A WORLD BANK COUNTRY STUDY FILE COPY Report No.:1 11245 Title: STRA Author: WORL Ext.: 0 Room: Dept.: BOOKSTORE OCTOBER 1992 China Strategies for Reducing Poverty in the 1990s FrILE COPY I '~ L '@ U A WORLD BANK COUNTRY STUDY China Strategies for Reducing Poverty in the 1990s The World Bank Washington, D.C Copyright © 1992 The International Bank for Reconstruction and Development/THE WORLD BANK 1818 H Street, N.W. Washington, D.C. 20433, U.S.A. All rights reserved Manufactured in the United States of America First printing October 1992 World Bank Country Studies are among the many reports originally prepared for internal use as part of the continuing analysis by the Bank of the economic and related conditions of its developing member countries and of its dialogues with the governments. Some of the reports are published in this series with the least possible delay for the use of governments and the academic, business and financial, and development communities. The typescript of this paper therefore has not been prepared in accordance with the procedures appropriate to formal printed texts, and the World Bank accepts no responsibility for errors. The World Bank does not guarantee the accuracy of the data included in this publication and accepts no responsibility whatsoever for any consequence of their use. Any maps that accompany the text have been prepared solely for the convenience of readers; the designations and presentation of material in them do not imply the expression of any opinion whatsoever on the part of the World Bank, its affiliates, or its Board or member countries concerning the legal status of any country, territory, city, or area or of the authorities thereof or concerning the delimitation of its boundaries or its national affiliation. The material in this publication is copyrighted. Requests for permission to reproduce portions of it should be sent to the Office of the Publisher at the address shown in the copyright notice above. The World Bank encourages dissemination of its work and will normally give permission promptly and, when the reproduction is for noncommercial purposes, without asking a fee. Permission to copy portions for classroom use is granted through the Copyright Clearance Center, 27 Congress Street, Salem, Massachusetts 01970, U.S.A. The complete backlist of publications from the World Bank is shown in the annual Index of Publications, which contains an alphabetical title list (with full ordering information) and indexes of subjects, authors, and countries and regions. The latest edition is available free of charge from the Distribution Unit, Office of the Publisher, Department F, The World Bank, 1818 H Street, N.W., Washington, D.C. 20433, U.S.A., or from Publications, The World Bank, 66, avenue d'Iena, 75116 Paris, France. ISSN: 0253-2123 Library of Congress Cataloging-in-Publication Data China : strategies for reducing poverty in the 1990s. p. cm. - (A World Bank country study) Includes bibliographical references. ISBN 0-8213-2248-6 1. Rural poor-China. 2. China-Rural conditions. 3. China- Economic policy-1976- 4. Social service-China. I. Series. HC430.P6C47 1992 362.5'8'0951-dc2O 92-34227 CIP iii Preface This report is based on the findings of a mission which visited Beijing, Yunnan, Guangxi and Gansu during April and May, 1991. Mission members were Alan Piazza (EA2AG, mission leader), Wang Yuan (EA2CH, public finance), Elaine Chan (EA2CO, labor and rural enterprise), Lee Travers (consultant, urban poverty and social services), Paula Harrell (consultant, poverty programs and southwest China case study), Elisabeth Croll (consultant, gender issues and village studies), and Suresh Tendulkar (consultant, social services and poverty measurement). The mission worked closely with the Leading Group for Poor Area Development (Leading Group), was accompanied by Messrs. Gao Hongbin (Deputy Director of the Leading Group) and Zhang Yiming throughout its stay in China, and was hosted by provincial and local level staff of the Leading Group in Yunnan, Guangxi and Gansu. The excellent support afforded the mission at all levels of the Leading Group system is gratefully acknowledged. In addition, the mission met with staff of the central, provincial and local l,evel offices of the Ministry of Civil Affairs, Ministry of Agriculture, Ministry of Public Health, State Planning Commission, State Education Commission, Minorities Commission, All China Women's Federation, and State Statistical Bureau. The mission also benefitted from interaction with UNICEF, WFP, UNDP and Ford Foundation staff. The report's analysis and recommendations are scheduled to be discussed at an international conference on "Poverty Issues in China" to be held in Beijing during late-1992. UNDP's financial support and active participation in organizing the proposed conference are gratefully acknowledged. Judith Banister (consultant, demographics), Bert Keidel (consultant, macroeconomics), and John Quinley (consultant, public health) each completed desk studies in support of the report. Mr. Yao Xianbin provided valuable research assistance. The mission also wishes to acknowledge the extensive review and comments on the report provided by Lyn Squire, Paul Glewwe, Terrice Bassler, Jeffrey Taylor and Martin Ravallion (peer reviewers) and Robert Parker (UNICEF). Helena Ribe, Samuel Lieberman, Barbara Herz, Zafer Ecevit, Shahid Yusuf, Joseph Goldberg, Paul Cadario, Richard Bumgarner, Ivy Chang, Caroline Jen, Himelda Martinez, Tony Ody, and Christine Wallich (Bank) also gave valuable suggestions and comments. iv OFFICIAL EXCHANGE RATE Currency Unit = Yuan (Y) Effective March 31, 1992: USs = Y 5.46 Y = US$ 0.18 Up to November 29, 1990: US$ = Y 4.72 Y = US$ 0.21 Up to December 15, 1989: US$ = Y 3.72 Y = US$ 0.27 CONVERSIONS 1 jin = 0.50 kg 1 kg = 2.00 jin 1 mu = 0.0667 ha 1 ha = 15.00 mu ACRONYMS ABC - Agricultural Bank of China CASS - Chinese Academy of Social Sciences COE - Collectively-Owned Enterprise ICBC - Industrial and Commercial Bank of China IMR - Infant Mortality Rate (first year) LEMO - Labor Export Management Office LGEDPA - Leading Group for the Economic Development of Poor Areas MOA - Ministry of Agriculture MCA - Ministry of Civil Affairs MOF - Ministry of Finance MOL - Ministry of Labor MOPH - Ministry of Public Health PADO - Poor Area Development Office PBC - People's Bank of China POE - Privately-owned Enterprise RCRD - Research Center for Rural Development RE - Rural Enterprise SEdC - State Education Commission SOE - State-Owned Enterprise SPC - State Planning Commission SSB - State Statistical Bureau STC - State Science and Technology Commission Y - Yuan (Renminbi) v Table of Contents Paae No. EXECUTIVE SUMMARY ........................................... ix-xviii 1. POVERTY ALLEVIATION AND MACROECONOMIC TRENDS IN THE 1980s ... 1 A. Absolute Poverty .................................... 1 B. Impact of Macroeconomic Trends ................. 5 -Economic Reform and Growth .... ............5 -Prices and Income ..... ................ 7 -Employment Levels and Demographic Change ......... 9 -Regional Growth Trends ............ .. ............. 12 -Public Finance and Redistributive Transfers ...... 16 -Macroeconomic Trends and Poverty Alleviation Work in the 1990s ............... .. ............... 19 2. INCIDENCE AND CORRELATES OF ABSOLUTE POVERTY .21 A. Incidence of Absolute Poverty . .22 -Income Levels ..22 -Urban Poverty ..26 -Rural Poverty ..32 B. Correlates of Absolute Poverty ..41 -Natural Resources ..42 -Infrastructure ..42 -Minorities ..43 -Involuntary Resettlement . .45 -Gender . .45 3. EMPLOYMENT AND PRODUCTIVITY .49 A. Demographic Change, Employment Levels and Labor Mobility . .49 -Demographic Change and Labor Supply . .50 -Labor Market Development . .52 -Expected Labor Market Trends . .54 -Labor Mobility and Poverty Alleviation . .57 B. Rural Enterprise ..60 -Overview ................................. 60 -National Trends ..62 -Rural Enterprise Development in the Poor Areas 65 C. Agriculture and Natural Resources . .66 -Employment in Agriculture 69 -The Loess Plateau . .72 vi 4. SOCIAL SERVICES IN THE POOR AREAS ........................... 81 A. Financing Social Services in the Poor Areas .... ..... 81 B. Education ........................................... 82 -The Poor's Access to Education ................... 83 -Returns to Education ............................. 84 -Education Financing and Services ................. 85 -Intervention Strategies .......................... 87 C. Health .............................................. 90 -Maternal and Child Health ........................ 90 -Disease Morbidity ................................ 93 -Health Financing and Services .................... 96 -Costs of Improving Health Services and Status in the Poor Areas ................................ 101 D. Relief .............................................. 107 -Introduction ..................................... 107 -Rural Relief ..................................... 109 -Financing Rural Relief: An Expanded Role for the Ministry of Civil Affairs ........................ 112 5. POVERTY ALLEVIATION POLICIES, PROGRAMS AND INSTITUTIONS ..... 115 A. Poverty Alleviation During the 7th Five Year Plan ... 115 -Strategy ......................................... 115 -Institutions ..................................... 116 -Defining the Poor ................................ 117 B. Development Assistance Programs ..................... 121 -Domestic Programs ................................ 121 -International Assistance ......................... 125 C. 8th Five Year Plan: Reaching the Chronic Absolute Poor ................................................ 127 -Strategy and Funding ............................. 127 -Monitoring and Targeting ......................... 128 -Institutional Issues ............................. 130 REFERENCES ....................................................... 131 Annexes 1. Incidence of Absolute Poverty ........... .. ............. 137 -National ....... .............. .................... 137 -Provincial ...... ........... .. .................... 141 -Conclusion .................... ................... 141 -Appendix 1: Rural Income Reporting Systems ....... 152 NAS 1. 1989 Incidence of Rural Poverty (SSB Data) 2. 1989 Incidence of Rural Poverty (MOA Data) vii Tables in the Main Report 1.1 Selected Indicators of Social Development: China and Other Countries ................................... 1 1.2 Macroeconomic Indicators and Estimated Incidence of Absolute Poverty .................................. 4 1.3 Provincial Average Annual Real Income Growth (%) ....... 14 1.4 Real Provincial Average Annual Budget Balances ......... 18 2.1 Distribution of Annual Per Capita Income, Poverty Lines, and Incidence of Absolute Poverty .... ............. 23 2.2 Provincial Indicators of Social Development .... ........ 37 2.3 MOA County Per Capita Income Data ...................... 40 3.1 Labor Supply and Employment Trends, 1978-2000 .... ...... 52 3.2 Rural Enterprise Development, 1978-90 .... .............. 62 3.3 Provincial Rural Enterprise Employment, 1980-90 ........ 64 3.4 Estimated Agricultural Labor Requirements, 1978, 1984, 1989 and 2000 ......................... 71 4.1 Rural Disaster and Poverty Relief -- Numbers Assisted and Funding ....................................... 113 5.1 PADO (1985) and MOA (1989) Poor Counties and SSB (1989) Poverty Incidence by Province .... ................. 119 Figures in the Main Report 2.1 Rural Income and Poverty ............................... 33 Boxes in the Main Report 2.1 The Women's Federation and Poverty Alleviation ......... 48 3.1 Jingtai Phase II -- Laudable Objectives, Questionable Economics ....... .............. .................... 80 4.1 "Self-Relief Through Development" Enterprises ... ....... 