro-climatic conditions (including factors such as the level of precipitation, temperature, slope, and the incidence of frost). At the bottom end (first percentile), grain yields are only about two- fifths of the yields at the top end (99th percentile). About 10% of all counties have grain yields of about 250 Yuan/mu or less. Figure 6.67 : Farm yields vary significantly across counties Predicted value of grain output (yuan/mu) of area sown with grain 373 401 415 433 400 329 279 300 232 252 175 200 100 0 1% 5% 10% 25% 50% 75% 90% 95% 99% Percentile of counties Source: World Bank estimates based on NBS Rural Household Survey 2003 and published county- level data on agro-climatic conditions. E n h a n c i n g s e c u r i t y 1 1 1 Low productivity of land also has consequences for household income, consumption and poverty. Controlling for a number of household and village attributes, both per capita income and per capita consumption of rural households are strongly affected by county farm yields, increasing by 1% for every 1% increase in yields.50 Similarly, farm yields also have significant implications for rural poverty. For instance, the average probability of a rural person being poor (using the World Bank poverty line) is just under 10% using per capita income and about 18% using per capita consumption. With a 10% increase in yields, these probabilities could be reduced to 7% and 14% respectively. Similarly, for someone facing a 50% probability of being income or consumption poor, a 10% increase in yields reduces these probabilities to 44 or 43 percent (Figure 6.68). Figure 6.68 : Predicted effects of a 10% increase in farm yields on poverty, 2003 100% 87% 90% 76% poverty yields 80% of 65% 87% in 70% tyi e 54% 60% 75% Income poverty 44% 50% 64% 34% increas 40% 53% Consumption probabil 25% 43% d %01 30% 16% poverty 20% 8% 33% 10% 24% with 7% 15% Predicte 0% 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% Predicted probability of poverty with observed yields Source: World Bank estimates based on NBS 2003 Rural Household Survey, using the World Bank poverty line. The share of farm income in rural income continues to decline With the structural change in the rural economy and the growing importance of the non- farm sector (including migration to urban areas), the share of agricultural income in rural household incomes has been steadily declining. Between 1985 and 2004, the share of income from the primary sector (including farming, forestry, animal husbandry and fishery) fell from 66 to 48 percent (NBS Statistical Yearbook, 2005). In 2004, while the average per capita income from the primary sector was 50% more than the dollar-a-day poverty line for the rural population as a whole, for the bottom quintile it was only about two-thirds of the dollar-a-day line. Income from farming alone was equal to only 54% of the dollar-a- day line for the bottom quintile (Table 6.57). Thus, farm-based income alone is unlikely to cover the basic needs of the poor and near poor in rural areas. Table 6.57 : Income from agriculture for the rural population, 2004 Per-capita income quintile of Rural rural population population Bottom 2nd 3rd 4th Top Income from primary sector ... as a proportion of net per capita income 0.48 0.63 0.59 0.54 0.48 0.35 ... as a proportion $/day poverty line 1.50 0.68 1.16 1.49 1.86 2.59 Income from cropping ... as a proportion of net per capita income 0.36 0.54 0.47 0.43 0.37 0.23 ... as a proportion $/day poverty line 1.13 0.59 0.94 1.18 1.43 1.72 Sources and notes: Computations based on NBS (2005a): 2005 China Yearbook: Rural Household Survey. Quintiles of population are defined in terms of net per capita income of rural households. _________________________________________________________ 50This is based on income and consumption poverty functions estimated using the national sample from the NBS' 2003 Rural Household Survey. E n h a n c i n g s e c u r i t y 1 1 2 Despite recent improvements, security of tenure is still limited by land readjustments and requisitions The effectiveness of land as a safety net also critically depends on farmers' land rights and security of their land tenure which have been the subject of intense policy debate. Since farmers do not own land but only have use rights to it, the main issue is the scope and security of these use rights. Insecurity of use rights has been shown to have inhibitive effects on (i) long-term investment in land (Jacoby, Li, Rozelle, 2002), and (ii) the development of a productivity-enhancing land rental market (Deininger and Jin, 2005). Both directly and through limitations they impose on land rental markets, insecure land rights further limit the mobility of labor for off-farm employment outside the village, especially to urban areas (Yang, 1997). For instance, in 2004, only 45% of villages reported a clear policy of not taking away land if migrant households rent out their land to others (Table 6.58). Most of the remaining either reported "no rule" (45%) or "depends on the case" (8%) in spite of clear provisions allowing such transfer of use rights both in the Land Management Law (LML) of 1998 and the Rural Land Contracting Law (RCRL) of 2003 (see Box 6.16). Thus, the majority of households considering or engaging in migration face at least some risk of losing their land if they were to lease out their use rights. Table 6.58 : Risk of land expropriation faced by migrant households, 2004 Village policy on whether land will be taken away if migrant households rent out their land to others % of villages Will not take away land 44.6 Will take away land immediately or after one or more years 1.8 Depends on the case 8.3 No rule 45.4 Total 100.0 Sources and notes: World Bank estimates based on the 2004 World Bank-NBS Village Level Survey of 3037 villages. There are two main sources of insecurity experienced by farmers: the policy and practices related to land readjustments, and those related to land takings. Readjustments refer to periodic administrative reallocations of land amongst village households. They could be small, affecting only a few households or large affecting all or most of the households.51 Lank takings refer to expropriation of farmers' land (compulsory withdrawal of their use rights) by the village collective for other, mostly non-agricultural, uses. The two are not unrelated, as land takings have been one of the reasons for readjustments that have been used to spread the burden of diversion of arable land more evenly within the village (Brandt et al., 2004). Box 6.16 : Key provisions of the Rural Land Contracting Law, 2003 After more than three years of drafting and deliberation, the Rural Land Contracting Law (RCRL) became effective on March 1, 2003. Building on many of the provisions of the Land Management Law of 1998, the RCRL aims to strengthen the legal basis for more secure land use rights for China's farmers. The main provisions of the law include the following: · Use rights to arable, grass and forest land are to be established through written contracts between the village collective and individual households (Article 21). County governments are required to issue land right certificates to farm households affirming these rights. · The duration of contracts is stipulated to be 30 years for arable land, 30-50 years for grassland, and 30-70 years for forest land. _________________________________________________________ 51Traditionally, land readjustments have been undertaken to periodically reallocate land amongst households following changes in population and household size (due to births, deaths, marriage, divorce, and out-migration). Other reasons have included consolidation of land parcels, introduction of larger scale farming, and natural disasters leading to loss of arable land. According to a 2005 survey undertaken by the Rural Development Institute (Seattle), Renmin University (Beijing) and Michigan State University in 1773 villages in 17 major agricultural provinces, 72% of the land readjustments were due to population change, 14% due to land takings, 7% for consolidation of parcels and higher-scale farming and 2% for natural disasters (Prosterman, et al 2006). E n h a n c i n g s e c u r i t y 1 1 3 · Administrative readjustments of land over the 30-year period are prohibited with only two exceptions: for natural disasters that seriously damage the contracted land, and small readjustments involving only a few families. · The small readjustments have to be approved by two-thirds of the village assembly or representatives as well as by the township and county governments. · The collective can not take back land (for non-agricultural use) over the 30-year duration of the contract without adequate compensation. · Farmers' land rights include "rights to use, profit from, and transfer land contracting and operation rights, and the right of autonomy over production operations and disposition of products". · In case of change of residence for women due to marriage, divorce or widowhood, their existing rights to land can not be revoked unless they receive an allocation in their new residence. · Violations of the above rights are punishable offenses subject to civil penalties, including monetary damages and restitution of rights. Sources: Li (2003); Deininger, Lin and Rozelle (2006). Neither of these issues is limited to the rural poor. For instance, data from the 2004 NBS- World Bank survey indicate that about 42% of villages in China have had at least one major land readjustment (involving more than half the households in the village) during 1998-2004; including 31% of designated poor villages and 44% of non-poor villages.52 While the incidence of major land readjustments has declined since 1998 ­ at least partly in response to the legislative reform of land use rights ­ the decline has been similar for both designated poor and non-poor villages (Figure 6.69). Figure 6.69 : Declining incidence of major land readjustments in designated poor and non- poor villages Non-poor villages Poor villages es 15 12.3 12.8 12.2 agl 10 5.2 vilfo 3.7 4.7 4.9 4.0 5 2.7 2.6 2.7 2.9 1.3 1.7 % 0 1998 1999 2000 2001 2002 2003 2004 Most recent year of major land readjustment Sources and notes: World Bank estimates based on the 2004 World Bank-NBS Village Level Survey of 3039 villages. Major land readjustment is defined as one affecting more than half the households in the village. Similarly, land takings are an issue in both designated poor and non-poor villages. More than one-fifth of all villages in China reported land takings over the 5-year period 2000- 2004.53 The incidence of takings was lower in designated poor villages (16%) relative to the non-poor villages (23%). However, amongst villages with land takings, about 6% of arable land was subject to such acquisition in both poor and non-poor villages (Table 6.59). _________________________________________________________ 52According the 2004 World Bank-NBS Village Level Survey of 3039 villages, 70% of the major land adjustments involved 100% of village households. 53This is comparable with the estimate of 29% of villages with land takings reported for 2001-2003 by Deininger, Jin and Rozelle (2006) based on a 1,100 village survey in 12 provinces undertaken during 2003 by the NBS in collaboration with the World Bank's Development Research Group. E n h a n c i n g s e c u r i t y 1 1 4 Table 6.59 : Prevalence of land taking and the extent of compensation paid, 2000-2004 Designated poor Non-poor All villages villages villages % of villages with land takings during 2000-2004 15.6 23.4 22.4 In villages with land takings: Total arable land (mu) 2120.4 1688.6 1729.9 Area of land taken (mu) 145.6 108.5 112.0 Share of arable land taken (%) 6.9 6.4 6.5 Average compensation (Yuan/mu) 6435 11294 10761 % of villages offering no compensation 17.5 10.3 11.0 Of the land taken % taken with no compensation 26.7 3.9 6.5 Sources and notes: World Bank estimates based on the 2004 World Bank-NBS Village Level Survey of 3039 villages. An important issue with land takings is the adequacy of compensation for the land taken. Here, there are important differences between designated poor and non-poor villages. Amongst villages reporting land takings, the average compensation per mu in poor villages is only about 57% of that in non-poor villages. To some extent, this may reflect differences in the quality of land. But it also appears to reflect a significantly greater likelihood of the lack of any compensation in poor villages. Compared to 10% of the non- poor villages acquiring land without compensation, 18% of poor villages with land takings provided no compensation. The contrast is even sharper in terms of the area of land: while no compensation was paid for about 4% of the land taken in non-poor villages, this proportion was 27% in poor villages. In sum, therefore, while land is clearly an important asset for the rural population, there are many factors that limit its adequacy as a rural safety net. Some of them have to do with the considerable heterogeneity in the quality of land. Others have to do with the changing structure of the rural economy with increasing importance of non-farm sources of income. Still others have to do with the security of land use rights of farmers which, while they have improved with the better legal framework provided by the recent legislation, continue to be an impediment to productive land use. Many of these factors apply to the non-poor and the poor alike, but some of them seem to matter more for the latter. b. Are rural social protection programs adequate? In broadest terms, the rural social protection policy in China may be seen to comprise of three sets of programs distinguished in terms of their primary functional objectives: (i) rural poverty alleviation programs, (ii) social assistance programs, and (iii) social insurance programs. While poverty alleviation programs have aimed to augment the income-earning capacity of the poor mainly through area-based investments, social assistance programs have sought to supplement incomes of the poor through direct transfers, and social insurance programs have sought to promote greater security against health shocks and old age. The distinction between the three types is however necessarily imperfect as (a) all programs potentially provide both short-term and long-term, as well as both transfer and insurance benefits, and (b) the successes or failures of one type of program may reduce or accentuate the need for others. Of these three main categories, the area-based programs ­ targeted to the 592 designated poor counties until 2001 and to the about 148,000 designated poor villages thereafter ­ have been the main thrust of social policy for the rural poor. However, as discussed in detail chapter 5, an evaluation of these programs suggests that they do not adequately address the problem of rural poverty. For instance, they do not cover nearly half the rural poor who do not live in the designated poor villages. And within designated poor villages, a careful impact evaluation further suggests that program effects on income and consumption growth for the poor have been insignificant. This sobering assessment of the area-based poverty alleviation program thus puts a greater responsibility for E n h a n c i n g s e c u r i t y 1 1 5 providing support to the poor on the other components of the rural social protection policy. The extent to which the latter fulfill this role is discussed below. There are five main social assistance programs and two social insurance programs that are currently operative in rural China. The social assistance programs include: Wu Bao, Tekun Hu, Di Bao, natural disaster relief, and the Medical Assistance programs. The two social insurance programs include: rural health insurance, and rural pensions. The rest of this section briefly discusses the main features of these programs and offers an assessment of the extent to which they provide adequate social protection to the rural population, especially the rural poor.54 Features of key social assistance programs in rural areas (1) Wu Bao The Wu Bao or the Five Guarantee program has a long history going back to the 1950s. It targets the "3 No's" population ­ those with no work capacity, no source of income, and no external support. The five guarantees the program aims to provide to the beneficiaries are for food, clothing, housing, medical care, and funeral expenses. Wu Bao households are seen to be at the very bottom rung of the socio-economic ladder, and there is usually a degree of social stigma attached to being identified as a Wu Bao household. Benefits may be provided under a centralized system (through government-operated welfare homes) or under a decentralized system (through cash or in-kind benefits provided by the village or township). The decentralized system traditionally relied on funds from township and village levies that were discontinued with the rural fee reform of 2002. In January 2006, the State Council issued regulations to shift financing responsibility for Wu Bao from village reserves to local fiscal budgets (although poor areas may receive support from the central government). The regulations also stipulate that instead of village committees or town governments, the provincial or municipal/county governments with the approval of the provincial government will be responsible for determining Wu Bao benefits. According to official estimates, there were 5.3 million Wu Bao support beneficiaries in 2007, 4.1 million under the decentralized system and 1.2 million under the centralized system (Table 6.60). Table 6.60 : Number of beneficiaries of social assistance programs in China, 2004-7 Year 2004 2005 2007 Number of Wu Bao support beneficiaries (million households) 4.53 4.18 5.30 ...of whom, number covered by decentralized support 3.94 3.50 4.20 ...of whom, number covered by centralized support 0.59 0.68 1.20 Number of Tekun Hu beneficiaries (million persons) 9.14 10.67 0.30 Number of Di Bao beneficiaries (million persons) 4.88 8.25 34.52 Number of Medical Assistance beneficiaries (million persons) 11.71 29.09 Assistance with NCMS contribution 8.67 23.06 Other 3.04 6.03 Sources and notes: National Bureau of Statistics: China Statistical Yearbook 2006 and 2007, and Ministry of Civil Affairs Statistical Yearbook 2007. (2) Tekun Hu This program in now in the process of being absorbed under the rural Di Bao program (discussed below). Traditionally, the Tekun Hu program catered to extremely poor households (Tekun Hu) who are below the official poverty line, and who may have lost their work capacity due to major illness, disasters, accidents or becoming handicapped. The benefits provided could be both in cash or kind. Income and asset thresholds generally used for determining eligibility are locally-established. To become eligible, a household needs to apply for a Tekun Hu status which has to be approved at the village and township levels. As a result of the nationwide expansion of the rural Di Bao program, _________________________________________________________ 54The discussion in this section draws heavily upon an analysis of primary data from the World-Bank NBS 2004 Village-Level Survey conducted specially for this poverty assessment. It also draws upon reviews and evidence presented in Lin (2004), E n h a n c i n g s e c u r i t y 1 1 6 the rural tekun hu are now being covered by Di Bao. This is reflected in the sharp decline in the number of Tekun Hu beneficiaries to only about 300,000 by the end of 2007 (Table 6.60). (3) Rural Di Bao The most significant development in the area of rural social assistance is the decision of the State Council (indicated in the July 2007 document) to roll out the rural Di Bao program on a nationwide basis. With this decision, the program has expanded very rapidly. By the end of 2007, all 31 provinces had established the rural Di Bao, and number of beneficiaries had climbed to 34.5 million (Table 6.60). With this expansion, the rural Di Bao has emerged as the single most important social assistance program in rural areas. The program is mainly financed by local governments at various levels, with certain inputs from the central government. The Rural Di Bao program is modeled after the Urban Di Bao program, and it shares with the latter many of its design features. The main criterion for eligibility is household per capita income being below an administratively set Di Bao threshold or standard. The state council decision lays out basic policies for eligibility criteria and means-testing mechanisms. However, the experience until recently indicates that the thresholds themselves were only loosely tied to the official poverty line and in practice varied a lot across different locations. Like the urban Di Bao, the income test has often been supplemented with other non-income criteria reflecting in part the difficulties of reliably measuring household incomes. "Assumed income" given by the average rural income for the previous year was often applied to working or able-bodied members of the household when income documentation was unavailable (O'Keefe, 2004). Eligibility is usually determined for a year and has to be renewed thereafter. In principle, the program follows a top-up design with per capita benefits determined by the difference between the threshold and household per capita income, but it is unclear to what extent this is followed in practice. Recent regulations raised the Di Bao benefit levels by 10 yuan per month (in February 2008), while also permitting better-off areas to implement higher increases.55 The average Di Bao standard during 2007 was 70 yuan per person per month, and the average benefit level was 41 yuan per person per month.. (4) Medical assistance (MA) Parallel to the Ministry of Health's roll-out of the new rural health insurance program (the National Cooperative Medical Scheme) in 2003, the Ministry of Civil Affairs launched the Medical Assistance (MA) program which aimed to reach poor rural households who, despite the existence of other transfer programs, needed additional support for meeting the financial burden of medical care expenses. In counties with the National Cooperative Medical Scheme (NCMS), MA benefits take the form of assisting households with their NCMS contributions or copayments or both. In non-NCMS counties, MA provides direct assistance with out-of-pocket medical expenses. MA has three target groups: households with large and potentially impoverishing medical expenses, poor households (typically Wu Bao, Tekun and Di Bao), and other deserving groups (most commonly those receiving special assistance and laid-off workers). The MA program is financed mostly from central, provincial and county-level sources, but is also supplemented with sources at the township level, lottery revenues, donations and development assistance.56 The share of these supplementary sources tends to be higher in richer counties.57 _________________________________________________________ 55Till recently, in many areas the same per capita Di Bao subsidy is applied to all eligible households in the area. Cash benefits are also supplemented with in-kind benefits in many regions including waivers or subsidies for education, utilities, medical expenses and housing. 56More than half the MA programs initiated by 2005 were unofficial ones according to the 2005 Medical Assistance Survey carried out by the World Bank Rural Health Team in collaboration with MOCA and the Institute of Social Development and Public Policy at Beijing Normal University. 57Government budget on MA (for both rural and urban areas) has increased rapidly. The central government transfer for MA increased to 3.3 billion yuan in 2007 (up from 1.4 billion yuan in 2006), while the local government budget increased to 3.8 billion yuan (up from 2.7 billion yuan in 2006). E n h a n c i n g s e c u r i t y 1 1 7 The coverage of the MA program has grown rapidly. According to data from the World Bank-NBS 2004 Village-Level Survey, 48.9% of all counties (33.5% of designated poor counties and 53.6% of non-poor counties) reported at least some presence of MA (positive number of beneficiaries) by the end of 2004. By the end of 2006, all rural counties had established the MA scheme. At the end of 2007, 29 million persons in rural areas benefited from MA, the number nearly tripling since 2005 (Table 6.60). (5) Natural disaster relief The devastating earthquake in Sichuan on May 12, 2008, is a grim reminder that China is amongst the most natural disaster-prone countries in the world. The tragic loss of life and livelihood associated with such natural disasters caused not only by earthquakes but also by floods, droughts, landslides, typhoons, oceanic disasters and forest fires has been an important concern motivating a large and long-standing government effort in both disaster reduction and disaster relief. In 2007, natural disasters are estimated to have affected more than 30% of sown area in the country (Table 6.61). The area affected by floods and droughts alone is about 70% of the total area affected by natural disasters. Table 6.61 : Official statistics on natural disasters and disaster relief in China Year Unit 2004 2005 2006 2007 Area affected by floods and droughts (`000 hectares) 24567 26960 28741 Share of sown area affected by floods and droughts (%) 16.0 17.3 18.3 Area affected by natural disasters (`000 hectares) 37106 38818 41091 48993 Share of sown area affected by natural disasters (%) 24.2 25.0 26.2 31.2 Direct economic damage (billion yuan) 160.2 204.2 252.8 236.3 Total disaster relief from all levels of (billion government yuan) 3.2* 5.29 5.93 6.56 Share of direct economic damage covered by government relief (%) 2.6 2.3 2.8 Sources and notes: National Bureau of Statistics: 2006 China Statistical Yearbook, and the Ministry of Civil Affairs website. *Central government expenditure only. Further evidence on the extent of natural disasters and disaster relief based on the World Bank-NBS 2004 Village-Level Survey is presented in Table 6.62. E n h a n c i n g s e c u r i t y 1 1 8 Table 6.62 : Village-Level evidence on natural disasters and disaster relief, 2004 % of villages affected In affected villages Average benefit per capita Fraction of Avg. % Avg. % of amongst village arable Non- of arable population recipients land affected poor Designated All land receiving (Yuan/person/ (%) villages poor villages villages affected some relief year) 0 63.0 49.2 61.1 1 to 24 21.0 28.9 22.1 9.8 8.7 23 25 to 49 6.9 9.1 7.2 35.5 15.5 20 50 to 74 4.2 5.6 4.4 62.0 23.0 12 75 to 99 3.1 3.6 3.1 86.3 21.3 17 100 2.0 3.7 2.2 100 37.2 14 Total 100 100 100 33.2 14.1 19 Non-poor villages 37.1 33.1 14.3 19 Designated poor villages 50.8 34.0 13.1 17 Coastal 29.4 31.3 9.0 41 Northeast 35.8 31.0 11.3 35 Central 43.7 38.1 17.7 14 Southwest 50.2 22.1 11.6 17 Northwest 39.0 42.3 21.3 10 Sources and notes: NBS-World Bank 2004 Village-Level Survey of 3039 villages. Nearly 40% of the villages were reported to be affected by some form of natural disasters in 2004. The proportion was nearly 50% for the designated poor villages which is indicative of their relatively greater vulnerability. Nearly 17% of all villages (22% of designated poor villages) reported a quarter or more of their arable land being affected. Thus, even while natural disasters do not discriminate between the rich and the poor, they remain an important source of vulnerability to poverty in rural areas. As noted above (section 2b), the proportion of rural population that is transitorily poor (poor at least once in three years) is about twice as high as the proportion that is poor in a single year. Disaster relief takes the form of both cash and in-kind transfers to populations in affected areas. The distribution of benefits is determined by the disaster relief plans prepared by county or higher levels of government based on the disaster situation and prognosis for the region. Upon approval of these plans, resources are allocated to townships and villages. The amount of government relief is related to the intensity of the disaster. Most of the funds (about 85% in recent years) go into the Living Relief Fund for food, clothing, reconstruction of homes, and medical care. A relatively small amount goes towards relocation of people in extreme disaster areas. Total government relief though large in absolute terms is less than 3% of the direct economic damage (Table 6.61). Additional government resources are thus needed to improve the coverage and benefit levels of disaster relief which remain limited relative to the scale of the problem. Moreover, most of the government response is ex post in nature and there is a conspicuous absence of disaster insurance. Recently, China Life (China's largest life insurer) launched the county's first accident insurance covering six types of natural disasters including earthquakes, floods, tsunami, typhoons, mudslides and landslides. However, the private uptake of such insurance may be limited, and there is a strong case for the government to consider public provision of disaster insurance with subsidized premia for the low-income groups. Coverage of rural social assistance programs has expanded but remains uneven relative to the levels of poverty The coverage of rural social assistance programs, in particular the rural Di Bao and the Medical Assistance scheme, has improved very rapidly since 2005. Between 2005 and 2007, the number of Di Bao beneficiaries increased more than four times to about 34.5 E n h a n c i n g s e c u r i t y 1 1 9 million, while the coverage of Medical Assistance scheme nearly tripled to about 29 million. At the same time, the incidence of rural poverty has also declined rapidly. The projections for 2007 suggest that a rural poverty rate of 5.4% or about 52 million persons (using the World Bank poverty line and per capita consumption as a measure of welfare). Thus, the gross coverage rates of these social assistance programs now compare favorably with the overall rate of rural poverty. For instance, as against a rural poverty rate of 14.3% in 2004 and a combined Di Bao-Tekun Hu-Wu Bao coverage rate of 2% of the rural population, by 2007 the combined coverage rate had reached 4.2% relative to the (projected) rural poverty rate of 5.4%. This rapid progress over a short span of just three years is indeed impressive. However, for lack of necessary data we do not know much about the targeting performance of these programs. Nonetheless, while rural poverty levels have declined, the current poverty counts also indicate that 52 million in rural areas continue to remain poor in spite of expanded coverage of rural social assistance. Moreover, there are many more who remain vulnerable to poverty. The estimates of risk of poverty presented earlier in chapter 2 suggest that the number of those vulnerable to poverty may be twice as high as the number of those currently poor. Hence, there is still considerable room for fuller coverage of both the remaining poor and others at risk of poverty. The target population for rural Di Bao may be around 100 million, nearly three times the current coverage. Similarly, despite the apparently large amount of central government resources going toward disaster relief (reported to be 6.6 billion Yuan in 2007 according to the Ministry of Civil Affairs), relief efforts are relatively modest in relation to the scale of the problem, and are characterized by limited coverage. In affected villages, the proportion of the population receiving any disaster relief assistance was significantly less than the proportion of land affected: 14% compared to 33% (Table 6.62). Even in villages where the entire arable land was hit by a disaster, only about 37% of the population received any relief. Evidence for 2004 based on the World Bank-NBS Village-Level Survey58 also indicates that coverage rates do not seem to have any systematic relationship with the extent of poverty. For instance, while the incidence of poverty in the designated poor villages is nearly three times higher than in non-poor villages, the coverage of Wu Bao and Di Bao is only 16 and 29 percent higher (Table 6.63). Tekun Hu has a 68% higher coverage in designated poor villages, but Medical Assistance coverage in poor villages is only 41% of that in non-poor villages. Similarly, regional variations in program coverage also appear unrelated ­ or worse, negatively related ­ to the regional variations in poverty. For instance, the lowest rate of Wu Bao coverage is observed for the poorest Northwest region. Di Bao coverage in the Southwest is about half (one-third) of that in the Coastal (Northeast) region, while it's income poverty rate is five times (twice) as high. _________________________________________________________ 58The World Bank-NBS Village-Level Survey was conducted in early 2005 by the NBS, with most of the questions relating to the year 2004. In particular, the survey gathered information from village officials on participation and average benefit levels for the main social assistance programs described above. The survey follows the NBS' Rural Household Survey design and is nationally-representative covering 3036 villages nationwide. For further details on this survey, see Annex on Data Sources. E n h a n c i n g s e c u r i t y 1 2 0 Table 6.63 : Coverage of social assistance programs and poverty in rural China, 2004 % of rural population receiving % of rural population benefits below the World Bank poverty line Wu bao Di Bao Tekun hu* Non-poor villages 11.9 1.1 1.4 3.5 Designated poor villages 31.6 1.3 1.8 5.8 Coastal 5.7 0.9 1.6 3.0 Northeast 13.5 1.2 2.3 4.3 Central 13.5 1.4 1.6 3.8 Southwest 21.3 1.2 0.8 4.4 Northwest 34.8 0.6 1.6 4.4 Total 14.3 1.1 1.5 3.7 Sources and notes: Poverty incidence based on NBS' 2004 Rural Household Survey and refers to consumption poverty using the World Bank poverty line; coverage of social assistance programs based on NBS-World Bank 2004 Village-Level Survey. *Numbers included under Tekun Hu include those receiving disaster/temporary relief. The qualitative poverty assessment in 12 study villages also found that in most research sites the quota for Di Bao determined by the local government was well below the number of households who should be included. For instance, the village leaders in two villages in Inner Mongolia reported that beyond the government quota of 1% of the population, 10 and 40 additional households respectively should have been included in the Di Bao program (CASS-IOS, 2006). Benefit levels for recipients are low for several programs especially in comparison with similar urban programs Average benefits offered by the social assistance programs are also low relative to both average income levels and the poverty line. For instance, based on the NBS-World Bank Village-Level Survey, the average per capita benefit for Wu Bao households is estimated to be 172 Yuan in 2004 (Table 6.64), which is a relatively small fraction of the World Bank poverty line of 931 Yuan per person per year at 2004 rural prices. The benefit levels for recipients of other social assistance programs are much lower: 65, 45 and 42 Yuan per person per year for Di Bao, Tekun Hu and MA respectively. Table 6.64 : Benefit levels of social assistance programs in rural China, 2004 Average benefit per capita amongst recipients Per capita (yuan/person/year) income Medical (2004 yuan) Wu bao Di Bao Tekun hu* assistance Non-poor villages 3215 179 68 45 47 Designated poor villages 2105 129 47 43 15 Coastal 4309 244 136 70 99 Northeast 3481 270 71 72 7 Central 2663 141 36 32 30 Southwest 2265 139 50 41 16 Northwest 1898 105 34 34 31 Total 3076 172 65 45 42 World Bank poverty line in 2004 rural price931 yuan per-capita Sources and notes: Per capita incomes based on NBS' 2004 Rural Household Survey; average benefits of social assistance programs based on NBS-World Bank 2004 Village-Level Survey. *Numbers under Tekun Hu include those receiving disaster/temporary relief. Per capita incomes are expressed in 2004 all-China rural prices. Similarly, the overall benefit levels for disaster relief are low averaging only about 19 Yuan per capita during 2004 (Table 6.62). The qualitative poverty assessment also found that E n h a n c i n g s e c u r i t y 1 2 1 the amount of relief was often inadequate for many households to be able to recover from disasters especially in villages highly dependent on agriculture. For instance, in D village in Sichuan Province, the affected households were given 200 Yuan each after a destructive landslide in 2005, although both the villagers and the township government believed that the loss of each household reached 4200 Yuan (CASS-IOS, 2006). With the recent expansion of the rural Di Bao program average benefit levels have also increased. During 2007, however, the average annual Di Bao expenditure per beneficiary during 2007 was about 302 Yuan. To be sure, these benefits are underestimated to some extent as they are likely to exclude (or only partially include) any in-kind benefits that are sometimes associated with these programs. 59 However, the cash benefits certainly remain modest, especially in comparison with the urban social assistance programs. For instance, per capita cash benefits for urban Di Bao recipients are estimated to be 501 Yuan per year for 2003 (based on the NBS' Urban Short-Form Household Survey data).60 And for 2007, the average urban Di Bao expenditure per beneficiary at about 1210 Yuan was about 4 times that for rural Di Bao at 302 Yuan (Civil Affairs Statistical Yearbook 2007). Benefit levels and coverage vary from areas to area, not with need, but with local fiscal resources Benefit levels also often tend to be lower in poorer regions and villages. For instance, benefit levels in the coastal region are 2-3 times higher than those in western provinces. Similarly, with the exception of Tekun Hu, benefits per recipient are significantly higher in non-poor than in designated poor villages. Levels of per capita benefit for those affected by natural disasters are similar for poor and non-poor affected villages, although they are 2-3 times higher in coastal and northeastern provinces (Table 6.62). There are also concerns with horizontal equity of the social assistance programs as can be illustrated in the case of the rural Di Bao program. The income thresholds reportedly used to determine eligibility for Di Bao vary enormously across villages. As seen in Figure 6.70, while at the bottom end 5% of the villages report thresholds of 120 Yuan or less per person per year, the 95th percentile is 13 times higher at 1560 Yuan. The horizontal inequities are also systematic: richer villages and regions tend to have higher thresholds. For instance, the average threshold in designated poor villages was 713 Yuan compared to 482 Yuan in designated poor villages. Figure 6.70 : Rural Di Bao thresholds vary enormously across villages, 2004 (Percentiles of Di Bao income thresholds, Yuan/capita/year) Average annual per- 1800 capita dibao income 1560 1600 1440 threshold 1400 Non-poor 1200 villages 713 1000 840 Poor villages 482 800 600 600 Coastal 991 400 320 Northeast 486 145 200 120 Central 465 0 Southwest 490 5% 10% 25% 50% 75% 90% 95% Northwest 545 % of villages Total 683 Source: NBS-World Bank 2004 Village-Level Survey of 3039 villages. _________________________________________________________ 59It is unclear to what extent the village officials' responses to questions on program benefits in the Village-Level Survey factored in the value of in-kind benefits. 60See section 6d below for a detailed assessment of the urban Di Bao program. E n h a n c i n g s e c u r i t y 1 2 2 With the exception of disaster relief, there is very little financing of the rural social assistance programs from the central and provincial governments, and these programs rely heavily on local finances. This dependence has had two effects. First, with recent reforms in the local financing system (in particular, the decline and eventual elimination of local fees and agricultural taxes), local revenues, and with them the ability of local governments to finance social assistance schemes, have been decimated. In principle, the financing gaps should be bridged by greater transfers of funds from higher levels of government. However, this has not happened in an adequate or equalizing manner, leading to the second (related) effect: per capita spending on social assistance programs is strongly related the level of local finances, and inequities in the latter are reproduced in the former.61 This is borne out by the evidence from the World Bank-NBS 2004 Village- Level Survey. As shown in Figure 6.71, total per capita spending in the village on the four main social assistance programs (Wu Bao, Di Bao, Tekun Hu and MA) increases with village revenue per capita. Poorer villages that typically have lower per capita revenues spend lower amounts on social assistance, despite their greater need. This effect holds across all villages in the national sample, across villages within counties, as well as between counties. Put differently, transfers from higher levels of government have not been able to eliminate the unequal provision of social assistance across villages in rural China. Figure 6.71 : Social assistance benefits depend on village revenues, 2004 80 60 Across all villages atpiacr Within counties petif 40 neeB Between counties 20 0 0 500 1000 1500 2000 2500 Revenue per capita Source and notes: NBS-World Bank 2004 Village-Level Survey. Benefits per capita refer to total benefits received for the Wu Bao, Di Bao, Tekun Hu and Medical Assistance programs per person in the village. Coverage of rural health insurance has improved dramatically but challenges of limited coverage of rural pensions and low benefit levels of all programs remain (1) Rural health insurance _________________________________________________________ 61This problem of the drying up of the village collectives' income and inadequate transfer payments from the central government to support the social assistance programs was also widely noted in the 12-village qualitative poverty assessment (CASS-IOS, 2006). E n h a n c i n g s e c u r i t y 1 2 3 Prior to the economic reforms of the 1980s, almost everyone in rural areas was covered by some form of health insurance organized under the commune-based Cooperative Medical System (CMS).62 The CMS operated village and township health centers staffed by practitioners with only basic medical training (the so-called "barefoot doctors"), and it is estimated that during the 1970s the CMS system covered about 90% of the rural population. This near-universal coverage of the health care system is credited with major gains in health outcomes in China during the pre-reform period. With the decollectivization of agriculture and the move to the Household Responsibility System during the late 1970s and 1980s, the financial and organizational basis of the CMS was completely dismantled. By 2003, only about 20% of the rural population had health insurance, while out-of-pocket expenses for health care rose rapidly growing in real terms at about 16% per annum during 1978-2003 (World Bank, 2006). The vast majority of the rural population was thus left increasingly exposed to uninsured health shocks that have been shown to result in substantial reductions in household disposable incomes and labor supply (Lindelow and Wagstaff, 2005). The 12-village qualitative poverty assessment also identified illness to be amongst the most important factors leading to poverty for rural households. For instance, in Q village in Inner Mongolia, more than 40% of the households deemed poor had someone in the household who was seriously ill; in X village in Jiangxi Province, 16 of the 18 poor households had sick family members. The cost of medical care for any serious illness is unaffordably high for many households; the qualitative assessment gives numerous examples where households have either had to incur considerable debt to finance medical expenses or had to simply forgo treatment in many instances (CASS-IOS, 2006). The main government initiative in response to this situation is the National Cooperative Medical Scheme (NCMS), a new rural health insurance program that was introduced on a pilot basis in 2003, and has rapidly expanded since. At the end of 2007, 2451 counties (86% of total counties in the country) have established the NCMS, covering about 730 million rural residents, as compared with 641 counties (22% of all counties) covering 163 million rural residents in June 2005. The progress is impressive. The NCMS is a voluntary health insurance scheme operating at the county level. Household contributions are fixed at 10 Yuan per person per year supplemented with a local government contribution of 20 (formerly 10) Yuan per person and a central government contribution also of 20 (formerly 10) Yuan per person for counties in the poorer central and western provinces. In many counties, however, NCMS budgets exceed the total planned contributions. As a result, in practice member contributions are often higher (sometimes up to four times higher) than 10 Yuan (World Bank, 2006). The target for 2008 is to double the three-party contributions to 40-40-20 Yuan. While the recent rapid expansion of the NCMS has gone a long way towards realizing the objective of universal health insurance, some challenges remain. First, the reimbursement benefits under NCMS are relatively low.63 Once the varying schedules of deductibles, reimbursement rates and ceilings are applied, on average, the amount reimbursed is estimated to be only about one-sixth of the out-of-pocket expenses for medical care (World Bank, 2006). There is also inadequate reimbursement of outpatient expenses; more than 80% of reimbursements are for inpatient costs (mostly incurred at county or provincial level hospital rather than township health centers). Concerns also remain about the often cumbersome reimbursements (see Box 6.17 ), perceived or actual inaffordability of contributions or copayments, and adverse selection (healthier households opting out of the scheme). Participation rates of poor households _________________________________________________________ 62Within the collectivized system, village communes took the responsibility not only for organizing production and distribution but also for providing social services including health care. Financing for CMS came from three sources: a contribution by rural households equivalent to 0.5-2% of their annual income, a contribution by the village of a fraction of its income from agriculture and rural enterprises, and a subsidy from higher-level government used mostly to pay for health workers' salaries and medical equipment. 63NCMS is primarily a patient reimbursement scheme where the insured pay the costs of medical care upfront and get reimbursed later. Reimbursements are subject to varying deductibles, reimbursement rates and reimbursement ceilings. E n h a n c i n g s e c u r i t y 1 2 4 have been high, partly as a result of the Medical Assistance program that (as discussed above) has aimed to cover the NCMS contributions or copayments for poor households. Evidence from a 10-county impact evaluation also suggests that NCMS has been successful in increasing use of health care services, but it has been less successful in reducing poverty through a reduction in out-of-pocket medical expenses, which in many instances have risen partly as a result of greater utilization of health services (World Bank, 2006). Box 6.17 : Reimbursements under the new rural health insurance scheme are very low Low levels of reimbursement were also noted as an important factor influencing farmers' participation in NCMS during the qualitative poverty assessment, as illustrated by the following responses during fieldwork in J village in Sichuan Province: "We did not participate in NCMS because the percentage of reimbursement is small. My friend spent more than 1000 Yuan on an appendectomy and was only reimbursed 100 Yuan. In order to get reimbursed he had to pay transportation costs of nearly 100 Yuan. At the same time, his farming time was lost. It is meaningless for us." "I did not get reimbursed even one cent last year. Therefore, we did not participate this year. " "I spent about 700 Yuan in the county hospital, and I was told that only 7 Yuan can be reimbursed. I did not do it. 7 Yuan is not enough even for transportation." "Last year my older brother suffered a nosebleed and we paid more than 1000 Yuan for it to be treated. Less than 10 Yuan was reimbursed." Field research also found that the complex process of reimbursement is also another factor limiting participation. Farmers cannot get reimbursed quickly when they are in dire need of money. Farmers' responses also indicated that in the absence of competition and effective management, there is an escalation of medical costs under the rural health insurance system: "In the drugstores which have been designated to provide medicines for patients who have been included in NCMS, the prices of medicines are 30% higher than those in common drugstores. ...... The NCMS is good. But it is unreasonable that its medicines are more expensive than those in drug stores. The prices of medicines which participants of NCMS can get reimbursed for are always more expensive than other medicines." Source: CASS-IOS (2006). (2) Rural pension system Started in early 1990s with the aim of providing old-age security to rural residents aged 20-60 years, the rural pension system in China has remained very small-scale in nature. After an initial period of growth which saw participation climb to about 75 million (16% of the rural labor force), the scheme stalled due to lack of supporting policy, low payment capability and difficulties in pension fund mismanagement. The scheme was restructured and scaled down in 1999 and participation has remained fairly stagnant at about 54 million or about 10-11% of the rural labor force in recent years (Table 6.65).64 _________________________________________________________ 64In January 2006, the Ministry of Labor and Social Security launched a 3-year pilot for a rural pension scheme in 8 counties/districts. E n h a n c i n g s e c u r i t y 1 2 5 Table 6.65 : Rural pension insurance system, 1997-2007 Year 1997 2002 2003 2004 2005 2006 2007 Contributors Number of rural laborers contributing (millions) 75.42 54.62 54.28 53.78 54.42 53.74 51.71 Number of rural laborers (millions) 462.34 485.27 489.71 496.95 503.87 510.89 518.00 Share of rural labor force contributing (%) 16.3 11.3 11.1 10.8 10.8 10.5 10.0 Beneficiaries Number of beneficiaries (millions) 1.23 1.98 2.05 3.02 3.551 3.92 Number of rural elderly: age 60-plus (millions) 86.07 87.61 89.34 92.37 92.98 93.60 Fraction of rural elderly receiving pensions (%) 1.4 2.3 2.3 3.3 3.8 4.2 Sources and notes: NBS: China Statistical Yearbooks and China Population Statistics Yearbooks (several years). The scheme is mainly funded by personal contributions, collective subsidies and government policy support. Participants above the age of 60 are eligible for pension payments. The average benefit per pensioner was about 1000 Yuan per year (or about 85 Yuan per month) in 2007; this is about one-twelfth of the average benefit levels for urban pensioners (see section 6d). The contribution of collective allowances, though significant (54% of insurance revenues in 2002; the remaining 48% coming from individual contributions), varies enormously from province to province, with the predictable pattern of higher collective allowance per participant in richer provinces. The number of rural pension recipients has grown in recent years, but still remains very limited at about 3.9 million in 2007, or about 4% of the rural population aged 60 and above (Table 6.65). Coverage of the farm-based elderly population appears to be particularly limited. According to the China Urban and Rural Elderly Survey of 2000, most (more than 60%) of the rural pensioners have non-agricultural hukou (China Center on Aging, 2002). The distribution of beneficiaries is also highly concentrated in a handful of mostly wealthy provinces. For instance, in 2002, the five provinces of Shanghai, Jiangsu, Zhejiang, Shandong and Sichuan accounted for 76% of all pension recipients and nearly 90% of total pension benefits.65 Overall, therefore, given the limited and uneven reach of the current system, pensions are not a source of old-age security for the rural elderly who continue to rely predominantly on their children and family members for support during the later years of their life. Finding ways to address this will be particularly important in a context where demographic transition together with the cumulative effects of the one-child policy is fast propelling the country towards a rapidly aging population with an increasing dependency ratio. Policy implications Over the last 3-4 years, there has been significant progress in the expansion of several elements of the rural social protection system, which is beginning to address its longstanding underdeveloped state. The recent initiatives are consistent with the recognition that access to land alone, despite the relatively egalitarian distribution of use rights, cannot be relied upon to provide an effective rural safety net. The main thrust of these initiatives has been on expanded coverage, and as a result, the coverage of key programs, in particular the rural Di Bao, rural heath insurance and the Medical Assistance scheme, has rapidly improved. At the same time, however, the coverage of other programs such as rural pensions and disaster relief remains limited, and the benefit levels of most rural programs are quite low. This is in contrast to the urban sector where following the major restructuring of the state-owned enterprises during the 1990s, a much greater level of government effort and resources have been devoted to the development of a relatively modern non-enterprise-based social welfare system. The recent progress is starting to bridge the gap between the rural and urban systems, but much still remains to be done. Overall, there remain four critical challenges for the rural social protection system. _________________________________________________________ 65These five provinces also account for about half of all participants in the scheme (NBS, 2003). E n h a n c i n g s e c u r i t y 1 2 6 · The coverage (in terms of the number of beneficiaries) of several programs needs to be expanded to more fully cover the rural poor and vulnerable population. · The benefit levels of several programs, which are only a fraction of the benefits under urban programs, will also need to rise if they are to offer effective protection to those covered by the programs. · There is a need for coordination and harmonization of different programs within the rural social protection system as well as between the rural and urban systems. A harmonized national system is important not only for bridging the rural-urban gap in the social protection, but also for ensuring adequate protection to migrant workers who are currently only imperfectly covered by the rural and urban systems. · All this will require further infusion of funds from the central and provincial governments. Such commitment of resources from higher levels of government is also needed to break the dependence of social protection spending on local fiscal capacity, and hence address the wide disparities in both the coverage and benefit levels of programs within rural areas due to the large differences in the revenue base of local governments. There are specific challenges for individual programs. · Rural Di Bao: The rural Di Bao has clearly emerged as the main focal rural assistance program. Its rapid expansion since 2005 has been impressive with gross coverage rate now equivalent to about two-thirds of the number of rural poor. However, as it expands further, the program will need to be streamlined with a more consistent approach to coverage and targeting. Here, many of the challenges will be of a similar nature to those already being encountered with the more established urban Di Bao program (see section 6e for further discussion), namely, realizing the scope for better targeting and hence improved coverage of the poor, rationalization of Di Bao lines and eligibility criteria across different local governments so that program coverage is better aligned to needs rather than local fiscal capacity, and development of a benefit design framework that limits adverse work incentives. · Rural health insurance: The recent rapid expansion of rural health insurance is indicative of its recognition as an important priority for the government. With an estimated participation rate of 86% at the end of 2007, the coverage issue has been largely addressed. The next challenge will be to ensure the provision of an adequate level of benefits that can secure a sizeable reduction in the burden of private health costs. The targeted increase in the budget through a doubling of contributions in 2008 will help provide more resources for raising benefit levels, but further increments of funding from central and provincial governments will be needed for net reimbursements to cover a larger fraction of the out-of-pocket health costs. Coverage of outpatient services also remains a further challenge. Another major challenge is the coverage of migrant workers who seem to be left out by both the urban and rural health insurance systems, and this raises the broader challenge of the coordination and harmonization of two systems, which is all the more pertinent now as the NCMS is beginning to operate at a national scale. · Medical Assistance: The Medical Assistance program is an important part of the overall rural health insurance system that has been useful in extending the latter's coverage to the rural poor. However, despite its national scale, its coverage in terms of the number of beneficiaries is still relatively limited; the total number of beneficiaries is equivalent to about half the number of rural poor. Moreover, its relationship with the Di Bao program will also need to be clarified. There is potential for rural Di Bao to serve as a unifying umbrella program with medical assistance as one of its components as both programs aim to target the poor. The government may also want to consider using the MA scheme to introduce a means-tested safety net provision to health insurance whereby after reaching a certain threshold of per capita healthcare expenses in any year, the insured are entitled to higher (even 100%) reimbursements. Such safety net provisions have been used to cushion the impact of large health shocks on household welfare in several countries. · Disaster relief: Additional government resources will be needed to improve the coverage and benefit levels of disaster relief which remain limited relative to the scale of the problem. However, most of the government response is ex post in nature and E n h a n c i n g s e c u r i t y 1 2 7 there is a conspicuous absence of disaster insurance. Recently, China Life (China's largest life insurer) launched the county's first accident insurance covering six types of natural disasters. However, the private uptake of such insurance may be limited, and there is a strong case for the government to consider public provision of disaster insurance with subsidized premia for the low-income groups. · Rural pensions: The reach of rural pension program remains limited and uneven. A significant expansion of the program will almost certainly require a high degree of "subsidization' with a much larger government contribution (into individual pensions), which in turn will need to come from higher levels of government. Such a highly subsidized program however makes it more like a transfer or assistance program for the rural elderly. While such an approach may be quite reasonable, there is need for a careful evaluation of different options for providing support to the elderly while recognizing that social insurance programs need to be considered together with social assistance programs. c. Promoting re-employment in a changing urban labor market Beginning in the mid-1990s, China's urban labor market experienced a rapid diversification of ownership types. The most striking change was the rapid decline of the state and collective sectors. According to official employment statistics published in the China Statistical Yearbooks, employment in the SOE sector, which had actually grown in absolute number from 1990 to 1994, fell at a mean annual rate of 6.4% from 1995 to 2000, a total loss of 31.5 million jobs or 15% of the urban labor force. The collective sector, which had begun declining earlier, had already shed over 10% of the 1990 workforce by 1995, shrank by 13.8% per year from 1995 to 2000, losing 16.5 million jobs over the period. As a share of all urban employment, employment in the state and collective sector fell from 76% in 1995 to 41% in 2000 to only 25% in 2006. The employment shares of other enterprise ownership forms, especially the private sector and limited liability corporations, increased significantly. Two important developments were related to this major restructuring of the urban labor market: the rapid growth of informal employment, and the emergence of high unemployment and declining labor force participation. Figure 6.72 : Rising share of private and unregistered businesses in urban employment (% of urban employment in various enterprise types) Private registered Self-employed 100 80 State and collective 60 40 Other unregistered 20 0 1978 1985 1990 1992 1994 1996 1998 2000 2002 2004 2006 Sources and notes: NBS 2006 China Statistical Yearbook. The "other" category is obtained residually as the difference between total urban employment as estimated from labor-force surveys and that reported by state and collective units, and registered private businesses and self-employed individuals. It primarily relates to those in some form of informal employment in urban areas. Nearly half of all urban employment is informal in nature As seen in Figure 6.72, the category of urban workers that witnessed the fastest rate of increase was "other" workers, derived residually as the difference between total urban E n h a n c i n g s e c u r i t y 1 2 8 employment and the sum of urban employment in different ownership types. Starting from a low base in 1995, "other" workers increased by 24% per year and accounted for 39% of all workers in 2003, falling slightly to 35% in 2006. These workers can be described as "missing" workers, because they reflect differences in the numbers of workers reported through different statistical reporting systems. Total urban employment comes from labor force surveys, while the number of workers employed by different ownership types comes from administrative reporting by all independent accounting units. Estimates of self- employed workers are based on the number of such individuals registered with the Industrial and Commercial Bureau. In principle, missing workers include unreported urban workers and unregistered informal employment, including undocumented work by migrants in urban areas. The large increase in the number of such workers thus suggests that informal urban employment may have increased significantly in recent years. Although the term "informal" lacks precise definition, the ILO emphasizes that informal employment is characterized by a lack of stability or security. To some extent, the very fact that workers are not registered by their employers itself is prima facie evidence that such work is informal, and so there is little if any monitoring of how such workers are treated by their employers (e.g., job termination, social insurance). While part of the increase in informal employment may be "statistical" reflecting some of the previously unrecorded economic activity, especially in the tertiary sector,66 most of it seems to be on account of the rapid growth in the unregistered and imperfectly monitored private sector, unreported migrant workers, as well as some informal employment in the state and collective sector. All enterprises, including those in the state and collective sector, probably will hire workers on an informal basis to avoid having to pay required payroll taxes for pensions, unemployment insurance, medical insurance, and work injuries, which can be quite costly (e.g., in Shanghai in 2004, the regular employer contribution rate for these types of insurance totaled 36.5% of wages). Hiring workers on an informal basis also gives employers greater flexibility to adjust the size of the labor force to respond to changes in economic conditions. Cai,Fang and Wu (2006) analyze a 66-city survey of workers conducted by the Ministry of Labor in 2002, and find that in the state and collective sectors 23% of employment is informal (Table 6.66). The share of informal employment is higher for women, amongst youngest and oldest workers, for migrants and amongst less educated workers. Table 6.66 : Share of informal employment for local residents in urban China in 2002 Share of Share of Share of informal informal informal employ- employ- employ- ment ment ment (%) (%) (%) All 45.3 By age: By education level: By sector: 16-24 57.2 Primary and below 78.4 State and collective 22.9 25-34 44.6 Middle school 62.6 Other sectors 56.3 35-44 42.2 High school 48.6 By sex: 45-54 42.2 Junior technical school 35.9 Male 42.2 55-64 55.6 Specialized college 24.6 Female 49.5 >65 83.9 College and above 14.8 Source and notes: Cai, Fang and Wu (2006). The authors use nine criteria to define informal employment, the most important of which include: self-employed workers, those working on a temporary or hourly basis, those without a labor contract and not considered officially registered workers. Table 6.67 provides further evidence on the extent of formal and informal employment among different types of employers, drawing upon survey data from the 2005 China Urban Labor Survey (CULS) conducted in five large and five smaller cities in different regions of _________________________________________________________ 66Based on the Economic Census of 2005, in January 2006 the National Bureau of Statistics made a 50% upward revision of the GDP for the tertiary sector for 2004 (with corresponding adjustments going back to 1994) raising the share of tertiary sector in GDP from 32 to 41%. E n h a n c i n g s e c u r i t y 1 2 9 China. An advantage of the CULS data is that it also surveyed migrants in the same cities. The Table reports the share of workers with formal labor contracts, broken down by type of employer. Lacking a labor contract in written form is of course only one indicator of informal employment, and it is even possible that some permanent workers will not recall signing an official document and so report not having a labor contract even though they are permanently employed. According to the survey data, 62% of urban employees signed labor contracts in urban enterprises while only 13% of migrants had signed labor contracts. As expected, the share of those with labor contracts is highest in the state sectors. It is very low in the private sector, with 27% of urban employees and only 7% of migrants having labor contracts. Across all ownership categories, the share of workers with labor contracts is lower for migrants (who are mostly in the private sector) than for local urban employees. Table 6.67 : Share of urban workers with formal contracts, 2005 Local residents Migrants Employment % with labor Employment % with labor share (%) contract share (%) contract All 100.0 61.8 100.0 12.9 Government 23.6 76.2 3.0 43.2 State-owned 30.6 88.5 3.2 55.6 Collective 5.2 71.7 1.6 47.4 Private 28.1 26.5 84.7 7.0 Foreign/joint ventures 5.7 74.9 2.3 57.8 Other 6.8 28.4 5.4 18.0 Source and notes: World Bank estimates using data from the 2005 China Urban Labor Survey 2 of 5 large and 5 small cities, equally weighted. Thus, factoring in that migrants account for about 30% of all urban employment, the share of informal employment in urban areas based on the 2005 CULS data is around half, even higher than the 45% estimate of Cai, Fang and Wu (2006). The informalization of the Chinese labor market has both positive and negative implications for the poor. On the positive side, the informal sector creates jobs and allocates labor based on market principles, thus furthering the development of the labor market. On the negative side, informalization creates new policy challenges because the government is poorly informed about informal economic activities and has difficulty regulating such activity. This makes it hard to enforce protective regulations for workers that ensure safe working environments and fair treatment of workers. It also makes it difficult to develop sustainable social insurance systems that cover most of the population and which are adequately financed. High unemployment and declining labor force participation The breaking of China's "iron rice bowl" for state-sector workers has created new income insecurity for urban households. Aggressive restructuring of loss-making state-owned enterprises starting in the mid-1990s led to the layoffs of tens of millions of urban workers. According to official data, at least 37 million workers were laid off by 2004 (Cai, Park, and Zhao, 2008). For the first time since the founding of the PRC, China's urban areas experienced high rates of unemployment, causing substantial dislocation and hardship for many urban families. Table 6.68 provides different estimates of China's urban unemployment rate since 1996. The officially reported unemployment rate, calculated as the number of administratively registered unemployed workers divided by the total number of employed and registered unemployed workers, increased from 3.1% in 1996 to 4.3% in 2003, falling slightly to 4.1% in 2006. Most agree that official figures significantly underestimate the true rate of unemployment since many unemployed workers are not officially registered as such. Standard international practice is to measure unemployment rates from sample surveys rather than administrative data. Internationally, unemployment is defined as not working but available and looking for work in the past week. Alternative estimates that try to approximate this definition yield higher estimates of the unemployment rate. E n h a n c i n g s e c u r i t y 1 3 0 Table 6.68 : Alternative estimates of China's urban unemployment rate Official NBS labor CULS-based CULS-based registered: all force surveys: extrapolations: extrapolations: NBS-UHS: Year urban all urban urban residents all urban all urban 1996 3.0 3.9 6.8 4.5 1997 3.1 4.3 7.7 5.0 1998 3.1 6.3 8.5 5.6 1999 3.1 5.9 9.0 5.9 2000 3.1 7.6 10.0 6.5 2001 3.6 5.6 10.8 7.0 2002 4.0 6.1 11.1 7.3 2003 4.3 6.0 8.8 2004 4.2 5.8 2005 4.2 7.0 6.7 4.4 2006 4.1 6.1 Sources and notes: 1.China Labor Statistical Yearbook (2005); 2.Cai, Park, and Zhao (2006); 3.Giles, Park, and Zhang (2005); 4. World Bank calculations based on 2003 NBS Urban Household Survey. One way to estimate urban unemployment is to calculate the difference between urban employment and the urban economically active population, which includes employed workers and those not working but actively looking for work at the time of the survey. The urban economically active population is calculated by subtracting the rural labor force from the total economically active population. All figures come from the annual labor force survey conducted by NBS, which includes both local residents and migrants. Using this method, the estimated unemployment rate increased from 3.9% in 1996 to 6.3% in 1998, and remained in a range of 5.6% to 7.6% thereafter with no clear trend (2nd column of Table 6.68). The high estimate of 7.6% in 2000 may be due to the fact that it is based on the 2000 census rather than sample surveys. A third set of estimates are from Giles, Park, and Zhang (2005), who analyze the China Urban Labor Survey data collected in 2001 and 2002 in 5 large cities using an internationally standard definition of unemployment, and then extrapolate national estimates over time using the 2000 census data and retrospective survey data. This report extends the estimates to 2005 using new survey data from the second wave of the CULS in the same 5 cities. In the five large cities, the unemployment rate of local residents increased from 7.7% in 1996 to 14.0% in 2002, then declined to 8.5% in 2005. Using the 2000 census data to compare the unemployment rate in the 5 large cities with that of all of urban China, the national rate of unemployment of local urban residents is estimated to be 6.8% in 1996, 11.1% in 2002, and 6.7% in 2005. In addition to these figures, the 2003 NBS urban survey data (which includes very few migrants) is used to calculate an independent estimate of the urban unemployment rate of local residents of 8.8%, which is consistent with falling unemployment from 2002 to 2005. The unemployment rate estimates for local residents is larger than those for all residents because the unemployment rate of migrants is much lower than that of urban residents. Analysis of the 2000 census data produces unemployment rate estimates of 4.7% for migrants and 12.7% for local residents. Based on this relative difference, Giles, Park, and Zhang (2005) estimate that the unemployment rate for all of urban China was 7.3% in 2002, compared to 11.1% for local residents. Using the same assumption, the urban unemployment rate for all residents in 2005 fell to 4.4% in 2005 (Table 6.68). In addition to rising then falling unemployment, China's recent period of economic restructuring was characterized by a steadily declining labor force participation rate. Figure 6.73 plots the labor force participation rate defined as the urban economically active population divided by the total urban population aged 16 and older. The main decline occurs after 1999, with the labor force participation rate falling by 10 percentage points from 1999 to 2003. Because the decline was associated with restructuring, there should be concern that many of those leaving the labor force were discouraged workers. E n h a n c i n g s e c u r i t y 1 3 1 Figure 6.73 : China's declining urban labor force participation rate: 1996 to 2006 (% of population aged 16 and older who are economically active) 80 73 72 73 71 70 66 67 66 63 64 65 64 60 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 Sources and notes: China Statistical Yearbook, various years. Table 6.69, breaks down the labor force participation rates and unemployment rates for those age 16 to 60 by demographic and education groups using the 2003 NBS urban household survey data. Women now have a much lower labor force participation rate than men (70.9% compared to 83.6%), and this difference increases with age and decreases with the level of education. The unemployment rate was higher among the less educated, younger workers, women, and workers in the northeast and southwest. Table 6.69 : Disaggregated urban labor force participation and unemployment rates in 2003 Labor force participation rate Unemployment rate Men Women All Men Women All All age 16 to 60 83.6 70.9 77.1 6.6 11.2 8.8 By education level Primary school 78.4 39.1 52.3 9.8 20.7 15.2 Middle school 81.5 61.5 70.7 9.4 17.6 13.3 High school 81.0 75.0 77.8 7.9 11.4 9.7 College and above 89.7 84.5 87.5 2.7 3.9 3.2 By age group 16-30 50.8 55.3 53.1 23.8 21.8 22.8 30-40 99.3 94.2 96.5 4.3 10.6 7.6 40-50 98.6 88.2 93.3 3.6 9.0 6.2 50-60 81.7 35.5 58.7 4.6 4.5 4.6 Sources and notes: World Bank calculations from national sample of NBS 2003 UHS. Urban poverty is linked to the work status of household members The linkage between unemployment, labor force participation, and relative poverty in urban areas was already noted in chapter 4. That analysis indicated that an adult member being unemployed or being out of the labor force significantly increased the risk of an urban household being relatively poor. Table 6.70 refines that analysis by examining whether having a member who is unemployed or out of the labor force has different consequences for family welfare in terms of relative poverty depending on the demographic group of the non-working member. Here, relative poverty, or disadvantaged status, is defined as having income per capita below three times the World Bank (urban) poverty line. Families are hurt the most by having an unemployed worker in their prime earning years (age 40 to 50), especially if it is a man. Also, for younger workers (age below 40) being out of the labor force is as correlated with poverty status of household members as is being unemployed. Such workers may be discouraged workers whose families are vulnerable without their earned income. The weak overall relationship between poverty and having an adult household member out of the labor force is due to the fact that retired persons, many with pensions, comprise a large share of those out of the labor force. Thus, while unemployment is the clearest indicator of poverty, being out of the labor force also may be an indicator of poverty for younger adults. E n h a n c i n g s e c u r i t y 1 3 2 Table 6.70 : Urban relative poverty rate with unemployed and out of labor force family members Relative poverty rate of Relative poverty rate of population with unemployed population with out of labor family member who is... force family member who is... ...male... 29.9 8.3 ...16 to 30 years old 25.4 20.9 ...30 to 40 years old 37.0 41.3 ...40 to 50 years old 38.9 21.0 ...50 to 55 years old 31.7 10.2 ...more than 55 years old 22.9 7.7 ...female 28.6 11.2 ...16 to 30 years old 28.5 36.1 ...30 to 40 years old 30.2 34.9 ...40 to 50 years old 29.6 18.7 ...50 to 55 years old 22.2 8.9 ...more than 55 years old 48.5 9.1 Sources and notes: World Bank calculations from national sample of NBS 2003 UHS. Relative poverty is defined in terms of per capita incomes below three times the World Bank urban poverty line. Promoting re-employment in urban areas The challenge of employing dislocated urban workers has both a macroeconomic and microeconomic dimension. Some wondered why China was having such a difficult time creating new jobs during a period (since the mid-1990s) in which the rate of economic growth remained very high. Part of the difficulty undoubtedly was due to the sheer number of workers who lost jobs during the period of economic restructuring. But industrial and financial policies also played a role by encouraging capital- rather than labor-intensive industrial growth. China's industrial development strategy emphasized the development of capital-intensive industries such as automobiles, machinery, and steel, which are viewed as keys to modernization. The banking system has continued to steer credit to large SOEs at below-market interest rates, distorting the relative price of capital and thus reducing labor demand. Private enterprises, many of which are small or medium in size and highly labor-intensive, continue to find it difficult to obtain credit from state commercial banks. This is despite the fact that the private sector has accounted for the majority of new job creation since the mid-1990s (Rawski, 2003). How successful were laid off workers in finding new jobs? According to the China Urban Labor Survey in five large cities, 34.8% of individuals experiencing job separations between January 1996 and November 2001 were employed again within 12 months of leaving their jobs, and 44.7% were employed again by the end of the period (Giles, Park, and Cai, 2006a). Other research found that by the year 2000, 47% of retrenched workers had found new jobs (Appleton et al., 2002). These two studies, as well as one by Maurer- Fazio (2005) found that out-of-work duration was shorter for men, the young, and the better educated, the same groups that were less likely to have lost their jobs in the first place. Two policy issues related to re-employment are the possible disincentive effects of government subsidies to those who lose jobs (e.g., xiagang subsidies, unemployment insurance, urban minimum living standard payments), and the effectiveness of training programs in providing skills that increased the skills and employment prospects of dislocated workers. Research analyzing the CULS individual work histories from 1996 to 2001 found that access to such subsidies reduced the probability of reemployment within a year by 34% for men, but did not significantly influence the reemployment probability of women (Giles, Park, and Cai, 2006b). Beginning in 1998, the government mandated that state-owned enterprises and local government labor bureaus provide training as well as job referral assistance to laid off workers, analysis of the CULS data found that only 12% of laid-off (xiagang) workers reported that they were requested to participate in training activities, and many of these E n h a n c i n g s e c u r i t y 1 3 3 did not bother attending (Giles, Park, and Cai, 2006b). A survey conducted by the Ministry of Labor found that only 8.79% of dislocated workers reported participating in training activities, while 20.9% received some type of non-financial assistance from the local government (e.g., consultation, vocational guidance). It is difficult to evaluate the impact of training on reemployment and wages, because participation in training is a choice that may reflect the unobserved ability and motivation of the participants. One systematic evaluation of training programs in two cities in the late 1990s found that in Wuhan training increased the probability of employment by 9-12% but did not affect wages, while in Shenyang training did not affect reemployment probability but increased wages by 10-20% (Bidani et al, 2005). The differences in outcomes could reflect differences in labor demand, training content, and who the trainings targeted. International experience has found that the effectiveness of active labor market policies depends critically on the how programs are designed and implemented. Successful programs tend to be those that are responsive to labor market demand (if possible by involving employers), provide supporting services, especially in job search, and target those who are most likely to benefit from the training. Access to information about new job opportunities also plays a key role in facilitating reemployment of dislocated workers. The study of reemployment using the CULS data found that the size of social networks significantly influenced the likelihood of finding a new job for both men and women (Giles, Park, and Cai, 2006b). Indeed, over 40% of workers finding new jobs relied on referrals from relatives and friends. In contrast, only 20% of new jobs were found through reemployment centers or other government agencies. According to the CULS survey data, employment referral services were also provided by about two thirds of urban residence committees (juweihui, or shequ, are neighborhood organizations organized by municipal governments), and individuals living in neighborhoods with such services were more likely to find new jobs. Policy implications The foregoing discussion suggests following policy implications for the new employment challenges in urban areas described in this section: · Urban poverty is strongly associated with having adult household members who are unemployed, and having younger workers who have exited the labor force. A high priority should therefore be placed on accurate, timely, and publicly available measurement of labor force outcomes, especially the unemployment rate. · Industrial and financial policies may be limiting job creation by encouraging capital- intensive production, and may thus need to be reviewed. · Subsidies to laid-off and unemployed workers appear to have created a disincentive to work for men but not for women. While the xiagang subsidies have now been replaced by the unemployment insurance system, there is still need to consider how unemployment insurance can contribute to promoting employment and re- employment. · Active labor market policies have not played a major role in the reemployment of dislocated workers, and must be designed appropriately to be successful. These policies may need to be better tailored to the less educated, women, and older workers who have been particularly disadvantaged with respect to losing jobs and encountering difficulties finding new employment. Access to information and referral services can be a useful component of such policies. d. Replacing the iron rice bowl: urban social insurance programs Since the mid-1990s, China has completely revamped its urban social security system. The main change has been the shift from an enterprise-based system where workers' social welfare was the prime responsibility of their work units to one that is largely administered by the government. In many respects, this was the logical consequence of the massive restructuring of the economy especially since the mid-1990s, with a dramatic reduction in the size of the state-sector. As noted above, the share of state-owned and E n h a n c i n g s e c u r i t y 1 3 4 collective enterprises in total urban employment shrunk from 96% in 1985 to 76% in 1995 to only about 25% in 2006 (Figure 6.72). The massive lay-offs, rising unemployment and declining labor force participation associated with this process along with the growing inability of the remaining restructured state sector to fulfill its social security obligations meant that an enterprise-based structure was no longer viable. The shattering of the "iron rice bowl" made the reform of the urban social protection system both necessary and urgent. Features of the main urban social insurance programs The trends in the Chinese labor market discussed in the previous section together with the demise of the "iron rice bowl" system of guaranteed state-sector jobs and social security introduced new vulnerabilities into the lives of China's urban population that persist to the present, creating new challenges for the urban social protection system. The reformed but still evolving urban social protection system has five main social insurance programs and two social assistance programs, namely the urban Di Bao program and the urban medical assistance scheme. The core social insurance programs include pension insurance, basic medical insurance, unemployment insurance, work injury insurance, and maternity insurance. The rest of this section discusses the main features of this package of programs and the key policy issues relevant to their effectiveness in offering adequate social protection to the urban population. The urban Di Bao program is discussed in detail in section 6e, and the medical assistance scheme ­ although a social assistance program ­ is discussed together with the urban Basic Medical Insurance program as the assistance it offers is closely linked to the urban health insurance program. (1) Pension program Following a series of reforms since the 1990s, the current urban pension system in China has the following structure: (i) a basic defined-benefit pension plan whose standard is based on the monthly average wage of employed workers and individual average salary (1% is paid with each year that the worker pays into the pension), financed entirely by enterprise contributions of 20% of the wage bill; (ii) a mandatory defined-contribution plan via individual accounts financed through individual contribution of 8% of wages, entitling the pensioner to a monthly payment based on the account balance divided by the number of months as determined by the life expectancy for retired urban and township employees, retirement age, deposit interest and other factors; and (iii) a voluntary supplementary pension plan (enterprise annuity) managed by enterprises. The voluntary plan is still at an early stage of development, but has been expanding, covering over 10 million employees from over 24,000 enterprises by the end of 2006. The pension system also extends to the self-employed in urban areas who, according to standard, pay 20% of the average local wages for the employed workers of the previous year. The current retirement age is 60 years for men and 55 years for women, and retirees become eligible for pensions after 15 years of contribution into the pension plan.67 The expected replacement rate (based on both the basic and the individual account plans) is estimated at about 58% for an individual contributing for 35 years (Trinh, 2006). In 2007, about 152 million urban workers (about 52% of the total urban employed) participated in the pension program (Table 6.71). While the absolute number of participating workers increased by 74% since 1995, the proportion of the urban work force covered by the pension program has increased more gradually from 46% in 1995 to its current level of 52%. On the other hand, the number of retirees receiving pension benefits has been increasing faster, by 110% since 1995 to about 50 million in 2007. The average benefit level per pensioner was about 12000 Yuan in 2007, up from 9250 Yuan in 2005. _________________________________________________________ 67For women workers as opposed to cadres, retirement age is 50 years. E n h a n c i n g s e c u r i t y 1 3 5 Table 6.71 : Coverage of urban social insurance programs in China, 1995-2007 Number of Benefit benefi- per Number of participating workers ciaries beneficiary (Yuan /person Total (million) As % of urban employment (million) /year) 1995 2000 2005 2007 1995 2000 2005 2007 2007 2005 2007 Pension insurance Total 109.8 136.2 174.9 201.4 Employees and workers 87.4 104.5 131.2 151.8 45.9 45.1 48.0 51.7 Of which, migrant workers 18.5 Retirees 22.4 31.7 43.7 49.5 49.5 9251 12041 Basic Medical Insurance (BMI) Total BMI 7.5 37.9 137.8 223.1 Urban Employee BMI Employees and workers 7.0 28.6 100.2 134.2 3.7 12.4 36.7 45.7 180.2 783 867 Of which, migrant workers 31.3 Retirees 0.4 9.2 37.6 46.0 Urban Resident BMI 42.9 42.9 Unemployment insurance Total 82.4 104.1 106.5 116.5 43.3 45.0 39.0 39.7 2.86 5711 7622 Of which, migrant workers 11.5 Work Injury insurance Total 26.1 43.5 84.8 121.7 13.7 18.8 31.0 41.5 0.96 7296 9167 Of which, migrant workers 39.8 Maternity insurance Total 15.0 30.0 54.1 77.8 20.5 34.1 52.3 70.0 1.13 4419 4956 Sources and notes: NBS China Statistical Yearbook (various years); NBS (2008). The number of participating workers for Basic Medical Insurance (BMI) does not include those covered under the earlier GIS and LIS health insurance programs. Per capita benefits for BMI are calculated by dividing the total BMI expenditure by the sum of contributing employees and workers and the retirees who are nonetheless covered by the scheme. For maternity insurance, the participation rate is expressed as percentage of female urban employment estimated using the share of female employment in urban units as reported in the China Labor Statistical Yearbooks. (2) Basic Medical Insurance Starting at the end of 1998, China established the new Basic Medical Insurance (BMI) program in urban areas which has gradually subsumed the pre-reform Labor Insurance Scheme (LIS) for state-owned enterprise workers and the Government Insurance Scheme (GIS) for civil servants and other government employees. BMI is financed by contributions from both employers and employees with the former contributing between 6-10% of the wage bill and the latter contributing 2% of individual wages. Retirees are usually exempt from BMI contributions. All individual contributions and 30% of employer contributions go into individual medical savings accounts (MSA) and the remaining 70% of employer contributions go into a social pool account (SPA). Medical expenses are shared by the individual and the BMI. Outpatient expenses are usually met from MSA funds, while inpatient costs are mainly paid from SPA funds. Reimbursements for medical expenses are subject to varying levels of co-payments, deductibles and ceilings. The scheme's coverage has rapidly expanded over the last decade. It currently (at the end of 2007) covers about 134 million urban workers (and about 46 million retirees) accounting for about 46% of the total urban employed (Table 6.71). Survey-based evidence points to a somewhat higher though comparable coverage rate: according to the CULS in 12 cities in 2004 and 2005, 52% of urban permanent residents had medical insurance. Average benefit levels have increased from 783 Yuan in 2005 to 867 Yuan in 2007. An important recent policy development is the introduction in September 2007 of the Urban Resident Basic Medical Insurance (URBMI), a voluntary health insurance scheme that aims to cover some of the urban residents who are not covered by the existing BMI which mainly covers urban formal sector workers. URBMI is an attempt to reach in E n h a n c i n g s e c u r i t y 1 3 6 particular students, children, and unemployed urban and township residents. Similar to the BMI and NCMS, URBMI focuses largely on inpatient expenses and catastrophic outpatient expenses. The scheme is managed at the municipal level by MOLSS. It is financed mainly from household contributions supplemented with subsidies from different levels of government. For every enrollee, there is a provision for government subsidies of at least 40 Yuan, up to 80 Yuan in 2008. For