110 Working Papers (available upon request) 1. Demographic Aspects of Poverty 2. A Program to Improve Health Services in the Poor Areas 3. Economic Analysis of the Loess Plateau Land Rehabilitation Program 4. Poverty in Southwest China 5. Rural Poverty: Characteristics, Causes and Policies in Baise Prefecture ix EXECUTIVE SUMMARY -Incidence and Determinants of Absolute Poverty (i) Incidence of Absolute Poverty. Broad participation in strong rural economic growth brought about a tremendous reduction in absolute poverty in China during 1978-85. World Bank estimates show the number of absolute poor to have declined from roughly 270 million in 1978 to 97 million in 1985, or from about one-third to less than one-tenth of the total population. However, no further reductions in poverty were achieved during the second half of the 1980s -- the proportion of total population living in absolute poverty remained roughly constant at about 9%. The stagnation of poverty during 1985- 90, which contrasts the strong overall economic growth of those years, is consistent with the modest observed increase of rural income disparity. The poverty gap also increased slightly during the second half of the 1980s -- indicating a modest deepening of poverty -- but remained extremely low by developing country standards. The summary poverty assessment figures shown below document China's impressive overall achievements in poverty alleviation since the initiation of economic reform in 1978: 1978 1985 1990 Total Population (million) 963 1059 1143 Urban 172 (17.9%) 251 (23.7%) 302 (26.4%) Rural 790 (82.1%) 808 (76.3%) 841 (73.6%) Average Per Capita Income (1978 Yuan) Urban -- 557 685 Rural 134 324 319 Poverty Line (Yuan/year) Urban -- 215 319 RuraL 98 190 275 Incidence of Poverty (million) \a Total 270 (28.0%) 97 (9.2%) 98 (8.6%) Urban 10 (4.4%) 1 (0.4%) 1 (0.4%) Rural 260 (33.0%) 96 (11.9%) 97 (11.5%) Gini Coefficient Rural -- 0.28 0.29 Poverty Gap (Index) \b Rural -- 2.1 2.5 _________________________ \a Sensitivity analysis shows that these trends in the incidence of poverty are robust to all reasonable changes in the assumptions underlying the estimated poverty lines (see paras 2.05 and 2.06). \b The poverty gap is a measure of the additional income that would be necessary to bring a poor person up to the poverty line. Summed across the poor population and normalized against the poverty line and total population size, the "poverty gap index" can be used to assess the relative depth of poverty among subpopulations, across time, or internationally. x Recently released unofficial State Statistical Bureau (SSB) estimates show a sharp decline in the number of rural poor from 124.1 million in 1985 to 84.1 million in 1990. The SSB has reported the poverty lines used in their analysis--Y206 in 1985 and Y268 in 1990--but has not released detailed documentation of the methods and data used in their calculations. In contrast to the 60% increase in the rural retail price index during 1985-90, the SSB's 1990 poverty line is just 30% greater than their 1985 line--suggesting an optimistic bias to the SSB estimates. By comparison, the 1990 poverty line used in the table above is 45% greater than its 1985 poverty line. (ii) Correlates of Poverty. Absolute poverty In China is almost entirely restricted to resource constrained remote upland areas -- it is extremely limited among the registered urban population and Is not a problem of landless laborers. The government has effectively limited urban poverty through a system of full employment, linked to narrow ranges of earned income, coupled with heavy subsidization and price controls on basic foodstuffs and housing and near universal access to education and health care. The incidence of urban poverty, which was already contained to very low levels by the late- 1970s, is estimated to have declined to less than 1% of the registered urban population by 1984 and to have remained at about that level thereafter. This optimistic portrayal, however, fails to reflect the status of the unregistered urban population. Those people, who number at least ten million, formally hold rural residence but work in urban areas without the job, food, health, and other rights normally accorded urban people. Most earn a decent living, but many do not and all are extremely vulnerable if struck by illness, accident, or other calamity. By convention, the Chinese statistical system continues to treat these people as rural residents, therefore those among them in poverty are counted as rural poor, leaving the total estimates of poverty unchanged but shifting the apparent sectoral composition. (iii) Rural poverty is estimated to have declined from about one-third of the rural population in 1978 to roughly one-tenth by 1985. Since virtually all of China's rural population received land use rights as part of the implementation of the production responsibility system during the early-1980s, there are few if any landless laborers. Instead, the majority of the rural poor are now concentrated in resource deficient areas, and comprise entire communities located mostly in upland sections of the interior provinces of northern, northwestern and southwestern China. Although these poor have land use rights, in most cases the land itself is of such low quality that it is not possible to achieve subsistence levels of crop production. Consequently, most poor are net consumers of grain and other subsistence foods, and are negatively affected by price increases for these products. The poorest households are typically those further disadvantaged by high dependency ratios, ill health and other difficulties. Minority peoples are known to represent a highly disproportionate share of the rural poor. Available evidence does not suggest that women are overrepresented among the poor, though poverty does certainly exacerbate society-wide problems of lower rates of female participation in education, higher relative female infant mortality rates, and higher rates of maternal mortality. (iv) The educational and health status of China's remaining absolute poor is deplorable. At least 50% of the boys in many of China's poorest towns xi and villages and, particularly in some minority areas, nearly 100% of the girls do not: attend school and will not achieve literacy. Infant and maternal mortality rates in very poor counties -- which exceed 10% and 0.3% respectively -- are at least 50% to 100% greater than the national average, and are much greater yet in the poorest townships and villages. Incidence of several infectious and endemic diseases, including tuberculosis and iodine deficiency disorders, is concentrated in poor and remote areas. Roughly 50% of children in households at or below the absolute poverty line are at least mildly malnourished (stunted), and iron, vitamin A, and other micronutrient deficiencies remain a severe problem among the poor. As many as 90% of poor children suffer chronic helminthic infection. (v) Determinants of Trends in Poverty. Broad participation in reform- driven agriculture sector growth played the key role in the tremendous two- thirds reduction in absolute poverty during 1978-84. Rural per capita income grew at an average annual rate of 15% in real terms during this period, and increased a total of more than 130%. The failure to achieve further reductions in poverty during the second half of the 1980s, despite modest agricultural growth and very strong industrial growth, is more difficult to explain. A number of macroeconomic developments stymied efforts to reduce poverty during 1985-90: (i) sharply increased prices for grain and other subsistence goods adversely affected the real incomes of the majority of the rural poor; (ii) rapid growth of the working age population exceeded the expansion of employment opportunities, contributing to a worsening of rural underemployment; and (iii) economic growth was greater in the higher income coastal prcvinces than in the lower income inland northwestern and southwestern provinces. In addition, fiscal decentralization passed an increasing share of the costs of rural social and relief services to local governments. While successful in better off areas, this fiscal reform put pressure on the limited revenues of poor area governments. In the absence of adequate levels of appropriately earmarked funding from higher levels of government, poor area local governments have been unable to support either adequate social services or economic growth. (vi) The key determinant of the stagnation of poverty during 1985-90, however, was the absence of meaningful levels of agricultural growth and rural enterprise development in the upland areas. Many of the rural poor in 1978 resided in less remote and less hilly areas, where increased application of fertilizer, better seed and other modern inputs could bring about rapid productivity gains, and so were able to participate in the rapid agricultural growth of 1978-84. However, the quick reductions of poverty through agricultural growth were largely exhausted by end-1984. Most of the residual poor have remained trapped in more remote upland areas where agricultural productivity gains have proven far more problematic. Measured on a per capita basis, output of grain and subsistence foods in such areas failed to sustain any significant increase during the 1980s. Although the agriculture sector did expand in real terms during 1985-89, increased output of nongrain crops and animal and aquatic products, products which few of the poor either produce or consume in significant quantities, accounted for more than all of the modest growth which did occur. By comparison, the annual growth rate of per capita production of grain and oilseeds -- which are the subsistence crops of xii most immediate importance to the poor -- declined from 2.8% and 14.5% respectively during 1978-84 to -0.4% and -1.6% during 1985-89. (vii) Rural enterprises, an important source of employment in the rural economy as a whole, have developed very slowly in poor areas. In the early 1980s such employment actually decreased in poor areas as the commune system was dismantled and workers could no longer be paid in work points instead of cash. Employment rose only after a 1984 policy initiative supporting private rural enterprise opened opportunities for small, family firms particularly suited to the small market niches available in poor areas. Despite private enterprise gains, by 1990 only 4% of the rural labor force in China's 120 poorest counties had found employment in rural enterprises, in contrast to the 22.1% finding such employment in the nation as a whole. Not only are fewer employed, but rural enterprise wages and profits are lower in poor areas than the rest of the country. Rural enterprise growth will continue to be concentrated in the wealthier areas with inexpensive access to national and international markets. If the poor are to share in that growth, it will be largely through migration to take advantage of employment opportunities where they occur. (viii) Government's Poverty Alleviation Program. The central and provincial governments sustained their strong commitment to poverty alleviation efforts throughout the 1978-90 economic reform period. While continuing pre-existing rural social and relief services, the poverty alleviation strategy adopted during the Seventh Five Year Plan introduced a new emphasis on economic development programs in the poor areas. The Ministry of Civil Affairs provides disaster relief and income maintenance support, and coordinates the distribution of relief grain with the Ministry of Commerce's Grain Bureau. The State Education Commission and the Ministry of Public Health administer some special programs to improve the education and health status of the poor. The Agricultural Bank of China and several other banks offer subsidized loans for poor area development through a variety of funds administered by provincial bank branches and their networks of county and lower level banks. The Regional Office of the State Planning Commission administers a Food-for-Work Program which assists with the building of roads and riverine transport, drinking water systems, irrigation works and other capital construction in poor areas. In addition, each of 27 central ministries and agencies has its own special poor area project and every province has its own specially-funded programs. The Leading Group for the Economic Development of Poor Areas (LGEDPA) was established in 1986 in part to provide greater coherence to these many poverty alleviation initiatives and, in particular, to expedite economic development in the poor areas. Since its establishment, LGEDPA has emerged as the principal advocate of China's rural poor and in now the key agency responsible for coordinating the nation's more than Y4 billion in annual funding for poverty alleviation programs. -Strategies for the 1990s (ix) Achieving further reductions in the number of absolute poor will prove a more difficult challenge than in the early-1980s. However, the transition and significant reduction of absolute poverty, from large numbers of poor spread widely across the countryside in the late-1970s to pockets of xiii poverty in remote resource deficient areas by the mid-1980s, makes it easier and fiscally less burdensome to target increased development and social services assistance to the absolute poor. Most importantly, in order to guarantee a iminimum safety net while improving the productivity of the poor over the longer term, revitalized social services must be integrated with improved agriculture and rural enterprise development programs. In order to realize the poverty alleviation potential of the expected reduction of rural underemployment during the 1990s, it is also essential that all opportunities for expanding out-migration of surplus labor from the poor areas be explored. (x) Economic Reform and Growth. Steady adoption of a range of economic reforms is expected to sustain overall growth of the Chinese economy at about 7% per annum during the 1990s (World Bank, 1992b). While this is undoubtedly essential to further reductions in poverty, the analysis of this report shows that growth alone would not be sufficient to achieve desired levels of poverty reduction. Reducing poverty requires that economic growth be coupled with expanded and new social service and rural development programs directed to the poor. Fortunately, the additional funding demands for these programs come at a time when planned fiscal and price reforms promise to cut the central government deficit and, over the medium term, free resources that will allow government to match with a financial commitment its moral commitment to poverty alleviation. Price reform, including the expected decontrol of distorted raw material and