the central-western region, the central government will provide 20 Yuan per enrollee, and for the poor and vulnerable groups (such as Di Bao, disabled children or students, poor elderly) the government will provide additional subsidy. Municipal governments have the discretion over the total contribution per person based on local economic conditions. Following some preparatory pilots by local governments in 2005, the scheme was rolled out for a nationwide pilot in 88 urban cities and townships in September 2007. By the end of 2007, nearly 43 million were already enrolled in URBMI, and there is an ambitious plan to roll out the scheme to 229 cities in 2008. In 2005, the State Council also decided to introduce the Medical Assistance (MA) program in urban areas to provide financial assistance for medical expenses to the urban poor and disadvantaged. Analogous to the rural program, the urban MA also assists the target population with contributions and copayments for the new urban resident BMI. Since the initial pilots of 2005, urban MA has grown rapidly, and by the end of September 2007, 86% of urban cities/counties had established the program. In 2007, the number of beneficiaries under the program had grown to 4.07 million, up from 1.63 million in 2005. (3) Unemployment insurance Though its origins date back to 1986, the unemployment insurance (UI) program was introduced on a more standardized basis in 1999 and its coverage extended to all enterprises and government institutions in urban areas. A key impetus in promoting the improvement of unemployment insurance came from the large numbers of laid-off or xiagang workers from state-owned and collective enterprises as part of the large-scale economic restructuring of the state sector undertaken by the government since the mid- 1990s. To ease the burden of labor adjustment, the government had set up a system of support for xiagang workers which entitled them to basic living subsidies for up to three years (based on 60% of their final wage) as well as exemptions for their pension and health insurance contributions. The xiagang benefits were funded from multiple sources including contributions from enterprises, the local and provincial government budget, as well as the unemployment insurance fund. Thus, when the UI system was standardized in 1999, laid-off workers whose three-year xiagang subsidies had expired immediately became eligible for UI. Xiagang subsidies have now been phased out and laid-off workers were gradually included into the UI program. The UI program is funded through employer and employee contributions, the former at 2% of the wage bill and the latter at 1% of the workers' individual wages. Farmer-turned contract workers are exempt from paying the UI premium. To qualify for benefits, an unemployed person should satisfy three conditions: not have terminated their employment voluntarily, should have paid the unemployment premium for at least one year, and be registered as unemployed and willing to be re-employed. The benefits include an unemployment allowance at a flat rate set somewhere between the minimum living allowance under the urban Di Bao program and the local minimum wage. In practice, 60- 70% of the minimum wage is commonly used. Benefits are available for a maximum of two years depending on how long the worker or their employers have paid their contributions to the UI program; the maximum benefit period of two years requires participation in UI for more than 10 years. Benefits also include medical subsidies if the unemployed person falls sick while receiving an unemployment allowance. The government is now in the process of amending the 1999 regulations on UI. The UI program currently covers about 40% of the urban workforce or 116.5 million workers (Table 6.71). The actual number of UI beneficiaries in December 2007 was 2.86 million, falling from 3.62 million in December 2005 while per capita benefits rose from 5711 to 7622 Yuan over the same period. E n h a n c i n g s e c u r i t y 1 3 7 (4) Work-injury insurance The work-injury insurance scheme is intended to cover employees of all enterprises and individual businesses against risks of work-related injuries or occupational disease. Benefits under the scheme include medical expenses for work-related injuries, injury and disability subsidy allowances and nursing fees according to the extent of work disability suffered, and in case of death due to a work-related accident, a funeral subsidy, a lump- sum death allowance and a pension for family members of the deceased. Compensation for work-related injury is subject to a labor ability assessment, applications for which can be submitted by the employing unit, the injured employee or his/her family. The work- injury insurance scheme is funded entirely from employer contributions of about 1% of the payroll, though the rates vary by different sectors and also within sectors according to the relative hazards of such work injuries. Recent regulations from MOLSS aim to extend work injury insurance to all rural migrant workers employed in high-risk industries, and require that construction companies make work injury contribution for rural migrant workers on time and in full. In 2007, about 122 million workers (about 42% of the urban work force) were covered by the scheme, about 40 million of whom were migrant workers (Table 6.71). About three-fourths of the increase in coverage since 2005 (from about 85 million) has been on account of higher coverage of migrant workers. The number of beneficiaries in 2007 was about 960 thousand. (5) Maternity insurance According to the government's Social Security White Paper (Government of the Peoples' Republic of China, 2004a), this program aims to cover women employees of urban enterprises, and in many places, those in government agencies, public institutions and mass organizations. The program provides for paid maternity leave of 90 days and additionally for medical expenses incurred during pregnancy to be covered by the employer. The program is funded entirely through employer contributions of up to 1% of the total wage bill. The program covered about 78 million workers or about 70% of the urban female work force in 2007 (Table 6.71). The actual number of beneficiaries under this program was about 1.1 million in 2007. Policy issues The development of the urban social insurance system has progressed rapidly in China over the last decade. The country has traveled a fair distance towards replacing the iron rice bowl, but there remain several areas of concern in relation to the ability of the core programs to serve as an effective safety net for the urban population. While each program has its own set of specific design and implementation issues, there are some cross-cutting concerns from a poverty and inequality reduction perspective. The most significant of these include: uneven and in many instances limited coverage of social insurance programs, in particular inadequate coverage of the relatively poor and marginal groups, the low level of pooling of social insurance funds and very limited portability of benefits, and the need to balance provision of greater protection with maintaining incentives and competitiveness. Limited coverage, especially of the poor and lower income groups and those in informal employment. Despite the expansion over the last decade, the coverage of social insurance programs still remains limited. Currently the key programs cover between 40-50% of urban workers (Table 6.71) which indicates that there is a lot of ground to be covered still. Information on the coverage of social insurance programs by level of income is generally limited. However, recent evidence from the 2005 China Urban Labor Survey in 5 large and 5 small cities in China shows that the coverage of major social insurance programs is significantly more limited amongst the poor and the lower income groups. For instance, pension participation rates amongst urban resident workers in poor households (those with per capita incomes less than twice the World Bank poverty line) are about 23 (33) percentage points lower than for workers from non-poor households in the 5 large (small) cities (Figure 6.74). Similarly, health insurance participation rates amongst the poor are 32 (26) percentage points lower than those amongst the non-poor in the 5 large (small) E n h a n c i n g s e c u r i t y 1 3 8 cities; unemployment insurance participation rates amongst the poor are only about half of those amongst the non-poor; and work injury insurance rates for the poor are a little more than a quarter of those for the non-poor. Similar differences in participation rates are also notable across urban workers from different quintiles of per capita income with rates at the bottom end often only a fraction of those for the top quintile (Figure 6.74). Overall therefore there is a clear positive association across all cities between social insurance coverage and workers' income level, and the relative exclusion of the poor is particularly striking. Figure 6.74 : Lower coverage of social insurance for the poor and the bottom income quintiles in Chinese cities, 2005 (% of workers participating, by poverty status and by quintiles of per capita income) Pension insurance 5 Large cities 5 Small cities 90 84 86 89 79 80 81 76 79 80 66 68 60 56 59 47 30 0 Poor Non-poor Bottom 2nd 3rd 4th Top quintile quintile Health insurance 5 Large cities 5 Small cities 8689 90 83 76 75 75 69 65 66 60 55 58 50 37 42 30 0 Poor Non-poor Bottom 2nd 3rd 4th Top quintile quintile Unemployment insurance 5 Large cities 5 Small cities 90 60 60 46 48 49 35 37 42 24 27 31 32 35 30 17 21 0 Poor Non-poor Bottom 2nd 3rd 4th Top quintile quintile E n h a n c i n g s e c u r i t y 1 3 9 Work-injury insurance 90 5 Large cities 60 33 42 51 23 28 30 9 13 0 Poor Non-poor Bottom 2nd 3rd 4th Top quintile quintile Sources and notes: China Urban Labor Survey, 2005, in Shanghai, Wuhan, Shenyang, Fuzhou, and Xian, and 5 smaller cities in surrounding areas. The poor are defined as those with per capita incomes less than twice the World Bank urban poverty line. Work injury insurance coverage is not available for the small city sample. The more limited social insurance coverage of relatively poorer groups is also mirrored in the low coverage rates for informal sector workers whose share in urban employment, as noted above, has been rising rapidly. According to the MOLSS 66-city survey for 2002, the share of informal workers with pension, unemployment, and health insurance were 37, 21, and 14 percent compared to 85, 73, and 62 percent for formal workers (Figure 6.75). Figure 6.75 : Urban social insurance coverage under informal and formal employment in 2002 (% of workers with coverage) 90.0 Informal workers Formal workers 80.0 70.0 60.0 50.0 40.0 84.5 72.5 30.0 61.5 20.0 37.0 10.0 20.7 14.1 0.0 Pensions Unemployment Health Sources and notes: Cai and Wu (2006) based on Ministry of Labor and Social Security 66-city survey in 2002. Limited coverage of migrant workers. Large-scale rural to urban migration continues to pose challenges for the urban social protection system. The total number of migrants from rural areas is estimated to be about 120 million (Sheng and Peng, 2005) or about one-fifth of the urban population in 2004 (and a larger fraction of the urban work force). Even though China's 1994 Labor Law allows for all city workers to join social insurance schemes, only a select few of the migrants succeed in accessing them. In the 12-city China Urban Labor Survey of 2004-5, the coverage rates for pension, health and unemployment insurance for migrants are estimated to be about one-seventh of those for local urban residents (see chapter 7 for evidence on this). Evidence from Jiangsu province for 2003 further suggests that amongst migrant workers, those with higher past incomes are more likely to receive pension, unemployment and medical insurance (Nielsen et al. 2005). In the last two years, there has been a significant momentum on the extension of work injury and health insurance which, according to administrative data, covered about a third and a quarter of migrant workers respectively in 2007 (Table 6.71). However, much still remains to be done in these two areas. The inclusion of migrants in pension and unemployment insurance is significantly more limited and poses a bigger challenge as the design of these programs is not well-suited to serving a mobile labor force that changes employers and jobs more frequently. Low level of pooling and very limited portability of benefits. While the social insurance schemes have all moved in the direction of funded systems, the contributions E n h a n c i n g s e c u r i t y 1 4 0 into social insurance funds are typically pooled only at the local government (municipality or city) level. The lower level of pooling has the advantage of creating greater incentives for local agencies to collect social insurance contributions. However, local governments may fear that high contribution rates and strict enforcement will drive away businesses to other jurisdictions. Low level pooling is inefficient from a risk management perspective and limits the insurance benefits of the programs. It also has the effect of creating inequities in the system by tying local social insurance expenditures to local revenues. Many local governments face potential deficits as they find it difficult to finance their social insurance obligations from contributions within their jurisdiction. While we do not have municipality or city-level information on this, some of the effects of the low level of pooling are notable also at the provincial level. Figure 6.76 plots per capita revenues and expenditures on pensions and BMI against the per capita GDP for each province for 2005. Two points are notable. First, provincial per capita expenditures on pensions and BMI are largely limited by per capita contributions within the province. As a consequence, large disparities in social insurance spending persist. Per capita spending at the top end (Shanghai province) is 6.4 times higher than at the bottom end (Jiangxi province). Second, these disparities are also regressive in nature. There is a clear positive association between per capita pension and medical insurance spending and provincial per capita income. Figure 6.76 : Higher income levels are associated with higher per-capita contributions and benefits of urban BMI and pension programs across provinces, 2005 ) r on 3500 fo rs eun 3000 Revenue: y = 0.0464x + 453.9 n/pe 2500 R-squared = 0.75 ever/.pxe uaY(I 2000 MB 1500 ati dna Expenditure: y = 0.0423x + 288.2 1000 R-squared = 0.71 capreP ions 500 ns 0 pe 5000 15000 25000 35000 45000 55000 Provincial GDP per capita (Yuan/person) Sources and notes: NBS (2006), China Statistical Yearbook 2006. The hollow squares denote per capita expenditures while diamonds indicate per capita revenues for pensions and BMI for each province. However, higher levels of pooling can be unattractive to richer local governments who may view it as an implicit tax. The reform of the system is thus likely to be difficult. Low level of pooling goes hand in hand with very limited portability of benefits. This is not only a significant barrier to improved coverage of the programs, but is also a major impediment to labor mobility and the development of a flexible labor market. Balancing protection with incentives and competitiveness. Even as the reach and breadth of social insurance programs remains limited in several respects, China's social insurance system is costly with its financing by employers and employees accounting for about 43% of the wage bill. The employers' contribution for the five key programs alone comes to about 32% of the payroll. Some of this may be passed on to the workers in lower wages, but private incentives for non-compliance remain high. Incentives for non- compliance also exist for the poorer workers. As discussed above, this is already contributing to the informaiization of the urban labor market. The existing system also has important implications for maintaining competitiveness of Chinese business and industry. Greater effort at enforcing compliance, though desirable, is unlikely to be enough to tackle the dual problems of informalization and ensuring competitiveness. A part of the solution E n h a n c i n g s e c u r i t y 1 4 1 will almost certainly involve greater public funding of the social insurance system. To protect the interests of the poorer workers, the government may also consider reducing or waiving the employee contribution and reducing even the employer contribution for workers below a certain income threshold. e. Replacing the iron rice bowl: the urban Di Bao program A major component of the urban safety net in China is the "Minimum Livelihood Guarantee Scheme", commonly known as the Di Bao program. Starting with Shanghai in 1993, the program was replicated extensively in the following years before being officially adopted as national policy in 1999. Described as the third line of defense (with pension and unemployment insurance as the first and basic livelihood support for laid-off workers as the second) for guaranteeing basic necessities of life to those living in urban areas (Tang, Sha and Ren, 2003), the program is part of the government's initiative to revamp the earlier enterprise-based urban social security system. The program is administered by the Ministry of Civil Affairs (MOCA). Table 6.72 summarizes key indicators of the growth of the program. Three points are notable. First, starting in 2001 the program was scaled-up substantially, but since 2003 the number of beneficiaries has stabilized at about 22-23 million. The total budget (excluding administrative costs) has continued to rise reaching about 27 billion Yuan (about 0.1% of GDP) in 2007. Second, the expansion was enabled by a large infusion of central government funding, the latter's share in total program spending rising from about 20% in 1999 to over 56% in 2004. Third, the transfer benefit per beneficiary fluctuated in the early phase of the program, but there has been a steady increase in recent years. For 2007, the average benefit level is 1210 Yuan which is just short of the World Bank urban poverty line of 1251 Yuan per person per year at 2007 all-China urban prices.68 Table 6.72 : Basic information on the urban Di Bao program in China, 1996-2007 1996 1998 2000 2001 2002 2003 2004 2005 2007 Urban di bao beneficiaries (millions) 0.85 1.84 4.03 11.71 20.65 22.47 22.05 22.34 22.71 % of urban residents receiving di bao benefits 0.28 0.56 1.19 3.42 5.91 6.33 6.12 6.11 6.04 Urban di bao expenditures (billion Yuan) 0.30 1.20 2.96 5.42 10.47 15.31 17.27 19.07 27.48 Expenditures per beneficiary (Yuan/year) 353 652 734 463 507 681 783 854 1210 Sub-national government share (%) 100.0 100.0 73.0 57.6 56.1 38.7 44.3 Central government share (%) 0.0 0.0 27.0 42.4 43.9 61.3 55.7 % of GDP 0.00 0.01 0.03 0.05 0.09 0.11 0.11 0.10 0.11 Sources and notes: MoCA (2007) Civil Affairs Statistical Yearbook 2007, and MoCA administrative data cited in O'Keefe (2004). Key features of the program The urban Di Bao (hereafter UDB) is a means-tested social assistance program limited to households with local urban (non-agricultural) hukou. Eligibility for Di Bao is determined by whether the per capita income of the household is below a locally determined minimum livelihood threshold, referred to as the Di Bao line. National and provincial regulations provide only limited guidance on the determination of these minimum thresholds. While the Di Bao line is not a poverty line in a strict sense, national regulations indicate that it should be based on local minimum livelihood costs for "the basic needs of food, clothing, and housing, and expenditures on children's compulsory education" (Government of PRC, 2004). The threshold should in principle be below the level of unemployment insurance benefits, the minimum pension benefit, and the minimum wage. However, the regulations also go on to note that "consideration must also be given to the level of local socio- _________________________________________________________ 68See chapter 1 for further discussion of the World Bank poverty line. The urban poverty line of 1124 Yuan per person per year is updated by the official urban consumer price index. Note that the benefit levels are obtained by dividing the total Di Bao budget by the number of beneficiaries, and is generally lower than the benefit level report by MOCA (182 Yuan per person per month for 2007) as many of the beneficiaries receive may receive benefits for less than the full year. E n h a n c i n g s e c u r i t y 1 4 2 economic development, the number of people eligible for receiving the minimum living allowance and the local government's fiscal capacity" (Government of PRC, 2004b). The provincial regulations also do not provide more specific guidance on how Di Bao lines should be determined (O'Keefe, 2004). In practice, the program is implemented in a highly decentralized fashion, and there is wide variation in Di Bao lines across cities and even districts within cities, and the variation is not attributable to differences in cost of living. Decentralized implementation is also manifest in the identification of eligible beneficiaries. While the program is means-tested, per capita income is not the only criterion used in practice, reflecting in part the difficulties of measuring household incomes accurately and also the consideration in many cases that current incomes are an imperfect indicator of household needs or welfare. A wide range of additional eligibility criteria and screening measures are applied to varying degrees in different cities or districts. These include: · asset criteria (e.g. per capita living space), used as part of the assessment of "actual living conditions" of households by neighborhood committees; cadres can also interview neighbors to collect information on the living standards and lifestyle of applicants, including such things as eating in restaurants, wearing jewelry, and even ownership of pets or mobile phones; · informal assessment of extraordinary expenditures incurred by some households reflecting circumstances beyond their control (e.g. to meet high medical costs due to serious illness) or otherwise deemed socially desirable (e.g. to pay education fees for further education of children); · informal screening by the community ­ in addition to the role of neighborhood committees in the initial determination of eligibility, the identified households are further subject to scrutiny or objections by the local community through the posting of their names in a public place; and · further exclusion conditions sometimes applied for initial or continued eligibility; for instance, households may be excluded if they receive unemployment benefits (Shanghai), do not participate in birth control programs where relevant (Guangdong), refuse more than a limited number of job or training offers, or if their economic circumstances are deemed to be the result of alcohol, drugs or gambling dependence. Thus, there is considerable local discretion in the determination of eligibility. However, the overall assessment process is still considered to be fairly stringent. Field interviews with neighborhood committee staff suggest that formal refusal rates are very low as applicants are warned of this rigorous screening process (O'Keefe, 2004). Di Bao benefits follow a "top-up" design that aims to fill the deficit of an eligible household's per capita income below the Di Bao line. 69 However, given the actual process of assessment of household eligibility and the difficulties of accurately measuring household incomes, it remains unclear how the benefit levels are determined in practice. Moreover, cash benefits are supplemented with in-kind benefits such as subsidies or exemptions for education, utilities, health care payments, or access to low-income housing. The valuation of these in-kind benefits is difficult, and it is unlikely that the overall package of benefits follows the top-up design envisioned for the program. Anecdotal information suggests that beneficiaries place considerable weight on the "golden card" value of Di Bao eligibility rather than on the cash benefits alone (O'Keefe, 2004). There is no limit in principle on how long a household may receive Di Bao benefits. In practice, only those who are unable to work are likely to be provided with long-term assistance. Others are provided short-term benefits for three months subject to regular review for continuation. Those who are able to work have to register at employment agencies for jobs and retraining. Refusal can lead to termination of benefits. For the _________________________________________________________ 69Spouses or dependants from rural areas or other provinces without local hokou are excluded from the calculation of benefits. E n h a n c i n g s e c u r i t y 1 4 3 unemployed, extension of assistance is conditional on satisfactory participation in community work ranging from 10 to 40 hours per month. Characteristics of urban Di Bao beneficiaries resemble those of the relatively poor in urban areas Administrative data from the Ministry of Civil Affairs sheds some light on the composition of Di Bao beneficiaries in recent years. In 2004, xiagang/laid-off workers accounted for just under two-fifths of total beneficiaries excluding "others", their share having declined since 2002. The share of on-job workers also declined. On the other hand, the share of unemployed workers increased from about one-fourth to over a third between 2002 and 2004. This reflects in part the policy change since 2001 to integrate the basic livelihood guarantee for laid-off workers into the unemployment insurance program. However, since the coverage of unemployment insurance still remains limited despite its recent growth, there seems to be a growing recourse to Di Bao in recent years. Table 6.73 : Composition of urban Di Bao beneficiaries, 2002-04 Share excluding "others" Categories 2002 2003 2004 2002 2003 2004 On-job workers 9.0 8.0 6.4 14.6 13.8 11.7 Xiagang/laid-off workers 26.9 23.1 21.3 43.2 40.0 39.0 Retired workers 4.4 4.0 3.3 7.1 7.0 6.1 Unemployed workers 17.4 18.2 19.2 27.9 31.5 35.2 Others 37.9 42.3 45.5 Three nos 4.5 4.4 4.3 7.2 7.7 7.9 Total 100.0 100.0 100.0 100.0 100.0 100.0 Sources and notes: Ministry of Civil Affairs. "Others" includes the dependents of all categories of workers/households except "three nos". The "three no" (san wu) households are those with no ability to work, no income source, and no family or social support. These households were outside the earlier work unit-based social security system. It is also possible to construct a profile of UDB beneficiaries using survey data collected by the National Bureau of Statistics in 2003/4 (see Box 6.18 for a description of these data). The survey data indicate somewhat lower levels of participation in the Di Bao program than the administrative data. This profile is summarized in Table 6.74 which compares the average characteristics of Di Bao beneficiaries with non-beneficiaries as well as the urban population as a whole, using a 35-city sub-sample of the survey data. Box 6.18 : Data on urban Di Bao Apart from administrative data available from the Ministry of Civil Affairs, an important recent source of information on the Di Bao program comes from the Urban (Short-Form) Household Survey (USFHS) for 2003-4 conducted by the National Bureau of Statistics. Using a relatively brief questionnaire, this survey covered about 260,000 urban households nationally. In particular, the survey had questions on participation in UDB and Di Bao benefits received during 2003. While being a highly valuable source of information on the Di Bao program, these data are also restrictive in some respects. First, the questions on Di Bao benefits include cash benefits only; thus various in-kind concessions and entitlements also offered by the program are not covered. Second, the questionnaire had a single question on the total income of the household. A single question on income is likely to be subject to greater measurement error relative to the detailed set of income questions that are typical of the regular Urban Household Survey (UHS) conducted by the NBS every year. This renders the income measure from the USFHS non-comparable with that from the UHS. In general, income measures derived from abridged questionnaires tend to be lower. A comparison of (nominal) mean per capita income for 2003 from the two sources indicates that the USFHS measure is 6.4% lower than the UHS measure. There are also likely to be differences in the measures of dispersion of income from the two sources. In view of this problem, most of the analysis presented in this chapter does not use information on the absolute levels of income from the USFHS; instead, the income data are used only to rank households (from the poorest to the richest). E n h a n c i n g s e c u r i t y 1 4 4 Table 6.74 : Di Bao beneficiaries have different characteristics from non-beneficiaries Household characteristics Receiving di bao? No Yes Population Household size (persons) 2.95 3.08 2.95 Area of dwelling (sq. m.) 68.46 49.84 67.76 Year of dwelling 1988 1984 1988 Health shock [a] 0.12 0.2 0.12 Number of household members who are: Children 0.45 0.57 0.45 Disabled 0.03 0.27 0.04 Have a long-term illness 0.17 0.39 0.18 Retired 0.66 0.28 0.64 Unemployed 0.28 0.89 0.3 Self-employed 0.09 0.19 0.09 Students 0.44 0.65 0.45 Household head's years of schooling 11.17 8.42 11.07 Male household head 0.71 0.63 0.71 Age of household head 50.65 51.1 50.67 Computer (number) 0.43 0.08 0.42 Wage ratio (% of income) 68 46 67 Di Bao receipts per person (yuan per year) 0 270 11 Net income per person (yuan per year) 10237 2928 9951 Sample size 73920 2888 76808 Source and notes: Chen, Ravallion and Wang (2006). Except for the age of household head, all differences between the beneficiaries and non-beneficiaries are statistically significant. This profile is based on the 35-city sub-sample of USFHS 2003-2004. [a] A household is considered to have experienced a health shock if medical expenditures are more than one-third of income. The Table indicates that relative to non-beneficiaries, households receiving Di Bao payments, on average, have a slightly larger family size due to the comparably larger number of children in the household. They have smaller dwellings, are more likely to have a disabled person or someone with a long-term illness in the household. They are also more likely to be supporting someone who is unemployed or a student, and are less likely to have a retiree. They are more likely to be headed by females, and their household heads (whether male or female) have significantly fewer years of schooling. Further details on the profile of beneficiaries are shown in Table 6.75 in terms of their education, health, labor force and employment status based on the national sample of the Urban (Short-Form) Household Survey. The findings are complementary to those in Table 6.74. · In terms of education status, those below (above) junior high school have a higher (lower) share amongst Di Bao beneficiaries than amongst non-beneficiaries, and the proportion of those receiving Di Bao benefits declines with the level of education. · In terms of self-reported health status, those without a healthy status are over- represented amongst the beneficiaries relative to the non-beneficiaries. · For the urban population aged 16 and above, the employed are under-represented amongst the beneficiaries as also are the retired, while the laid-off, the unemployed as well as those outside the labor force are over-represented. · Amongst those who are employed, the self-employed and those in the "other" category ­ which is likely to include various forms of informal employment ­ are over- represented amongst the beneficiaries, while those in more formal employment in the government, public sector undertakings, state-owned enterprises, the corporate (shareholding) or the joint venture sector are under-represented. Employment in the private (non-corporate) sector is also more common for the beneficiaries than non- beneficiaries. E n h a n c i n g s e c u r i t y 1 4 5 Table 6.75 : Profile of urban Di Bao beneficiaries, 2003 Share in total Share in total Proportion UDB non- Share in total receiving UDB beneficiaries beneficiaries population (%) benefits (%) (%) (%) All urban population Education status Illiterate 10.0 4.3 4.6 11.4 Elementary 15.9 9.8 10.1 8.2 Junior-high 40.0 27.4 28.0 7.4 Senior-high 22.5 23.3 23.2 5.0 Vocational 5.8 10.7 10.5 2.9 Junior college 3.7 14.5 14.0 1.4 College 2.0 9.4 9.1 1.2 Masters & above 0.1 0.6 0.6 0.5 Health status (self-reported) Healthy 81.6 94.0 93.3 4.6 Disabled 7.0 1.0 1.3 27.2 Other 11.4 5.0 5.3 11.2 Total 100.0 100.0 100.0 5.2 All urban individuals age 16 and above Labor force status Employed 30.7 51.5 50.5 3.1 Retired 10.3 23.2 22.5 2.3 Household work 9.3 5.0 5.2 9.1 Laid-off 11.4 3.5 3.9 15.1 Unemployed 13.9 4.5 5.0 14.4 In school 11.1 7.7 7.9 7.2 Other 13.3 4.5 5.0 13.8 Total 100.0 100.0 100.0 5.1 Individuals who are employed Employer Government 1.7 9.4 9.2 0.6 PSU 4.3 18.7 18.3 0.7 SOE 11.1 25.9 25.5 1.4 Collective 6.0 3.9 3.9 4.8 Shareholding 4.7 8.6 8.5 1.7 Private 17.3 11.0 11.2 4.9 Joint venture 0.9 1.9 1.8 1.6 Self-employed 27.9 14.9 15.3 5.7 Other 26.0 5.7 6.4 12.7 Total 100.0 100.0 100.0 3.1 Source and notes: World Bank estimates from national sample of NBS Urban (Short-Form) Household Survey 2003-2004. Overall, therefore, the profile of Di Bao households seems to resemble in many ways that of the relatively poor in urban areas. Many of the attributes that are positively (negatively) associated with Di Bao participation are also those that are typically negatively (positively) associated with income. It is thus not surprising to find that the average net per capita income of Di Bao participants is much lower than (only about 30% of) that of the non- participants (Table 6.74). All this is thus suggestive of the program being reasonably well- targeted. The extent to which this is indeed the case is examined below where evidence on targeting as well as other aspects of the program's performance as an effective safety net for the urban poor is further discussed. Assessing Di Bao's performance Several key questions can be posed in assessing Di Bao's performance as a social assistance program for the relatively poor in urban areas. · Is the program well-targeted? · Is the program's coverage of the poor or the target population adequate? · Is the variation in Di Bao lines and horizontal inequity a problem? E n h a n c i n g s e c u r i t y 1 4 6 · Does the program have adverse incentive effects? These questions and the evidence in relation to them are discussed below. The assessment relies primarily on the information available from the Urban Short-Form Household Survey (USFHS) for 2003. One caveat should be noted at the outset. The assessment is limited to Di Bao's cash benefits only as data on in-kind benefits are not available. While this obviously leads to an underestimation of the overall benefits from the program, it remains unclear how precisely it may affect the assessment of the program's targeting, coverage and incentive effects. However, insofar as non-cash benefits are bundled together with the cash benefits, an analysis of the latter is likely to be informative of the overall program's performance. By international standards, urban Di Bao has excellent targeting performance The targeting performance of a program can be measured in various ways. Commonly- used indicators are measures of the extent to which participation or total benefits from the program are concentrated amongst the target group. Thus, a program can be said to be well-targeted if a large fraction of program beneficiaries belong to the target group or if a large fraction of benefits accrue to the target group. The target group itself may be defined in terms of the eligibility criteria intended to be used by the program or more generally in terms of the relatively poor population. Based on survey data from the USFHS, Di Bao program's targeting performance in terms of the relative poverty criterion is illustrated in Figure 6.77. As is evident from the Figure, both the beneficiaries and the total benefits from Di Bao are highly concentrated amongst the poorest urban groups. For instance, 64% of the beneficiaries are in the poorest decile. Similarly, about half of all Di Bao benefits accrue to the poorest 5% of the population, and about 70% accrue to the poorest decile. By international standards, this is excellent targeting performance. For instance, it outperforms Argentina's Trabajar workfare program for the urban poor, for which it is estimated that the poorest decile accounted for about 60% of total program benefits (Jalan and Ravallion, 2003). The Trabajar program itself was judged to be the best-targeted amongst 122 programs across 48 countries by Coady, Grosh and Hoddinott (2004). Thus, in terms of its success in being able to direct program benefits to the bottom end of the income distribution, the Di Bao program is amongst the world's best. The program's targeting performance in terms of its own eligibility criteria seems to be less spectacular. Before considering the survey evidence on eligibility, however, some caveats on assessing program eligibility from the USFHS survey data should be noted. First, as mentioned above there are concerns with how incomes are measured in the USFHS, being based on a single income question (see Box 6.18). Second, the income measures used by program administrators in the field to identify potential beneficiaries are not based on a household survey or census, and are thus likely to differ from survey-based income measures. Third, we know that program eligibility is not based on the income test alone, and various other non-income criteria are also deployed in practice to determine who qualifies for the program. E n h a n c i n g s e c u r i t y 1 4 7 Figure 6.77 : Targeting performance of Urban Di Bao, 2003 Share (%) of all di bao beneficiaries in each income decile 70% 63.8% 60% 50% 40% 30% 19.5% 20% 7.5% 10% 4.5% 1.9% 1.0% 0.8% 0.4% 0.3% 0.3% 0% Population ranked by income before di bao: 10% 10% Poorest from the poorest 10% to the richest 10% ichestR Share (%) of all di bao payments going to each income ventile 100% 80% 60% 40% 20% 0% Poorest 5% Richest 5% Population ranked by per-capita income before di bao: from poorest 5% to richest 5% Sources and notes: World Bank estimates from national sample of NBS Urban (Short-Form) Household Survey 2003-2004. Subject to these caveats, what do the survey data tell us about Di Bao's ability to target those eligible for the program? The data indicate that 5.22% of the urban population received Di Bao benefits in 2003. Of this 5.22%, 2.98% met the eligibility criterion of their pre-transfer per capita incomes being below the local Di Bao line (Table 6.76). Thus, about 57% of the program beneficiaries were in the target group. As for the share of total benefits accruing to the target group, this is estimated to be about 58%. As noted in Box 6.18, the single income question in the USFHS may lead to some underestimation of household incomes. To assess how much of a difference that could make, alternative estimates were also made assuming actual household incomes to be 10% higher than those reported in the survey. These estimates indicate that 52% of beneficiaries belong to the target group. Thus, the program's targeting performance in terms of reaching those eligible, while good, is not as impressive in terms of reaching the bottom income decile of the urban population. This is partly because all those who are eligible are not necessarily in the bottom decile, in turn reflecting the large underlying variation in Di Bao lines (see below). E n h a n c i n g s e c u r i t y 1 4 8 Table 6.76 : Targeting and coverage under Di Bao (% of urban population, 2003) Using pre-transfer per capita Using 10% higher pre-transfer per incomes as reported in USFHS capita incomes Eligible for di bao? Eligible for di bao? Yes No Total Yes No Total Receiving Yes 2.98 2.24 5.22 42.9% 2.71 2.51 5.22 48.1% di bao? No 4.85 89.93 94.78 leakage 4.06 90.72 94.78 leakage Total 7.83 92.17 100 6.77 93.80 100 38.1% coverage 40.0% coverage Sources and notes: World Bank estimates from the national sample of the NBS 2003-2004 Urban (Short Form) Household Survey. The leakage rate is defined as the percentage of beneficiaries who are not eligible, and the coverage rate is defined as the percentage of the eligible who are beneficiaries. But the coverage of the program could be increased A well-targeted program may nonetheless still fail to cover all those who need or are eligible for the program. For instance, even if a 100% of program benefits accrue to the target population, the overall size of the program could be too small to cover all or most of the target population. Thus, low coverage can still be an issue even for well-targeted programs.70 Any leakage to non-target population further lowers the coverage rate. As already noted, based on USFHS data, it is estimated that 5.22% of the urban population (or about 18.5 million persons) benefited from the Di Bao program in 2003. This is somewhat lower than the coverage rate of 6.3% (or about 22.5 million persons) according to administrative data from MoCA. However, the difference may well be accounted for by sampling error and differences in methodology between how survey and administrative data define and aggregate the number of beneficiaries over the year. There are also indications (based on MoCA administrative data) that the number of beneficiaries is stabilizing in recent years; the estimates for 2004 and 2005 are also around the 22 million mark. Can this level of program coverage be considered adequate? The evidence on coverage from the perspective of those eligible and those who are relatively poor or disadvantaged is reviewed below. It is estimated that in 2003 about 7.83% of the urban population were eligible for the Di Bao program insofar as their pre-transfer per capita incomes were below the local Di Bao lines. Compared with the overall size of the program covering 5.22% of the urban population, it is evident that even if all beneficiaries met the income eligibility condition, only about two-thirds of those eligible would have been covered by the program. However, on account of the targeting errors noted above, only about 58% of the beneficiaries were deemed eligible, thus implying an actual coverage rate of about 38% of all those eligible. Even if it were assumed that per capita incomes were 10% higher than those reported in the survey (to account for possible underestimation of incomes in USFHS), the coverage rate improves only marginally to about 40% of those eligible. Thus, under-coverage seems to be a significant issue with the program. Coverage is also limited in terms of how much of the aggregate Di Bao gap (i.e. the aggregate income shortfall of all those households whose incomes are below the Di Bao lines) is filled by the program. Based on the USFHS data, it is estimated that for 2003 total Di Bao benefits were about 46.1% of the aggregate Di Bao gap. However taking into account leakages to those with incomes higher than Di Bao thresholds (and excess of Di Bao payments over Di Bao gaps in some cases), the total benefits only filled about 25.6% of the total Di Bao gap. About 72% of unfilled Di Bao gap is on account of non-coverage of those with pre-transfer incomes below Di Bao lines; the rest is on account of the _________________________________________________________ 70Note that the term targeting here is being used in the narrower sense of concentration of program participation or benefits amongst the target population. In a wider definition of targeting, sometimes the extent of coverage is also seen to be an element of targeting. E n h a n c i n g s e c u r i t y 1 4 9 average Di Bao benefits being less than the Di Bao gaps for those who are eligible and covered. Relatively low coverage is also apparent when considered from the perspective of the relatively poor in urban areas. As seen in Figure 6.78, less than half (about 45%) of the poorest 5% of the urban population receives Di Bao, and similarly, only about a third of the bottom decile are Di Bao recipients. For an urban poverty rate of about 6% (similar to the consumption poverty rate noted earlier for urban areas), only about 40% of them are covered by the program. Figure 6.78 : Does Di Bao reach the poorest population in urban areas? Share (%) of each income ventile receiving di bao benefits 50 45.5 40 30 21.7 20 12.6 7.9 10 4.9 3.0 2.9 1.8 1.1 0.9 0.6 0.5 0.5 0.3 0.3 0.1 0.1 0.2 0.1 0.2 0 estr 5% Poo Population ranked by income before di bao: from the poorest 5% to the richest 5% Sources and notes: World Bank estimates from the national sample of the NBS 2003-2004 Urban (Short Form) Household Survey. Another important issue in relation to coverage is that the program is limited to those with local non-agricultural hukou, thereby excluding the growing migrant population in urban areas from the potential benefits of social protection offered by the program. Since migrants retain their rural household registration status under the current system, in the event their employment in urban areas is disrupted for any reason, they are likely to return to rural areas. In rural areas, they could be covered by the rural Di Bao program, but the current coverage and benefit levels of the rural program remain limited. Thus, while a case can be made for extending the coverage of urban Di Bao to migrant workers, any such extension will also need to deal with the issue of coordination of the urban and rural Di Bao programs in particular to resolve issues of parity and portability. Eligibility thresholds vary with local fiscal resources rather than with need Reflecting its decentralized structure, the Di Bao lines deployed by the program to determine eligibility vary significantly across cities and towns, and sometimes even across districts within cities. For instance, for 2003, the lines ranged from 936 Yuan per person per year for Rizhao city in Shandong province to the 4.4 times higher line of 4128 Yuan for Shenzhen in Guandong province. Only a relatively small part of the variation can be attributed to differences in cost of living.71 When normalized by spatial cost of living indices for urban areas, the maximum Di Bao line is 3.4 times higher than the minimum.72 This raises concerns about horizontal inequity under the program, i.e. the extent to which people with similar standards of living are treated differently across locations. The large variation in Di Bao lines suggests that such horizontally dissimilar treatment of similar households is indeed a point of concern for the program. _________________________________________________________ 71Based on the list of Di Bao lines published by MoCA (2003) for 3153 cities, towns and districts. 72In the absence of city-specific spatial price indices, urban inter-provincial cost of living indices from Brandt and Holz (2006) are used to deflate the Di Bao lines to 2003 urban all-China prices. E n h a n c i n g s e c u r i t y 1 5 0 This is especially so because the variation in Di Bao lines also has some systematic elements. The evidence indicates that Di Bao lines tend to decline with the incidence of relative poverty; thus, the thresholds tend to be lower where needs are higher (Figure 6.79). Similarly, the thresholds also tend to increase with the level of per capita income of the municipal area, indicating that richer (poorer) cities with greater fiscal capacity tend to set more (less) generous thresholds. Figure 6.79 : Horizontal equity of Di Bao program Local Di Bao lines in relation to per capita incomes, 2003 )raeyr 00 40 pe ati capr- 00 pe 30 n (yua enilo baid 00 aeralapciin 20 Mu 00 10 5 0 0 0 1 0 0 0 0 1 5 0 0 0 2 0 0 0 0 2 5 0 0 0 M u n ic ip a l a re a m e a n a n n u a l p e r-c a p ita in c o m e Local Di Bao lines and % of vulnerable population across municipal areas, 2003 )raeyrep 00 40 atip ca-r 00 pe 30 n uay( enilo baid 00 aeral 20 paicinu M 00 10 0 .1 .2 .3 .4 % o f m u n ic ip a l a re a p o p u la tio n v u ln e ra b le Sources and notes: World Bank estimates from the national sample of the NBS 2003-2004 Urban (Short Form) Household Survey. The % of vulnerable population in the area is defined as the percentage of population with per capita income below the national average Di Bao line of 1928 Yuan per person per year at 2003 all-China urban prices. The national Di Bao line itself is the urban population-weighted average of local Di Bao lines deflated by the urban spatial price index for the province to which the local municipal area belongs. The inter-provincial urban prices indices are based on estimates by Brandt and Holz (2006). Another issue with the setting of Di Bao lines is that there seems to be no clear procedure for the indexation of these lines to take account of price changes over time. The annual negotiations between local civil affairs and finance authorities have been described as a process of "guided ad hocism" rather than the application of a clear indexation formula (O'Keefe, 2004). Thus, there is scope for rationalization of Di Bao lines such that they are less susceptible to local fiscal considerations and more reflective of the needs for social assistance in different areas. E n h a n c i n g s e c u r i t y 1 5 1 At least in terms of its design, if not in practice, work disincentive effects of the program are a concern Benefits under the Di Bao program are designed to follow a top-up scheme whereby the per capita benefit for an eligible household is given by the amount by which the household's per capita income falls short of the Di Bao line. In principle, the idea is to bring everyone's income up to the Di Bao line. Such a design however implies a 100% marginal tax rate on incomes below the Di Bao line since any increase in pre-transfer incomes (below the Di Bao threshold) is penalized by a reduction in the benefits by the same amount. If literally followed, such a scheme would have highly deleterious work incentive effects; those whose pre-transfer incomes are below the Di Bao line would have no incentive to because all of the income gains from this work would be taxed away in the form of lower benefits. There is also the related issue of whether Di Bao benefits are engendering a longer-term dependency on the program. Based on data from the second China Urban Labor Survey (CULS2) for five cities for 2005, it is estimated that on average a Di Bao beneficiary received benefits for 33 months; 60% of the beneficiaries received benefits for more than 2 years, 25% for one to two years, and about 15% for less than a year. Thus, there is a concern about whether the top-up design for program benefits is creating negative work incentives amongst those the program intends to benefit. However, in practice, there seem to be significant departures from the top-up benefit design. For instance, it has been reported that a participating household may continue to receive benefits for some time after a working-age member finds employment even if the resulting household income exceeds the Di Bao threshold. Negative incentive effects are also reduced by by targeting those with disabilities, and by the aforementioned restrictions on refusal of job or training offers by participants (though it is unclear how widespread such restrictions are). Similarly, it has also been reported that an imputed household income based on the minimum wage and the number of members with work capacity is sometimes used to determine eligibility (Chen, Ravallion and Wang, 2006). All these practices seem geared to limiting the potentially negative work incentive effects of the program. What can we learn about these incentive effects from the available survey data? An estimate of the actual marginal tax rate implied by the program may be obtained by looking at how actual Di Bao benefits increase with the income gaps (the difference between Di Bao lines and pre-transfer incomes) for those who are eligible for the program. The estimated relationship between per capita benefits and income gaps for 2003 is shown in Figure 6.80. As the Figure shows, the estimated marginal tax rate is not 100%, but about 28%. However, this positive marginal tax rate seems to be entirely on account of variation across cities, i.e. cities with larger Di Bao income gaps tend to offer higher benefits on average though the increment in benefits is only a little over a quarter of the increment in income gaps. Once cross-city variation is controlled for, the evidence indicates that within a city actual benefits are largely invariant to household income gaps; the implied within-city marginal tax rate is only about 1%. E n h a n c i n g s e c u r i t y 1 5 2 Figure 6.80 : Incentive effects of the urban Di Bao program t 4000 benefi 3000 Projected per bao capita benefit di Estimated actual at 2000 under top-up Without city- design per capita di bao effects capi 1000 benefit Per With city-effects 0 0 500 1000 1500 2000 2500 3000 3500 4000 Per-capita income gap (di bao line minus pre-transfer per capita income) Sources and notes: World Bank estimates from the national sample of the NBS 2003-2004 Urban (Short Form) Household Survey. The estimated benefit schedule without city-effects is: Benefits = 15.8 + 0.27*Income-gap. When city-effects are included, the estimated benefit schedule is: Benefits = 213.5 + 0.01*Income-gap However, this estimate is likely to understate of the true marginal tax rate since the participants' pre-transfer incomes are likely to underestimate their counterfactual incomes in the absence of Di Bao program to the extent that without the program (and its implied tax on incomes) households would have exerted greater work effort. However, attempts to control for this effect indicate that the extent of such underestimation is not substantial.73 Thus, overall, the evidence suggests that the top-up benefit design for Di Bao is generally not followed in practice, so that per capita benefits within cities are largely unrelated to household income gaps. Thus, negative work incentive effects of the program as currently implemented are unlikely to be a significant concern. However, these findings do raise the issue of whether the top-up design should continue to be a guiding principle for the program. While various actual practices (mentioned above) do appear to succeed in containing the disincentive effects, it may be useful to consider an alternative benefit design framework which more explicitly builds in a lower marginal tax rate. Policy issues Di Bao is the main social assistance program for the urban disadvantaged in China. As noted, by international standards, the targeting performance of the program is excellent in terms of directing benefits to those at the bottom end of the urban income distribution. With 64% of the program beneficiaries in the bottom decile of the urban population who account for about 70% of program benefits, the urban Di Bao is certainly amongst the world's best-targeted social transfer programs. While such performance is surely commendable and holds lessons for other countries, there are still several policy challenges facing the program. One significant challenge is to improve program coverage which remains limited in at least three respects: (i) limited (about 40%) coverage of those who may be considered eligible by the program's income criterion, (ii) limited (also about 40%) coverage of those who may be considered relatively poor in urban areas (with per capita incomes below the World Bank's relative urban poverty line), and (iii) non-coverage of the large and growing mass of migrant population in urban areas. Some of the limited coverage reflects the program's reliance on other non-income criteria to select beneficiaries. But, perhaps more _________________________________________________________ 73For the 35-city sub-sample of the 2003-2004 Urban (Short-Form) Household Survey, Chen, Ravallion and Wang (2006) report only a slightly higher estimate of marginal tax rate when they instrument pre-transfer incomes with a number of household characteristics. E n h a n c i n g s e c u r i t y 1 5 3 significantly it also reflects the overall size of the program which ­ despite its recent growth ­ is still small relative to needs. With the current targeting performance, the size of the program may need to grow to as much as 2.5 times its present size in order to achieve complete coverage of the urban disadvantaged. Here, a caveat on the otherwise excellent targeting performance of the program is in order. While targeting is outstanding in terms of concentration of benefits amongst the bottom decile (ranked by per capita income), it is less spectacular if one were to focus on the narrower target group of the urban disadvantaged who comprise bottom 5-6% of the urban population. Only a little over half the benefits accrue to this target group. If targeting to this narrower group could be improved, their universal coverage (of the urban disadvantaged) need not require a two-and-a-half fold expansion of the program. Coverage of the migrant population needs special attention. Migrants include both individual migrant workers and migrant families. As a first step, extending Di Bao coverage to migrant families who have been resident in an urban area for some minimum period of time and meet the program's means test ought to be considered. Until such time that the urban and rural Di Bao programs can be coordinated, individual migrant workers may have to be covered under rural Di Bao through their families in rural areas. A second area of potential reform is in relation to the rationalization of Di Bao lines whose variation across cities implies horizontal inequities as well as differential program coverage across different urban areas. There is a case for eliminating horizontal inequities by establishing a set of eligibility lines that refer to comparable standards of living across locations. If it is not possible to develop city-specific lines with the available data, consistently-estimated provincial eligibility lines could be a useful starting point. The de-linking of Di Bao eligibility lines from considerations of local fiscal capacity will require transfers from higher levels of government to poorer municipalities. The central government's share in overall Di Bao funding is already substantial, around 60% in recent years. Even if a similar share is maintained in the future, there will be a need to direct central funding to provincial and local governments based on the estimated number of beneficiaries using consistent eligibility lines. Third, there is a case for the considering a proxy-means targeting mechanism for the selection of beneficiaries. Household incomes are difficult to observe and measure accurately, and some form of implicit proxy-means targeting is already being practiced as is evident in the use of various non-income criteria by the neighborhood committees. While the implicit proxy-means mechanism appears to have worked reasonably well thus far, many of the urban poor remain excluded, and this is not mainly due to the overall small size of the program. There is thus a case for developing a more explicit proxy- means targeting system based on relatively easily-observed household characteristics that are correlated with income. The list of household characteristics could include indicators related to household demographics, labor force participation, any work disability, educational attainment of adult members, ownership of specific assets, and characteristics of dwellings, amongst others. There is now a good deal of international experience with proxy-means targeting that could be drawn upon in developing a system tailored to China (Box 6.19). Box 6.19 : Proxy means targeting systems Countries have used a variety of targeting mechanisms for social programs, including (a) household (or individual) assessments, (b) categorical targeting, which refers to selecting broad groups of households based on a common characteristic such as geographic location), or (c) self-targeting, which makes benefit available to all but the design discourages participation by the poor while discouraging participation by the non-poor (for instance, public works employment offered to all who are willing to work at a relatively low wage). The choice of a particular targeting mechanism depends on the objectives of the program, the information available, the administrative capacity of implementing agencies, the overall budget for the program and the distribution and characteristics of the poor who are intended to be the program beneficiaries. And there is often scope for combining different methods. Most cash transfer programs require household assessments to determine eligibility (and sometimes also the level of benefits). Three main types of household assessments include: (i) Verified Means Testing (VMT) which relies on documented incomes, assets or expenditures of applicants to determine eligibility; (ii) Unverified Means Testing (UMT) which relies on self-reported incomes or assets with little or no verification; (iii) Proxy Means Testing (PMT) which is based on predictions of household welfare E n h a n c i n g s e c u r i t y 1 5 4 from a range of readily observed household characteristics. Given the difficulties developing countries face in meeting the prerequisites for operating effective VMT systems and measurement problems inherent in UMT, many countries have opted for the feasible and practical alternative of proxy means targeting. Proxy means targeting generates a composite score for applicant households based on fairly easy to observe household characteristics ("proxies") that are associated with poverty, such as the location and quality of the household's dwelling, its ownership of durable goods, demographic structure of the household, and the education and employment of adult members. Table shows range of "proxy" variables used for PMT systems in several countries. The indicators used in calculating this score and their weights are derived from statistical analysis (usually regression analysis or principal components) of data from detailed household sample surveys. Once the indicators are established, information on those indicators is collected from all applicants or potential program beneficiaries. Using the established weights, a household score is calculated, and eligibility is determined by comparing the household's score against a predetermined cutoff. Table 6.77 : Variables commonly used for proxy means targeting Chile Colombia Costa Rica Mexico Variables FICHA CAS SISBEN SIPO OPORTUNIDADES Housing quality In-house water x x x x Electricity x Waste disposal x Sewage disposal x x x WC with bath x x x Materials of floor x x x x Materials of walls x x x Materials of roof x x x Ownership of durables Color TV x x Refrigerator x x x x Washing machine x x x Water heater x Motor vehicle x Gas stove x Education Yeas of schooling x (head) x (oldest wage earner) x (head) x (head) Years of schooling (all > 12 years) x School attendance x Occupation and income x (share of members Occupation x (head or spouse) employed) x Income variables x (2 variables) x (1 variable) x (2 variables) Household size and demographics Number of persons in household x Age and gender of household head x Number of children x x Others Health variables x x Over-crowding x x x Ownership of site x x Source and notes: Castaneda and Lindert (2005). x denotes the variables used for proxy means scores. Proxy means scores can be especially useful in many developing countries where a high degree of informality, seasonality, in- kind or multiple-source earnings means that incomes are not easily quantifiable or verifiable. The experience of PMT in many countries suggests several advantages that make them potentially attractive household targeting instruments: · Targeting outcomes of PMT are often as accurate as VMT and in some cases more accurate than UMT. · PMT systems are significantly cheaper than VMT systems and their costs are in line with UMT systems. · The use of multi-dimensional indices is more politically appealing in many countries. Moreover, because it does not measure income itself, PMT may discourage work effort less than a means test would. · PMT mechanisms are more transparent and the use of a well-established proxy means index limits the scope for discretionary error and corruption. · The administrative burden of PMT systems is reasonable for many developing countries, and less than that for VMT systems. PMT mechanisms also have some drawbacks. The administrative burden though often manageable still requires a large number of literate and probably computer-trained staff, and moderate-to-high levels of information and technology. There is also an inherent inaccuracy at household level, since the formula is only a prediction, though good results on average have been observed. Moreover, the formula and results may seem mysterious or arbitrary to some households and communities. PMT methods are most appropriately used in contexts where there is a reasonable level of administrative capacity, but where E n h a n c i n g s e c u r i t y 1 5 5 there are significant concerns with the feasibility, cost or targeting performance of other targeting methods. PMT systems can also be particularly useful when there is potential for using the same targeting mechanism for several interventions. The household targeting systems used in some of the best-known conditional cash transfer (CCT) programs constitute major `institutional capital' for the country, and the same system is often used to target many programs, sometimes with different thresholds or ancillary criteria. For example, in Chile, the first country to use proxy means tests extensively, it is used not just for the recent Chile Solidario program, but for much older child allowance and social pensions, for water price subsidies, for housing subsidies. In Colombia, the same proxy means test (the SISBEN) is used to determine eligibility for the CCT program, the public workfare program, a youth training program, subsidized health insurance, hospital fees waivers and a social pension. Even in countries with more recently established proxy means test, they can be used in multiple programs. Jamaica established its proxy means test expressly for the CCT program, but now uses it to grant fee waivers in the health system, for secondary education, textbook rentals and school lunches as well. Such an investment will pay off sooner for programs that are generous in coverage or benefit levels and for countries that will, at least eventually, use them for multiple programs. Source: Castaneda and Lindert (2005) Fourth, there is also a case for reconsidering the top-up design formula that has been used as a guiding principle for the determination of Di Bao benefits. As noted above, such a design implies a 100% marginal tax rate on incomes of those eligible for the program. The reason why the program has managed to avoid serious adverse effects on work incentives thus far is that the top-up design appears not to have been followed in practice. Thus, the actual implementation practices are already pointing to the need for an alternative benefit structure. The main consideration here is to avoid negative work incentive effects. One alternative is uniform per capita benefits. While this is attractive for its simplicity and transparency, uniform benefits may not be considered sufficiently progressive. Another alternative could be to base benefits on the proxy-means scores of households, with lower benefits for those with higher scores but calibrated so as to maintain relatively low marginal tax rates. Fifth, the rationale for non-cash benefits provided by the Di Bao program should also be clarified. As noted, non-cash entitlements currently take the form of exemptions or concessions for education, health services or utilities. This partly reflects the intention of other line agencies to ensure access to these services by relatively poor households. But there is a lack of clarity as well as a lot of variation in the precise nature of these non-cash entitlements. The main issue is whether Di Bao eligibility can serve as a "backbone" targeting mechanism for the provision of other services to the poor. For Di Bao to serve such a core targeting function, it is imperative that the program maintains its good targeting performance, and in that regard, the policy issues discussed above in relation to targeting are important. To limit effects on work incentives, non-cash entitlements, if any, should kept minimal, and should be factored in while determining the overall benefit design of the program. Successful implementation will also require greater coordination and administrative linkages between the Ministry of Civil Affairs and the line agencies responsible for the delivery of other services included in the non-cash entitlements. E n h a n c i n g s e c u r i t y 1 5 6 7.Fostering equity: reducing disparities and building human capacities a. The rise in income inequality is a concern but it is important to be clear on the reasons why The rise in inequality is The rise in income inequality is understandably a concern. But it is important to be clear understandably a on the economic reasons why it should be a concern. concern but it is important to be clear on There are both "good" and "bad" inequalities. Good inequalities are those that reflect and the reasons why it should be reinforce market-based incentives that are needed to foster innovation, entrepreneurship and growth. The evidence presented later suggests that the rise in inequality with the introduction of market reforms in China is at least in part a reflection of newly-unleashed "good" market-based incentives at work, in contrast with the earlier period of artificially low levels of inequality brought about by regulatory distortions and interventions that suppressed incentives for individual effort and innovation. China has also been simultaneously undergoing a process of development and structural transformation. It needs to be recognized that some rise in inequality is perhaps inevitable during this process. Over five decades ago, Sir Arthur Lewis, the Nobel Laureate in Economics, observed that the defining feature of structural transformation in economies with large pools of surplus labor--a description that fits China well--is the gradual transfer of surplus labor from "traditional" low-productivity activities to "modern" high-productivity activities (Lewis, 1954). Lewis argued that this process is inevitably accompanied initially by rising levels of inequality as some make the transition and others are, at least temporarily, left behind. As Lewis put it: "Development must be inegalitarian because it does not start in every part of the economy at the same time." China is in such a process of structural transformation, and it may only be a matter of time before those left behind catch up. The rise in inequality would then be a transitional phenomenon, although because the transition is occurring on a decadal scale (even for a rapidly changing society and economy such as China's), inequality might continue to rise for several more years. And even when the transition is complete, because of the good drivers of inequality set in motion by the reforms, there will almost certainly be an increase in the steady-state inequality relative to that in the pre-reform period. But not all inequalities are benign. Geographic poverty traps, patterns of social exclusion, unequal opportunities for enhancing human capital, lack of access to credit and insurance, corruption and uneven influence can all conspire to simultaneously fuel rising inequality and prevent certain segments of the population from making the transition out of traditional low-productivity activities. These "bad" inequalities--rooted in market failures, coordination failures and governance failures--prevent individuals from connecting to markets and limit investment in human and physical capital. Bad inequalities are doubly harmful. First, they directly reduce the potential for growth because segments of the population are left behind, lacking the opportunity to contribute to the growth process. Second, on top of these direct human and economic costs, persistent bad inequalities in a setting of heightened aspirations can yield negative perceptions about the benefits of reform. Because it is difficult for citizens to disentangle the sources of the aggregate inequality in observed outcomes--to determine whether the underlying drivers are good or bad--societal intolerance for inequality of any kind emerges. And that can trigger social unrest or harden resistance to further needed reforms, thereby (indirectly) threatening the sustainability of growth. In effect, the persistence of bad inequalities drives out the good ones. Whether the rise in income inequality in China should be a concern therefore depends on the extent to which the rise is due to "good" incentive-promoting inequalities or "bad" opportunity-retarding inequalities. The evidence presented below suggests that both good and bad inequalities have played a role in China's growth and poverty reduction experience thus far, and continue to do so. Good incentive-promoting inequalities remain F o s t e r i n g e q u i t y 1 5 7 important for China's continued growth. At the same time, the evidence suggests that bad opportunity-retarding inequalities are becoming increasingly dominant in preventing certain segments of the population from connecting to the growth process. This should be of concern and the causes of such inequalities need to be addressed. But the mechanisms and policies that are introduced to do this should be carefully selected. Efforts to attenuate the bad inequalities should not undermine the drivers of good inequality to the point where the longer-term growth and the living standards of poor are threatened. The challenge China faces is to identify the mix of policies that directly target the bad inequalities without undermining the good ones. b. Inequalities in human capital underlie the three components of income inequality that have risen the most Differences in the human capital endowments of rural and urban workers explain much of the rural-urban labor productivity gap Understanding the rural- In proximate terms, the rural-urban income differences stem from the large differences in urban income divide labor productivity (value-added per worker) between those working in rural areas and those in urban areas. Estimates indicate that in 2004, value-added per worker (in real terms) in urban areas was 3.3 times that in rural areas (Figure 7.81). Some of that difference is undoubtedly due to differences in the sectoral shares of employment in rural and urban areas and the associated sectoral differences in productivity. In particular, agriculture, which continues to be the main source of employment for a majority of rural workers, has a very low value-added per worker compared to employment in services or industry, whether in rural or urban areas. Figure 7.81 : Labor productivity differentials between sectors and rural and urban areas (Index of labor productivity: 1978 labor productivity in agriculture=1) 80.0 76.4 70.0 60.0 50.0 Urban industry 40.0 36.6 Urban 30.0 24.3 20.0 23.6 Rural industry & Rural 10.0 services 11.4 Urban services Agriculture 3.3 0.0 1978 1983 1988 1993 1998 2003 Source and notes: Zhang and Tan (2006) for estimates up to 2001, updated to 2004 through forecasts and data from NBS Statistical Tearbooks. Labor productivity defined as value added per worker in constant 1978 Yuan. Index of labor productivity derived by taking the ratio relative to 1978 labor productivity in agriculture. But with migration increasingly offering a pathway into higher productivity work off the farm and in urban areas, and with education levels being a critical factor in determining access F o s t e r i n g e q u i t y 1 5 8 to such work, a more fundamental reason for the continued large rural-urban income gap turns out to be differences in human capital endowments of rural and urban workers. A decomposition of the rural-urban income differences in 1995 and in 2002 using a nationally representative sample reveals that differences in the educational attainment of a household's workers explains 25% of the mean difference in rural-urban incomes in both years (Sicular et al, 2007). In fact, differences in endowments of human capital are found to be the largest single observable component of the rural-urban gap apart from unobservable differences associated with a household's location. Moreover, in 2002, the contribution of differences in human capital endowments is the only observable characteristic that remains significant in explaining rural-urban income differences, suggesting that the importance of this component has risen over time. Data from the 2003 national samples of the RHS and UHS confirm the disparities in human capital endowments as well as the differences in the returns to these endowments between rural and urban areas. In rural areas, 32% of the population lived in households where the average years of schooling was greater than 9 years; the corresponding figure for urban areas was 76% (Figure 7.82). Figure 7.82 : Differences in human capital endowments between rural and urban areas in 2003 60% Rural 51.6% Urban 45.6% iont 40% 30.5% populafo 20.6% 23.6% 20% 16.1% 8.7% % 3.3% 0% Less than 6 Between 6 and 9 Between 9 and 12 More than 12 Average years of education of adult household members Source and notes: World Bank estimates from national samples of NBS' 2003 Rural and Urban Household Surveys. These differences in human capital endowments between rural and urban areas were magnified by the differences in per-capita incomes, controlling for human capital endowments. So, for instance, mean per-capita income for rural households with 9-12 years of average schooling was 4850 Yuan per year, whereas for urban households with similar levels of human capital, mean per-capita income at comparable prices was 24% higher at 6007 Yuan per year (Figure 7.83). Figure 7.83 : Differences in per-capita incomes between rural and urban areas in 2003, controlling for human capital endowments larur 9,000 8,290 8,457 Rural Urban in 6,007 6,000 4,963 4,850 meo se 4,560 3,040 inc a pric3,000 2,330 pitac-r 0 Pe Less than 6 Between 6 and 9 Between 9 and More than 12 12 Average years of education of adult household members Source and notes: World Bank estimates from national samples of NBS' 2003 Rural and Urban Household Surveys. All per capita incomes are expressed in 2003 all-China rural prices. F o s t e r i n g e q u i t y 1 5 9 Differential access to off-farm employment due, increasingly, to differences in schooling is the most visible source of inequality within rural areas The Gini index of income inequality within rural areas rose from 24.73 in 1981 to 37.26 in 2003. The rise in rural inequality in the early years of reform provides an example of the role that good incentive-promoting inequalities have played in China's growth. Under the Household Responsibility System (HRS) introduced in the early 1980s, rural households were assigned plots of land and became the residual claimants on the output from that land. This significantly enhanced the incentives for production. Prior to that, land had been farmed collectively, with all members sharing the output more-or-less equally. Incentives for individual effort in this setting were naturally very weak, and the reforms to this system implied by the HRS were critical in stimulating rural economic growth (and the associated poverty reduction) at the early stages of China's transition (Lin, 1992). But because there is wide spatial variation across rural areas in land productivity, agro-climatic conditions and access to markets, as well as differences across households in farming acumen and entrepreneurial bent, at the same time that these reforms stimulated growth and reduced poverty, they put upward pressure on inequality within rural areas. Over the years, however, as the share of agricultural incomes in rural incomes has declined, the contribution of the local agroclimatic factors to income inequality within rural areas has correspondingly decreased. Location continues to be important but increasingly, differences in income can be traced to the specific characteristics of individual households, not just of the villages in which they live (Benjamin, Brandt and Giles, and Wang 2008). And of these characteristics, the schooling level of the household's workers is perhaps the most critical. Growing inequality within Basic schooling was far more widespread in rural China at the outset of the reform period rural areas can also be than in many other countries, such as India, with similar levels of income. And China has traced to differences in achieved close to universal primary education. But inequalities in educational attainment human capital beyond primary school remain, and these have become an increasingly important source of disadvantage because a junior high school education, and in some instances, a senior high school education, has become a de facto prerequisite for accessing non-farm work particularly in urban areas, where wages far exceed the shadow wages in farming. In terms of the sources of rural incomes, studies have found that income from local non- farm enterprises and wage income from employment outside the village contribute disproportionately to differences in income levels across rural households, even within a village, and that differential access to employment off the farm has been the most visible proximate source of the rise in income inequality within rural areas (Benjamin, Brandt and Giles, and Wang 2008). Some of the differences in access stem from where households are located. In villages in remote interior areas, both local off-farm employment and migrant work opportunities are harder to access, the first because it is rare, the second because of the distances involved. But as the earlier analysis of the determinants of migration indicated, and as other studies of rural China have also found, education is the dominant factor underlying whether rural laborers are able to find more lucrative off-farm jobs (Zhao,1997; de Brauw et al., 2002). Fees for middle school A major reason why differences in educational attainment have persisted and in some are very high for many cases widened within rural areas is that the burden of educational expenditures, rural households particularly post-primary, is very high for many households. As local governments have had to rely more and more on financing schools through tuition and other fees charged to parents, in some villages, the cost of a middle school education has risen to the equivalent of more than 60 days of work at the prevailing daily wage for male laborers (Figure 7.84). For many poor households, this is unaffordable. And that appears to be why nearly twenty years after its passage, the aims of the Compulsory Education Law calling for universal 9-year education have yet to be fully attained. F o s t e r i n g e q u i t y 1 6 0 Figure 7.84 : The cost of schooling is high for many rural households (Days of work at local male daily wage needed to pay middle school fees per child per year) seeflooh 80 scelddi 60 myapotdedeen 40 rkowfo 20 ysaD 0 0 20 40 60 A verage male daily w age in village (yuan) Source and notes: Estimates from the 2004 World Bank-NBS Special Purpose Survey in 3037 villages. Income inequality within urban areas stems primarily from disparities in labor incomes attributable in large part to differences in educational attainment The rise in income inequality within urban areas is arguably the most notable feature of the evolution in income inequality in China over the last quarter-century. The Gini index of income inequality within urban areas nearly doubled in this period, from 16.2 in 1982 to 30.8 in 2003. And because of a number of measurement issues, even this likely underestimates the true extent of the increase (see Box 6.16). Box 7.20 : Why income inequality in urban areas may be underestimated Measured inequality in urban areas is likely to be under-estimated for two reasons. The first is that the most reliable source of data on the distribution of income in urban China, the Urban Household Survey of the National Bureau of Statistics, may be under-sampling unregistered non-hukou rural migrants who live and work in urban areas. The second is that sample surveys do not usually pick up those at the extremes (both the top and the bottom) of the income distribution. These reasons are not unique to China. Similar difficulties in adequately covering itinerant populations (who are likely to be at or near the bottom of the income distribution) as well as urban elites at the top of the distribution are faced by sample survey organizations in almost every country. In the case of China, however, the pace of income growth and the size of the floating population of migrants in urban areas suggest that these difficulties in estimating urban inequality may be particularly a concern. Increasing dispersion in Increasing dispersion in labor incomes is perhaps the most important reason that urban labor incomes are the inequality has gone up. That dispersion in earnings should have increased in urban China primary source of rising in the last quarter century is not surprising given the transformation of the urban labor income inequality in market that has taken place during this time. At the outset of that period, urban China had urban areas a system of fixed wage scales, allocation of labor by government and (hence) low returns to schooling (Fleisher and Wang, 2004). There were few incentives for work-effort or skill- acquisition. From this legacy of wage compression and low labor mobility, China moved gradually in the 1990s to a market-based system featuring a dynamic non-state sector and an increasingly open labor market. The pace of the transition quickened in 1997 when the government moved forward with aggressive restructuring of state-owned enterprises, leading, as noted earlier, to substantial layoffs, retirements, and exits from the labor force F o s t e r i n g e q u i t y 1 6 1 (Giles, Park, and Cai, 2006). These changes have contributed to rising inequality in labor incomes in two ways. First, the increase can be traced to the increased incidence of unemployment and declining labor force participation associated with the massive restructuring of state-owned enterprises. Second, the increasing dispersion in incomes also reflects the increasing dispersion in wages among those who are employed. Educational attainment has played a central role in both processes. In terms of the former, recent studies have found that educational attainment distinguished those who were most likely to retain their jobs in the face of widespread layoffs as well as those who were most likely to be re-employed after an unemployment spell. Interestingly, studies have also found that returns to education were higher amongst those re-employed compared to those who were continuously employed, an indication that urban labor markets are increasingly rewarding skill. This change in China's urban labor markets turns out to be central in explaining the rising dispersions in wages among those who are employed. Using a nationally representative sub-sample of the NBS Urban Household Survey, Park et al. (2006) find that by almost every measure, inequality in annual wages (at the household level) rose between 1988 and 2003. Table 7.78 : Rising wage inequality in urban China 1988 1992 1994 1997 2000 2003 Standard deviation of log wages 0.45 0.48 0.61 0.63 0.66 0.67 Percentile differentials: 90-10 1.06 1.30 1.66 1.64 1.74 1.81 90-75 0.28 0.44 0.47 0.47 0.46 0.48 90-50 0.52 0.75 0.90 0.89 0.92 0.96 75-50 0.24 0.30 0.42 0.42 0.47 0.48 75-25 0.50 0.59 0.82 0.82 0.88 0.92 50-10 0.54 0.56 0.76 0.75 0.81 0.85 50-25 0.26 0.28 0.39 0.39 0.42 0.44 25-10 0.27 0.27 0.37 0.35 0.40 0.41 Sources and notes: Estimates by Park et al (2006) from sub-samples of the NBS Urban Household Survey. It needs to be noted, however, that this rise in wage inequality occurred in the context of substantial growth in wages across the entire distribution. All percentiles experienced wage gains during the period, but higher percentiles experienced larger wage gains. Thus, the rise in inequality has not been a story of the rich getting richer and the poor becoming poorer, but rather of the rich getting richer faster than the poor. The difference between the bottom and top of the wage distribution in the annualized rate of wage growth over the period from 1988 to 2003 was about 8 percent per year in urban China (Park et al., 2006). To put this magnitude in perspective, consider that Juhn et al. (1993) found that the difference in wage growth at the top and bottom of the U.S. wage distribution from 1964 to 1988 to be less than two percent per year, a quarter of the difference in China. The rising wage inequality in urban China is another example of good incentive-promoting inequalities at work. With reforms that expanded the scope for employment in a growing private sector and the emergence of a competitive labor market, wage dispersion within skill categories and experience cohorts has increased considerably and returns to schooling have also risen. Among urban wage-earners in 6 surveyed provinces, the returns to a year of schooling increased from 4.0% in 1988 to reach 11.4% in 2003 (Figure 7.85). The high rate of return is confirmed for 2003 for all provinces. Using the national sample of the NBS urban household survey, the return to a year of schooling in urban areas was estimated to be 11.6%.74 From an international comparative perspective, over _________________________________________________________ 74These were calculated from wage regressions using a basic Mincer specification (education, experience and experience-squared, supplemented with gender and region dummies. F o s t e r i n g e q u i t y 1 6 2 these 15 years, China has switched from having very low returns to schooling to having relatively high rates of return. Figure 7.85 : The rising skill premium in urban China: 1988 to 2003 (Returns to an additional year of schooling as a % of annual earnings of employed workers) 12.0 10.0 11.4 8.0 10.1 6.0 7.3 8.1 4.0 6.8 2.0 4.0 4.7 4.7 0.0 1988 1990 1992 1994 1996 1998 2000 2003 Source and notes: Park et al. (2007), analyzing NBS urban household surveys in 6 provinces. Based on Mincer regressions of annual earnings on years of schooling, sex, potential experience, potential experience squared, and provincial dummy variables. The increases were particularly pronounced for higher education (Zhang et al., 2005). The annual wage premium for the average college-educated worker relative to a worker with only a primary education was about 25% in 1988. By 2003, this had increased to over 70%. Table 7.79 examines the returns to schooling for men and women and by region, and also by level of schooling. The table also shows how the returns to schooling change when different job characteristics are included as control variables. Women have higher returns to schooling than men (13.0% versus 10.7%), the returns vary by region but not significantly so, ranging from a high of 12.8% in the Southwest to 10.7% in the Northeast, and the returns to middle school, high school, and college are 30.5%, 30.1%, and 42.7%. Even after controlling for ownership, industry, and occupation, the returns to a year of schooling are still relatively high at 6.9% in 2003. It is also the case that over time, the returns to schooling have witnessed convergence (less variance) with respect to ownership groups, experience cohorts, and regions (Zhang et al., 2005). Overall, the evidence suggests that the value of education (or the cost of lack of education) has magnified. Table 7.79 : Estimates of the returns to education in urban China in 2003 (% of annual earnings of employed workers) All Women Men Returns to a year of schooling: national Basic 11.6 13.0 10.7 Controlling for ownership and sector 8.2 8.9 7.7 Controlling for ownership, sector, and occupation 6.9 7.5 6.5 Returns to a year of schooling: by region (basic) Coast 12.0 13.3 11.1 Northeast 10.7 11.7 9.9 Central 10.9 12.4 9.7 Southwest 12.8 13.5 12.0 Northwest 11.5 13.8 9.8 Returns to each level of schooling: national (basic) Junior high compared to primary 30.5 33.0 28.5 Senior high compared to junior high 30.1 35.6 26.0 College or above compared to senior high 42.7 47.4 39.9 Sources and notes: World Bank calculations from national sample of NBS 2003 UHS. The rising returns to education and their contribution to rising urban wage inequality have occurred even as urban labor markets have become more national in scope, reducing the the component of wage inequality that is attributable to regional