agricultural prices and, in particular, the further liberalization of grain prices, need not adversely affect the real incomes or welfare of either the urban or rural poor. Though the reform of grain prices in recent years was associated with some erosion of transfers to the rural poor, it is unlikely that completing grain price reform would further erode such transfers since the poor now pay the equivalent of market prices for any grain they need purchase (in excess of relief grain provided to them by government relief agencies at no cost). Reform of state-owned enterprise, another priority for the 1990s with important implications for the poor, could be associated with some increase in unemployment in cities where there is a large concentration of state enterprises. However, it should be possible to maintain the welfare of the urban population at current levels without incurring additional social welfare costs by (i) shifting welfare functions from urban enterprises to specialized government agencies and (ii) making adequate provision, in advance, for unemployment insurance. On balance, the completion of the economic reform agenda should prove consistent with, and even help finance, the realization of poverty alleviation objectives. (xi) Poverty Alleviation under the 8th Five Year Plan. Two key poverty alleviation initiatives introduced under the government's Eighth Five Year Plan (1991-95) are (i) the extension and strengthening of assistance to the poorest of the poor residing in the worst physical environments and (ii) the integration of production, education, health, family planning and transport programs into comprehensive local intervbntion packages. The first initiative will be supported by individual programs tailored to the special needs of minorities, communities residing in remote, high altitude, and karst areas, and other disadvantaged groups. Other initiatives include (i) an additional Y500 million in annual support for the roughly 200 counties newly included to the national roster of poor counties, (ii) expansion of China's Food-for-Work Program, with the bulk of incremental assistance directed to construction of xiv terraced agricultural lands, and (iii) increased support for agricultural extension and training and farmers' marketing systems in the poor areas. (xii) Demographic Trends and Labor Mobility. Pronounced demographic changes are expected to enhance employment and income prospects during the 1990s. Driven by fertility trends of the 1960s and 70s, annual growth of the total working age population is projected to decline by more than half from 3.1% during 1978-90 to 1.3% during 1990-95 and then to 1.1% during 1995-2000. Employment growth is also expected to drop sharply to about 1.4% annually during the 1990s. Even if urban enterprise reform induces efficiency gains sufficient to reduce employment growth below these relatively modest expectations, increases in labor demand would still exceed the slow expansion of labor supply. Contrary to government projections of a very sharp increase, rural underemployment should therefore subside somewhat during the 1990s. The early success of family planning measures in urban areas in the 1970s, in particular, has already resulted in a sharp reduction in the number of urban- born entrants to the urban work force. A steady stream of migrants from rural areas will therefore be needed during the 1990s to offset the declining numbers of urban-born work force entrants and replace large numbers of retiring urban workers. (xiii) The greatest potential for the absorption of surplus rural workers is believed to be in the multitude of small and medium sized cities and urban towns. Those urban areas have experienced rapid development in recent years, often fostered by a relaxation of controls on in-migration by rural people. While many urban jobs require a middle school education, construction and service sector jobs often do not. Those jobs, and peri-urban agriculture, provide employment opportunities to the poor, who are typically less well educated. Realizing the potential for improved employment and income prospects for the poor will require maximizing the range of economic activities available to them, both on and off the farm, within and, in particular, outside the poor areas. Efforts to facilitate the out-migration of excess labor from the poor areas could play a key role in translating this potential into reality. (xiv) Recommended Strateuies for the 1990s. Meeting the challenge of poverty reduction in the 1990s will, in addition to overall economic reform, require: (i) investment in the development of human capital, including greater central government funding for education, health and relief services in the poorest areas; (ii) institution building and policy reform, including a strengthening of the institutions responsible for implementing explicit poverty alleviation programs, the establishment of an independent and objective poverty monitoring system, and an improvement in the access of the poor to employment opportunities outside the poor areas; and, (iii) infrastructure development, with poor area development assistance programs revised to include only those agricultural, rural enterprise, xv road and other rural infrastructure development projects which generate reasonable market-determined returns to investment. More specifically, the government should undertake the following initiatives during the 1990s: Increase Funding for Social Services: Now that the incidence of absolute poverty has been reduced to less than 10% of total population, the government is in the position to address the most critical education, health, and relief needs of the poor at acceptable costs. Recent experimental programs to improve access of the poor, including poor minorities and females, to education and health services provide strong evidence that effective interventions exist at a modest per capita cost. Annualized incremental funding requirements for programs to improve education and health services and status in the poor areas are estimated to be about Y2 billion. With this, it should be possible to universalize slx-year primary education for 15 million poor children and to reduce currently excessive levels of infant, child and maternal morbidity and mortality in the poor areas by at least one-third. Failure to adopt such programs will preclude realization of the government's goals of universal nine-year primary education and "Health- For-All in 2000." Incremental funding for the programs, which must be provided by the central and provincial governments, could be partially defrayed through efficiency gains in planned expenditure on education during the 1990s. Even at full cost, however, incremental funding requirements would represent only about 0.5% of current total government expenditure and only 4% of the government's fiscal losses on consumer food subsidies it now provides almost exclusively to the well-off registered urban population; ' Stren then Poverty Alleviation Aaencies: The government should recognize that sustaining the effectiveness of poverty reduction efforts during the 1990s and beyond will require a permanent and much stronger institutional structure. Accordingly, the permanent status of the LGEDPA system should be formalized and its central, provincial and local offices expanded to include staff trained in applied poverty research and assessment, project appraisal, supervision and monitoring, and, consistent with its mandate to coordinate all poverty alleviation programs, education and health. The LGEDPA should also take a more I/ Although education, health and welfare standards are nationally mandated, social services rely primarily on local finance. This situation leaves program availability and quality much lower in poor than well-off rural areas. Relief programs in the poorest villages are capable of sustaining life, but only at the most minimal levels -- principally or solely through the provision of relief grain. Additional resources must be made available if adequate education, health, and other social consumption by the poorest and most vulnerable groups is to occur. Some of those resources can be found through efficiencies to be gained in, or reorienting of, existing programs, but the bulk must come from increased central government transfers to poor areas. xvi active role in facilitating the movement of surplus labor out of the poor areas, and new staff should be appointed to research and implement meanB to this end; Upgrade Poverty Monitoring: Currently, the LGEDPA has principal responsibility for both monitoring absolute poverty at the national and provincial levels and coordinating the implementation of programs to reduce poverty. This conflict of interest is associated with an optimistic bias in the LGEDPA's estimates of the incidence of poverty. More importantly, the LGEDPA's assessment of the location of poverty indicates that nearly two-thirds of the absolute poor reside in southwestern China -- a finding which is inconsistent with more reliable information for the late-1980s from the State Statistical Bureau's independent large-scale rural household income and expenditure surveys. The LGEDPA's assessment, which is based on Ministry of Agriculture county-average rural income data, plays a key role in determining the focus of the nation's poverty alleviation effort and targeting of government assistance. 2 It is essential that an independent and objective poverty monitoring system be established and the true location of the poor be accurately determined as soon as possible. Fortunately, the State Statistical Bureau has both the necessary data and skills to undertake objective monitoring of the incidence of poverty; Improve Labor Mobility: obstacles to rural labor mobility, including resistance by county and municipal governments, inadequate availability of labor market information, and social and cultural prohibitions, have heretofore limited intercounty and interprovincial out-migration of surplus labor -- particularly from the poor areas. County governments have used administrative and punitive economic measures to impede the inflow of labor from lower-income areas in order to maximize local employment and incomes. Similarly, the household registration system and the urban grain ration system have been used to limit rural-to-urban migration in order to keep urban unemployment below planned targets. These obstacles notwithstanding, organized and unorganized labor migration did increase somewhat during the 1980s, and demographic trends and government policy favor a continuation of that increase during the 1990s. Favorable experience with provincial labor export programs during the 1980s has documented the feasibility of improving labor mobility, particularly from the poor areas. The State Council also recently approved 22 experimental labor export and mobility projects, lending high level support to efforts to expand and improve rural labor mobility. Consistent with overall plans to decontrol the labor market and remove remaining barriers to labor mobility, the LGEDPA should take an active role in facilitating a 'market friendly* increase in out- migration of surplus labor from the poor areas to better off rural and urban areas; and, 2/ The State Statistical Bureau sample survey data, by comparison, indicate that at least half of the poor reside in northwestern and northern China. xvii Foster Rural Enterprise and Aaricultural Development in the Poor Areas: Small--scale private enterprise (PE) appears to be best suited to the undercdeveloped markets of the poor areas. Government initiatives to augment the development and poverty alleviation impact of rural enterprise in the poor areas have included the introduction of appropriate technologies, the subcontracting of simple manufacturing or production activities to poor households and villages, and the provision of subsidized credit to enterprises which agree to employ a specified minimum number of poor. The recently enacted Regulations on Rural Collective Enterprises, which seek to limit local government's excessive extractions from collective enterprises and protect the rights and interests of such enterprise, are also of particular importance to creating a favorable environment for poor area rural enterprise development. 