wage differences. F o s t e r i n g e q u i t y 1 6 3 Regional wage differences still do exist in urban China. Estimates based on the 2003 NBS urban household surveys for all of China show that, controlling for individual characteristics, coastal wages remain about 20% higher than elsewhere in real terms, though the differences among non-coastal regions are relatively small (Figure 7.86). Figure 7.86 : Regional urban wage differences in 2003 (% difference in wages relative to coastal region) Northeast Central Southwest Northwest 0% -10% -20% -30% -40% Nominal With spatial price adjustments -50% Sources and notes: World Bank calculations from national sample of NBS 2003 Urban Household Survey, based on Mincer regressions of annual earnings on years of schooling, sex, potential experience, potential experience squared, and provincial dummy variables. But these differentials are relatively modest. And evidence from the 6-province sub- sample of the UHS suggests that inter-regional wage differences have started to come down from their peak in late 1990s, after increasing dramatically in the early 1990s (Figure 7.87). That is indicative of greater labor market integration over time. Figure 7.87 : There are signs of growing integration of urban labor markets (Provincial wage differences relative to Sichuan as a % of annual earnings of employed workers) 1.2 1 0.8 Guangdong 0.6 Zhejiang 0.4 0.2 Beijing Liaoning 0 -0.2 Shaanxi -0.4 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 198 198 199 199 199 199 199 199 199 199 199 199 200 200 200 200 Source and notes: Park et al. (2006), analyzing NBS urban household survey data for 6 provinces. Based on Mincer regressions of annual earnings on years of schooling, sex, potential experience, potential experience squared, and provincial dummy variables. Earnings are deflated by provincial CPIs to 1988 prices. For 2003, the relative importance of different factors in explaining the variation in wages across individuals can be illustrated by a simple analysis of variance using the 2003 NBS urban survey data. The variance decomposition results presented in Table 7.80 show that of the four observable factors explaining variation in wages, education explains the greatest amount (45.6%) followed by province (30.2%), experience (16.8%), and gender (7.0%). Overall, the two factors related to productivity account for much more of the wage variation than the two non-productive factors. F o s t e r i n g e q u i t y 1 6 4 Table 7.80 : Analysis of variance of urban wages in 2003 Factors underlying variation in wages Percent of explained Percent of all variance variance Factors related to productivity 62.3 15.3 Education 45.6 11.2 Experience 16.8 4.1 Factors unrelated to productivity 37.7 9.2 Gender 7.3 1.8 Province 30.2 7.4 Unobservable factors 75.5 Sources and notes: World Bank calculations from national sample of NBS 2003 UHS. Employment and income prospects of workers depend upon the overall integration of the labor market. This is particularly true for the poor, for whom local employment opportunities are often much less attractive than job opportunities in more distant markets. As labor markets become increasingly market-oriented and labor becomes increasingly mobile, the importance of workers' productivity (e.g., education, experience) in wage determination should increase while the importance of non-productive factors (gender, location of origin) should decrease. By altering the returns to education, labor market integration are also likely to influence the incentives for human capital investment.These developments in urban labor markets, while they have contributed to rising inequality are for the most part "good" because they reflect incentive-enhancing inequalities. c. Disparities in other aspects of human development mirror, and can be traced to disparities in income, raising the possibility of an inequality trap Disparities in other China's successes have not only been in raising incomes. China's progress in human aspects of human development has been equally, if not more, impressive. In terms of various human development mirror the development indicators, China compares favorably to levels achieved in middle-income disparities in income... countries. But as in the case of income growth and poverty reduction, the progress has been uneven. And like income, disparities in many non-income aspects of human development have grown in recent years. For instance, using provincial-level aggregates, for rural and urban areas from published Census reports, Zhang and Kanbur (2005) document rising inequality in illiteracy and infant-mortality rates between 1981 and 2000. Changes in institutional arrangements have led to a marketization of public service delivery, particularly in rural areas ...because of the Incomes matter more now than they used to because the change in the institutional marketization of public arrangements for the provision of health and education has led to a marketization of service delivery service delivery. The dismantling of the communes in rural areas and the move to the Household Responsibility System was arguably the single most important reason for the rapid decline in poverty China experienced in the first half of the 1980s. However, a byproduct of this institutional transformation has been the increasing marketization of public service delivery in rural areas. In place of the communes, local governments were charged with administering and financing these services, which in poorer areas, they were unable to fully do because of a lack of fiscal resources. Schools and health facilities have therefore had to increasingly rely on charging user fees to rural households in order to cover their costs. Incomes matter more now than they used to for access to health and education ...because incomes That disparities in human development mirror disparities in income is largely due to the matter more now than fact that incomes matter more now than they used to in determining access to health and they used to for access to education. This is evident both at the household-level and across provinces. For instance, health and education as Figure 7.88Error! Reference source not found. indicates, at the provincial level, the differences in per-capita GDP and income are associated, in recent years, with much greater differences in infant survival rates and high school enrollment rates than they used to in 1990. In 1990, a 10% higher provincial per-capita mean income was, on average associated with a 0.9 percentage point higher high school enrollment rate; in 2003, a similar income difference translated into a 2.5 percentage point difference in high school F o s t e r i n g e q u i t y 1 6 5 enrollment rates. Income differences seem to matter even more for differences in infant mortality rates. Figure 7.88 : Health and education outcomes at the provincial level are more closely tied to incomes than they used to be Infant survival rate (logit) High school enrollment rate (%) 5.5 100 5.0 2003 80 4.5 60 4.0 1981 2000 40 3.5 estimated 1990 1990 best-fit line 3.0 20 2.5 0 5.25 6.25 7.25 8.25 9.25 10.25 5.5 6.5 7.5 8.5 9.5 Provincial GDP per-capita (log) Provincial mean per-capita income (log) Sources and notes: For infant survival, estimated best-fit lines in 1981 and 1990 are reproduced from Grigoriou, Guillamont and Yang (2005); for 2000, World Bank estimates from published Census reports. Provincial high-school enrollment rates are from the 1997 and 2005 China Human Development Reports, which in turn rely on data from various years of the China Statistical Yearbook. ...and the burden of health and education expenditures has increased for households ...and that has increased A direct consequence of this is that the burden of health and education expenditures has the burden of health and increased for rural households (Figure 2.39). By one estimate, in the 10 years up to 2004, education expenditures out-of-pocket medical expenditures of households have increased 6-fold, and by forty for the poor times since 1980. Similarly, it is estimated that between 1997 and 2000, about 50% of the operational fees for basic education was paid by the farmers. In the survey of 3037 villages undertaken in connection with this report, primary school fees per year for a fifth grade student averaged 260 Yuan in 2004 while fees per student per year for middle school averaged 442 Yuan, or about 40% and 70%, respectively of the official poverty line. Because health and education determine income prospects, this raises the possibility of an inequality trap The educational attainment of a household's adult workers is the clearest observable predictor of poverty in China, in both rural and urban areas. It is also the most visible factor underlying income inequality, whether in terms of rural-urban income differences or in terms of income inequality within rural and urban areas. The rising importance of human capital in explaining income differences in both rural and urban China are a sign of market-based incentives playing a role in rewarding skill and providing increased incentives for work and skill-acquisition. That is a good thing. But those with relatively little schooling and few assets, or little access to credit are less able to respond these incentives and are less well positioned to take advantage of the new opportunities unleashed by market-oriented reforms. Income inequalities combined with credit market failures and governmental failures in service delivery therefore translate into inequalities in human capital endowments. In an environment where incomes increasingly reflect human capital endowments, these inequalities in turn reproduce income inequalities and propagate the inter-generational persistence of inequality and poverty, leading to an "inequality trap", which can worsen over time (World Bank, 2005). Only public action to ensure that every child has an equal opportunity to acquire the requisite human capital to contribute productively to a growing economy can break this vicious cycle. The health of a household's adult workers, and the ability or inability to work that it implies, is also a strong predictor of poverty, and health shocks and the burden of medical expenses and the loss of income they entail, are one of the most common reasons why households fall into poverty. The NBS reports that in China's 592 poor counties, among those households who fell into poverty in 2004, 26% had disabled or seriously ailing family F o s t e r i n g e q u i t y 1 6 6 members. With only the well-off being able to afford reasonable quality healthcare, inequalities in income are transmitted to those in health outcomes. There are thus large disparities in health outcomes between urban and rural areas and between the rich and the poor within both urban and rural areas. This too raises the prospects of an intergenerational inequality trap whereby households who have inadequate access to health care or coverage from health-related income shocks are unable to provide adequate human capital for their children, who, in turn, grow up with limited earnings opportunities and livelihood prospects. While policy initiatives and challenges in relation to affordable healthcare have already been discussed in chapter 6 as part of the discussion on the evolving social protection system in China, those for basic education are discussed in the following section. d. Compulsory education finance reform has helped improve access to affordable basic education but many challenges remain A major recent initiative to address the goal of universal primary and junior secondary education has been the compulsory education finance reform. Introduced in March 2006, and written into the revised Education Law in September 2006, the reform repeals tuition/miscellaneous fees and textbook fees for all students, and provides boarding subsidies to poor students (the so-called "two exemptions and one subsidy" policy) to reduce the households' financial burden of educational expenses, improve access and raise completion rates. The reform is financed through large-scale intergovernmental transfers to compensate local governments for the loss of school revenues resulting from the fee exemptions. A recent assessment of the impact of the reform indicates significant achievements75: · Provincial, county and school-level level data suggest a substantial increase in per student revenues and expenditures at both primary and junior secondary levels. The increases in budgetary revenues (transfers) more than made up for the loss in fees and surcharges. The transfers have also helped reduce the rural/urban gap in per student revenues, as well as the gap in spending per student across the primary, junior and senior secondary levels, thus contributing to a pro-poor shift of education financing. · Fee reduction has improved enrolment probabilities for girls, poorer students, and those in higher grades. Gross enrolment rates at the junior secondary level are estimated to have reached 98% in 2007. · Fee reductions are progressive, and are estimated to have accounted for 2.8% of the average expenditure for all households, and 6.2% for the bottom one-third of households. These savings appear to have been used by households to increase their voluntary education spending, thereby contributing to further human capital development. However, there also remain several policy challenges: · Transfers from central and provincial governments to counties still account for a relatively small share of total education revenues, for instance, only about a quarter in Gansu in 2006. There is also evidence that central government transfers have displaced transfers from the provincial governments, leading to continued disparities in allocation of expenditures subject to varying levels of county revenues. · The progress on boarding subsidies appears to have been much slower partly reflecting the fact that these are financed by county governments without corresponding transfers from the central or provincial governments. · The formula for compensation for miscellaneous fees is based on enrolments multiplied by previously set per student miscellaneous fees, which favors richer _________________________________________________________ 75This draws upon the study by World Bank (2008) on China's Rural Compulsory Education Reform. F o s t e r i n g e q u i t y 1 6 7 provinces (which had set relatively higher fees) at the expense of poorer ones. Due to economies of scale, the per-student formula also favors bigger schools at the expense of smaller ones. There is certainly scope for more equitable and needs- based allocations. · The budgeted expenditure for compulsory education has not been adequate to address the problem of existing school debts. Barred from charging any miscellaneous fees as well as from using the transfers received to repay debts, the schools have no feasible means of paying off their outstanding debts. · Besides miscellaneous and textbook fees, there are other direct costs of schooling including notebooks and writing materials, school uniforms, costs of providing bedding, boarding fees, and food for boarders, fees for computers, examinations and co-curricular activities. Evidence from Gansu indicates that 40% of students continue to pay some form of fees even after the elimination of miscellaneous fees. Also, some schools in financial difficulty are resorting to in-kind contributions from parents and fundraising activities which are not prohibited. These additional direct costs plus the indirect opportunity costs of the students' time in school can still be substantial for many poor households. · Concurrently with the compulsory education reform, there has also been a policy of school consolidation by closing teaching points with high unit costs, and transferring the students and teachers to bigger schools. There is evidence that the closure of smaller schools is compromising access in many areas with potential negative effects on attendance, dropouts and school performance (Box 7.21). Although revised guidelines require community consent before school closure, there is a need to monitor the implementation of the policy to mitigate any adverse impact. Box 7.21 :Consolidation of schools is making access to education more difficult in many areas The qualitative poverty assessment found many instances of the adverse effects of the policy of consolidation of schools on access to education, as illustrated by the following reports from several study villages. A village in Inner Mongolia: "The implemented policy of school centralization has increased the households' burden. The village is 10 miles away from the central school. Each student needs to spend 50 yuan on transportation a semester. The cost of accommodation is 260 yuan a semester, plus 10 kilogram millet and 7.5 kilogram beans. Thus the costs for a semester are more than 300 yuan a semester." A village in Sichuan: "The distance between the village and the school is at least 4 miles. It is not safe for the children on lower grades and thus parents or grandparents have to pick up or send them everyday. Some old people do not know how to ride a bike. Thus, they carry their children by handcart. It has become a new burden on peasants." A village in Yunnan: "It is a problem for rural households to support their children's accommodation in school. The costs for children to accommodate in school are greater than those if they accommodate at home. At the same time, parents are worried about young children's safety when they go between the village and school." A village in Jiangxi: "After the school was removed, most students must go to another village school. One way distance between the school and village is 7.5 kilometers. The central school in the township is even further. They have to accommodate in the school or parents have to rent a house in the township and stay with children. This had increased the risk of students' safety and the costs of education." Source: CASS-IOS (2008). Thus, while compulsory education finance reform has been a very important policy initiative in promoting universal compulsory education, there is a continuing need to monitor the implementation and impact of the policy, and future policy development will need to address some of the issues mentioned above. At the same time, there is also a need to look beyond compulsory education and turn attention to the next challenge of promoting greater high-school education. With senior secondary school enrolments well below half of those at the junior secondary level, there F o s t e r i n g e q u i t y 1 6 8 is a clear need for greater public support of high school education (Figure 7.89). Since poorer households have the greatest need for such support, there is a case for introducing some form of targeted subsidies so that there is greater opportunity to build on the investments into compulsory education through a successful progression to the next stage. Figure 7.89 : Senior secondary enrolments are less than half of enrolments at the junior secondary level 120 illion) 107.1 m( 100 0062 80 59.4 in 60 ent 40 25.1 17.4 20 enrolmlat 0 To Primary Junior secondary Senior secondary Higher education Sources and notes: NBS: China Statistical Yearbook 2007. e. Publicly-financed interventions can ensure equity of opportunity but under China's current fiscal system do not adequately do so A more adequate and In China, as in most of the world, local governments are at the frontlines of service- equitable allocation of delivery, social protection, and poverty reduction. However, unlike in much of the world, resources for local within China's highly decentralized fiscal system, local governments at the county level governments is needed and below, as well as village collectives, also bear the primary responsibility for financing these efforts. But local governments in China vary widely in terms of their revenue-raising capacity. These large disparities in fiscal resources across local jurisdictions translate into large disparities in levels of public spending. And that in turn leads to large differences in the quality and level of services provided, in the extent of social protection and unevenness in the pace of poverty reduction, thereby contributing to the inter-generational persistence, and even magnification, of bad inequalities. Implementing the broader poverty reduction agenda and ensuring that local governments all across China have the resources to provide a basic nationally determined level and quality of services and social protection to those living within their jurisdictions will require a more equitable and adequate allocation of resources for local governments. This is a two-fold task. There is certainly a need to move towards a more equitable allocation of resources through further equalizing transfers. The data indicate that the pattern of public expenditures favors richer provinces, and within provinces, richer local governments. But the challenge of financing the broader poverty reduction agenda is not simply one of redirecting resources from richer to poorer areas. Existing benefit levels and coverage are so low that a simple budget neutral spatial re-allocation of existing resources, even if it were feasible or advisable, would not be adequate. Therefore, there is a need as well to devote substantial additional resources to development-oriented poverty reduction, service delivery and social protection. And in doing so, because of existing disparities it will be critical to target all incremental resources to the localities where funding is currently inadequate. Very rough estimates suggest that given China's tremendous growth in recent years, and the relatively good shape public finances are in, even a substantial scaling up is in principle affordable. Expenditures on poverty-reduction-related activities are primarily the responsibility of local governments Expenditures on poverty- China has one of the most highly decentralized fiscal systems in the world. In 2005, sub- reduction-related F o s t e r i n g e q u i t y 1 6 9 activities are primarily the national governments accounted for nearly three quarters of the total budgetary responsibility of local expenditure by all levels of government. The sub-national share of expenditures on items governments that can broadly be construed as contributing to poverty reduction was even higher. Excluding expenditures on capital construction, such expenditures constituted 22% of the total budgetary expenditures by central and local governments combined, and local governments were responsible for nearly 95% of them (Table 7.81). Even if expenditures on capital construction (some of which go towards rural infrastructure and construction of schools, etc.) are included, the share borne by local governments was still 85%. Table 7.81 : Expenditures on poverty-reduction related activities are primarily borne by local governments, 2006 Main Items of Budgetary Expenditure of Share (%) Central and Local Governments in 2005 Combined Central Local of total expenditure govt. govt. budgetary (billion RMB) share (%) share (%) expenditures Expenditure for Supporting Agricultural Production 216.1 9.0 91.0 5.4 Operating Expenses for Education 478.0 6.2 93.8 11.9 Operating Expenses for Health 132.0 1.8 98.2 3.3 Pensions and Relief Funds for Social Welfare 90.8 0.6 99.4 2.3 Social Security Subsidiary Expenses 212.4 11.4 88.6 5.3 Expenditure for Supporting Undeveloped Areas 22.0 1.9 98.1 0.5 Expenditure for Price Subsidies 138.8 39.7 60.3 3.4 Combined budgetary expenditure on poverty reduction-related activities excluding capital construction 1290.1 10.2 89.8 32.1 Expenditure for Capital Construction 439.0 33.8 66.2 10.9 Combined budgetary expenditure on poverty reduction-related activities including capital construction 1729.2 16.2 83.8 43.0 Total budgetary expenditure of central and local governments 4022.3 24.8 75.2 100.0 Sources and notes: Table 8-13: Main Items of Budgetary Expenditure of Central and Local Governments, 2007 China Statistical Yearbook. That local governments bear the primary responsibility for financing poverty reduction efforts would not be a concern if local governments everywhere had the fiscal resources to do so adequately. But as is well known, because of varying economic circumstances, the revenue raising capacity of provincial and sub-provincial governments in China varies widely and so many local governments are unable to satisfactorily meet their spending obligations. Per-capita revenues at the provincial level ranged in 2006 from a low of 518 and 542 RMB in Tibet and Gansu to a high of 8684 RMB in Shanghai, a multiple of nearly 17 times. Even if the province-level cities of Beijing, Tianjin, and Shanghai are excluded, the per-capita revenue of the most revenue-rich province, Zhejiang, was, at 2607 RMB, five times as great as that of the Gansu (Figure 7.90). Such differences are unusual from a cross-country perspective. In Brazil, the richest state has 2.3 times the revenues per capita of the poorest state. In the US, the poorest state has about 65 percent of the revenues of the average state, and in Germany, any state falling below 95 percent of the average level gets subsidized. F o s t e r i n g e q u i t y 1 7 0 Figure 7.90 : Large disparities in local government revenues (provincial government revenue, yuan per capita, 2006) 9000 8000 7000 6000 5000 4000 3000 2000 1000 0 jinn ng sug n ng ia ng n ix n n n n i u t nghai haS ijingeB jia ilinJ ngxi ixg hou ns Tibe Tia an oning Fujia ndong beieH ubeiH an nhuA Zhe ngdong unaH naeH Ga Ji xinahS Lia ongolia haS ingxN inaaH unnaY Qinghai Ji Gua Mre injiaX aanhS huaciS hongqingC Gua Guiz ilongjiaeH Inn Sources and notes: NBS: China Statistical Yearbook 2007. Despite recent initiatives to enhance equalizing transfers, public expenditures tend to favor richer provinces and localities Because disparities in Recognizing the need to reduce these disparities, the government has in recent years, provincial revenues have introduced several initiatives to enhance the role of equalizing transfers. These initiatives grown, despite recent have had an impact. Compared to 1995, transfers to the provinces from the central initiatives to enhance government were much more equalizing in 2004 in the sense that the distribution of equalizing transfers, provincial expenditures provincial expenditures was, in 2004, much more equal relative to the distribution of in 2004 still favored richer revenues, than was the case in 1995. For instance, in 1995, the poorest 60% of the provinces as much as population lived in provinces that had a cumulative share of revenues of 40% and they did in 1995 transfers brought the expenditure share up to 43%. In 2004, the poorest 60% were in provinces with a revenue-share of only 33%, but transfers raised the expenditure share to 42% (Figure 7.91). But, as this example also suggests, because the disparity across provinces in revenue-raising capacity has grown noticeably since 1995, provincial expenditures in 2004 still favored richer provinces as much as they did in 1995. F o s t e r i n g e q u i t y 1 7 1 Figure 7.91 : The provincial distribution of expenditures in 2004 favored the richer provinces (Concentration curves of cumulative % of expenditures/revenues) 100 2004 expenditures 2004 revenues 1995 expenditures 1995 revenues 80 cial nues reve/ provin 60 of resu % 40 lativeu expendit Cum 20 public 0 0 20 40 60 80 100 Cumulative % of population ordered in terms of 2004 provincial per-capita income from poorest to richest Sources and notes: World Bank estimates based on data from the 2005 China Statistical Yearbook. The province-level concentration curves were constructed by first ordering provinces in terms of per- capita household income, and then aggregating up the population, expenditure and revenue shares of the provinces from the poorest to the richest. Tibet is excluded. Poverty-related The province-level distributions of expenditures on poverty reduction-related activities expenditures tended to tended to be slightly more pro-poor than the distribution of overall expenditures. But they be slightly more pro-poor still favored richer provinces. Figure 7.92 displays province-level concentration curves for but still favored richer various categories of poverty-related expenditures in 2004. Curves that are further below provinces the 45 degree line reflect more unequal expenditures across provinces. Operating expenditures on education and health do not appear in the figure because they favor richer provinces just as much as overall expenditures do--the concentration curves are indistinguishable. Expenditures on social assistance tend to be slightly more pro-poor but still favor richer provinces. In the case of expenditures on agriculture and related activities, however, there is no sign of a bias towards richer provinces. But neither is the pattern of expenditures markedly pro-poor. F o s t e r i n g e q u i t y 1 7 2 Figure 7.92 : Poverty-related expenditures in 2004 were slightly more pro-poor than total expenditures (Concentration curves of cumulative % of expenditures/revenues) 2004 provincial expenditures 100 2004 social assistance 2004 agriculture l 80 ince-levev nuese 60 pro of % e ditures/rev 40 expen 20 Cumulativ 0 0 20 40 60 80 100 Cumulative % of population (from poorest to richest province) Sources and notes: World Bank estimates based on data from the 2005 China Statistical Yearbook. The province-level concentration curves were constructed by first ordering provinces in terms of per- capita household income, and then aggregating up the population, expenditure and revenue shares of the provinces from the poorest to the richest. Tibet is excluded. Disparities in spending can be even more pronounced at lower levels of government Disparities below the Spending disparities below the provincial level tend to be even more pronounced. Official provincial level tend to be data indicate that health and education expenditures are more unequally distributed at the more pronounced, even county level, the level where most expenditures on social assistance, human development for poverty-related and poverty reduction take place, than at the provincial level. There are no easily available expenditures official data on disparities at the level of individual villages, where the overwhelming majority of China's poor reside. Data from the Special Purpose Survey of 3036 villages, undertaken jointly by the NBS and the World Bank for the purposes of this poverty assessment, indicate that the village-level distribution of social assistance and infrastructure investment expenditures is much less pro-poor than the provincial distribution would suggest. This can be seen clearly in Figure 7.93, which super-imposes the province-level concentration curves for social assistance expenditures, and expenditures related to agriculture in 2004, on the village-level concentration curves for social assistance, also in 2004, and infrastructure investment expenditures, averaged over the four years from 2001 to 2004. The village-level concentration curves were derived by first ordering the villages based on per-capita village revenues. In the absence of a direct measure of the incidence of poverty in each village, per-capita village revenues were taken to be a suitable proxy. Figures on social assistance expenditures were constructed from information on the number of beneficiaries of the various rural social assistance programs, and the average level of benefits that each received. These benefits were funded partly through resources provided by higher levels of government, and partly through contributions made by the village committee itself. Infrastructure investment numbers only included that which was directly funded financially by the government or village committee. In-kind and matching contributions from the village residents were not included. F o s t e r i n g e q u i t y 1 7 3 Figure 7.93 : The village-level distribution of expenditures in 2004 was even less pro-poor Village-level social assistance Province-level social assistance Village-level investment Province-level agriculture 100 90 turesi 80 70 expend 60 of 50 % 40 30 lativeu 20 10 Cum 0 0 20 40 60 80 100 Cumulative % of population from poorest to richest Source: World Bank estimates based on the 2004 World Bank-NBS Village Level Survey of 3037 villages. The data presented above indicate that there is still much to be done in terms of making poverty-related public expenditures more equitable and pro-poor. Equalizing transfers from the center need to continue and be scaled up, but attention has to be paid as well to how equitable the allocation of resources is within provinces, and going one step further down, within counties. f. Group-based inequities are particularly visible along certain axes The rural-urban income gap is only one manifestation of the great divide that separates China's urban and rural areas The rural-urban income The rural-urban income gap is only one manifestation of the great divide that separates gap is only one China's urban and rural areas. On almost every dimension of welfare, the rural population manifestation of the great lags behind the urban population, in part because, until recently, government policies and divide that separates public expenditures had a pronounced urban bias. Literacy rates, enrollment rates China's urban and rural areas (especially past primary school), and educational attainment rates (especially past junior high school) are all much lower in rural areas, and while there have been improvements in all these indicators in rural areas in absolute terms, the relative gap between rural and urban areas has continued to widen. Infant mortality rates, maternal mortality, and other indicators of health status are much worse in rural areas. There are signs, even, of a growing difference in the average heights of rural and urban children because of differences in nutrition. What perhaps has made these rural-urban differences so salient and a source of considerable controversy is the fact that these differences in outcomes can be traced to rural-urban differences in the levels of public expenditure and support. In terms of publicly financed per-capita expenditures on education and health, rural areas lag far behind urban areas. Social assistance funds, too, are disproportionately directed in support of the urban population. At the same time, it needs At the same time, it needs to be kept in mind that not all rural areas are the same. There to be kept in mind that are in fact pronounced regional differences in the extent of the rural-urban disparity, not not all rural areas are the just in terms of incomes, but also on the other dimensions of well-being, and in terms of same levels of public support. The rural-urban income Regional differences in the contribution of rural-urban income differences to overall gap as the source of inequality within a region provide one indication of the regional dimension to the rural- inequality is small in urban divide. In the western regions nearly half--exactly half in the northwest, 44% in the coastal areas... F o s t e r i n g e q u i t y 1 7 4 southwest--of the overall level of inequality stems from differences in mean incomes between rural and urban areas. On the other hand, the component of overall inequality attributable to the rural-urban income gap is only 12% in the coastal region (Figure 7.94). ...so in focusing on rural And so, while an overall focus on rural poverty reduction is warranted, and needs to be areas, there is still a need recognized as the top priority, the approach and the allocation of funds needs to be for the approach be regionally differentiated. In particular, a greater share of incremental resources have to be regionally differentiated targeted to those rural areas, mostly in the central and western regions, that are the most disadvantaged. Figure 7.94 : The contribution of rural-urban income differences to overall inequality varies by region (Region-wise Theil decomposition of income inequality in 2003) ytilau 40.0 Within rural Within urban Between rural-urban 30.0 eqnifo 3.0 4.7 17.9 13.7 20.0 9.2 9.8 6.8 8.3 exd 9.3 5.4 10.0 inl 12.5 8.5 10.3 10.9 9.8 eihT 0.0 Coastal Northeast Central Southwest Northwest Source and notes: World Bank estimates from NBS 2003 Rural and Urban Household Surveys. Estimates of inequality derived from household-level income data adjusted for spatial price differences. The gender gap in China begins at birth and continues through the working years and into old age The chapter on basic facts about the poor in China indicated that children, and in particular, girls are at a higher risk of poverty than working-age adults and even the elderly. But the data also reveal a more fundamental indicator of bias against girls, biases that may originate at the time of birth. And that is the ratio of girls to boys amongst children below the age of 16. For the nation as a whole, the ratio is 873 girls for every 1000 boys. Ironically, amongst the poor, the sex-ratio is less adverse for girls. As shown in Figure 7.95, while the overall sex ratio for those below 16 is 873, the ratio for the non-poor is 864 while in contrast the ratio for the poor is 924. Thus, poorer households on average have 60 more girls per thousand boys than non-poor households. Figure 7.95 : Child sex ratio amongst the poor and non-poor in China, 2003 (number of girls less than 16 years old for every thousand boys below 16) 950 924 900 873 864 850 800 Poor Non-poor Total Sources and notes: World Bank estimates from national samples of the NBS 2003 Rural and Urban Household Surveys. Note that the poor are defined as those with per-capita consumption below the World Bank poverty line. How does one interpret a higher female-to-male sex ratio amongst children of the poor relative to the non-poor? It could imply either of two things. First, practices such as sex- selective abortions, abandonment and neglect of female children are more common amongst the non-poor than the poor. Or second, there is general sex-preference for a F o s t e r i n g e q u i t y 1 7 5 male child in planning family size such that families with more children have a higher ratio of girls to boys. But larger families on average tend to be poorer in terms of their per capita income or consumption. Without further analysis, it is not possible to determine the relative roles of these two factors, but in either case the evidence points to significant underlying discrimination against female children. The basic inequity suggested by the biased sex ratio continues into many different aspects of the lives of Chinese women including education, health and employment (Table 7.82). Table 7.82 : Women fare worse than men in education, health and employment Female to male Year Female Male ratio Illiteracy rate (age: 15 and above) 2006 13.7 4.9 2.82 Mean years of schooling (age: 16 and above) 2003 7.1 8.6 0.83 Infant mortality rate (per thousand) 2000 33.7 23.9 1.41 Labor force participation rate (%) (Female: 16-54 years; Male: 16- 59 years) 2000 78.3 87.7 0.89 Unemployment rate (%) 2000 9.0 7.6 1.18 Average wage in enterprises (Yuan) - manufacturing 2002 13544 16652 0.81 - wholesale and retail trade and catering 2002 13385 16705 0.80 Insurance paid by work unit (%) - medical insurance 2000 45.6 54.5 0.84 - pension insurance 2000 57.1 62.1 0.92 - unemployment insurance 2000 22.4 26.3 0.85 - work injury insurance 2000 29.7 40.7 0.73 Sources and notes: National Bureau of Statistics. In rural areas, despite the principle of gender equality in land rights espoused in the Land Management Law (LML) and the Rural Land Contracting Law (RCRL), women's land rights within a patrilocal and patriarchal system remain precarious. Evidence from the 2004 World Bank-NBS survey sheds some light on this issue in terms of the reported land allocation norms for women marrying into or out of the village, and for widowed or divorced women (Table 7.83). Table 7.83 : Village policies in relation to women's land rights, 2004 (percentage of villages) Women Women marrying marrying Widowed Divorced in out women* women* Receive/lose land: ... immediately 3.7 3.3 Get land automatically 44.5 8.6 ... at the next big/small adjustment 58.1 58.4 Get land only if family agrees 4.9 17.3 ... only when land becomes available 4.9 Don't get any land 1.5 3.8 ... won't get land 6.5 No rule 49.1 70.3 ... won't lose land 11.0 ... depends on case 4.8 4.5 ... no rule 22.0 22.7 Total 100.0 100.0 Total 100.0 100.0 Sources and notes: World Bank estimates based on the 2004 World Bank-NBS Village Level Survey. * Between age 25-50 and if they do not have land in their maiden village. First, despite the existence of a clear policy in the RCRL in relation to women relocating to their husband's village after marriage (viz., that they will not lose land rights in their maiden village until they are granted rights in their spouse's village), more than a quarter of villages report "no rule" or "depends on case" as the village policy. About 60% of the villages report that women marrying out of the village will lose their land rights at the next big or small land readjustment. Lack of a clear policy is more acute for widowed or F o s t e r i n g e q u i t y 1 7 6 divorced women with nearly 50 and 70% of the villages respectively reporting "no rule" as their policy. Thus, in the prevailing legal and social environment, women have much lesser assurance of retaining their land use rights. The security of women's land rights have become particularly salient in recent years as the higher propensity of men to migrate to urban areas for work has led to what has been termed the "feminization" of agriculture. Working-age and child-bearing women in particular, who are less likely to migrate, face the triple burden on child-care, elder-care, and farm work. In urban areas as well, working women confront a gender gap. Analysis of data from the Urban Household Survey indicates that women are more likely to be in lower paying occupations and sectors, and that even controlling for these, and for educational attainment and experience, female wages tend to be lower (Park et al, 2006). Even in old age, women face unequal treatment in some areas. Most strikingly, the China Urban and Rural Elderly Survey of 2000 indicated that the rural pension system, the coverage of which is low to begin with, largely bypasses older women. Only 10% of pensioners were women even though women actually comprise a larger fraction of the elderly. Integrating rural migrant workers and their families in urban areas The large scale of rural-urban migration has led to increasing concern about whether migrants living in cities are becoming a new vulnerable population. According to the 2000 population census, migrants, defined as those whose hukou is in a different county or city than their place of residence, accounted for 14.6 (19.6) percent of China's city population (employment), 8.0 (11.2) percent of the population (employment) in townships, and 5.8 (7.0) percent of China's total population (employment) (Table 7.84). Nearly 60% of all migrants live in cities. Table 7.84 : Migrant population and workers in urban and rural China Cities Townships Villages All % of total population 23.4 13.2 63.4 100.0 % of migrant population 58.8 18.3 23.0 100.0 % of population that are migrants 14.6 8.0 2.1 5.8 % of workers that are migrants 19.6 11.2 2.5 7.0 Sources and notes: 2000 National Census. Migrant is defined as persons whose hukou is in a different county or city than their place of residence. Lacking local residence permits (hukous), migrants frequently are treated as outsiders by municipal governments and local residents alike, and lack the political voice to stand up for their interests (Solinger, 2005). Understanding the situation of migrants in cities is an important policy question not only because migrants are a large and growing group of citizens whose needs and welfare merit policy attention, but also because the treatment of migrants in cities affects the incentives of other rural residents to migrate, and so affects the opportunities of the rural poor. Reducing barriers to migration may increase labor market integration which arguably furthers both growth and equity objectives. Because no national surveys systematically sample migrants, knowledge of the economic and social conditions of migrants living in urban areas remains limited. Improving the quantity and quality of data on this rapidly growing population group itself deserves high priority, because lack of such information could lead to inaccurate measurement of urban incomes and expenditures, urban-rural income gaps, and overall inequality and poverty. The goal of this section is to present recent evidence on the situation of migrants in urban China, drawing upon existing literature as well as analysis of recent data from the China Urban Labor Survey in 2005, which surveyed both migrants and local residents in urban areas (see the annex on data sources). The following sections examine income poverty, employment opportunities, housing, educational access, social insurance programs, and social assistance. F o s t e r i n g e q u i t y 1 7 7 Migration and urban What is the extent of poverty among migrants in urban areas and how does that compare poverty with the poverty and living standards of urban residents? Unfortunately, NBS urban household surveys significantly undersample migrants and so cannot be used to answer these questions reliably. NBS did begin including migrants in their urban sampling frame in 2002, but only about 2% of the NBS sampled urban population are migrants, far less than migrants' actual share of city populations, which was estimated in the 2000 census to be 14.6% (Table 7.84). The NBS Urban Household Survey, like nearly all urban surveys in China, primarily samples migrants who are registered as temporary residents of urban city districts. This excludes many migrants who are unregistered, or who live in dormitories or in places of work. A few surveys with broad coverage have measured the income of migrants as well as urban residents. A report on urban poverty by the ADB uses data from a one-time urban survey conducted by NBS in provincial capital cities in 1999 to analyze the poverty situation of migrants and local residents (ADB, 2004). The report estimates income poverty headcount rates of 10.3% for local residents and 15.2% for migrants. A recent study using data from the China Income Distribution Survey (CIDS) conducted in 2002 in cities in 6 provinces estimates income poverty