3 The sheer numbers of the rural poor make it inevitable that most will continue to reside in remote mountainous lands during the 1990s and be forced to rely primarily upon agricultural production on those lands for their subsistence. Therefore, every effort must be made to take advantage of opportunities to augment productivity in poor area agriculture. Though they are generally not easy to find nor necessarily suitable for large- scale government support, real opportunities for improving agricultural productivity and incomes in the upland areas do exist., While increasing the incomes and productivity of poor families and poor areas is a laudable objective, the government must recognize the limits to the effectiveness and efficiency of assisting the poor through rural enterprise and agricultural development projects within the poor areas. (xv) Agenda for International Assistance. International assistance can play an important role in China's poverty alleviation program through support for operatLonal research and rural development programs. Operational research priorities,, to both evaluate and improve the effectiveness of China's poverty alleviation program, include (i) the establishment of an independent poverty monitoring system, (ii) the development of effective programs of education and health interventions for the poor areas, (iii) an examination of measures to augment out-migration from the poor areas, (iv) a careful assessment of explicit poverty alleviation programs, and (v) an accurate appraisal of the size, composition and well-being of the growing unregistered urban and rural migrant population (para ii). First, the poverty monitoring system (para xiv), once established, should be extended to include education, health and other indicators of welfare. Monitoring should also be extended to cover the unregistered urban population and migrant laborers more generally. It is essential that national and provincial poverty assessments be validated through detailed field checks in a number of villages around the country on a regular basis. Second, the most efficient and effective programs of education, health and nutrition interventions for the poor areas must be identified through careful experimentation. Extension to progressively wider segments of the poor population would be undertaken as these intervention packages are refined on the basis of demonstrated cost-effectiveness and 3/ Similar regulations covering private enterprises have now been separately issued. xviii popular acceptance. Third, the remaining obstacles to poor area labor mobility must be assessed and the effectiveness of the 22 experimental labor export and mobility projects evaluated (para xiv). It is also important that the 1986 Chinese Academy of Social Sciences survey (of 222 villages and 11 provinces), which provides the best documentation of rural labor transformation and mobility, be updated and expanded. Fourth, there is a need to undertake a detailed examination of the poverty alleviation impact and economic viability of China's various poverty alleviation programs. Initially, attention could be directed to the State Planning Commission's Food-for-Work Program and an evaluation of relative benefits to investment in road and riverine transport, drinking water and sanitation systems, small- scale irrigation and soil conservation works, primary schools and village health stations, and other capital construction (para viii). Last, there is an urgent need for an improved understanding of the well-being of the migrant population. Though most migrants may enjoy improved living standards, there are signs that a significant number are not faring well. (xvi) In balance with the priority attached to combating poverty, international assistance for rural development programs should be focussed on the inland regions where most of China's poor reside. Until now, however, relatively little international assistance has reached the inland poor. Greater efforts could therefore be made to target international assistance to poor provinces and, while recognizing the inherent difficulties and costs to reaching small communities situated in distant areas, to poor counties and even poor townships and villages. Assistance for education, health and other social sector development programs should be concentrated primarily in the poorer rural areas where achieving and maintaining minimal welfare standards is most at risk. A substantial share of agricultural development programs should be directed to poor upland areas to resolve the interrelated problems of low productivity and environmental degradation. -Reader's Guide (xvii) This report provides an analysis of the incidence of absolute poverty during the 1978-90 reform period (Chapter 2 and Annex 1) and proposes macroeconomic and sectoral strategies (consistent with the overall reform agenda) necessary to further reductions of poverty in the 1990s. The macroeconomic developments which helped determine the reduction of poverty during 1978-84, and its stagnation thereafter, are reviewed in Chapter 1. Trends in labor supply and demand during 1978-90 and expected during the 1990s are examined in Chapter 3. Strategies for augmenting labor mobility, rural enterprise development, and agricultural productivity in the uplands are also explored. Education, health and relief services in the poor areas are discussed in Chapter 4. Financial support for these services during the 1980s is analyzed, and a program to bolster education and health in the poor areas during the 1990s is proposed. The government's poverty alleviation programs during the 1980s and under the Eighth Five Year Plan are described and evaluated in Chapter 5. The report is supported by the following five working papers (available upon request): "Demographic Aspects of Poverty," "A Program to Improve Health Services in the Poor Areas," "Economic Analysis of the Loess Plateau Land Rehabilitation Program," "Poverty in Southwest China" and "Rural Poverty: Characteristics, Causes and Policies in Baise Prefecture." 1. POVERTY ALLEVIATION AND MACROECONOMIC TRENDS IN THE 1980s A. Absolute Poverty 1.01 Overview. Both as an accompaniment of overall economic growth and also through a strong commitment to improving standards of education, health and nutrition for all, China has compiled an impressive record in reducing absolute poverty over the last four decades. Perhaps the most telling indicator of the improvement in the overall well-being of the Chinese population is the increase in expectation of life at birth, which nearly doubled from 34 years in the early-1950s to 67 years at present. Other indicators of social development, reported in Table 1.1, also document tremendous improvement. The crude death rate and infant mortality both declined by about three-quarters since the early-1950s, and illiteracy is estimated to have dropped from 80% of the adult population in the early-1950s to about 30% at present. Average per capita availability of food energy increased by nearly 40%, from about 90% of food energy requirements in the Table 1.1: Selected Indicators of Social Develooment: China and Other Countries Infant Expectation Crude MortaLity Per Capita of Life Death Rate AduLt AvaiLabiLity GNP at Birth Rate \a (under age 1) Illiteracy of Food Energy Per Capita Country/Region years per 1000 per 1000 % caLories/day 1988 USS China: 1988 67 8 58 31 2637 330 1950-55 34 31 236 80 1894 na India: 1988 56 12 91 57 2104 340 Indonesia: 1988 59 11 79 26 2631 440 Brazil: 1988 64 9 71 22 2703 2,160 1988 Average: Low Income Countries 60 10 72 44 2442 \b 320 MiddLe Income Countries 66 8 52 26 1,930 Sources: World Bank (1990a) for aduLt illiteracy, GNP, and averages for Low and middle income countries, FAO for food energy, Banister (1992) country-specific data for 1988, and World Bank (1984 and 1991a) and Piazza for figures for China for 1950-55. \a Figures for China, India, Indonesia and Brazil for 1988 are age standardized. \b Average for alL devetoping countries (incLuding China). 2 early-1950s to 115% by the late-1980s. These indicators of social development in China compare favorably with the estimates shown in Table 1.1 for India, Indonesia, Brazil, and the averages for other low income and middle income countries. 1.02 Urban Poverty. L The Chinese government has limited urban poverty to extremely low levels through full employment linked to narrow ranges of earned income, price controls on basic foodstuffs, and near universal access to education and health care. Full employment has been maintained by severely restricting migration to urban areas, a restriction strengthened until the mid-1980s by the lack of a rental market for housing and the requirement that grain coupons be used when purchasing either grain or restaurant meals. The urban food ration system has insured that China's urban population enjoys preferential access to staple foods, especially grain and vegetable oil, at below market prices. Urban enrollment ratios at the primary and secondary school levels have risen faster and are now higher than those of most other developing countries. Widespread availability of primary health services, provided at little or no cost, brought the most serious contagious and parasitic diseases under control in urban areas by the early-1960s. At present, the urban population exhibits few signs of absolute poverty and enjoys more than twice the per capita income, significantly better access to education and health services, and greater levels of food intake than do rural inhabitants. 1.03 Rural Poverty. In rural areas, the redistribution of productive assets, adoption of a basic human needs program, and greatly increased food production have been of key importance in reducing absolute poverty. In the early-1950s, the National Land Reform redistributed large amounts of arable land, livestock, and other farm assets, and established the basis for significantly increased income among (previously) landless laborers and other elements of the rural poor. The basic human needs program adopted in the 1950s sought to guarantee floor levels of food consumption, income and health services adequate to maintain minimum levels of physical well-being. The objective of the rural relief component of the basic human needs program was to prevent outbreaks of starvation or severe malnutrition, and to guarantee the welfare of the elderly. Rural relief was for the most part funded locally, and regional or state funding was made available only in cases where local poverty was so severe and widespread that it precluded the possibility of local self-reliance. The rural primary health care component of the basic human needs program, which has included support for childhood immunization, accessible primary health care, infectious disease control, sanitation and safe water supply, became increasingly effective beginning in the late-1950s 1/ In this report, China's urban population is defined to include only (i) the registered population of city and town neighborhood committees and (ii) the unregistered long term (that is, one year or longer) residents of those neighborhood committees (para 1.19). The "floating" population of unregistered migrants is considered part of the rural population for the purpose of estimating the incidence of absolute poverty. 3 and contr:ibuted to significant reductions in morbidity and mortality during the 1960s and 70s. 1.04 The positive impact of these rural programs has been complemented by large increases in national average food availability and, especially in the grain deficient provinces, by increased grain production. Increased agricultural production helped push up average per capita daily food energy and protein availability by 35% and 25% respectively, from less than 1900 kcal of food energy and 53 grams of protein in 1950-52 to 2570 kcal and 66 grams in 1980-82. An analysis of provincial per capita grain production shows that seven provinces (Hebei, Henan, Jiangsu, Liaoning, Shandong, Shanxi, and Qinghai), with a total population of 200 million, were grain deficient (less than 275 kg of unprocessed grain production per capita) in 1953/57. By comparison, only five provinces (Gansu, Guizhou, Qinghai, Tibet, and Yunnan), with a population of 85 million, were grain deficient in 1979/82. Overall, the number of people residing in grain deficient provinces declined from 35% to less than 10% of total population (Piazza). The improvement in the physical well-being of the rural population, however, was punctuated by severe dislocations. The most striking failure occurred during the agricultural crisis of 1960-61, which brought about an estimated loss of life of between 16 to 30 million excess deaths (Coale and Ashton. et al). These serious difficulties notwithstanding, it is certain that China's rural population in the early-1980s enjoyed much higher standards of health and nutrition than they did in the early-1950s. Overall, China's efforts to reduce the number of poor and to ameliorate the most extreme manifestations of residual poverty must be judged to have been highly successful. 1.05 Poverty During 1978-90. This report's estimates of the incidence of absolute poverty in China, which are presented in Table 1.2 and detailed in Annex 1, show that the number of poor declined sharply from about 265 million in 1978 to 97 million in 1985, or from about 28% to less than 10% of total population. However, the reduction in poverty was not sustained during the second half of the 1980s -- the number of poor stagnated between 86 to 103 million during 1985-90 and the proportion of total population living in absolute poverty remained roughly constant. These temporal trends in the incidence of absolute poverty have been matched by trends in per capita availability of grain and average levels of nutrition. Availability of grain increased steadily from 305 kg per capita in 1978 to 401 kg in 1984, then dipped to between 359 kg to 385 kg during 1985-89 before recovering to 400 kg in 1990. v The overall reduction in poverty since 1978 was also matched by appreciable improvements in expectation of life and reductions in mortality (since China had achieved fairly low levels of mortality by the late-1970s, continued sharp declines in the 1980s could not be expected). Data from the 1990 census indicate that survival chances have either improved or stayed the same since 1981 for males and females at almost all ages in rural as well as urban areas. The 1988 nationwide fertility survey suggests that infant mortality for both sexes was approximately constant from 1977 to 1987, though 2/ Figures are calculated as domestic production net of international grain imports and exports. Table 1.2: Macroeconomic Indicators and Estimated Incidence of Absolute Poverty Year 1978 1979 1980 1981 1982 1983 1984 1985 1986 1987 1988 1989 1990 GNP: 1978=100 100 108 116 121 132 145 167 188 203 226 251 261 274 Annual Growth Rate (%) 7.6% 7.9% 4.4% 8.8% 10.4% 14.7% 12.8% 8.1% 10.9% 11.0% 4.0% 5.2% Per Capita Grain Availabilitv (kg): 326 354 338 340 367 391 401 359 368 385 368 378 400 Retail Prices CX annual change): NationaL Average 0.7% 2.0% 6.0% 2.4% 1.9% 1.5% 2.8% 8.8% 6.0% 7.3% 18.5% 17.8% 2.1% Urban (all food) 2.5% 1.9% 8.1% 2.7% 2.1% 3.7% 4.0% 16.5% 7.2% 12.0% 25.2% 14.4% -1.2% (grain) 2.0% -0.5% 0.2% 0.7% -0.8% 2.1% 0.0% 3.3% 4.1% 6.2% 14.1% 16.9% -6.5% Rural (all food) 0.3% 3.8% 7.1% 3.0% 4.4% 0.6% 1.4% 12.5% 7.5% 8.4% 20.9% 18.0% 1.7% (grain) 0.2% 5.7% 7.1% 1.1% 3.5% 0.2% -0.4% 17.6% 13.0% 6.1% 14.0% 25.0% -3.3% Per Capita Income (Yuan): Urban (current Yuan) -- -- -- 500 535 573 660 749 910 1012 1192 1388 1523 Rural (current Yuan) 134 160 191 223 270 310 355 398 424 463 545 602 630 Urban (1978 Yuan) \a -- -- -- 446 467 490 550 557 633 647 632 632 685 Rural (1978 Yuan) \_ 134 157 180 202 241 274 311 324 325 335 336 313 319 Emt)l2yMet (million people): Working Age Population \E 485 -- -- -- 567 583 602 621 641 656 670 684 697 Total Employment 402 410 424 437 453 464 482 499 513 528 543 553 567 Urban Economy \d 95 100 105 111 114 117 122 128 133 138 143 144 147 Rural Economy \e 306 310 318 327 339 347 360 371 380 390 401 409 420 -Agriculture \_ 275 278 283 290 301 304 301 304 305 309 315 324 333 -Nonagriculture \ 31 32 35 37 38 44 59 67 75 81 86 85 87 (-Rural Enterprise \f 28 29 30 30 31 32 52 70 79 88 95 94 93) Estimated Incidence of Absolute Poverty: Kegistered Urban (million peopLe) -- -- -- 4 2 1 1 1 1 1 1 1 1 (X of urban population) -- -- -- 1.9% 0.9% 0.6% 0.3% 0.4% 0.2% 0.2% 0.2% 0.3% 0.4% Rural (million people) 260 -- 218 194 140 123 89 96 97 91 86 103 97 (X of rural population) 33.0% -- 27.6% 24.3% 17.4% 15.2% 11.0% 11.9% 11.9% 11.1% 10.4% 12.3% 11.5% Total (million people) -- -- -- 198 142 125 90 97 97 91 86 103 98 (X of total population) -- -- -- 19.8% 13.9% 12.0% 8.6% 9.2% 9.0% 8.3% 7.8% 9.2% 8.6% Source: SSB (1991a) for GNP, prices, income and employment, World Bank (1991b) for grain, and Annex 1 for poverty estimates. \a DeTlited by the urban cost of living index. %i Deflated by the rural retail consumer price index. Xc Officially defined as women ages 16 to 54 and men ages 16 to 59 (excluding military, prisoners and the disabled). \a Includes urban staff and workers and urban individual laborers in urban and (while they retain their urban residency registration) rural areas. \e Excludes urban staff and workers employed in rural agriculture and rural nonagriculture. \T Rural enterprise employment, including some part time workers. A definitional change, to include all kinds of rural enterprise, explains part of the sharp increase in employment in 1984. 5 the rising sex ratios observed for cohorts born in the 1980s strongly suggest -- to the contrary -- that female infant mortality has been rising (Banister, 1992). 1.06 There are important differences in the incidence of poverty between sectors and regions. Absolute poverty amongst the urban population, which was already at extremely low levels by the late-1970s, is estimated to have declined from about 1.9% of the registered urban population in 1981 to only 0.3% by 1984, or from about 4 million to only 1 million urban residents, and to have remained at about that level throughout 1984-90. The number of rural poor is estimated to have declined from 260 million in 1978 to about 100 million by 1985, or from 33% to 12% of total rural population. The incidence of rural poverty then remained roughly constant at between 10.4% to 12.3% of the rural population during 1985-90, ranging between 86 to 103 million poor. Most of China's remaining absolute poverty is now concentrated in a number of resource-poor rural areas, primarily in the northern, northwestern and southwestern provinces. With the exception of a growing residual of households, disadvantaged by physical or mental disabilities, age structure, ill health or other factors, who have not fared well in the transition to a more market oriented economy, absolute poverty has been largely eliminated from better endowed rural areas. B. Impact of Macroeconomic Trends -Economic Reform and Growth 1.07 Economic Reforms. Leveraging investments in agriculture and industry made during the pre-reform period, economic reforms initiated in the late-1970s led to rapid economic growth and structural change during the 1980s. X Rural economic reforms, including the adoption of the production responsibility system (PRS), the dismantling of the commune system, agricultural product price increases and market liberalization, were key elements of the first stage of economic reform during 1978-84. Industrial reforms initiated in 1983, including the adoption of profitability and retained earnings as principal management criteria and the introduction of an enterprise profit tax, allowed enterprises more latitude in the disposition of resources and permitted greater retention of profits. Enterprise reform was furthered in late-1984 with the introduction of management contracting and increased autonomy for rural enterprise. Increased investment demand was accommodaLted through expanded credit as newly established specialized state banks began making long-term enterprise investment loans. 1.08 Growth. These reforms helped China sustain high rates of real economic growth throughout most of the 1978-90 reform period. Official figures i(Table 1.2) indicate that real GNP increased by 174%, at an average 3/ Economic performance and reforms during 1978-88 are discussed in detail in World Bank (1990b). 6 annual rate of nearly 9%, during 1978-90. As summarized below, significant mid-decade shifts in sectoral growth trends underlie this aggregate growth: Averaae Annual Growth Rates: GNP and Agriculture \a Gross Value of Agricultural GNP Output Per Capita Production ------------------------ -------------- (physical output) Agri- Non- --------------------- Total culture Industry Crops Crops Grain Oilseeds Meat ---------------------------------------------------------------__------------ 1978-89 9.2% 5.4% 10.7% 4.5% 9.4% 1.3% 7.9% 7.6% 1978-84 9.1% 6.6% 9.7% 6.2% 9.6% 2.8% 14.5% 8.2% 1985-89 8.9% 3.5% 11.2% 2.4% 8.2% -0.4% -1.6% 6.1% Source: SSB (1990a). \a Average annual real growth based on three year averages and constant 1980 prices. Spurred by the rural economic reforms, the agriculture sector enjoyed real average annual growth of 6.6% during 1978-84. Average annual growth in the sector declined to 3.5% during 1985-89, or 1.9% annual growth in per capita terms. Within agriculture, the growth rate of the gross value of crop production slowed from 6.2% during 1978-84 to 2.4% during 1985-89. The growth rate of the gross value of non-crop output (including animal husbandry, forestry, fisheries and sideline production), however, declined only modestly from 9.6% to 8.2% during these same periods. The impact of industrial reform, by comparison, was felt several years after that of the rural reforms. Average annual industrial growth accelerated from 7.8% during 1978-83 to about 14% during 1983-88. Offsetting the slowdown in agriculture sector growth, this strong industrial growth helped drive a surge in GNP growth from 7.7% during 1978-83 to more than 11% annually during 1983-88. 1.09 Broad participation in reform-driven agriculture sector growth played the key role in the tremendous two-thirds reduction in absolute poverty during 1978-84. Rural per capita income grew at an average annual rate of 15% in real terms during this period, and increased by a total of more than 130%. In contrast, modest, but still positive, agricultural growth in combination with strong industrial growth was not associated with any significant further reductions in poverty during the second half of the 1980s. The inability to reduce poverty during 1985-90 is in part explained by mid-decade changes in subsectoral growth in agriculture. The average annual growth rate of per capita production of grain and oilseeds -- which are the subsistence crops of most immediate importance to the poor -- declined from 2.8% and 14.5% respectively during 1978-84 to -0.4% and -1.6% during 1985-89 (see figures in para. 1.08). Increased output of other products which few of the poor either produce or consume in significant quantities, including nongrain crops and 7 animal and aquatic products, accounted for more than all of net agricultural output growt:h during 1985-89. A number of other macroeconomic developments, discussed in the remaining sections of this chapter, also help explain the failure to reduce poverty despite strong overall economic growth during 1985- 90: (i) sharply increased prices for grain and other subsistence goods adversely affected the real incomes of the rural poor; (ii) rapid growth of the working age population exceeded the expansion of employment opportunities, contributing to a worsening of rural underemployment; and (iii) economic growth was 4greater in the higher income coastal provinces than in the lower income inland northwestern and southwestern provinces. In addition, fiscal decentralization passed an increasing share of the costs of rural social and relief services to local governments. While successful in better off areas, fiscal reform pressured the limited revenues of poor area governments. In the absence of meaningful levels of funding from higher levels of government, poor area local governments have been unable to support either adequate social services or economic growth. 1.10 An equally important determinant of the stagnation of poverty during 1985-90, discussed in detail in Chapter 3, is that the quick and easy reductions of poverty through agricultural growth were exhausted by end-1984. Many of the rural poor in 1978 resided in less remote and less hilly areas, where increased application of fertilizer, better seed and other modern inputs could bring about rapid productivity gains, and so were able to participate in the rapid agricultural growth of 1978-84. By 1985, however, most of the remaining poor were trapped in more remote upland areas where agricultural productivity gains are much more problematic. -Prices andl Income 1.11 Though achieving unprecedented rates of national income growth throughout the 1980s, policy makers failed to adequately guard against inflation. Excessive credit expansion, and the failure to sufficiently augment supply responsiveness through either national market integration or greater competition, contributed to price spirals first in 1985 and then again beginning in late-1987. As shown in Table 1.2, the growth rate of national average retail prices increased from an annual average of less than 3% during 1979-84 to nearly 9% in 1985, declined moderately in 1986-87, and then surged to about 18% in 1988 and 1989. The acceleration of inflation during 1985-89 adversely affected the rural poor, both because it included sharp increases in the prices of grain and other subsistence foods and by necessitating adoption of anti-inflationary austerity programs which brought about a retrenchment of off-farm jobs in rural enterprises and the urban economy during 1988-89. 1.12 Prices. Price changes during the 1980s significantly affected the poor, first by improving their terms of trade through 1984 and then by worsening them in critical dimensions through the rest of the decade. As shown below, the growth rate of national average retail prices, which averaged about 7% per annum during 1978-89, increased from an annual average of less than 3% during 1979-84 to more than 12% during 1985-89: 8 Price Indices: Averaqe Annual Change (%) Rural Urban Retail \a Rural Retail \a Free Market All ------------------ ------------------ ------------ Farm Retail Total Food Grain Total Food Grain Total Grain Procurement -----------------------------------------------------------__----------------__---- 1978-89 6.7% 7.4% 8.7% 4.1% 6.2% 7.8% 8.2% 7.8% 5.6% 9.9% 1978-84 2.8% 3.1% 3.7% 0.3% 2.3% 3.4% 2.8% 1.2% -4.9% 7.4% 1985-89 11.6% 13.1% 14.5% 10.2% 11.9% 13.5% 14.3% 16.6% 24.8% 13.9% ---------------------------------- Source: SSB (1990a). \a Consumer retail prices. Overall, the increase in urban consumer prices modestly exceeded that in rural areas. Food prices also increased more rapidly in urban areas. However, urban consumers have been insulated by grain subsidies (para 1.02), and retail grain prices in urban areas have increased at a significantly lesser rate than in rural areas. 