headcount rates of 3% for local residents and 10% for migrants using urban Di Bao lines, and 6% for local residents and 16% for migrants using a higher poverty line (Du, Gregory, and Meng, 2006). However, the CIDS surveyed only rural migrants and in both the NBS and CIDS surveys, incomes of local residents are based on self-recorded diaries while the incomes of migrants are based on one-time surveys that are known to produce lower income estimates. The 2005 China Urban Labor Survey sampled migrant and urban resident households in 5 large cities (Shanghai, Fuzhou, Shenyang, Xian, and Wuhan) and in 5 smaller cities in the same regions using a consistent sampling and survey methodology. Unlike the other migrant surveys just described, the CULS found that the poverty headcount rate of migrants was similar to that of urban residents in the 5 large cities; it was lower than that of urban residents in the 5 small cities using the relative poverty lines equivalent to twice and thrice the World Bank line (Table 7.85). Using the more austere World Bank line itself, the poverty headcount rate of migrants was higher than that of urban residents in small cities, and as a result of this, measures of the depth of poverty (the poverty gap indices) for migrants in small were higher also for twice the World Bank line. Overall, the results point to comparable levels of poverty amongst migrants and urban residents. Table 7.85 : Urban poverty headcount rates for local residents and migrants (%) Poverty line in terms of income WB poverty Twice the WB Thrice the WB line poverty line line 5 Large Cities Urban residents 1.4 3.8 9.8 Migrants 1.4 3.3 9.6 5 Small Cities Urban residents 2.8 7.6 14.6 Migrants 5.3 6.6 10.5 Source: China Urban Labor Survey 2005. See Park, Wang, and Cai (2006) for greater detail. From Table 7.86, we can also see that there is considerable variation in the poverty rate of migrants versus local residents in different cities. Using twice the World Bank poverty line as the threshold, the poverty rate of migrants is higher than that of urban residents in 4 of 5 large cities and in 3 of 5 small cities. The main point is that overall the difference in poverty rates is not very great between the two groups. F o s t e r i n g e q u i t y 1 7 8 Table 7.86 : Income poverty headcount rates of local residents and migrants, by city (using twice the World Bank poverty line) Province City Local residents Migrants Large cities Shanghai Shanghai 0.6 0.7 Hunan Wuhan 6.7 7.8 Liaoning Shengyang 4.9 6.4 Fujian Fuzhou 1.5 2.8 Shaanxi Xian 6.4 2.5 Small cities Zhejiang Wuxi 2.7 4.0 Hubei Yichan 16.5 22.9 Liaoning Benxi 4.7 11.9 Guangdong Zhuhai 9.7 6.4 Shaanxi Baoji 8.3 5.1 Source and notes: China Urban Labor Survey 2005 (5 large cities). See Park, Wang, and Cai (2006) for greater detail. Local residents in small cities were surveyed in 2004. The absence of a higher rate of income poverty among migrants might seem surprising since migrants generally have less human capital and receive lower wages than urban residents. However, as seen in Table 7.87, even though the hourly earnings of local residents are on average more than twice as great as that of migrants, compared to local residents, migrants are much less likely to be out of the labor force or unemployed, work much longer hours (over 300 hours per month compared to less than 200 hours for local residents), and live in households with lower dependency ratios (1.6 compared to 2.2). It seems that migrants who cannot find gainful employment in cities are likely to return to rural areas, and the dependents of low income earners often remain in rural villages. A single migrant can work for a wage as low as 200 Yuan per month and still be above the relative income poverty line (equivalent to twice the World Bank line). Thus, the poverty of migrants in urban areas is low because migration in China remains a selective process. Table 7.87 : Factors explaining relatively low poverty rate of migrants in 5 large cities Local residents Migrants Labor force participation rate (%) 56 89 Unemployment rate (%) 8.5 1.6 Hourly earnings (yuan) 7.74 3.83 Monthly hours worked 198 304 Dependency ratio 2.22 1.63 Mean income per capita 12369 9685 Median income per capita 10140 7721 Source: China Urban Labor Survey 2005 (5 large cities). See Park, Wang, and Cai (2006) for greater detail. Migrants work in An important policy question is whether migrants have the same work opportunities as segmented labor local residents in urban labor markets. Measuring discrimination in any context is markets? challenging, because differences between groups (i.e., migrants and local residents) in occupational attainment or earnings could simply reflect unobserved productivity differences. In China, many municipal governments passed regulations restricting the occupations for which migrants are eligible to work. Such restrictions increased in the late 1990s when millions of state-owned enterprise workers were laid off. At the same time, the rapid growth of migration and open policies in some coastal cities have led some to question the relevance of hukou status in determining employment outcomes. The most complete source of information on migrant employment in urban areas remains the 2000 census. Table 7.88 summarizes the characteristics of migrant and local resident workers in cities, as well as their occupations and sectors of work. In comparison to local residents, migrants are younger (mean age of 29.5 versus 36.3), less likely to be female (40.6 versus 42.5 percent), and are less educated. The vast majority of migrants have a middle school degree or lower (75.5%) while the majority of local residents have high school or college degrees (57.0%). In terms of occupations, migrants are most likely to be production or transport workers (51.0%) followed by commercial or service workers (36.4%). This is also true for local residents, but the share of workers in white collar jobs F o s t e r i n g e q u i t y 1 7 9 is much higher for local residents (38.8%) than for migrants (12.5%). Similarly, migrant jobs are concentrated in manufacturing (40.3%) and retail and wholesale trade (28.8%), while local residents are spread more evenly across sectors and are much more likely to be working in high-skill sectors such as finance or education. Table 7.88 : Characteristics, occupations and sectors of local residents and migrants in Chinese cities, 2000 Local residents Migrants Basic characteristics Age (years) 36.3 29.5 Female 42.5 40.6 Primary schooling or less 8.2 19 Middle school 34.9 56.5 High school 35 18.9 College and above 22 5.7 Occupations: % of residents % of migrants Government, party, and managers 6.0 2.7 Technical workers 19.8 4.8 Clerical workers 13.0 5.0 Commercial and service workers 24.5 36.4 Production and transport workers 36.7 51.0 Other 0.1 0.1 Sectors % of residents % of migrants Mining 2.2 1.0 Manufacturing 31.9 40.3 Utilities 2.5 0.4 Construction 5.6 10.7 Geological survey, water management 0.5 0.1 Transport, storage, post, communication 7.8 4.2 Retail and wholesale trade 17.8 28.8 Finance and insurance 2.6 0.5 Real estate 1.3 0.8 Social services 7.2 8.5 Health, sports, social welfare 3.6 1.0 Education, culture, arts 7.1 1.8 Scientific research and technical service 1.3 0.3 Government, party, NGOs 7.8 1.2 Other 0.9 0.3 Source and notes: 2000 Census. Migrant is defined as persons whose hukou is in a different county or city than their place of residence. Using the 2000 census data, Park, Zhao, and Huang (2006) construct indices of labor market segmentation in different provinces by estimating the probability that migrants and local residents work in different occupations or sectors after controlling for differences in individual characteristics. They find that coastal provinces are much less segmented than central and interior provinces, and that the extent of segmentation is positively correlated with the unemployment rate of urban residents and negatively correlated with GDP per capita and the importance of state-owned enterprises in industrial output. Separate analysis of rural survey data finds that rural residents are less likely to migrate to provinces with higher segmentation indices even after controlling for the level of development. Because the census did not include questions about income, the data cannot be used to examine earnings differences between migrants and local residents. Using data from surveys in Shanghai in the mid-1990s, Meng and Zhang (2001) analyzed earnings differences between local residents and rural migrants in and found that local residents' hourly earnings were nearly twice that of migrants. Nearly all of the difference was due to unexplained differences in pay within the same occupations, suggesting that discrimination within occupations may be substantial. A similar decomposition exercise was conducted using the CULS 2005 data from 5 large cities. Table 7.89 describes the distribution of local residents and migrants across occupations. The concentration of migrants in the commercial and service work is much F o s t e r i n g e q u i t y 1 8 0 more pronounced than in the census data, suggesting that migrant workers in manufacturing (who typically live in dormitories provided by the workplace) were undersampled in the CULS. The CULS also found that over 60% of migrants were self- employed. Table 7.89 : Occupations of local residents and migrants in 5 large cities in 2005 Local residents Migrants All Wage workers Government, party, and managers 9.2 8.4 3.9 Technical workers 15.2 5.3 4.8 Clerical workers 14.5 3.5 9.5 Commercial and service workers 18.1 71.4 61.2 Production, transport, and other 23 11.5 20.7 Source and notes: China Urban Labor Survey 2005 (5 large cities-Shenyang, Wuhan, Xian, Shanghai and Fuzhou). See Park, Wang, and Cai (2006) for greater detail. The results of the decomposition of earnings differences between migrants and local residents are presented in Table 7.90. Among all workers, the hourly earnings of local residents on average was about 70% higher than that of migrants. Of this difference, 78.2% was within-occupation differences and 21.8% was between-occupation differences. The share of the earnings difference that cannot be explained by differences in observable characteristics, and so is possibly associated with discrimination, is 39.9%, with nearly all of this unexplained difference being within occupations, just as found by Meng and Zhang (2001). Restricting attention to hired workers only (excluding the self-employed), the results are roughly similar, except the earnings gap is smaller and the unexplained share of the gap is greater. Overall, there is an earnings gap between local residents and migrants of nearly 30%, nearly all within occupational categories, that cannot be explained by differences in education, age, gender, or location. Table 7.90 : Decomposition of hourly earnings differences between migrants and local residents All workers Hired workers Total differential in log hourly earnings 0.694 0.505 Of which (percent): Intra-occupational 78.2 67.5 Explained 44.1 32.9 Unexplained 34.1 34.6 Inter-occupational 21.8 32.5 Explained 16.0 19.7 Unexplained 5.8 12.8 Total Explained 60.1 52.6 Total Unexplained 39.9 47.4 Source and notes: China Urban Labor Survey 2005 (5 large cities-Shenyang, Wuhan, Xian, Shanghai and Fuzhou). See Park, Wang, and Cai (2006) for greater detail. To summarize, evidence suggests that significant differences persist in how migrants and local residents are treated in urban labor markets. Whether these differences reflect discriminatory policies, discriminatory attitudes of employers or consumers, or unobserved productivity differences merits more in-depth investigation. Maintaining vigilance in promoting the equal treatment of migrants by employers can help promote the welfare of migrants and their families in both urban and rural areas. Access of migrant One major concern of migrants is the schooling of their children, which some feel perhaps children to education provides the greatest incentive for migrants to obtain local hukou. In urban areas, children of non-local residents are frequently excluded from attending local schools or are asked to pay extra fees. In some cities, migrants have established their own schools with privately hired teachers. However, often such schools lack public financing so must charge relatively high fees. Most migrant schools are unable to meet the quality standards required for recognition by local educational bureaus. In 2004, the government passed a new regulation requiring that urban governments provide educational access to the children of migrant families. However, it remains unclear to what extent the new law has been implemented. F o s t e r i n g e q u i t y 1 8 1 The China Urban Labor Surveys in 2001 and 2005 asked migrants detailed questions about the costs of sending their children to school. Specifically, parents with children in school were asked to report three estimates of educational costs: the costs of schooling in the city, the estimated costs of schooling in the city if the parents had a local hukou, and the costs of schooling in the place of hukou. As reported in Table 7.91, in 2005 school fees for migrant children would have been 35% lower on average if migrants had local hukou, compared to 43% lower in 2001. In 2005, 69% of migrants report that school fees would be lower with a local hukou, compared to 74% in 2001. Although these results suggest slight reductions in educational discrimination against migrants from 2001 to 2005, breaking down the results by city reveals that the gains were entirely due to the substantial reductions in educational discrimination against migrants in Shanghai. By 2005, only 22% of migrants in Shanghai reported having to pay extra school fees because they lacked a local hukou compared to 80% in 2001, and the differences in mean school fees with and without local hukou were negligible. In the other three cities with data for both years, both the percentage differences in school fees associated with hukou status and the percentage of migrants reporting having to pay extra fees because of lack of hukou actually increased slightly from 2001 to 2005. Overall, it appears that in most Chinese cities, little progress has been made in reducing educational discrimination against migrants. Table 7.91 : City educational costs of migrant children with and without local hukou % reduction in mean city school Mean city Mean city school fees with local % for whom city school fees fees if local hukou school fees is (yuan) hukou (yuan) greater without Year (a) (b) ((a)-(b))/(a) local hukou All 2001 1628 931 43 74 2005 1571 998 35 69 Shanghai 2001 1744 926 47 80 2005 1343 1311 2 22 Wuhan 2001 1402 893 36 65 2005 1193 700 42 69 Fuzhou 2001 1973 967 51 73 2005 1892 820 55 78 Xian 2001 1515 954 37 80 2005 1793 980 42 89 Source and notes: 2001 and 2005 China Urban Labor Survey in 4 large cities (Wuhan, Xian, Shanghai and Fuzhou), based on estimates by migrant parents with children in school. Prices deflated to 2003 Yuan. Shenyang is excluded because the questions were not asked properly in 2005. Table 7.92 reports differences in schools fees of migrant children in the place of hukou and in the cities where migrants reside. In 2005, mean school fees in the cities were twice as great as in the place of hukou. This difference was somewhat lower than in 2001, as rural school fees increased at a faster rate than school fees in the cities. In 2005, nearly 80% of migrants reported that school fees were higher in the city, compared to 85% in 2001. Table 7.92 also shows that city school fees for primary school were much higher than in the place of hukou in comparison to secondary school fees, and that the differences in primary school fees increased from 2001 to 2005 while the differences for secondary school fees fell. Again, the results also differ by city. In particular, Shanghai appeared to make significant strides in reducing city school fees of migrants, while the results in other cities were mixed and showed relatively little change over time. Also, it is worth pointing out that the city school fees for migrants reported in Table 7.92 are much lower than average annual school fees paid by local residents, which were 2882 Yuan in 2005. Local residents generally pay higher fees to send their children to higher quality schools. F o s t e r i n g e q u i t y 1 8 2 Table 7.92 : School fees of migrant children in place of hukou and in cities School fees at School % by which mean % for which place of fees in school fees in city school fees in hukou city exceed mean school city are higher (yuan) (yuan) fees in place of hukou than in place of Year (a) (b) ((a)-(b))/(a) hukou All 2001 695 1568 126 85 2005 768 1554 102 79 Primary school 2001 482 1189 147 87 2005 403 1090 171 86 Junior/senior high 2001 1112 2421 118 82 2005 1357 2312 70 68 Source and notes: 2001 and 2005 China Urban Labor Survey in 4 large cities (Wuhan, Xian, Shanghai and Fuzhou), based on estimates by migrant parents with children in school. Prices deflated to 2003 Yuan. Shenyang is excluded because the questions were not asked properly in 2005. An obvious policy question is whether the high costs of educating children in the city act as a barrier to migration. Based on the estimates presented in Table 7.91 and Table 7.92, eliminating educational discrimination against migrants could substantially reduce the cost difference between sending children to school in the city and in the place of hukou. Parents also need to consider differences in school quality, the costs of housing and food, and the ability of parents or other individuals to take care of children when not in school. One interesting finding from the 2001 and 2005 CULS surveys in 4 large cities was that the percent of migrant children attending school in the local city increased significantly from 38% to 57% despite the lack of progress in reducing educational discrimination. This trend is likely associated to the increasing permanence of household migration. In 2005, migrant parents with children in school on average had lived in the local city for 9.5 years, compared to 7.0 years in 2001. Housing situation of With the rapid increase in the flow of migrants to the cities, and more of them staying on in migrants the cities for longer periods, the issue of their housing conditions has become increasingly challenging. As noted earlier (Figure 2.42), data for 2005 indicated that nearly half of the migrant population have lived in their current city of residence for more than 5 years, and 20% have lived there for more than 10 years. Housing reforms in the late 1990s have deepened the market for rental units. However, because most local residents were allowed to purchase housing at subsidized prices while migrants must pay market rental rates, housing reforms have increased wealth and income inequality in urban areas (Sicular et al, 2007). Using 1999 survey data from 13 cities in 6 provinces, Sato (2006) found that 67% of migrant households in urban areas rented their housing, and the average annual rent was 2281 Yuan ($285), or 26% of average migrant household expenditures. In contrast, local residents only spent only 7% of expenditures on housing. The 2005 CULS asked questions about the housing situation of migrants. As seen in Figure 7.96, compared to local residents, migrants live in housing with much less space (10.7 m2 versus 18.4 m2), and less access to drinking water (78.1% versus 98.8%), toilets (45.5% versus 88.2%) and less heating (31.8% versus 61.7%). Sato (2006) and Logan, Fang and Zhang (2006) using earlier data also find significant differences in the quality of housing of migrants versus local residents. F o s t e r i n g e q u i t y 1 8 3 Figure 7.96 : Housing conditions of local residents and migrants 98.8 Urban residents Migrants 100.0 88.2 78.1 80.0 61.7 60.0 45.5 40.0 31.8 18.4 20.0 10.7 0.0 Construction area Has drinking water Has toilet (%) Has heating (%) per capita (sq.m) (%) Sources and notes: China Urban Labor Survey 2005 (5 large cities-Shenyang, Wuhan, Xian, Shanghai and Fuzhou). With housing prices rising rapidly in many large cities in China, finding affordable housing may become a greater challenge for migrants in the future. Migrant "villages" in outlying urban areas already have a long history in many Chinese cities, but municipal governments often do not feel obligated to provide public services to such communities. Local governments will need to consider seriously whether policy actions are required to provide adequate living environments for migrant communities in urban areas. Access to social To what extent do migrants have access to key social insurance and social assistance insurance and social programs in urban areas? Figure 7.97 presents tabulations of social insurance coverage assistance programs rates from the CULS surveys in 2004 and 2005 in 12 cities. The differences are dramatic. The percentage of migrant workers who report participating in the main social insurance programs are 8.3% for pensions, 6.8% for medical insurance, and 4.4% for unemployment insurance. The comparable coverage rates for local residents are 61.7, 52.3, and 18.8 percent. Figure 7.97 : Social insurance coverage of local residents and migrants (% of individuals with coverage) 80 Urban residents Migrants 61.7 60 52.3 40 20 18.8 8.3 6.8 4.4 0 Pension Medical Unemployment Sources and notes: China Urban Labor Survey 2005 of 12 cities. Local residents in small cities were surveyed in 2004. In light of the fact that the main social insurance programs are now linked to employment rather than hukou status, these large differences might at first appear surprising. However, the lack of social insurance coverage for migrants is better understood when one considers that most migrants are self-employed or are employed in the private sector or on an informal basis (see discussion of informalization in section 6c). In fact, the government should be concerned at the relatively low coverage rates for local residents, let alone migrants. A major difficulty in expanding worker coverage, whether local F o s t e r i n g e q u i t y 1 8 4 residents or migrants, is caused by the incentives of employers to evade mandatory payroll taxes to fund social insurance programs by hiring workers informally, and the lack of enforcement of such payroll contributions. Access to social assistance programs is even more discouraging for migrant families. By design, migrant families are ineligible for benefits from the urban Di Bao program, the main poverty assistance program in urban areas. Local governments are perhaps rightly concerned that expanding social assistance benefits to migrants could induce a large inflow of migrants to urban areas who would become dependent on local government expenditures to maintain their livelihoods. Just as for local residents, lack of access to social insurance and social assistance programs significantly increases the vulnerability of migrant families to unexpected shocks to employment or health, and to lack of sufficient support in old age. In conclusion, the evidence presented above on the situation of migrants in urban China suggests the following implications for policy: · Migrants are not much more likely to be income poor than local residents. To date, migration has primarily been a selective process in which those who lack adequate income support return to their home villages. Currently, migrants do not appear to constitute a new poor population in urban cities, at least as measured by income per capita though the important caveat that existing surveys may not adequately capture the situation of an itinerant migrant population also needs to be kept in mind · The nature of migration is changing. Migrants increasingly are resettling permanently in urban areas, often with their families, and more and more migrants are sending their children to school in urban areas despite the difficulties of educational access. These changes mean that in the future migration may become more permanent rather than selective, creating greater and more diverse needs for government policy. · Migrants fare considerably worse than local residents in nearly all non-income measures of welfare, including the quality and costs of housing, access to education, and access to social insurance and social assistance programs. Improving equal treatment of migrants in these areas merits high priority. As quickly as possible, access to education services and residential support services should be determined by place of residence rather than place of hukou. · Extending access by migrants to social insurance programs should focus on greater participation in and enforcement of employment-based coverage, and additional programs to increase coverage of the self-employed. This can be accomplished as part of more general reforms to improve the coverage and effectiveness of China's social insurance schemes. · Increasing the coverage of migrants by social assistance programs or other government programs should be accomplished gradually. Initial efforts should focus on expanding coverage to long-term or permanent migrants. By capping urban benefits at current levels and gradually raising rural benefit levels over time, the two systems can eventually be integrated and harmonized. F o s t e r i n g e q u i t y 1 8 5 PART III WHAT THE REVIEW SUGGESTS ABOUT THE POVERTY REDUCTION AGENDA 1 8 6 8.What the priorities are and what is needed to implement them Recently, the 17th Party Congress endorsed the application of the Scientific Outlook on Development. It was recognized that while the socialist market economy is in place, there remain structural and institutional obstacles slowing down development. A relatively comfortable standard of living has been achieved for the people as a whole, but the trend of growing inequality in income distribution has not been reversed, and there remain a considerable number of impoverished and low-income people in both urban and rural areas. The country also faces the arduous task of narrowing the urban-rural and interregional gaps in development and promoting balanced economic and social development. In the light of this, the scientific outlook on development emphasizes the four principles of (i) taking development as the top priority, (ii) putting people first as the core of development, (iii) pursuing comprehensive, balanced and sustainable development, and (iv) persisting in the overall consideration of achieving balance between urban and rural development, development amongst regions, economic and social development, relations between man and nature, and domestic development and opening to the outside world. This report has focused on, and attempted to systematically analyze, many of China's current development challenges from a perspective that is consistent with the issues and principles highlighted by the scientific outlook to development. The range of recommendations for policy, programmatic and institutional reform that are suggested below, it is hoped, can be helpful for a thorough application of the scientific outlook to development as envisioned by the 17th Party Congress. This final chapter thus focuses on what needs to be done. It attempts to distill the key priorities emerging from the foregoing analysis for China's current and evolving poverty reduction agenda and what kind of programs and policies will be needed to implement them. Given the nature of this report, the discussion is inevitably pitched at the level of broader directions for policy, programmatic and institutional reform rather than at specific operational details of individual interventions and initiatives, although some of these will also be touched upon in course of the discussion. Also, while this concluding chapter tries to bring together the main policy messages, it is not a substitute for the detailed discussion of all the policy issues presented earlier in the report. a. Adopting a broader conception of poverty and an adequate threshold for identifying and targeting the poor It may be time for China To begin with, it may be time for China to consider establishing a higher threshold to to consider adopting a measure poverty that could be consistently applied to both rural and urban areas. The higher threshold for current official poverty line of 785 Yuan per person per year at 2007 rural prices (or about identifying and targeting 71 cents a day in terms of the 1993 PPP dollars or about 57 cents a day in terms of the the poor new 2005 PPP dollars or about 31 cents a day using the current exchange rate) is arguably low both by international standards (in fact, it is one of the lowest in the developing world) and relative to average incomes and growing aspirations within China. Recent calculations further suggest that even by the objective standard that a poverty line aims to capture--the minimum expense necessary for meeting basic food and nonfood needs--the poverty line may be too low. The principle of a higher poverty threshold was endorsed by President Hu Jintao in his speech to the 17th Party Congress in October 2007. And recently, there has been some active discussion of an upward revision of the poverty line in the academic and policymaking circles. 76 Adopting a higher poverty threshold is however only an initial step in the formulation of a forward-looking poverty reduction strategy. But it is an important first step as it can be critical for focusing policy attention on the right target group ­ one that is more appropriate for China's current stage of development. The new threshold should form the basis of not only the measurement and monitoring of poverty in the country, but more importantly, the development of a _________________________________________________________ 76 Reported by Xinhua News Agency (April 12, 2008). Higher thresholds are also already being reflected in some elements of government policy, as for instance, in the eligibility for the Di Bao social assistance program and in the higher exemption limit for personal income taxes. W h a t t h e p r i o r i t i e s a r e 1 8 7 consistent operational approach to targeting a range of poverty alleviation and social protection programs to the poor in rural and urban areas. It may be also worth considering measuring poverty in terms of consumption rather than income. While this report has presented both income and consumption-based analysis, there are well-known reasons for preferring consumption over income, the main one being that consumption is a more stable measure of welfare. The emphasis on consumption is also in keeping with the current national goal of raising the share of consumption in GDP. Moreover, in the case of China, while many of the poor consume below their income levels, it is arguable that this is mostly in the nature of saving for a rainy day with a view to maintaining their consumption levels at a relatively stable level even as their incomes may fluctuate. This too suggests that consumption may be a more stable and reliable measure of welfare. But the conception of poverty could also be broadened in two other respects. First, beyond income and consumption, access to affordable education and healthcare also ought to be seen as part of the poverty reduction agenda. Higher income and consumption can be a means for improving this access; however, income and consumption can themselves be limited by the lack of such access. The evidence suggests that inability to cope with health shocks is often an important factor in households falling into poverty, and limited human capital is often the reason for many households being unable to make use of the opportunities provided by economic growth to move out of poverty. Second, since the contribution of (uninsured) risks to observed poverty is quite significant (estimated at about 40% for consumption poverty) and there are indications that this contribution tends to rise with further progress in poverty reduction, offering social protection to groups vulnerable to poverty also ought to be seen as an integral part of the poverty reduction agenda. Thus, the poverty reduction agenda is broader than may be conventionally recognized. There are indications that in practice the government is already viewing the poverty challenge in such wider terms, but as discussed further below much still remains to be done in implementing this broader agenda. b. Retaining rural poverty reduction as the top priority Reduction of rural Even with a broader conception of poverty, the evidence and analysis presented in this poverty should continue report suggests that the reduction of rural poverty should remain the top priority. Whether to be recognized as a top measured in terms of income or consumption, the poor in China are predominantly in or priority from rural areas. Even if migrant workers are not included in the rural population, as is commonly done in the national household surveys, rural poor account for about 90% of all those living below the poverty line. In contrast, depending upon the precise urban-rural cost of living differentials used to measure real incomes and consumption, urban residents (excluding migrants) account for 1-3% of the total number of poor in the country. The rural-urban income While income inequality within rural areas is somewhat higher than that within urban gap has grown in recent areas, the difference in rural and urban poverty rates largely stems from the widely-noted years... rural-urban gap in mean incomes, which has been steadily growing in recent years in both absolute and relative terms. Despite large-scale migration from rural areas, this gap is now at a historical high. The rural-urban income The rural-urban income gap is however only one manifestation of the great divide that gap is only one separates China's urban and rural areas. On almost every dimension of welfare, the rural manifestation of the great population lags behind the urban population. Literacy rates, enrollment rates (especially divide that separates past primary school), and educational attainment rates (especially past junior high school) China's urban and rural areas are all much lower in rural areas, and while there have been improvements in all these indicators in rural areas in absolute terms, the gap between rural and urban areas has continued to widen. Thus, rural poverty reduction must continue to be a top priority. What perhaps has made these rural-urban differences so salient and a focus of considerable policy attention is the fact that in contrast to the significantly poorer W h a t t h e p r i o r i t i e s a r e 1 8 8 outcomes for the rural population, the country's highly decentralized fiscal system continues to harbor an urban bias. On a per capita basis, let alone a per poor person basis, urban areas have much greater fiscal resources and these are reflected in correspondingly higher levels of public spending. For instance, in terms of publicly financed per-capita expenditures on education and health, rural areas lag far behind urban areas. Spending on social assistance and social insurance, too, is disproportionately higher for the urban population. Several recent Recently, the government has taken several steps towards addressing the rural-urban government initiatives gap. The major initiatives include: (i) the elimination of agricultural taxes and fees, and an are beginning to address increase in direct agricultural subsidies, (ii) the waiving of tuition fees for compulsory the rural-urban imbalance education in rural areas, (iii) a major expansion of rural health insurance to about 86% of counties nationwide, and (iv) a nationwide roll-out of the rural Medical Assistance scheme and the rural Di Bao program of minimum income support. These are all important initiatives which reflect increasing commitment of the government to addressing the rural- urban imbalance, and they are already beginning to be reflected in noticeable reductions in rural and hence national poverty levels. While they are unlikely to make a big dent in the rural-urban gap as urban areas continue to grow faster than the countryside, It will be important to maintain the momentum on these initiatives in future while also addressing several outstanding challenges (as discussed below) that will help deepen their impact on rural poverty reduction. But it would be imprudent Also, while an emphasis on tackling rural poverty is entirely appropriate, it would however to ignore the urban be imprudent to neglect the situation and problems of the relatively disadvantaged groups disadvantaged groups in urban areas. For several reasons. First, rural and urban poverty issues are now increasingly linked due to the large and growing flow of labor from rural to urban areas. Second, there could a degree of underestimation of urban poverty due to the inadequate coverage of migrant households in national surveys. Third, the urban labor market itself has undergone a profound transformation with the economic restructuring of the state and collective sector since the mid-1990s that has exposed the urban working population to new vulnerabilities which if left unaddressed could lead to the emergence of an urban underclass or future urban poverty. With increasing urbanization, the development of a nationally integrated approach to both rural and urban poverty reduction and social protection is an increasingly important policy imperative. c. Promoting opportunity by raising the returns to labor In both rural and urban In both rural and urban areas, an inability to work is only a relatively small part of the areas, inability to work is problem. While it is true that households with someone who is unable to work ­ due to only a small part of the disability or old age ­ are more likely to be poor, nearly three-quarters of China's rural poverty problem poor and all but 2% of the urban disadvantaged are in households where there is no lack of work capacity. Raising the returns to What these households lack are adequate returns to their main asset, which is their labor, labor--the main asset of and what they need is greater and more rewarding employment. Raising the returns to the the most of the poor--is labor of the rural poor and the urban disadvantaged is, therefore, critical to promoting their therefore critical opportunities for increased incomes and improved livelihoods. This can and is being done in several ways. Realizing the potential of migration for poverty reduction The scale of migration is A promising avenue for rural poverty reduction has been through migration to urban areas. large and has been The scale of migration from rural areas has increased rapidly during the 1990s. While growing rapidly estimates differ depending upon data sources and definitions, the size of migrant population could be as high as 150 million making this the largest peacetime movement of people in world history. An idea of the scale and significance of this phenomenon could be had from the following basic facts (for 2003): · Roughly one in every five rural workers is a migrant worker. · About 43% of the rural population lives in households that have one or more migrant workers. W h a t t h e p r i o r i t i e s a r e 1 8 9 · The share of migrant income in total rural income is nearly 30%. ...and it has been poverty While it is difficult to precisely estimate the contribution of migration to poverty reduction, reducing there seems little doubt that internal migration and remittances have played an important role in poverty reduction in rural areas. For instance, evidence from the NBS Rural Household Surveys indicates that households with migrant workers are (30%) less likely to be poor than households without any migrant workers. It is estimated that for 2003 alone, an additional 11.8 million persons would have been income poor in the absence of migration, or an increment in the total number of rural poor by 13%. And this is likely to be an underestimate as it does not take into account the indirect benefits of migration through productive investments supported by remittances or through the positive effects of migration on rural wages. But many of the However, migration's effects on poverty are limited by the fact that many of the poor are remaining poor are less unable to migrate. The relationship between migration probability (defined as the able to migrate... probability of a household having a migrant worker) and per capita income has an inverted-U shape. Both the poorest and richest are less likely to migrate although for different reasons. The rich have fewer incentives to migrate, while many of the poor are not able to make use of migration opportunities due to their limiting circumstances and endowments. Some of the key factors limiting migration include: the level of educational attainment of rural labor, limited access to information and labor recruitment networks especially where the informal (family and friends-based) networks are not well-established as in the case of certain minority communities, the expenses of movement and other costs associated with finding a job in cities, higher costs of health and education for migrants' children in urban areas which have been especially limiting of family as opposed to individual migration, and a degree of occupational and wage discrimination against migrants relative to local urban residents in relatively segmented urban labor markets. ... and migration is also While migration has been an important means of raising incomes of rural labor, it has also creating some problems had some adverse consequences in source areas, including negative effects on in source areas educational attainments amongst the rural youth, greater vulnerability of households to work injury or disability of migrant workers, the problems of the left-behind children and the elderly, and the erosion of village cohesion and community participation. While some effects such as those on community participation may be difficult to mitigate in villages that have experienced a large-scale labor exodus, most of the others are indicative of policy failures related to factors such as relatively limited human capital of migrant workers, limited social insurance for migrants and their families, discriminatory rules and practices restricting migrant workers' access to urban services and ultimately limiting the scope for family migration. In view of both the impediments to mobility that are particularly relevant to the poor, and some of the adverse effects in source areas, the potential of migration for poverty reduction remains under-realized. To better realize this potential, policy initiatives in several areas will be helpful. · Improving the education and skill base of potential migrants especially those from poor households who are more deficient in these respects. Ensuring junior high school level of education for all in rural areas is a necessary first step. And in this regard, the recent introduction of free compulsory education is a very useful initiative. But beyond that there is also a need to increase public support for senior/post- secondary and technical education for the rural population if migrants are to be able to access better-paid higher skill jobs in urban labor markets. · Providing assistance for reducing information and search costs associated with migration by supplementing informal social networks with more organized placement and intermediation services. A special effort is likely to be needed to reach some of the minority groups, and those in some of the poor areas where existing informal networks are underdeveloped. · Improving investment climate in cities in the main sending regions in Central and Western provinces. The pattern of migration in China in many ways mirrors the W h a t t h e p r i o r i t i e s a r e 1 9 0 geographic concentration of economic activity. The sending areas are concentrated in the Center and the West and the receiving areas along the coastal belt in the East. Most migration is inter-provincial and over long distances. However, like elsewhere in the world, distance increases the costs of migration, and reduces its magnitude. At the same time, the industrial clusters in the coastal belt are becoming increasingly skill and capital-intensive. Hence, developing the relatively more labor-intensive segments of industry in the central and western provinces can play a useful role in promoting further absorption of rural labor. This in turn will require several steps to improve the investment climate in these areas. · Improving labor market conditions and the provision of public services and social security for migrants in the cities. Despite its large scale, the current system of migration could with some justification be described as a `guest worker system'. A range of discriminatory practices in destination areas abridge migrants' entitlements as workers or citizens. There is an increasing need to eliminate these practices which are limiting both the scale of migration, especially family migration, and the potential welfare benefits from migration. In particular, this will require better enforcement of the provisions of labor laws for migrant workers, access to basic services (education, health and housing) to be determined on the basis of place of residence rather than the place of hukou, and the extension of the coverage of social protection programs to migrant workers. Insofar as this promotes family migration, this will also help address the problems of the left-behind children. Government initiatives in some of these areas are already underway as, for instance, laid out in the Opinions of the State Council on Issues Concerning Rural Migrant Workers (SC Document No. 5, March 2006), and the MOLSS Circular on the Implementation of SC Document No. 5 (April, 2006) as well as the action plans of the inter-ministerial Joint Conference on Rural Migrants Issues established in 2006. The long list of policy guidelines for migrant workers cover many areas including worker rights and protection, public services and living conditions, social security, employment services and skills development. However, not much systematic information is available on the actual implementation of these wide-ranging guidelines. It will be important to monitor the progress on these new initiatives not only to assess if they are being implemented as intended, but also to refine the different initiatives based on the implementation experience of what seems to be working, where and in what ways. Promoting agricultural and rural development in poor areas Promoting agricultural