1.13 Prior to 1985, the rural population benefitted from average annual increases in farm procurement prices of more than 7%, or more than twice the growth rate of rural consumer prices during 1978-84. Prices in the rural free markets, which are "prices of last resort" for the rural poor, increased at the very modest annual rate of only 1% during 1978-84 -- and grain prices in the rural free markets declined by about 5% annually during the period. The bumper grain harvest of 1984 depressed free market prices for grain and other staple foods to their lowest level of the decade. During the second half of the 1980s, however, rural consumer prices increased at a rate more closely approximating the growth of farm procurement prices. Increased demand for feed grain, in the context of reduced total availability of grain {para 1.05), tightened supplies, and rural grain prices increased at a rate greater than overall rural inflation during 1985-89 -- and jumped 18% and 25% in 1985 and 1989 respectively. The growth rate of the rural free market price of grain, in particular, increased at an average annual rate of 25% during 1985-89 -- or more than twice the rate of overall rural inflation. Most of the rural poor are net purchasers of food grain, and were adversely affected by increased prices of staple foods. 4 The rural poor suffered an additional price shock in 1985, furthermore, when rural "resales" of grain were increased from the old quota price to the new contract price (World Bank, 1990d). Those poor not entitled to receive sufficient quantities of free relief grain paid an 4/ However, a significant proportion of the poor are surplus producers of grain. Zhang (1992), for example, has shown that the poor of Baise Prefecture in Guangxi are net producers of grain and hence benefitted from the increase in free market grain prices. 9 additional 35% for resale grain beginning in 1985. 5 Further increases in contract procurement prices for grain in 1989 and 1992 were also associated with substantial increases in the planned sales prices paid by some of the poor for resale grain. @ 1.14 Income. While there is no doubt that rapid expansion of GNP raised living standards throughout the country, it is also evident that the increase in retail prices during the second half of the 1980s slowed, and even brought to a halt, increases in average per capita income. As shown in Table 1.2, nominal average per capita income data, from the State Statistical Bureau's (SSB) annual urban and rural household income and expenditure surveys, indicate steady increases in income levels throughout the reform period. However, deflating by the urban cost of living and the rural retail consumer price index, the official data show that real: -urban income increased a total of 45% during 1981-87, stagnated during 1988-89, and then increased in 1990 to about 6% above the 1987 level; and, -rural income increased a total of more than 130% during 1978-84, at an average annual rate of 15%, and then increased by a total of less than 3% during 1985-90. A -Employment Levels and Demoaraphic Chanae 1.15 Employment. Underemployment has been a persistent problem in rural and, to a lesser extent, urban China. Previously camouflaged by the commune administrative structure and the work-point system, the full extent of the rural labor surplus became apparent with the introduction of the PRS in the early-1980s. A number of sources have reported that roughly one-third of the rural work force was underemployed during the 1980s, with absolute estimates ranging between 70 million to 156 million underemployed rural workers (Taylor, 5./ The majority of rural resales are provided as disaster relief and to farmers specializing in the production of cash crops and other nongrain commodities. About 7 million tons of grain are provided annually in disaster relief and to the chronic absolute poor at no cost or at the contract price (para 4.55 and 4.64). 6/ The ]L992 increase brought contract grain procurement prices, and hence rural resale prices, closely in line with both domestic free market and world market grain prices. Consequently, any further revision or decontrol of grain prices should at most have a minimal adverse effect on the rural poor since they now pay the equivalent of market prices for any grain purchased in excess of the relief grain they receive from the government at no cost. 7/ Using the implicit GNP deflator (which increases less than the rural retail consumer price index during 1978-90), real rural income increases each year 1978-89 before declining modestly in 1990. Nevertheless, the mid-decade shift in real growth of rural income remains dramatic -- using the implicit GNP deflator, average annual real rural per capita income growth declined from about 15% during 1978-84 to 2% during 1984-90. 10 1988). Similarly, though the government's full employment policies have kept open unemployment in urban areas to very low levels, disguised underemployment is widespread, perhaps amounting to as much as 20% of the work force of urban state enterprises (World Bank, 1992a). 1.16 Official employment and working age population figures, summarized in Table 1.2, do not evidence growing underemployment during the 1980s. Total employment reportedly increased at an average annual rate of 2.9%, from 402 million in 1978 to 567 million in 1990, and remained nearly constant as a share of the total working age population. Employment in the urban economy increased at an average annual rate of 3.7% from 95 million to 147 million during 1978-90. Employment in the rural economy increased at the more moderate average annual rate of 2.7% during 1978-90, from 306 million to 420 million. Within the rural economy, employment growth in agriculture, increasing at an average annual rate of 1.6%, was significantly below that in nonagriculture, which increased at an average annual rate of 9.0%. The official figures show that, of the 114 million "new jobs" created in the rural economy during 1978-90, 58 million were in agriculture and 56 million were in the nonagricultural pursuits of industry and agro-processing, construction, transport, commerce and other activities. A careful review of farm labor norms and the scale of farm operations (para's 3.39 and 3.40), however, indicates that agricultural labor requirements increased by at most 10% during 1978-90, or less than half the increase suggested by the official figures. If true, even the dynamic growth of urban and rural off-farm employment would not have been sufficient to prevent growing underemployment of the rural working age population during the 1980s. Assuming agricultural employment increased by only 10% during 1978-90, total employment declined from 83% of the total working age population in 1978 to 77% in 1990. 1.17 The sharp increase in inflation in late-1987 (para 1.11) prompted the government to introduce strong deflationary measures beginning in late- 1988 including (i) a decrease in state investment in services, office construction, processing industries and other (perceived) low priority sectors, (ii) contractionary monetary policy, in part through selective tightening of administratively allocated credit, and (iii) a stiffening of direct controls on prices and marketing. 81 These measures were instrumental in reducing the annualized rate of inflation from 26% in December 1988 to under 1% in the first quarter of 1990. However, they also resulted in a decline in industrial growth rates beginning in the final quarter of 1988 and negative growth during the final quarter of 1989. Though industrial growth recovered to moderately positive levels by the third quarter of 1990, significant declines in rural nonagricultural employment have been reported.9' 8/ Economic developments in late-1988 and 1989 are reviewed in World Bank (1990c). 9/ An alternative measure, the rural enterprise employment series reported in parentheses in Table 1.2, indicates that employment in rural off-farm activities increased at the more rapid pace of 13% annually during 1978-88 before declining by about 2% during 1989-90. 11 The government specifically targeted rural enterprises (RE), which were believed to be a major source of excess demand, for selective credit tightening. Though the actual reduction in credit to REs was not as great as originally intended, one report has suggested that as many as 8 million people lost their nonagricultural jobs and were "sent back to the land" in 1989. t° The official data reported in Table 1.2, on the other hand, indicate (i) a less severe loss of about 1 million rural nonagricultural jobs in 1989, and (ii) that this loss was more than offset by 9 million new jobs in agriculture and 1 million new jobs in the urban economy during that year. 1.18 Dermoaraphic Change. Demographic changes during the reform period include a significant increase in the urban population's share of total population, substantial rural-to-rural and rural-to-urban migration, and a 30% decline in the growth rate of the total working age population during the late-1980s. The SSB recently released a new disaggregation of total population by residence which shows that China's urban population increased from 17.9% of total population in 1978 to 26.4% in 1990 (SSB, 1991a). These new figures correspond to the population of city and town neighborhood committees, and provide a consistent estimate of the urban population which is reasonably congruous with internationally accepted definitions. I'' The disaggregation also accords almost exactly with the definitions of urban and rural used in the SSB household income and expenditure surveys, and so provides the precise figures for total urban and rural population used in this report's estimates of the incidence of absolute poverty. Since the incidence of poverty amongst the urban population is much less than that of the rural population, the sharp increase in the urban population's share of total population accounts for about 10% of the decline in the number of absolute poor during 1981-89. Iv 1.19 Rural-to-urban migration, which amounted to at least 20 million people during the 1980s, has accounted for a significant share of the increase 10/ China Daily, January 6, 1990. Development of China's REs during 1986-90 is reviewed in detail in Odv. 11/ Prior to the release of these newly compiled figures, official SSB statistics had indicated that the urban population had increased from 18% of total population in 1978 to 52% in 1989 (see 1990 edition of SSB, 1991a). The previous statistics were widely considered to overstate the true increase in urban's share, and led to considerable speculation and confusion over the true size of the urban population. 12/ Holdinig urban's share of total population constant at the 1981 level, the number of absolute poor in 1990 would have been 106 million instead of the 98 million reported in Table 1.2. The increase in urban's share also explains part of the difference between the real growth rates of per capita GNP (98%) and urban (54%) and rural (58%) per capita income during 1981-90. Accounting for the increase in urban's share of total population and the fact that average per capita urban income is more than twice that of rural areas, the weighted average of urban and rural real per capita income increased nearly 70% during 1981-90. 12 in total urban population. People living outside of the place where their residence records are maintained are known as China's "floating population," and remain outside their place of residence recordation on either a "long- term" or seasonal basis. The 1990 census indicates that in July, 1990 the long-term floating population numbered about 30 million, of which at least 10 million are presumed to have been living in urban areas. Available evidence suggests that the size of the seasonal floating population at least equals that of the long-term floating population (paras 2.18 and 2.19). Unfortunately, very little is known about the income levels and well-being of the members of the floating population. It is certainly possible that some significant share of this growing segment of the population is absolutely poor, and it is known, for example, that rates of neonatal tetanus are significantly greater among the infants of the floating population than the rates for other segments of the population (UNICEF). 1.20 In contrast to the increase in the annual population growth rate from less than 1.3% during 1978-80 to more than 1.5% during 1985-89, the growth rate of the total working age population declined from 3.6% during 1978-85 to 2.3% during 1985-90 (para 3.05). A delayed response to fertility trends of the 1970s, the absolute annual increase in the total working age population declined from about 19 million during the first half of the 1980s to about 14 million during the second half. Growing underemployment of the rural working age population would have been considerably worse during the late-1980s without this decline in the growth rate of the total working age population. -Regional Growth Trends 1.21 Coastal Development. Adoption of the "open coastal strategy" at the outset of the reform period provided special privileges, incentives, and investment allocations to coastal provinces and counties. 13/ The relationship between the coastal and interior provinces was formalized in the Seventh Five Year Plan (1986-90), which established three "economic belts." The Plan document stated that most foreign investment and technical transfer would benefit the "coastal belt" comprising all the coastal provinces. The "central belt" of provinces, stretching from Heilongjiang through Henan south to Anhui and Jiangxi, would receive a "trickle through" of foreign investment benefits. The "western belt" was, for the time being, to depend largely on 13/ In 1979, Guangdong and Fujian were given special authority to accept large foreign investments without central government approval, and 4 special economic zones (SEZs) were established in coastal counties facing Hong Kong, Macao, and Taiwan (China). The strategy was extended in 1984 to 14 additional coastal cities, and in 1985 to the Yangtze and Pearl River Deltas as well as to a "southern coastal triangle" in Fujian. In 1988 an additional 140 coastal cities and counties were given "open" status, bringing the total to 288. Hainan was provided SEZ status in 1988, and Guangdong and Fujian were given special authority to establish foreign exchange markets and development banks, privatize housing, and undertake price, wage, and labor reforms. Most of the open coastal initiatives were at least partially curtailed in 1989-91. 13 its own resources. The coastal strategy took advantage of China's many harbors on one of the world's busiest great-circle shipping routes and encouraged foreign technology acquisition. But it also left more isolated provinces and counties in the interior with fewer resources than would have been the case under a development strategy giving less emphasis to the coast. Furthermore, greater local policy autonomy on the coast permitted greater retention of local tax and foreign exchange revenues and greater freedom to use expansionary bank loans for local investments. In interior provinces, state pricing and procurement procedures remained relatively authoritarian, maintaining the flow of raw materials and other real resources to the coast -- often at below market prices. 1.22 Not surprisingly, the open coastal strategy has encouraged a much greater role for international trade in the economies of the coastal provinces relative to their inland neighbors. The coastal provinces accounted for 77% of China's international exports of US$41 billion in 1989. Export value from these provinces amounted to US$76 per capita in that year. By comparison, the northwestern and southwestern provinces accounted for only 4% and 6% respectively of 1989 exports, and per capita exports averaged about US$12 -- less than one-sixth that of the coastal provinces. 1.23 Impact on Regional Growth. China's different regions shared the national experience of rapid real economic growth in the 1980s, but more isolated and poorer regions grew markedly more slowly than coastal areas. As reported in Table 1.3, real national income increased at an average annual rate of 9.1% during 1978-89, having slowed from an average of 9.5% annual growth during 1978-85 to 8.3% during 1985-89. National average economic growth per capita during 1978-89 was 129%. Real per capita national income grew 142% in the coastal provinces during 1978-89, or about 30% more than in the central and western belts. Overall, average annual growth in the coastal region was more than 1% greater than in the central and western belt provinces. Total per capita growth was greatest in the higher income provinces of south China (178%) and along the middle and lower reaches of the Yangtze River (138%), slightly less than the national average in north China (126%), and least in northeastern (102%) China and the lower income provinces of northwestern (107%) and southwestern (110%) China. 1E 1.24 The regional growth patterns of the 1980s have led to provincial income variation akin to that experienced at other times in the previous three decades. Using newly available data series on net material product and 14/ These growth rates document increased interprovincial income disparity during the reform period -- growth in the low income interior provinces was less than in the higher income coastal provinces. Denny (1991) reports a convergence of per capita provincial output from 1977 to 1988, but his analysis gives equal weight to all provinces. The regional growth rates presented in this report are weighted by provincial population, and correct for the excessive weight attached in Denny's analysis to the slow growth in Beijing, Shanghai and Tianjin. 14 Table 1.3: Provincial Average Annual Real Income Growth M%) \a ----------------------------------------------------------------------__----- Geographic Sub-Regions \c Economic Belts \b ---------------------------------------- ------------------ North- North- Yangtze South- Total Coast Center West North east west River South west Average Annual Real Growth (%) 1978-89 9.1% 9.7% 8.5% 8.3% 9.1% 7.4% 8.3% 9.5% 11.5% 8.3% 1978-85 9.5% 10.0% 9.2% 8.6% 9.4% 7.4% 8.7% 10.5% 11.0% 8.9% 1985-89 8.3% 9.1% 7.2% 7.9% 8.6% 7.3% 7.8% 7.9% 12.3% 7.3% Total Per Capita Growth (%) 1978-89 129% 142% 112% 109% 126% 102% 107% 138% 178% 110% National Income Per Capita (1980 yuan) 1989 761 1077 633 498 771 1029 614 908 872 457 Average Annual Real Agriculture Growth (%) 1978-89 5.5% 5.9% 5.2% 5.4% 6.5% 4.0% 5.9% 5.1% 6.9% 5.0% 1978-85 7.4% 8.1% 7.1% 6.7% 9.0% 6.3% 7.4% 7.0% 7.8% 6.6% 1985-89 2.3% 2.1% 2.0% 3.0% 2.2% 0.0% 3.5% 1.7% 5.4% 2.2% Total Per Capita Agricultural Growth (%) 1978-89 57% 64% 52% 54% 72% 41% 62% 50% 76% 49% Population (million) 1989 1108 414 395 296 242 98 128 325 96 217 ----------------------------------------------------------------------__----- Source: Compiled from SSB (1990b). \a National income in 1980 constant prices. \b The coastal provinces are Liaoning, Hebei, Beijing, Tianjin, Shandong, Jiangsu, Zhejiang, Shanghai, Fujian and Guangdong. The central provinces are Heilongjiang, Jilin, Inner Mongolia, Shanxi, Henan, Anhui, Hubei, Hunan and Jiangxi. The western provinces are Ningxia, Gansu, Qinghai, Xinjiang, Tibet, Shaanxi, Sichuan, Guizhou, Yunnan and Guangxi. \c The northern provinces are Beijing, Tianjin, Hebei, Henan and Shandong. The northeastern provinces are Liaoning, Jilin and Heilongjiang. The northwestern provinces are Inner Mongo3ia, Shanxi, Shaanxi, Ningxia, Gansu, Qinghai, and Xinjiang. The Yangtze River provinces are Shanghai, Jiangsu, Zhejiang, Anhui, Jiangxi, Hubei, and Hunan. The southern provinces are Fujian, Guangdong, and Hainan. The southwestern provinces are Guangxi, Sichuan, Guizhou, and Yunnan. 15 utilized national income by province, several authors (e.g. Tsui, 1991; Lyons, 1991; and Denny, 1991) have analyzed patterns of interprovincial production and income variation from 1952 to 1985. &I The analyses show that in all periods government transfers have played an important role in holding down variations in income relative to variations in production, and that the importance of government transfers has varied over time, being less important in the early 1980s than during the 1970s, but of much the same magnitude as in the early and mid-1960s. Overall income variation in the early 1980s was at about the level of the mid-1970s and early 1960s, but higher than all other periods. 1a Sources of income variation can be further decomposed, and these show most of the variation to originate in industrial, rather than agricultural, production (Tsui). 1.25 The figures for agriculture shown in Table 1.3 evidence a somewhat dissimilar pattern of growth. The slowdown in growth was more pronounced in agriculture than in the economy as a whole -- agricultural growth declined from 7.4% during 1978-85 to 2.3% in 1985-89. Total per capita agricultural growth in the coastal provinces (64%) was greater than in the central (52%) and western (54%) belts, but not by as great a margin as for the economy as a whole. Corresponding to greater than national average growth in the northwest and to poor performance in the Yangtze River provinces, furthermore, agricultural growth in the western belt was greater than in the coastal provinces during the subperiod 1985-80. Total per capita agricultural growth was greatest in south (76%) and north (72%) China, somewhat greater than the national average in northwest China (62%), and least in the northeastern (41%), southwestern (49%) and Yangtze River (50%) provinces. 1.26 Reaional Per Capita Income Growth. Slower than national average rates of economic growth in the northeastern, northwestern, and southwestern provinces during the 1980s are consistent with slower than average growth of rural per capita income in these same provinces. SSB rural per capita income 15/ Net material product (NMP) measures only commodity production, thereby excluding output of the service sector. None of the authors speculate about the impact of the service sector on their results. National income utilized (NIU) is the sum of material product consumption and investment. The difference between NMP and NIU represents net interprovincial resource flows, which may be regarded as a proxy for government transfers (Tsui). The NMP and NIU series are incomplete and several poorer provinces, including Guangxi, Qinghai, and Anhui are omitted from the analysis at least in the earlier years, as are Tibet and Zhejiang. 16/ Note that the measure of income used here is not realized personal income, but rather the total commodity income available for consumption and investment. Differences in rates of investment and in the importance of the service sector, as well as urban-rural population balances and provincial income distribution policies all affect the variation in realized personal income among provinces. Therefore, income variation as presented in these paragraphs cannot be used to infer interprovincial personal income variation. For that, household income surveys must be used, as they are in Chapter 2. 16 figures for the 1980s document that real rural income levels increased in all provinces during the 1980s, but that regional imbalances became more pronounced. A disproportionate share of the gain. has accrued to the well- positioned and favorably endowed coastal provinces. By comparison, growth of rural per capita income in each of the interior provinces of northwestern and southwestern China was less than national average growth during 1980-89. These same interior provinces, furthermore, all started from positions of less than national average rural income in 1980 (Table 2.2). -Public Finance and Redistributive Transfers 1.27 Financing Social Services. Fiscal decentralization, introduced in stages since the late-1970s, has been associated with substantial changes in the sources of mobilized revenue and the structure of expenditure. 171 Within plan revenue mobilization declined from over 30% of GNP in 1978 to less than 20% by the end of the 1980., but this decline was almost fully offset by the increase in above plan revenues from less than 10% of 1978 GNP to nearly 17% in 1990. LE Total revenues remained roughly constant at about 40% of GNP during 1978-89. 19/ The reduction in within plan expenditure (as a share of GNP) has so far been matched by reduced budgetary outlay shares for capital expenditures, working capital, and military spending. Within plan government current expenditure on health and education, on the other hand, increased as a share of both total expenditure and GNP during the reform period. Above plan expenditure on education, health, relief and culture has remained roughly constant at 12% to 13% of total above plan expenditure. Consequently, the sum of within and above plan government expenditure on health and education increased as a share of total expenditure and GNP during the 1980s. 1.28 Nost of the real increase in government expenditure on health and education, however, has benefitted the vell-off urban population and middle and upper income rural inhabitants -- very little if any of the increase has reached the absolute poor. Central and provincial government support for education and health in China's poorest townships and villages is at present very limited. Furthermore, total government expenditure on rural relief and poverty alleviation programs declined in real terms and as a share of total 17/ China's tax system is reviewed in World Bank (1989). I8/ Above plan ("extrabudgetary") revenues comprise funds raised -- in excess of and outside the state plan -- by local government, enterprises, trades and administrative units. By law, these units retain control of the above plan revenues they raise. Most above plan funds are ear-marked for special purposes including basic construction, renovation and renewal of capital investment, expansion of working capital, and administrative operating costs. 19/ Central and local above plan revenue mobilization increased from Y35 billion in 1978 to 266 billion in 1989, or from 30% to 95% respectively of 1978 and 1989 within plan revenue mobilization. Summing within plan and above plan revenues, total revenue remained roughly constant at about 40% of GNP throughout the reform period. 