However, migration will not be the answer for all the rural poor. There is clearly a role for and rural development promoting agricultural and rural development within a poverty alleviation strategy, as also has a role within a suggested by the analysis of the taxonomy of rural poor presented in this report. And this poverty reduction has also been a key element of the government's approach to poverty alleviation which strategy from the outset has been development-oriented and targeted to poor areas. China's flagship poverty alleviation program, the Integrated Village Development Program (IVDP), is an area-based program that currently targets about one fifth of all Chinese villages (that are designated as poor villages) using a participatory approach to village investments. Other important developmental poverty alleviation programs include subsidized credit and development of agribusiness, the food-for-work program and a labor transfer program which in fact seeks to promote migration. But area-based programs The review of these programs in this report highlights several areas for potential are unable to target the improvement. In relation to the village investment (IVDP) program, the review highlights poor effectively and the two key issues: (i) the relatively poor targeting of the program (a large fraction of the poor impact on incomes of the simply do not live in designated poor villages), and (ii) the very limited impact of village poor who are targeted is small investments on incomes of the poor within designated villages (reflecting the relatively small scale of investments, the inability of many of the poor to make complementary investments, and a number of factors linked to limited participation of villagers, especially the poor, in the planning and implementation of projects). Besides the IVDP, the subsidized credit program, although accounting for about half of all poverty alleviation funds, is not set up in a way to deliver significant benefits to the poor as its main focus has been to support large agribusiness enterprises. And participation of the poor in the training programs for labor transfer has also been limited by the relatively high costs of W h a t t h e p r i o r i t i e s a r e 1 9 1 participation. To make them more effective instruments for rural poverty alleviation, further reform of these programs will be needed. For the IVDP, the following areas merit consideration: · Develop consistent, data-based criteria for selecting poor villages and a system for updating poor village designations; privilege poorest villages for earlier and larger investments. · Increase the size of village investments. This may require higher funding for the budgetary grant program, but the effective size of village investments could also be increased through improved coordination of fund disbursement for village plans by designing clear rules to guide the aligning of Food-for-Work investments and subsidized credit with village plans. There is also scope for releasing more funds per village by targeting fewer villages. · Develop institutional mechanisms to improve coordination of village investments with other line agencies. Since the larger part of investments in rural areas comes not through the poverty alleviation funds but through the budgets of other line agencies, the needs for better coordination of village-level investments goes well beyond the poverty alleviation funds themselves. The current institutional mechanism for inter- agency coordination through the local Poverty Alleviation and Development offices works only imperfectly, and needs considerable strengthening to be effective. · Place new emphasis on household-level targeting, as even with better identification of poor villages, it will be important to reach the poor within these villages. This can be done in several ways: by changing the mix of projects to include more household- oriented interventions, for instance, labor training to promote off-farm employment; by waiving the requirement for complementary investments by poor households or providing subsidized credit for this purpose; by developing new programs such as self-targeting public works programs or conditional cash transfers which could offer assistance to poor households conditional on other objectives, for instance, ensuring that their children do not drop out of school, or they themselves participate in labor training. · Conduct additional study of the village planning process to identify ways to increase participation of the poor, and experiment with new methods of implementing village plans. For the other developmental programs, too, there are several areas for potential reform that may be considered: · The subsidized credit program needs to be integrated with the village plans. New institutional arrangements will be necessary for this to happen; simply exhorting different agencies to coordinate better will not help. It also needs to be recognized that village-level coordination of funds and program management is unlikely to succeed without coordination at the county and provincial levels. There should thus be a single agency that is adequately empowered to ensure this coordination at the provincial and county levels. · There is a need to redirect subsidized credit from large agribusiness enterprises (which ought to be supported through the regular commercial banking networks) to poor farmers, in part through provision of cheap loans for complementary investments by them to better utilize the public goods created by village investments. · In relation to the labor transfer programs, boarding and other expenses should be waived for poor farmers to ensure their enhanced participation. Further gains from better Two further policy considerations about developmental poverty alleviation programs geographical targeting should also be noted. are limited ... hence the case for household- · Even with perfect village-level targeting, many of the poor will be left out (due to the oriented approaches and increasingly dispersed nature of poverty) and many of the benefits of investments in new targeting targeted villages will flow to the non-poor in these villages. There are thus limits to mechanisms such as further gains from area-based targeting and it will be important to supplement area proxy means targeting targeting with household targeting. But new mechanisms will need to be developed to target poor households, and in this regard approaches such as proxy-means targeting deserve to be actively explored. W h a t t h e p r i o r i t i e s a r e 1 9 2 More household-oriented · If, as suggested above, more household-oriented interventions are to be adopted as interventions will raise the stakes of inter-agency part of these developmental poverty alleviation programs, then the issue of coordination of coordination with social assistance and social insurance programs, which also developmental and social provide critical support at the household-level, will become increasingly important. protection programs For instance, there will be a need for a common approach to identifying poor households by both the developmental and social protection programs, which raises the stakes of inter-agency coordination not just to the Poverty Alleviation Offices, Departments of Finance and Development Reform Commission and the ABC (or Rural Credit Cooperatives), but also the Departments of Civil Affairs and Labor and Social Security. The development of a coordinating mechanism for this related set of poverty reduction interventions remains a big institutional challenge. d. Enhancing security by expanding and improving the coverage of the social protection system in rural and urban areas The evidence presented earlier in this report indicates that rural and urban households in China remain vulnerable to poverty because they are exposed to a variety of risks arising from natural disasters, ill-health, fluctuations in agricultural yields and prices, and job loss. In rural areas, for instance, the size of population vulnerable to poverty may be twice as large as the number of poor in any given year. Social protection systems ­ through social assistance and social insurance programs ­ help individuals cope with such risk and vulnerability, and hence the development of such systems is an integral part of the overall poverty alleviation strategy. China has come a long China has come a long way in reestablishing its rural and urban social protection systems way in reestablishing its after the collapse of the traditional commune and enterprise-based systems during the rural and urban social process of economic reforms and restructuring of the 1980s and 1990s. Major revamping protection systems, but of the urban system started in the mid-1990s while the rebuilding of the rural system has many policy challenges remain been the focus of more recent policy attention especially over the last five years. While the substantial development of the new social protection system over a relatively short period of time is impressive, this has happened in the context of massive structural change in the rural and urban economic systems, and there remain many policy challenges. Drawing upon the analysis presented in this report, the key challenges and potential areas for reform for the rural and urban systems are noted below. Improving rural social protection Over the last 3-4 years in particular, there has been significant progress in the expansion of several elements of the rural social protection system, which is beginning to address its longstanding underdeveloped state. The recent initiatives are consistent with the recognition that access to land alone, despite the relatively egalitarian distribution of use rights, cannot be relied upon to provide an effective rural safety net. The main thrust of these initiatives has been on expanded coverage, and as a result, the coverage of key programs, in particular the rural Di Bao, rural heath insurance and the Medical Assistance scheme, has rapidly improved. At the same time, however, the coverage of other programs such as rural pensions and disaster relief remains limited, and the benefit levels of most rural programs are quite low. This is in contrast to the urban sector where following the major restructuring of the state-owned enterprises during the 1990s, a much greater level of government effort and resources have been devoted to the development of a relatively modern non-enterprise-based social welfare system. The recent progress is starting to bridge the gap between the rural and urban systems, but much still remains to be done. Overall, there remain four major challenges for the rural social protection system. · The coverage (in terms of the number of beneficiaries as different to the number of implementing counties) of several programs needs to be expanded to more fully cover the rural poor and vulnerable population. · The benefit levels of most programs, which are only a fraction of the benefits under urban programs, will also need to rise if they are to offer effective protection to those W h a t t h e p r i o r i t i e s a r e 1 9 3 covered by the programs. · There is a need for coordination and harmonization of different programs within the rural social protection system as well as between the rural and urban systems. A harmonized national system is important not only for bridging the rural-urban gap in the social protection, but also for ensuring adequate protection to migrant workers who are currently only imperfectly covered by the rural and urban systems. · All this will require further infusion of funds from the central and provincial governments. Such commitment of resources from higher levels of government is also needed to break the dependence of social protection spending on local fiscal capacity, and hence address the wide disparities in both the coverage and benefit levels of programs within rural areas due to the large differences in the revenue base of local governments. There are specific challenges for individual programs. · Rural Di Bao: The rural Di Bao has clearly emerged as the main focal assistance program for rural areas. Its rapid expansion since 2005 has been impressive with gross coverage rate now equivalent to about two-thirds of the number of rural poor. However, as it expands further, the program will need to be streamlined with a more consistent approach to coverage and targeting. Here, many of the challenges will be of a similar nature to those already being encountered with the more established urban Di Bao program (see further discussion below), namely, realizing the scope for better targeting and hence improved coverage of the poor, rationalization of Di Bao lines and eligibility criteria across different local governments so that program coverage is better aligned to needs rather than local fiscal capacity, and development of a benefit design framework that limits adverse work incentives. · Rural health insurance: The recent rapid expansion of rural health insurance is indicative of its recognition as an important priority for the government. With an estimated participation rate of 86% at the end of 2007, the coverage issue has been largely addressed. The next challenge will be to ensure the provision of an adequate level of benefits that can secure a sizeable reduction in the burden of private health costs. The targeted increase in the budget through a doubling of contributions in 2008 will help provide more resources for raising benefit levels, but further increments of funding from central and provincial governments will be needed for net reimbursements to cover a larger fraction of the out-of-pocket health costs. Coverage of outpatient services also remains a further challenge. Another major challenge is the coverage of migrant workers who seem to be left out by both the urban and rural health insurance systems, and this raises the broader challenge of the coordination and harmonization of two systems, which is all the more pertinent now as the NCMS is beginning to operate at a national scale. · Medical Assistance: The Medical Assistance program is a useful supplement to the rural health insurance scheme in extending the latter's coverage to the rural poor. However, despite its recent scaling up to the national level, its coverage in terms of the number of beneficiaries is still relatively limited; the gross coverage rates are equivalent to about half the number of rural poor. Moreover, its relationship with the Di Bao program will also need to be clarified. As both programs aim to target the poor, there is a need for adopting a unified approach to targeting assistance. The government may also want to consider using the MA scheme to introduce a means- tested safety net provision to health insurance whereby after reaching a certain threshold of per capita healthcare expenses in any year, the insured are entitled to higher (even 100%) reimbursements. · Disaster relief: Additional government resources will be needed to improve the coverage and benefit levels of disaster relief which remain limited relative to the scale of the problem. However, most of the government response is ex post in nature and there is a conspicuous absence of disaster insurance. Recently, China Life (China's largest life insurer) launched the county's first accident insurance covering six types of natural disasters. However, the private uptake of such insurance may be limited, and there is a strong case for the government to consider public provision of disaster insurance with subsidized premia for the low-income groups. · Rural pension system: The reach of rural pension program remains limited and uneven. Currently, the central government is studying the implementation of a new W h a t t h e p r i o r i t i e s a r e 1 9 4 type of rural pension system with the aim of covering all rural residents by 2020. The system will combine basic defined benefit pension with mandatory defined contribution plan via individual accounts. The pension is proposed to equal 30-50% of net per capita of local rural residents Such a highly subsidized pension system however makes it more like a transfer or assistance program for senior citizens in rural areas. There is thus a need for a careful evaluation of different options for providing support to the elderly while recognizing that social insurance programs need to be considered together with social assistance programs. Improving urban social protection While the urban social protection system is substantially more developed than the rural system, there remain many areas of concern in relation to the ability of the core programs to serve as an effective safety net for the urban population. The review presented in this report identifies several challenges for both social insurance and social assistance programs in urban areas. Key policy challenges and areas for reform for urban social insurance programs include the following. · Limited coverage, especially of the poor and lower income groups and those in informal employment. Despite the expansion over the last decade, the coverage of urban social insurance programs still remains limited. Currently the key programs cover between 40-50% of urban workers which indicates that there is a lot of ground to be covered still. The system is also heavily focused on the formal sector, and coverage rates for informal workers are much lower still, often as low as one-fourth of those in the formal sector. Similarly, the coverage rates for the poor and low-income workers are a fraction of those at the upper end of the distribution. Further expansion of the system needs to pay particular attention to better coverage of these groups. · Limited coverage of migrant workers. Large-scale rural to urban migration continues to pose challenges for the urban social protection system. The coverage rates for pension, health and unemployment insurance for migrants are estimated to be about one-seventh of those for local urban residents. Since 2006, there has been a significant momentum on the extension of work injury and health insurance which, according to administrative data, covered about a third and a quarter of migrant workers respectively in 2007. However, much still remains to be done in these two areas. The inclusion of migrants in pension and unemployment insurance is more limited and poses a bigger challenge as the design of these programs is not well- suited to serving a mobile labor force that changes employers and jobs more frequently. · Low level of pooling of social insurance funds and very limited portability of benefits. The contributions into most social insurance funds are typically pooled only at the local government (municipality or city) level. Low level pooling is however inefficient from a risk management perspective and limits the insurance benefits of the programs. It also has the effect of creating inequities in the system by tying local social insurance expenditures to local revenues. Low level of pooling also goes hand in hand with very limited portability of benefits. This is not only a significant barrier to improved coverage of the programs, but is also a major impediment to labor mobility and the development of a flexible labor market. Progress towards higher level pooling and greater portability has remained limited. Higher levels of pooling can be unattractive to richer local governments who may view it as an implicit tax. The reform of the system is thus likely to be difficult, but remains an important area for future policy development. · Need to balance provision of greater protection with maintaining incentives and competitiveness. The urban social insurance system is costly with its financing by employers and employees accounting for about 43% of the wage bill. The employers' contribution for the five key programs alone comes to about 32% of the payroll. Some of this may be passed on to the workers in lower wages, but private incentives for non-compliance remain high. Incentives for non-compliance also exist for the poorer workers. These factors are already contributing to the informaiization of the urban labor market. The existing system also has important implications for W h a t t h e p r i o r i t i e s a r e 1 9 5 maintaining competitiveness of Chinese business and industry. Greater effort at enforcing compliance, though desirable, is unlikely to be enough to tackle the dual problems of informalization and ensuring competitiveness. A part of the solution will almost certainly involve greater public funding of the social insurance system. To protect the interests of the poorer workers, the government may also consider reducing or waiving the employee contribution and even reducing the employer contribution for workers below a certain income threshold. Urban Di Bao. The main social assistance program for the urban disadvantaged is the urban Di Bao program. By international standards, the program is exceedingly well- targeted with 64% of the program beneficiaries in the bottom decile of the urban population who account for about 70% of program benefits. While such performance is exemplary, the review of the program in this report suggests that it still faces a number of policy challenges, the most important of which include: (a) limited coverage of the urban poor (bottom 5-10% of the urban population) despite overall excellent targeting in terms of concentration of benefits amongst the bottom end of the distribution, (b) non-coverage of the migrant population, (c) eligibility thresholds (Di Bao lines) and benefit levels varying with local fiscal capacity rather than need, and (d) potential work disincentive effects of the program design. In view of these challenges, several areas of reform are suggested; many of these also have relevance for the rural Di Bao program. · One area of potential reform is in relation to the rationalization of Di Bao lines. There is a case for eliminating inequities in the system by establishing a set of eligibility lines that refer to comparable standards of living across locations. If it is not possible to develop city-specific lines with the currently available data, consistently-estimated provincial eligibility lines could be a useful starting point. This approach will also delink Di Bao lines from considerations of local fiscal capacity. However, such a reform can only be effective if it is supported by transfers from higher levels of government to poorer municipalities. Central funding to provincial and local governments will thus need to be based on the estimated number of beneficiaries using the consistent set of eligibility lines. · A related issue is the rationalization of Di Bao benefit levels which too vary significantly across cities and municipalities in a manner that is related not to differences in the depth of poverty but to what the local governments can afford or are willing to fund. Recent regulations (in February 2008) to raise the benefit levels, for instance, also permit better-off areas to implement higher increases. A more nationally consistent and equitable approach is clearly needed. · Improving the coverage of the poor will need a further expansion of the program in areas where coverage has been limited by relatively low thresholds. But it will also require improved methods of targeting such as proxy-means targeting. Household incomes are difficult to observe and measure accurately, and some form of implicit proxy-means targeting is already being practiced by the neighborhood committees as evident in the use of various non-income criteria. While the implicit proxy-means mechanism appears to have worked reasonably well thus far, many of the urban disadvantaged remain excluded. Hence, there is a case for developing a more explicit proxy-means targeting system based on relatively easily-observed household characteristics that are correlated with income. The list of household characteristics could include indicators related to household demographics, labor force participation, any work disability, educational attainment of adult members, ownership of specific assets, and characteristics of dwellings, amongst others. There is now a good deal of international experience with proxy-means targeting that could be drawn upon in developing a system tailored to China. · Coverage of the migrant population needs special attention. Migrants include both individual migrant workers and migrant families. As a first step, extending Di Bao coverage to migrant families who have been resident in an urban area for some minimum period of time and meet the program's means test ought to be considered. Until such time that the urban and rural Di Bao programs can be coordinated, individual migrant workers may have to be covered under rural Di Bao through their families in rural areas. W h a t t h e p r i o r i t i e s a r e 1 9 6 · There is also a case for reconsidering the top-up design formula for the determination of Di Bao benefits. Such a design implies a 100% marginal tax rate on incomes of those eligible for the program. The reason why the program has managed to avoid serious adverse effects on work incentives thus far is that the top-up design appears not to have been followed in practice. Thus, the actual implementation practices are already pointing to the need for an alternative benefit structure. One alternative here is uniform per capita benefits for those eligible. While this is attractive for its simplicity and transparency, uniform benefits may not be considered sufficiently progressive. Another alternative could be to base benefits on the proxy-means scores of households, with lower benefits for those with higher scores but calibrated so as to maintain relatively low marginal tax rates. Harmonizing the rural and urban systems As already mentioned, the overall social protection system is still more focused on the urban formal sector, despite the recent expansion of the rural system. At the same time, with the growth in the rural system, the issue of the integration of the rural and urban social protection systems has become increasingly important, not only to close the gap between the two but also to promote a more flexible labor market environment and to facilitate the coverage of migrant population. The harmonization of the two systems may need to proceed in a phased manner. By capping urban benefits at current levels and gradually raising rural benefit levels over time, conditions can be created for the eventual integration and harmonization of the two systems. e. Fostering equity and reducing poverty by ensuring secondary school education and basic healthcare for all A large part of the current The analysis presented in this report has repeatedly found education to be of critical poverty and inequality in importance in determining better welfare outcomes for China's population. China can be traced to the underlying disparities · The educational attainment of a household's adult workers is the clearest observable in educational attainment predictor of poverty in China, in both rural and urban areas. While the poor differ from each other in other ways, low levels of educational attainment seem to be common to virtually all of them. A high school education for the adult working members virtually guarantees a household's exit out of poverty. Even with middle school education of adult workers, consistent with the goal of ensuring universal 9-year education, the probability of being poor declines to negligible levels. At the same time, over 90% of the poor in rural areas and nearly 60% of the disadvantaged in urban areas live in households where this target has not yet been attained. Investing in middle and high school education for all is thus the key to making a substantial dent in remaining poverty. · In terms of explaining income inequality too, differences in educational attainment, and the returns to education, turn out to be critical. The components of inequality that contributed most directly to the rise in inequality--the rural-urban gap and increasing inequality within rural and urban areas--can all be traced to inequalities in human capital. · As already noted, at least a junior high school level of education is an important determinant of migration and off-farm employment for rural laborers, and higher level of education will be important for migrant workers to tap into higher-skill and better- paid jobs in the cities. · In urban areas, the rising inequality in labor incomes can in turn be traced to, on the one hand, the increased incidence of unemployment and declining labor force participation associated with the massive restructuring of state-owned enterprises, and on the other, to increasing dispersion in wages among those who are employed. Educational attainment has played a central role in both processes by determining who were most likely to retain their jobs in the face of widespread layoffs, who were most likely to be re-employed after an unemployment spell, and who were likely to be paid more when employed. · A large part of the gender income gap can be traced to differences in educational attainment of men and women. W h a t t h e p r i o r i t i e s a r e 1 9 7 While private returns to education have increased with economic reforms and rapid growth of the Chinese economy, the persistence of underlying disparities in educational attainment suggests that the improved private incentives for greater investment in human capital are not enough especially for the poor. Since social returns to education are even higher than private returns, there is a strong case for greater subsidization of education. Unequal access to basic Another major area of policy challenge that has been opened up by China's wide-ranging healthcare also remains economic reforms during the 1980s and 1990s is in relation to provision of basic another major policy healthcare. challenge · Despite significant income poverty reduction over the last 25 years, China's progress in improving health outcomes (for instance, child mortality and communicable disease mortality) has slowed down. · Private costs of healthcare have risen dramatically (even for those who are insured) and are impinging on the utilization of healthcare services. · Health shocks and the burden of medical expenses and the loss of income they entail, are one of the most common reasons why households fall into poverty. · With only the well-off being able to afford reasonable quality healthcare, inequalities in income are transmitted to those in health outcomes. There are thus large disparities in health outcomes between urban and rural areas and between the rich and the poor within both urban and rural areas. Ensuring that every household has access to education and healthcare thus remains an important part of the current poverty reduction agenda and is also critical fostering greater equity. The government is already responding to this challenge in a number of ways, but a lot more remains to be accomplished. Improving access to affordable education · Recently, good progress is being made in universalizing 9-year compulsory education. The waiving of school fees and the provision of boarding subsidies under the compulsory education finance reform, introduced in rural areas in 2006 and gradually extended to urban areas in 2007, is an important step in the right direction. The momentum on this initiative should be maintained so as to achieve the goal for 2008 of achieving free compulsory education nationwide. It will also be necessary to ensure that this policy can be implemented through China's decentralized fiscal system on a sustained basis. Poorer local governments (at the provincial and sub- provincial levels) will thus need to be adequately resourced, which in turn will be important for ensuring that free compulsory education does not come at the expense of a compromised quality of education. · There is also a need to review the formula for compulsory education transfers, which should be allocated more equitably based on needs rather than the pre-existing level of fees. · The impact of the policy of school consolidation and the closure of teaching points needs to reviewed. While the government has indicated the closures will be subject to prior community consultation, the implementation of this provision needs to be monitored and if the closures are likely to compromise school access in particular areas, the policy should be suitably amended. · A solution needs to be found for the problem of school debts perhaps by allowing a fraction of the transfer to go towards debt repayment. · China has achieved close to universal primary education. But significant inequalities in educational attainment beyond primary school ­ middle and high school, and eventually tertiary education ­ remain, and these are an increasingly important source of disadvantage for those who are not well-off. While middle school education is being covered by the compulsory education initiative, there is a need to turn attention to reducing the private costs of high school education especially for the poor. Means- tested subsidies for high school education could be an important future policy initiative, and these could be offered to poor households conditional on school attendance by their children, especially girls. W h a t t h e p r i o r i t i e s a r e 1 9 8 · There is also a case for developing a credit market for education where students (or their parents) can borrow for investments in education against future earnings. And the subsidies could of course be combined with loans for education. There are several examples of the application of this approach internationally and drawing upon that experience China could develop a model suited to its own conditions. · The education of children from migrant families in urban areas remains an area of particular concern. While steps have been taken over the last few years (especially since the State Council Document No. 5 of March 2006 and the inter-ministerial Joint Conference on Rural Migrants Issues) to reduce effective educational discrimination against migrants, progress remains uneven across urban areas and the implementation of non-discriminatory practices needs to be further monitored and consolidated. · There also remain large differences in the quality of educational inputs across regions, especially across rural and urban areas. Further investments into improving the quality of educational inputs in rural areas will help strengthen the demand and private incentives for education which in turn will be important for addressing the rural-urban education gap. Improving access to affordable healthcare · Significant progress has been made in recent years in extending the coverage of basic healthcare to the Chinese population. Important initiatives in (i) rapid expansion of the coverage of rural health insurance to nearly 86% of counties in the country, (ii) expansion of the Medical Assistance scheme in rural and urban areas, (iii) the introduction of the Urban Resident Basic Medical Insurance in 2007 have all helped the country move towards the eventual goal of universal healthcare provision. Many of the remaining challenges have already been referred to above, and most importantly, include the following: · While the coverage of rural health insurance has now expanded to include the bulk of the rural population, the level of benefits (net reimbursement) needs to be increased to achieve a sizeable reduction in out of pocket expenses. In 2008, the government contribution was raised to 100 yuan per capita. However, this is probably still not enough. For instance, in 2006, rural households on average spent 192 yuan per person on healthcare and medical expenses, and even the bottom quintile spent 118 yuan. Thus, the standard should be gradually raised further. At the same time, the coverage of rural health insurance needs to be broadened to include outpatient services. · The recent rapid expansion of the Medical Assistance program is also laudable. There is however scope for further expansion of coverage beyond the 29 million rural and 4 million urban participants in 2007. As noted above, the relationship of MA to the Di Bao program also needs to be clarified with a view to establishing a unified approach to targeting assistance to the poor and the disadvantaged. The MA program could also be used for a means-tested safety net provision under health insurance for higher or even full reimbursement of healthcare expenses above a threshold. · In urban areas, the introduction of the new Urban Resident Basic Medical Insurance should help improve the inclusion of those in the growing informal sector whose participation in the largely formal-sector focused health insurance has been quite limited. The 41 million enrollees under this scheme by the end of 2007 is encouraging but still limited relative to the size of the informal sector. · The effective inclusion of migrant workers in health insurance remains an area of concern. The directions from MOLSS in May 2006 to set up a separate medical insurance pooling fund to mainly cover inpatient expenses for migrants, and to improve the settlement method for those in the urban system who return to their hometown for treatment or others in the rural system who seek treatment in urban areas, are indicative of the government's intent to address this issue. But the case of the migrant workers highlights the larger policy issue of the integration of rural and urban health insurance. This issue is all the more pressing with the recent expansion of both the rural and urban systems. There is also a risk of excessive fragmentation W h a t t h e p r i o r i t i e s a r e 1 9 9 of heath insurance with different schemes with varying provisions covering different segments of the population and the lack of an effective mechanism for portability across schemes. Though a difficult issue, complicated further by disparities in fiscal capacity under the decentralized fiscal system, it is one that will need to be confronted in future policy development for universal healthcare. f. Supplementing area-based poverty reduction efforts with a household-oriented approach Supplementing area- Implementation of the broader poverty reduction agenda outlined above will require based poverty reduction supplementing the traditional area-based poverty reduction efforts with household- efforts with a household- oriented approaches and interventions. The case for a household-oriented approach has oriented approach makes already been mentioned in the context of the government's poor village-based IVDP sense... program. But a focus on household-oriented approaches has a broader cross-cutting relevance for the entire poverty reduction agenda. ...for several reasons There are several related reasons why such a focus makes sense. · The large and growing scale of labor mobility has diminished the promise of exclusively area-based approaches for poverty reduction. With greater prospects of migration, it has become relatively more important to invest in people (who are increasingly mobile) rather than the areas they come from. · The remaining poor are more dispersed throughout the villages of rural China, and as already noted this is one of the reasons why an area-based approach such as the one used in the IVDP program for designated poor villages is likely, even with a better identification of poor villages, to miss reaching many of the poor. · The poor are also a heterogeneous group. They are poor for multiple and varying reasons related to their particular household endowments and circumstances, not all of which are reducible to where they live. A taxonomy of poverty suggests that several different clusters of poor can be identified depending upon different mix of factors underlying their poverty. Some of these factors are common to most, if not all, of the poor. Almost all the poor live in households where the adult workers are poorly educated. Few of the poor have access to employment off the farm. But the other reasons for poverty are more varied, and are relevant only for particular sections of the poor. In the case of some, it is that they have limited land to work with; others may be poor because of the burden of family members who are incapable of working; for still others their minority status may be a source of additional disadvantage. A household-oriented approach is needed to address this hetereogeneity. · Virtually the entire social protection agenda consists of interventions that are intrinsically household-oriented. Whether it is social assistance programs such as the rural and urban Di Bao, or social insurance programs such health or unemployment insurance, they are all directly targeted to households. · Finally, even within the context of area-based development-oriented programs, there is a case for supplementary targeted household-level assistance to enhance the poverty impact of these programs. The evidence indicates that without such direct assistance, the poorest households do not benefit. It is thus recommended that China's area-based poverty programs place new emphasis on household-level targeting, and incorporate a broader range of instruments for providing direct development-oriented assistance to poor households. These might include subsidies for complementary investments to make public goods more accessible to the poor, as well as development of new programs such as conditional cash transfers to increase rural school attendance, or self-targeted public works programs. Investing in the Greater focus on household-oriented approaches to poverty reduction also raises the development of a unified importance of developing suitable household targeting mechanisms to reach the poor. targeting system such as The common challenge faced by current programs, whether it is area-based interventions, proxy-means testing labor training for migrants, subsidized credit, or various social assistance or insurance could have large payoffs programs, is their ability to reach the poor and the disadvantaged. As discussed at several points in the report, there is significant scope for improvement of current targeting mechanisms, and there is a need for a unified approach to targeting the wide range of ongoing programs, many of which have expanded substantially in recent years. In W h a t t h e p r i o r i t i e s a r e 2 0 0 particular, there is a case for considering mechanisms such as a proxy means targeting system at least on a pilot basis. Investing in the development of such a system could have large payoffs as the same underlying system could be employed to target multiple programs, perhaps with the use of different thresholds and ancillary criteria for individual programs. Even if a proxy means system is not directly used to determine program eligibilities, it could serve an important role in the verification of currently used forms of screening and means testing, and thus help improve the overall targeting effectiveness of different programs. g. Providing an adequate and equitable allocation of resources for local governments A more adequate and In China, as in most of the world, local governments are at the frontlines of service- equitable allocation of delivery, social protection, and poverty reduction. However, unlike in much of the world, resources for local within China's highly decentralized fiscal system, local governments at the county level governments is needed and below, as well as village collectives, also bear the primary responsibility for financing these efforts. But local governments in China vary widely in terms of their revenue-raising capacity, and the variation is only partially mitigated by transfers from higher levels of government. The resulting large disparities in fiscal resources across local jurisdictions translate into large disparities in levels of public spending. And that in turn leads to large differences in the quality and level of services provided, in the extent of social protection and unevenness in the pace of poverty reduction, thereby contributing to the persistence, and even magnification, of inequalities. Implementing the broader poverty reduction agenda and ensuring that local governments all across China have the resources to provide a basic nationally determined level and quality of services and social protection to those living within their jurisdictions will require a more equitable and adequate allocation of resources for local governments. This is a two-fold task. There is certainly a need to move towards a more equitable allocation of resources through further equalizing transfers. The evidence indicates that the pattern of public expenditures favors richer provinces, and within provinces, richer local governments. For instance, at the provincial level while transfers from the center have been equalizing, but because the disparity across provinces in revenue-raising capacity has grown noticeably during 1995-2004, provincial expenditures in 2004 still favored richer provinces as much as they did in 1995. Disparities below the Spending disparities below the provincial level tend to be even more pronounced. As provincial level tend to be noted in the report, data from the Special Purpose Survey of 3036 villages, undertaken more pronounced, even jointly by the NBS and the World Bank for the purposes of this poverty assessment, for poverty-related indicate that the village-level distribution of social assistance and infrastructure investment expenditures expenditures is much less pro-poor than the provincial distribution would suggest. Much, therefore, still remains to be done in terms of making poverty-related public expenditures more equitable and pro-poor. Equalizing transfers from the center need to continue and be scaled up, but attention has to be paid as well to how equitable the allocation of resources is within provinces, and going one step further down, within counties. There is a need for But the challenge of financing the broader poverty reduction agenda is not simply one of additional resources, not redirecting resources from richer to poorer areas. A simple budget-neutral spatial re- just a more equitable allocation of existing resources, even if it were feasible or advisable, would not be allocation of resources adequate to improve the coverage and benefit levels of several programs. Therefore, there is a need as well to devote substantial additional resources to development-oriented poverty reduction, service delivery and social protection. And in doing so, because of existing disparities it will be critical to target all incremental resources to the localities where funding is currently inadequate. The additional resources What would these additional resources finance? The key areas include: could help finance critical expansion and · further expansion of the rural Di Bao, improvements in many · expansion of health insurance to achieve universal coverage of rural and urban key areas of the broader populations, W h a t t h e p r i o r i t i e s a r e 2 0 1 poverty alleviation · expanded coverage of rural and urban Medical Assistance schemes, agenda · expansion of disaster relief and provision of disaster insurance, · targeted subsidies for high-school education, · higher benefit levels of rural social assistance and social insurance programs, · greater subsidization of urban social insurance programs to ensure better coverage of the informal sector, · scaling up of basic training and job placement services for migrant workers, and improved provision of education and housing services for their families, and · the development of a unified targeting system. The scaling up of the While it is difficult to estimate precisely the cost of the expanded initiatives mentioned poverty reduction above, with the total government revenue in 2007 of nearly 5 trillion RMB, the broader initiatives is affordable poverty reduction agenda is in principle affordable. This is especially so in light of the rapid growth in the combined central and local government revenues at about 20% per year in nominal terms since 2000, significantly faster than the growth in GDP (Figure 8.98). Figure 8.98 : Combined central and local government revenues have been growing very rapidly in recent years 30 27.6 Annual growth rate of 25 22.3 total govt. revenue (%)22.5 21.6 19.9 20 17.0 15.4 14.9 20.1 15 18.4 17.2 Total govt. revenue 14.9 15.7 16.0 16.5 13.5 as % of GDP 10 2000 2001 2002 2003 2004 2005 2006 2007 Sources and notes: NBS: China Statistical Yearbook 2007, and Statistical Communique on the People's Republic of China for 2007 (February, 2008). The broader poverty Table 8.93 presents very rough estimates of the incremental cost of the key additional reduction agenda is initiatives mentioned above. These estimates suggest an added cost of around 155 billion affordable RMB, which equivalent to 3.1% of the combined tax revenues of central and local governments and 0.6% of the GDP in 2007. The scaling up of the poverty reduction efforts along these lines thus certainly seems affordable. W h a t t h e p r i o r i t i e s a r e 2 0 2 Table 8.93 : A rough estimate of the additional cost of scaling up poverty reduction efforts Program Description of additional initiatives Cost (billion Yuan) Rural Di Bao Coverage of an additional 35 million beneficiaries and an average benefit 17.5 level of 400 yuan per capita for all beneficiaries Senior secondary education Targeted subsidy for 25 million students (equal to the total enrolment in 25.0 2006) to cover tuition and miscellaneous fees of 1000 yuan per student Rural health insurance 100% coverage of the population with rural hukou and per capita 71.2 government contribution raised to 100 yuan per capita Urban health insurance 100 yuan per capita subsidy to an additional 50 million beneficiaries 5.0 under the Urban Resient Basic Medical Insurance Medical assistance 200 yuan per capita subsidy to an additional 40 million rural and 18 11.6 million urban beneficiaries (to bring the total number of beneficiaries in line with expanded coverage of rural Di Bao and current coverage of urban Di Bao) Disaster insurance Per capita subsidy of 33 Yuan (about half the cost of private catastrophic 3.3 insurance) for 100 million insured Support for labor transfer 10 million rural workers at an average cost of 2000 yuan per worker 20.0 (including 1000 yuan for boarding costs) Unified targeting system Development of a unified targeting system with a registry of 100 million 1.0 households at the aveage cost of 10 yuan per household Total additional spending for poverty reduction 154.6 As % of total tax revenue (central and local governments in 2007) 3.1 As % of GDP in 2007 0.6 Sources and notes: World Bank staff estimates. h. Strengthening institutional arrangements to promote participation, enhance accountability and improve coordination But an even harder challenge remains, namely to improve governance--capacity, accountability and responsiveness--notably (but not only) at the local level. A concern about channeling larger volumes of central funds to local governments is the lack of adequate financial management systems and monitoring and oversight capacity to ensure that the funds are well-spent. The government has Experience in other countries suggests that local communities can play an important role recognized that local in ensuring the effective implementation and targeting of poverty alleviation efforts, and to communities can help in making programs responsive to the needs of the poor. The Chinese government effective implementation recognized the importance of this principle when it mandated that the design of poor and targeting of poverty reduction efforts village investment plans be based on a participatory process. According to official training materials, projects are supposed to be selected by a plenary session of the entire village, with the views of poor households given added weight. But genuine local However, as discussed in this report, the actual practice rarely adheres to all of the participation remains guidelines, and all too often, there was little genuine participation by villagers. The limited, and more needs evaluation of the IVDP in this report also found that the better quality of village governance to be learnt about ways of (e.g., village committees with more members, more frequent meetings) enhanced program increasing participation by the poor benefits for both rich and poor in program villages. However, the poorest households did even worse relative to less poor households in villages with highly educated village leaders, a finding consistent with elite capture of program benefits. In light of these findings, high priority should be given to additional study of the village planning process to identify ways to increase participation of the poor, and to experiment with new methods of implementing village plans. There is now considerable international experience with community-based development models, which can provide some lessons for improving village planning and local accountability in China, but an appropriate model for China will also need to consider local institutional and power arrangements in Chinese villages. W h a t t h e p r i o r i t i e s a r e 2 0 3 The need for inter-agency Another huge challenge is one of inter-agency coordination. A central theme of this report coordination for the has been that the China's current poverty reduction agenda is not limited to just the implementation of the developmental poverty alleviation programs, but also includes pro-poor interventions broader poverty across a wide range of areas covering education, health, agriculture, migration policies, reduction agenda cannot be overemphasized social assistance and social insurance. This broad spectrum of poverty reduction efforts raises the enormous challenge of institutional coordination across different agencies that are responsible for their implementation. However, as in the past, coordination among government agencies has proven difficult because each agency has its own priorities and is reluctant to relinquish control over resources. The lack of effective coordination limits the overall effectiveness of the diverse set of poverty reduction initiatives. The symptomatic effects of this are already visible in many forms. For instance: · The lack of coordination amongst different agencies responsible for developmental poverty alleviation programs has limited the size of village investments, and the diversion of funds to uses that are neither consistent with village plans nor are well- directed to supporting poor households. · There is a conspicuous absence of a unified targeting mechanism to reach the poor. The area-based programs follow one approach, the social protection programs follow another, not to mention the differences across individual social assistance and social insurance programs themselves. While the 5-year planning process does provide an overall coordinating framework, there is no effective institutional mechanism to ensure coordination at the implementation level. This gap was highlighted in the case of migration policies, for instance, where in recognition of the multi-sectoral nature of policies related to migrants, in 2006 the government established a Joint Conference on Rural Migrants with representation from Ministries of Labor and Social Security, Civil Affairs, Agriculture, Construction, Education, Health, Justice, Propaganda, Public Security, the National Development Reform Commission, the National Bureau of Statistics, All China Federation of Trade Unions, All China Women's and Youth Federations, the General Bureau of Safety Inspections, and the Population and Family Planning Commission. If this was deemed necessary for migration policies alone, the coordination requirements for the wide range of poverty reduction programs and policies (of which those related to migrants are but one part) are much greater still. And these requirements have become even larger with the 11th Five-Year Plan's renewed emphasis and intensification of a broad range of initiatives on social development and the reduction of disparities. Yet, there is absence of an effective body with a large enough mandate and practical authority over other government units to play such a coordinating role. While it is not clear whether any of the existing agencies can be groomed to play this role, establishing an institutional mechanism for inter-agency coordination is critically important to realize the full benefits of the ongoing and future poverty reduction efforts. i. Enhancing statistical monitoring and evaluation capacity China has a well- China has a well-developed statistical system that in many ways exceeds the capabilities developed statistical of other statistical systems around the world. One example of this capacity is its long- system with substantial standing tradition of conducting a wide array of annual, large-scale and nationally- capabilities representative household and enterprise surveys in the rural and urban areas of the country. It also produces a very large body of statistical information based on non-survey based reporting systems. Most of the analysis presented in this report would have been impossible in the absence of this rich data base. But there remain some However, the analysis undertaken for this report also highlighted some critical gaps in the critical gaps in the information base for the monitoring and policy analysis of poverty and poverty reduction information base for policies and programs. The implementation of the poverty reduction agenda as outlined monitoring and poverty above will, among other things, also require efforts at enhancing statistical capacity to fill analysis of poverty, including ... these gaps. The key gaps are in the following areas. W h a t t h e p r i o r i t i e s a r e 2 0 4 · One important gap relates to information on migrant population in the national ... information on the living and working household surveys. China's statistical system has two concurrent annual household conditions of migrant surveys conducted independently in rural and urban areas of the country. Starting population 2001, NBS' Rural Household Surveys (RHS) began collecting information on individual migrant workers from rural areas who were considered part of their households in the source areas. However, for obvious reasons the RHS is not equipped to gather information on migrants who moved out of the villages with their families. But few of them are picked up by the Urban Household Surveys either whose sample frame is primarily confined to urban residents. Thus, very little information is available on full-family migrants in urban areas, and this alone could potentially lead to some underestimation of urban poverty. More generally, there is a lack of systematic and nationally-representative information on the working and living conditions of migrant population in destination areas that is important for developing future policies in relation to labor mobility. In view of both the large (and growing) scale of migration and its key role in achieving poverty reduction, it will be important to fill this gap. · Another gap relates to information on spatial cost of living differences. In order to ... information on spatial cost of living differentials have comparable measures of income and consumption across rural and urban areas and across regions and provinces, it is necessary to adjust nominal magnitudes by spatial cost of living indices. While different researchers have attempted to estimate these indices using available price information, and one of the most comprehensive such attempts (Brandt and Holz, 2006) has been used for the analysis in this report, the magnitudes of spatial cost of living differentials remains a subject of some debate. As this is critical for a reliable assessment of poverty, greater effort devoted to a regular production of such spatial price differences will be important for both policy analysis of poverty and the monitoring of future progress in poverty reduction. · A third important gap relates to information on program participation. As mentioned ... information on program participation at a number of points in the report, monitoring and evaluation are key to further and utilization of public reform and refinement of a wide range of initiatives, whether in education or health, services development poverty alleviation or social protection programs. While rural and urban household surveys conducted by the NBS have many strengths (e.g. large sample sizes, detailed information on household income and consumption), they have very little information on households' participation in various government programs and their utilization of various public services. This seriously limits the potential of these national surveys for program and policy evaluation. To enhance this potential, the inclusion of one or more additional modules on program participation and service utilization ought to be given a priority in the future development of the main household surveys. · Finally, the idea of a unified targeting system and the potential use of proxy-means ... more detailed information on household targeting has been suggested at a number of places in the foregoing discussion of the characteristics that could current poverty reduction agenda. An implementation of the idea will require data on be used for proxy-means a range of household characteristics that can serve as proxies for income or targeting consumption as a measure of welfare. Some of these data will be available from the existing household surveys, but more information may need to be collected by augmenting data collection efforts in this regard. Once the means-proxies are identified, it will also be necessary to develop a unified household information registry to implement such a targeting system. W h a t t h e p r i o r i t i e s a r e 2 0 5 Data appendix The analytical work undertaken for this report made use of a variety of data sources, broadly falling into the following categories: · data from the Rural and Urban Household Surveys conducted by the National Bureau of Statistics, · data from a Special Purpose Village-Level Survey conducted for this poverty assessment, · data from the China Urban Labor Survey conducted by the Institute of Population and Labor Economics, Chinese Academy of Social Sciences, · information from the 2006 Qualitative Study of Poverty conducted by the Institute of Sociology especially for this poverty assessment, and · data and information from secondary sources. This appendix gives some details of each of these data sources. Rural and Urban Household Surveys (NBS) As part of a collaborative work agreement with NBS, the poverty assessment team was given access to unit-record data for select years of the Rural and Urban Household Surveys conducted by the NBS. The arrangement involved indirect access with NBS staff running statistical program on the unit-record data while working together with the poverty assessment team. In particular, the following datasets were utilized for the analytical work on the poverty assessment: · Full household survey data for the 2003 Rural Household Survey(RHS) · Survey data for a nationally-representative longitudinal (panel) data set from 2001, 2003 and 2004 RHS constructed especially for this poverty assessment · Full household survey data for the 2003 Urban Household Survey (UHS) · A subset of data from the 2001 and 2004 UHS · Full household survey data from the 2004 Urban Short-Form Household Survey (USFHS) 2003 RHS China's Rural Household Surveys (RHS) are conducted annually by the National Bureau of Statistics with nationwide coverage of households in rural areas. For the poverty assessment, the full sample from the 2003 RHS was utilized consisting of 68,183 households covering 863 counties in the 31 provinces. The poverty assessment made use of both household and individual level data. In addition, some village-level variables were also used in the analysis. It is notable that the RHS includes also includes migrant workers from rural areas who have been are engaged in their employment away from their home for over 6 months, but bring most of their income back to the household and therefore are economically integrated into the household, in addition to members who reside in the household for at least 6 months. Longitudinal data from 2001, 2003 and 2004 RHS For the purposes of the poverty assessment, a longitudinal (panel) data set was especially constructed. The panel data set was a subset of the annual Rural Household Surveys but had national representation. It covered 25,987 households (about 38% of the full sample) in 386 of the 847 counties included in the RHS (excluding Tibet). In particular, the panel included households in all of the 199 designated poor counties and in 187 of the 648 other W h a t t h e p r i o r i t i e s a r e 2 0 6 non-poor counties included in the RHS sample. The 187 non-poor counties were randomly selected from all non-poor counties, and the sampling weights were appropriately adjusted to ensure national representation. 2003 UHS Parallel to the RHS, China also conducts Urban Household Surveys (UHS) annually covering households in urban areas. The poverty assessment utilized the full sample of the 2003 UHS which covered 49,327 households nationwide in 226 sampling units, including 146 cities and 80 county towns. Both household and individual level data from the survey were used for the poverty assessment. Traditionally, the UHS only included urban residents registered locally, thus effectively excluding all migrants. In recent years, the NBS has relaxed this. Now in principle, besides the urban households living in the survey neighborhood committees who are registered as permanent residents of the city (county town), the UHS also includes those who are registered elsewhere in the country but have lived in the survey neighborhood committee for 6 months or longer. In practice, there are concerns that the UHS under- surveys migrants. For instance, the weighted sample proportion of 2003 UHS with non- local or agricultural hukou (registration) was only 2.8% which is well below the proportion of migrants in urban samples from other surveys. Subsamples from 2001 and 2004 UHS The poverty assessment also utilized sub-samples from the 2001 and 2004 UHS. The subsample for 2001 included 17000 households while that for 2004 included 10938 households. 2004 Urban Short-Form Household Survey (USFHS) The full name of this survey is "2004 Survey of the Basic Conditions of Urban Households." It covers 251,765 households and was to serve as the sampling frame for the UHS in following years. The survey is an abridged version of the UHS and contains information on housing conditions, ownership of consumer durables, and a household roster with individual information on demographics and employment. It also contains information on household incomes and expenditures. However, there is only a single question on household income (referring to income during the previous year 2003), and hence income measures from this survey may not be as accurate as those based on the regular UHS. Significantly, the UFSHS contains questions on participation in and receipts from the urban Di Bao program and thus provides valuable information for an assessment of this program. it also has questions on self-reported adequacy of household incomes which is also allows one to estimate subjective poverty lines and poverty rates. 2004 World Bank-NBS Special Purpose Village-Level Survey Especially for the purposes of the poverty assessment, the Rural Survey Organization of the NBS conducted a Special Purpose Village-Level Survey (VLS) in 2005. Most of the data referred to the previous year, 2004. The survey was nationally representative and covered villages in the same 386 counties as included in the longitudinal data from 2001, 2003 and 2004 RHS. The survey had four parts: (i) basic information (population statistics, social security programs, subsidized loan programs, village finance, governance, schools and clinics, local taxation, local wage levels, natural disaster, human development indicators, infrastructure, family migration and land acquisition), (ii) village-level public investment from 2001 to 2004, (iii) information on village poverty alleviation plans, and (iv) information on land adjustments and village-level policies on land adjustments. W h a t t h e p r i o r i t i e s a r e 2 0 7 The village-level of data from this survey were merged with the longitudinal data from 2001, 2003 and 2004 RHS, and this data set was used in particular to analyze the performance of the village investment program. China Urban Labor Survey (CULS) The The China Urban Labor Survey (CULS) is a repeated cross-sectional survey conducted in 2001 and 2005 by the Institute of Population and Labor Economics of the Chinese Academy of Social Sciences, in collaboration with international scholars and with the support of the World Bank, the Ford Foundation, Michigan State University, and the Harvard Weatherhead Center for International Affairs. The first survey (CULS1), completed in the fall of 2001, included surveys of urban resident households and individual migrants in 5 large cities. The second survey (CULS2), completed in the summer of 2005, surveyed urban resident households and migrant households in the same 5 large cities, and also included surveys of migrant households in 7 smaller cities where urban resident households had previously been surveyed in 2004 by IPLE as part of a social protection survey. Among these 7 cities, 5 were smaller cities located near the 5 large cities surveyed by the CULS. The 5 pairs of large and small cities are located in different regions of the country. Shanghai and Wuxi (Jiangsu Province) are located in the Yangtze River Delta near the coast; Wuhan and Yichuan are in Hubei Province in central China; Shenyang and Benxi are in Liaoning Province in the northeast; Fuzhou and Zhuhai are in Fujian and Guangdong provinces in the southeast; and Xian and Baoji are in Shaanxi Province in the northwest. The two other cities surveyed in CULS2 were Daqing, a city in Heilongjiang Province rich in oil resources, and Shenzhen, a city in Guangdong Province near Hong Kong, famous for its open labor market, large number of migrants, and private enterprise development. For the poverty assessment, comparisons of large and small cities often restrict attention to the 5 large and 5 small cities located in the same regions. In each city, representative samples of local residents and migrants were independently selected in a 2-stage procedure. Using previous year data (2000 and 2004) on the local resident population of each neighborhood, a fixed number of neighborhoods were selected in each city using probability proportionate to size (PPS) sampling. For 2004, cities had limited information on the number of migrants living in each neighborhood, in which case neighborhoods were first selected based on local resident populations, and weights are used to correct for differences in the relative sizes of migrant and local resident populations based on population estimates by neighborhood office staff. These staff also helped construct an updated list of households to serve as a sampling frame. Neighborhood office staff were often aware of unregistered migrants living in the neighborhood, especially those operating small businesses. Then a fixed number of households were randomly sampled in each neighborhood. In large cities, about 600 (500) local resident households and 500 migrant individuals (households) were sampled in CULS1 (CULS2). In the other cities the sample sizes were 400 or 500 of each type of household. Detailed work history and other information were collected for all adult members of each household surveyed. In addition to the very detailed information collected, one strength of the CULS is that it surveyed migrants (including rural migrants and urban migrants) and local residents in an identical fashion, and it collected enough observations per city to calculate city-level aggregates. Two disadvantages are that for the 5 smaller cities, migrants were surveyed in 2005 while local residents were surveyed in 2004, reducing comparability, and migrants were sampled mainly through neighborhood committees so that unregistered migrants and those living in collective forms of housing, such as dormitories provided by work units, are likely to be under-sampled. 2006 Qualitative Study of Poverty The 2006 Qualitative Study of Poverty was conducted by the Institute of Sociology at the Chinese Academy of Social Sciences in support of the work for the poverty assessment. The study was conducted in 12 villages in 6 provinces with two villages selected in each W h a t t h e p r i o r i t i e s a r e 2 0 8 province. The 6 provinces covered were: Gansu, Inner Mongolia, Sichuan, Yunnan, Jiangxi and Jiangsu. The fieldwork was carried out during February-April 2006. The qualitative research focused on the following key questions: · First, how do villagers in rural communities, poor or non-poor, male or female, understand poverty within their community? What are the most important factors leading to poverty or poverty reduction from their standpoints? What differences exist between villagers' understanding of poverty and government definitions of poverty? · Second, whether or under what conditions can external support facilities such as the Village investment project or policies for the industrialization of agriculture become more effective in reducing rural poverty? This is a crucial contemporary issue for addressing poverty in China. · Third, to what extent are related rural policies such as those currently being implemented in education and health, contributing to poverty reduction in particular rural communities? · Fourth, how have rural communities and households established family strategies and sought support from their own resources to reduce poverty? Have their efforts succeed or failed?? · Finally, to what extent is labor migration effective as part of poor household's coping strategies? To what extent can poor households adopt labor migration as a strategy for improving their livelihoods? Additionally, what results will labor migration have for the rural community, rural production and the migrants themselves? The Study used a variety of research methods including key informant interviews, focus group discussions, household interviews, life histories, ladders of life assessments, participatory observation and collection of secondary documents. As part of the Study, the Institute of Sociology produced a national synthesis report and individual case studies for each of the 12 village sites. Data and information from secondary sources Finally, the poverty assessment also utilized a variety of data and information from secondary sources. This included information from NBS' China Statistical Yearbooks, the Rural Household Survey Yearbooks, and a number of Statistical Communiques from the NBS. In addition to this, statistical reports and information from a numbers of Ministries were also utilized for the poverty assessment. W h a t t h e p r i o r i t i e s a r e 2 0 9 References 1 Appleton, Simon, John Knight, Lina Song, and Qingjie Xia (2002). "Labor Retrenchment in China: Determinants and Consequences," China Economic Review 13: 252-275. 2 Asian Development Bank (2004). "Poverty Profile of the People's Republic of China". Asian Development Bank. Manila. 3 Au, Chun-Chung and J. Vernon Henderson (2006) "Are Chinese Cities too small?" A review of economic studies, 73(3): 549-576. 4 Bardhan, Pradan and Dilip Mookherjee (2005). "Decentralising Antipoverty Program Delivery in Developing Countries" Journal of Public Economics 89: 675-704. 5 Benjamin, D., L. Brandt, J.Giles and S. Wang (2008). "Income Inequality During China's Economic Transition," in China's Economic Transition: Origins, Mechanisms and Consequences, Loren Brandt and Thomas Rawski, eds., Cambridge University Press. 6 Bidani, Benu, Niels-Hugo Blunch, Chorching Goh, and Christopher J. O'Leary (2005). "Evaluating Job Training in Two Chinese Cities," mimeo, World Bank, Washington DC. 7 Brandt, L., S. Rozelle and M. A. Turner (2004). "Local Government Behavior and Property Right Formation in Rural China, " Journal of Institutional and Theoretical Economics, 35(4): 689-716. 8 Brandt, Loren and Carsten Holz (2006). "Spatial Price Differences in China: Estimates and Implications," mimeo, Department of Economics, University of Toronto. 9 Cai, Fang and Yaowu Wu (2006). "China's Urban Informal Employment: Scale and Characteristics", mimeo, World Bank, Washington DC. 10 Cai, Fang, Albert Park and Yaohui Zhao (2008). "The Chinese Labor Market in the Reform Era" in China's Economic Transition: Origins, Mechanisms and Consequences, Loren Brandt and Thomas Rawski, eds., Cambridge University Press. 11 CASS-IOS (2006). Institute of Sociology ­ Chinese Academy of Social Sciences 2006: "Rural Poverty and Anti-poverty Policies in China in a New Era". Mimeo, Beijing. 12 Castaneda, Tarsicio and Lindert, Kathy (2005). "Implementing Means-Tested Welfare Systems in the United States," World Bank Social Protection Discussion Paper No. 0532, June, 2005. 13 China Development Research Foundation (CDRF, 2007). "The Elimination of Poverty Trough Development". China Development Report 2007, China Development Research Foundation, Beijing. 14 Chen, Shaohua, Martin Ravallion and Youjuan Wang (2006). "Di Bao: A Guaranteed Minimum Income in China's Cities," World Bank Policy Research Working Paper 3805, Washington DC. 15 Chen, Shaohua and Martin Ravallion (2008). "The Developing World is Poorer Than We Thought, But No Less Successful in the Fight Against Poverty". Poverty Research Working Paper 4703, World Bank, Washington DC. 16 China Disabled Persons Federation (2006). www.cdpf.org.cn, website for China Disabled Persons Federation, accessed February 18, 2006. 17 China Center on Aging (2002). "Data Analysis of the Sample Survey of Aging Population", Beijing: China Standard Press. 18 Coady, David, Margaret Grosh and John Hoddinott (2004). "Targeting Outcomes Redux," World Bank Research Observer, 19(1): 61-86. 19 Deaton, Angus (1997). "The Analysis of Household Surveys: A Microeconometric Approach to Development Policy", The Johns Hopkins University Press for the World Bank , Baltimore. 20 DeBraw Alan and John Giles (2008). "Migrant Opportunity and the Educational Attainment of Youth in Rural China" Policy Research Working Paper 4526, World Bank Development Research Group (February) Washington DC. R e f e r e n c e s 2 1 0 21 Deininger, Klaus and Songqing Jin (2005). "The Potential of Land Rental Markets in the Process of Economic Development: Evidence from China, " Journal of Development Economics, 78: 241-270. 22 Deininger, Klaus, Songqing Jin and Scott Rozelle (2006). "Dynamics of Legal Change in a Decentralized Setting: Evidence from China's Rural Land Contracting Law," mimeo, World Bank, Washington DC. 23 Du, Yang, Robert Gregory and Xin Meng (2006). Impact of the Guest Worker System on Poverty and Wellbeing of Migrant Workers in Urban China, mimeo, Research School of Pacific and Asian Studies, Australian National University, Canberra. 24 Du, Yang, Albert Park, Sangui Wang (2005) "Migration and Rural Poverty in China", Journal of Comparative Economics, 33: 688-709, 2005. 25 Duan, Chengrong and Fulin Zhou (2005). "A Study on Children Left Behind", Population Research 29(1) :29-36. 26 Fleisher, Belton M and Xiaojun Wang (2004). "Skill Differenials, Return to Schooling and Market Segmentation in a Transition Economy: The Case of Mainland China" Journal of Development Economics, 73: 315-328. 27 Gao, Hongbing (2001) "Study on Poverty Reduction and Development Planning", China Financial and Economic Press, Beijing. 28 Gao, Hongbing, Weimin Wang and Chengwei Huang (2001) "Study on Poverty Reduction and Development Planning", China Financial and Economic Press, Beijing. 29 Giles, John, Albert Park and Fang Cai (2006a). "How Has Economic Restructuring Affected China's Urban Workers?," China Quarterly, 177: 61-95. 30 Giles, John, Albert Park and Fang Cai (2006b). "Re-employment of Dislocated Workers in Urban China: the Roles of Information and Incentives," Journal of Comparative Economics 34: 582-607. 31 Giles, John, Albert Park, and Juwei Zhang (2005). "What is China's True Unemployment Rate?," China Economic Review 16:149-170. 32 Government of the Peoples' Republic of China (2004a). "State Council of the People's Republic of China Social Security White Paper", www.china.org.cn. 33 Government of the Peoples' Republic of China (2004b). "Report on the Work of the Government, delivered by Premier Wen Jiabao at the Second Session of the Tenth National People's Congress", March 5, 2004. 34 Grigoriou, Christopher, Patrick Guillaumont, Wenyan Yang (2005). "Child Mortality Under Chinese Reforms," China Economic Review, 16, 441-464. 35 Gustafsson, Bjorn and Shi Li (2004). "Expenditures on Education and Health Care and Poverty in Rural China," China Economic Review, 15, 292-301. 36 Hussain, Althar (2003). "Urban Poverty in China: Measurement, Pattern and Policies, mimeo, International Labor Office, Geneva. 37 Jacoby, H. G., Gui. Li, Guo and Scott Rozelle (2002). "Hazards of Expropriation: Tenure Insecurity and Rural Investment in Rural China," American Economic Review, 92(5): 1420-1447. 38 Jalan, Jyotsna and Martin Ravallion (1998). "Transient Poverty in Postreform Rural China," Journal of Comparative Economics, 26, 338-357. 39 Jalan, Jyotsna and Martin Ravallion (2000). "Is Transient Poverty Different? Evidence for Rural China," Journal of Development Studies, 36, 82-98. 40 Juhn, Chinhui, Kevin M. Murphy and Brooks Pierce (1993). "Wage inequality and the rise in returns to skill", Journal of Political Economy, 101(2): 410-442. 41 Khan, A. R. and C. Ruskin (2001). Inequality and Poverty in China in the Age of Globalisation, Oxford University Press, Oxford and New York. 42 Lanjouw, Peter, and Martin Ravallion (1995). "Poverty and Household Size," Economic Journal 105: 1415-1434. 43 Lewis, Arthur (1954). "Economic Development with Unlimited Supplied of Labor. Manchester School 28: 139-91. R e f e r e n c e s 2 1 1 44 Lindelow, Magnus and Adam Wagstaff (2005). "Health Shocks in China: Are the Poor and Uninsured Less Protected?" The World Bank, Washington DC. 45 Li, Shi and Chuliang Lou. (2006). "Re-estimating the Income Gap Between Urban and Rural Households in China", mimeo, World Bank, Washington DC. 46 Li, Wen and Sangui Wang (2004). "Analysis on the Determinants of the Allocation of Poverty Reduction Funds from the Central Government". Chinese Rural Economy, 8:44- 48. 47 Lin, Justin Yifu (1992). "Rural Reforms and Agricultural Growth in China". American Economic Review, Vol 82, No 1 (March) 34-51. 48 Lin, Yang (2004). "The Rural Social Protection System and the Rural Minimal Living Allowance: A Review of Existing Rural Schemes," Background Paper for the World Bank Report on Rural Social Protection in China, World Bank, Washington DC. 49 Logan, John, Yuping Fang and Zhanxin Zhang (2006). "Access to Housing in Urban China", Brown University, mimeo. 50 MacQueen, J. B. (1967). "Some Methods for Classification and Analysis of Multivariate Observations," Proceedings of the 5th Berkeley Symposium on Mathematical Statistics, Vol. 1: pp. 281­297. 51 Mansuri, Ghazala and Vijayendra Rao (2004). "Community-based and driven development: a critical review", The World Bank Research Observer , 19(1): 1-39. 52 Meng, Xin and Chuliang Luo (2006). "Impact of the `Guest Worker' System on Education of Migrants' Left-behind Children, Unpublished Paper, Australian National University, Canberra. 53 Meng, Xin and Junsen Zhang (2001). "The Two-Tier Labor Market in Urban China: Occupational. Segregation and Wage Differentials between Urban Residents and Rural Migrants in Shanghai", Journal of Comparative Economics, 29: 485-504. 54 Meng, Xin, Robert Gregory and Youjuan Wang (2005). "Poverty, Inequality and Growth in Urban China, 1986-2000," Journal of Comparative Economics, 33, 710-729. 55 Ministry of Civil Affairs (2003). Civil Affairs Statistical Yearbook, Beijing. 56 Ministry of Civil Affairs (2007). Civil Affairs Statistical Yearbook, Beijing. 57 National Bureau of Statistics (2003). 2003 China Labour Statistical Yearbook: 2003, China Statistical Press, Beijing. 58 National Bureau of Statistics. China Statistical Yearbook: various years, China Statistical Press, Beijing. 59 National Bureau of Statistics (2005a). China Rural Household Survey Yearbook: 2005, China Statistical Press, Beijing. 60 National Bureau of Statistics (2005b). Research of Rural Labor of China 2005, Department of Rural Survey, National Bureau of Statistics, Beijing. 61 National Bureau of Statistics of China (2006). Communique on Major Data of 1% National Population Sample Survey in 2005, Beijing. 62 National Bureau of Statistics (2007). Statistical Communique of the Peoples' Republic of China on the 2006 National Economic and Social Devleopment, Beijing 63 National Bureau of Statistics (2008). Statistical Communique of the Peoples' Republic of China on the 2007 National Economic and Social Devleopment, Beijing 64 Nielsen, Ingrid, Chris Nyland, Russell Smyth, Mingqiong Zhang and Cherrie Jiuhua Zhu (2005). "Which rural migrants receive social assistance in Chinese cities? Evidence from Jiangsu survey data," Global Social Policy, 5(3): 353-381. 65 O'Keefe, Philip (2004). Social Assistance in China: An Evolving System, International Research Conference on Poverty, Inequality, Labor Markets and Welfare Reform in China, 25-27 August 2004, Australian National University, Canberra. 66 Oxford Analytica (2006), China: Urban Poor Mar Poverty Alleviation Record (January 5). 67 Park, Albert and Changqing Ren (2001). "Microfinance with Chinese Characteristics" World Development 29(1): 3-62. R e f e r e n c e s 2 1 2 68 Park, Albert, Xiaoqing Song, Junsen Zhang, Yaohui Zhao and Xiaohua Li (2006). "Rising Returns to Skill, Labor Market Transition, and the Growth of Wage Inequality in urban China, 1988 to 2003," mimeo, Department of Economics, University of Michigan. 69 Park, Albert, Xiaoqing Song, Junsen Zhang, Yaohui Zhao, and Xiaohua Li (2007). "Rising Returns to Skill, Labor Market Transition, and the Growth of Wage Inequality in China", mimeo. 70 Park, Albert, Sangi Wang (2001). "China's Poverty Statistics", China Economic Review, 12: 384-398. 71 Park, Albert and Sangi Wang (2006). "Community-based Development and Poverty Alleviation: An evaluation of China's Poor Village Investment Program" Background Paper for The World Bank's China Poverty Assessment. 72 Park, Albert, and Sangui Wang.(2007). "Migration and Urban Poverty and Inequality in China", mimeo, World Bank, Washington DC. 73 Park, Albert, Sangui Wang and Fang Cai (2006). "Migration and Urban Poverty and Inequality in China", mimeo. 74 Park, Albert, Sangui Wang and Guobao Wu, (2002). "Regional Poverty Targeting in China", Journal of Public Economics 86: 123-153. 75 Park, Albert, Yaohui Zhao, and Guofang Huang (2006). "Migration and Labor Market Segmentation in Urban China," mimeo, Department of Economics, University of Michigan, Ann Arbor. 76 Platteau, Jean-Phillipe (2004). "Monitoring elite capture in community driven development" Development and Change 35(2): 223-246. 77 Pradhan, Menno and Martin Ravallion (2000). Measuring poverty using qualitative perceptions of consumption adequacy, Review of Economics and Statistics, 82(3): 462- 471. 78 Prosterman, Roy, Li Peng, Zhu Keliang and Jeffrey Riedinger (2006). "The Rural Land Question in China: A Snapshot from a 17-Province Survey, Rural Development Institute and Michigan State University. 79 Ravallion, Martin (1988). "Expected Poverty under Risk-induced Welfare Variability," Economic Journal, 98: 1171-1182. 80 Ravallion, Martin (1994). Poverty Comparisons, Harwood Academic Publishers, Chur. 81 Ravallion, Martin and Shaohua Chen (2007). "China's (Uneven) Progress Against Poverty," Journal of Development Economics, 82 (1): 1-42. 82 Rawski, Thomas (2003). "Recent Developments in China's Labour Economy," Working Paper No. 11 of the International Labor Office, International Policy Group, Policy Integration Department, Geneva. 83 Sato, Hiroshi (2006). "Housing Inequality and Housing Poverty in Urban China in the late 1990s," China Economic Review, 17(1): 37-50. 84 Sheng, Laiyun and Peng, Liquan (2005). The Population, Structure and Characteristics of Rural Migrant Workers" in: NBS (2005b) Research of Rural Labor of China 2005 Department of Rural Survey: National Bureau of Statistics, Beijing. 85 Sicular, Terry, Yue Ximing, Bjorn Gustafsson and Li Shi. (2007) "The Urban-Rural Income Gap and Inequality in China." Review of Income and Wealth 53, 1 (2007): 93-126. 86 Solinger, Dorothy (2005)."The Creation of a new underclass in china and its implications" Centre for the study of Democracy, University of California, Paper No 05'10, Irvine. 87 Tang, J., L. Sha and Z. Ren (2003). Report on Poverty and Anti-Poverty in Urban China, Huaxia Publishing House, Beijing. 88 The Leading Group for Poverty Reduction and Development (1994). "The 8-7 Poverty Reduction Plan", Beijing 89 The Leading Group for Poverty Reduction and Development (2001). "China's Rural Poverty Reduction and Development Compendium (2001-2010)", Beijing. 90 Trinh, Tamara (2006). "China's Pension System: Caught Between Mounting Legacies and Unfavorable Demographics," Deutsche Bank Research, Current Issues: China Special, Frankfurt. R e f e r e n c e s 2 1 3 91 Wang Mei and Yaohui Zhao (2004). "Change in Labor Migration Pattern in China", International Research Conference on Poverty, Inequality, Labor Markets and Welfare Reform in China, 25-27 August 2004, Australian National University, Canberra. 92 Wang, Sangui (1994). "Poverty Problems and Economic Development". Rural Reading Materials Press, Beijing. 93 Wang, Sangui, Li Wen and Li Yun (2004). "Allocation of the Poverty Reduction Funds in China: Sectoral Distribution and its Impacts", Journal of Agrotechnical Economics, 5: 45- 49. 94 Wang, Sangui (2006). "A Note on New Initiatives of the Leading Group for Poverty Alleviation and Development", Institute of Agricultural Economics and Development, Chinese Academy of Agricultural Sciences, mimeo. 95 Wang, Sangui, Shubham Chaudhuri, Gaurav Datt, Albert Park (2006) Poverty Targeting of China's Integrated Village Development Program, mimeo, World Bank, Washington DC. 96 Whalley, John and Ximing Yue (2006). "Rural Income Volatility and Inequality in China," mimeo World Bank, Washington DC. 97 Wishart, D. (1982). Supplement, CLUSTAN User Manual, Third Edition, Program Library Unit, Edinburgh University, Edinburgh. 98 Wishart, D. (2001). "k-Means Clustering with Outlier Detection," GfKl 2001, The 25th Annual Conference of the German Classification Society, University of Munich, March 14-16, 2001. 99 Women's Federation PRC (2006). National Conference on Rural Left-behind Children 2006, available from www.women.org.cn. 100 World Bank (1997). Sharing rising incomes, China 2020: Disparities in China, World Bank Washington DC. 101 World Bank (2001). China: Overcoming Rural Poverty, World Bank, Washington DC. 102 World Bank (2002). Social Funds: Assessing Effectiveness, Operations Evaluation Department, World Bank, Washington DC. 103 World Bank (2005a). World Development Report 2006: Equity and Development, World Bank, Washington DC. 104 World Bank (2005b). World Development Indicators: 2005, World Bank, Washington DC. 105 World Bank (2006). Health Reform in Rural China: Where Next? World Bank, Washington DC. 106 World Bank (2007a). World Development Indicators: 2007 , World Bank, Washington DC. 107 World Bank (2007b). Will Resilience Overcome Risk? Special Focus: Agriculture for Development. East Asia & Pacific update (November), World Bank, Washington DC. 108 Ye, Jingzhong, James Murray and Wang Yi Huan, eds (2005). "Left-behind Children in rural China: Impact Study of Rural Labor Migration on Left-behind in Mid-West China", Social Sciences Academic Press, Beijing. 109 Yue, Ximing, and Shi Li (2002). "Analysis on Targeting Accuracy of Poverty Reducing Programs in Rural China," mimeo, Institute of Economics, Chinese Academy of Social Sciences, Beijing. 110 Xue Jinjun and John Knight (2004). "How High is Urban Unemployment in China?" International Research Conference on Poverty, Inequality, Labor Markets and Welfare Reform in China, 25-27 August 2004, Australian National University, Canberra. 111 Yang, D. T. (1997). "China's Land Arrangements and Rural Labor Mobility," China Economic Review, 8(2): 101-115. 112 Yue, Ximing, Shi Li and Pingping Wang (2005). "The Causes of Transient Poverty and its Implication to Poverty Reduction Policy in Rural China," mimeo, Institute of Economics, Chinese Academy of Social Sciences, Beijing. 113 Zhang, Junsen, Yaohui Zhao, Albert Park, Xiaoqing Song (2005). "Economic Returns to Schooling in Urban China, 1988 to 2001," Journal of Comparative Economics 33: 730- 752. R e f e r e n c e s 2 1 4 114 Zhang, Xiaobo and Ravi Kanbur (2005). "Spatial Inequality in Education and Health Care in China," Chinese Economic Review, 16 (2):189-204. R e f e r e n c e s 2 1 5