17 expenditure during the second half of the 1980s. Total government support for education, health, relief and other social services in the poor areas, in the form of grants, subsidized credit and other programs, is estimated to have totaled under Y10 billion in 1990. By comparison, government expenditure on urban consumer food subsidies, which almost exclusively benefit the well-off registered urban population, increased sharply throughout the reform period. Urban consumer food subsidies totaled more than Y50 billion in 1990, or about 15% of total within plan government expenditure and at least five times government expenditure on poor area social services and poverty alleviation programs. R 1.29 Fiscal Redistribution. Fiscal decentralization has also been associated with increased provincial expenditure, though the expected increase in provincial revenue mobilization did not fully materialize. Weakening provincial revenue performance has been partially compensated by the center which, a net recipient of provincial funds during 1980-85, began making net transfers to the provinces in 1986. In the absence of official interprovincial fiscal transfer figures, one method of investigating such transfers is to compare trends in provincial budget deficits and surpluses.2' Available statistics show growing deficits in, and hence imply increased transfers to, the interior provinces during the second half of the 1980s. As shown in Table 1.4, for the central belt, what had been slight surpluses or negligible deficits early in the decade became much stronger deficits during 1985-89. For the western belt, where transfers in from the outside had already been significant, provincial deficits also increased in real terms in the late-1980s. It should be noted, however, that not all regions in the western belt benefitted equally from the transfers. In particular, the four southwestern provinces of Guangxi, Sichuan, Guizhou and Guangxi received a much smaller increase (in per capita terms) than did the northwestern provinces. 20/ Urban food subsidies, which are significantly understated in official Ministry of Finance figures, are discussed in detail in World Bank (1990d). 21/ With the sudden growth of bank loans for government supported projects during the 1980s, provinces where money supply and purchasing power increased most rapidly would have been in a position to command more resources from other provinces. Indirect evidence suggests that loans and the money supply grew most rapidly in the higher income coastal provinces. 18 Table 1.4: Real Provincial Average Annual Budget Balances \a -----------------------------------------------------------------__---------- Geographic Sub-Regions \c Economic Belts \c ---------------------------------------- Total ------------------ North- North- Yangtze South- \b Coast Center West North east west River South west ------------------------------------------------------------------__--------- Yuan Billion 1978-89 15.3 24.1 -2.5 -6.2 5.4 2.4 -5.5 15.7 0.1 -2.6 1978-84 26.1 30.7 0.0 -4.6 8.0 4.3 -4.3 19.3 0.6 -1.7 1985-89 0.3 14.7 -6.0 -8.5 1.7 -0.3 -7.1 10.6 -0.6 -3.7 Yuan Per Capita 1978 38 100 9 -12 51 107 -24 68 15 -9 1989 -4 22 -13 -26 0 -7 -46 23 -7 -16 (Difference) 1978-89 -41 -78 -22 -14 -51 -114 -22 -45 -22 -7 ----------------------------------------------------------------------__----- Source: Compiled from SSB, (1990d). \a Expressed in constant 1980 yuan using the national net material product deflator. \b Total of provincial budgets only (excludes items in the central government's budget). \c As defined in Table 1.3. 1.30 Incomplete Ministry of Finance (MOF) figures for the 1980s confirm that nearly all of the poor northwestern and southwestern provinces were net recipients of contracted central government fiscal transfers: Contracted Central to Provincial Government Fiscal Transfers (current Y billion) 1979 1985 1990 -------------------------------------------------------------------__--------_ Northwestern Provinces Inner Mongolia 1.06 1.78 2.16 Shaanxi na na 0.12 Ningxia 0.27 0.49 0.60 Gansu na na 0.13 Qinghai 0.37 0.61 0.74 Xinjiang 0.81 1.45 1.75 subtotal 2.51 4.33 5.50 Southwestern Provinces Guangxi 0.27 0.72 0.87 Guizhou 0.48 0.74 0.90 Yunnan 0.30 0.64 0.77 Tibet 0.45 0.75 0.91 subtotal 1.50 2.85 3.45 Total 4.01 7.18 8.95 -------------------------------------------------------------------__--------_ 19 These figures indicate that central government transfers to these ten inland provinces remained roughly constant in real terms during 1979-90. Total contracted transfers to the ten provinces amounted to nearly Y9 billion in 1990. In that year, four other provinces (Jilin, Jiangxi, Fujian and Hainan) together received an additional YO.35 billion in central government contracted fiscal transfers. Shanxi (northwest) and Sichuan (southwest) apparently did not receive central government contracted fiscal transfers in 1990. 1.31 These contracted transfers are matched in some cases by substantial earmarked transfers from the central government. Earmarked central government transfers to Yunnan in 1990, for example, amounted to YO.87 billion -- more than the contracted transfer in that year. Nearly half, some Y400 million, of the earmarked transfers to Yunnan was used to cover urban consumer food subsidies. By comparison, only Y140 million was used for poor area development programs (Y60 million), education and health (Y30 million, only some of which may have benefitted the rural poor), and relief (Y50 million, only some of which may have benefitted the rural poor). It should also be noted that substantial losses of potential revenue on mandatory sales of Yunnan's plentiful output of timber, sugar, tobacco, and nonferrous metals to other provinces at below market prices (para 1.21) partially offset these central government transfers. Nevertheless, of the central government's total contracted and earmarked fiscal transfers to Yunnan of Y1.64 billion in 1990, some YO.5 billion was passed on from the provincial government to 102 counties (accounting for 80% of all of Yunnan's counties) experiencing fiscal deficits. In the poorest counties, these provincial government transfers covered fully one-half of average county government total expenditure. In sum, while much of the central government's transfers to Yunnan are either offset by lost revenue on mandatory sales of primary commodities at below market planned prices or diverted to the well-off urban population, a significant portion is used to bolster poor county expenditure capacity and for poor area development and social services. -Macroeconomic Trends and Poverty Alleviation Work in the 1990s 1.32 The macroeconomic and fiscal trends and their impact on poverty in the 1980s reviewed in this chapter demonstrate that high overall growth rates alone will not lead to further substantial reductions in rural poverty. Instead, further reductions in rural poverty will require new policy initiatives and increased government commitment to targeted interventions, including greater population mobility (para 3.16), improved market integration of poor areas through infrastructure development (para 2.39), and increased labor productivity through universal primary education and basic health care (para 4.06 and 4.21). The task of fostering greater mobility aside, the needed programs will require a marked expansion of government resources targeted to the poor. 1.33 The current national poverty alleviation program, discussed in detail in chapter 5, has suffered decreased real funding since its inception in 1985. The central government's decision to hold program allocations constant in nominal terms over a period of high inflation has substantially eroded the program's ability to have an impact on poverty. The static funding 20 comes not from a judgment that the program is ineffective, but rather is driven by central government budget deficits that have proven difficult to finance. However, recent policy decisions, if implemented, will ease budget pressures over the next 5 years, leaving government in a position to make a financial commitment matching its moral commitment to poverty alleviation. The most important policy decisions bearing on improvements in the budget are those for price and fiscal reform (World Bank, 1992b and 1992d). 1.34 Price reforms will not only reduce a subsidy burden now running over Y90 billion per year, or 28% of revenues, but will also enhance the efficiency of a broad array of industrial and other enterprises (World Bank, 1992b). The reduced subsidy burden alone would leave room sufficient to fund a substantially expanded anti-poverty program, but will not be the only source of net revenues. An array of proposed fiscal reforms, even if only partially implemented, will markedly increase the responsiveness of government revenue to overall economic growth (World Bank, 1992b). That growth itself will be encouraged by price and fiscal reforms, as well as by enterprise management reforms that release enterprises from their welfare burden and promote optimal use of labor and other resources (paras 2.14 and 2.15). Improvements in productivity expected from the reforms will increase enterprise profits, hence government revenues, and provide a major contribution toward meeting the goal of a continued high rate of GNP growth. 21 2. INCIDENCE AND CORRELATES OF ABSOLUTE POVERTY 2.01 Introduction. This report measures well-being by comparing actual per capita, consumption of commodities and services against a standard of living chosen to represent the absolute poverty line. Those whose consumption falls below the standard are deemed poor. The criticisms of this consumption- based approach are well known -- consumption does not always relate in a very consistent way to health and nutrition outcomes, education, and other physical and social attributes, and measurement problems can be severe when prices, externalities, and public goods all vary by place and social strata and are not captured in the surveys normally used to measure income or consumption. ' While not denying the importance of these challenges, the alternative forms of welfare valuation are subject to equally forbidding measurement difficulties. Furthermore, the standard of living approach adopted here is supplemented by information on social participation, morbidity and mortality, and other relevant indicators. The poverty pattern in the other indicators is broadly consistent with that in income and affirms the usefulness of income as a measure of poverty in China. 2.02 The poverty line used in this report derives its initial monetary value from a 2,150 kcal/day food consumption basket deemed to supply adequate nutrients for normal human functioning. v It then adds a sum for expenditures on non-food commodities and services based on the average expenditure pattern of the poor. Once established, the poverty line measures welfare acainst the ability to buy the initially determined commodity and service bundle. 3L The line accords well with the Chinese estimate of a poverty line for 1985, the base year for China's national poverty alleviation program. While the poverty line provides a point estimate of the numbers of poor in any year and is the basis for estimating the poverty gap, this report is primarily concerned with the comparison of magnitudes of poverty over time, 1/ Sen (:1981) summarizes the debate on poverty measurement through the 1970s, in which he was a major force. Ravallion (1992) provides a current general discussion of related methodological issues. 2/ This same standard of subsistence food energy intake is adopted in studies of poverty in other countries (e.g., Rao for Indonesia), and is about 100 kcal less than estimated average per capita food energy requirements for the population as a whole (Piazza, 1986). 3/ This report uses income, rather than consumption, as the measure of the standard of living in China. The choice of income was made out of necessity-- disaggregated consumption data for rural Chinese are not published, while income data are. However, the income data published by the Chinese include the value of self-consumed production and of transfers, so are partially free of biases often encountered in such surveys. Tests in other countries suggest that the isubstitution of income for consumption in poverty measurement is most unlikely to alter the trends and relative magnitudes described in this paper (World Bank, 1990a). 22 across sectors, and geographically. It is comparative magnitudes, not absolute levels, that are key to the results in the report. Sensitivity analysis indicates that the quantitative analysis of the trends in the national incidence of poverty presented in this report are robust to reasonable changes in the composition and energy value of the subsistence food basket, the budget share of non-food commodities and services, and all other underlying assumptions and parameters. Nonetheless, the report also strongly recommends than an independent poverty monitoring system be established for the purpose of regular and objective assessments of the severity and location of poverty at the national, provincial and local levels. A. Incidence of Absolute Poverty -Income Levels 2.03 Income Distribution Data. The estimates of the incidence of absolute poverty presented in this report have been calculated from official urban and rural income distribution data and estimated absolute poverty lines. Available income distribution data, shown in the top section of Table 2.1, are from the State Statistical Bureau's (SSB) annual large-scale household income and expenditure surveys. The rural data are known to substantially understate true income levels by valuing that proportion of grain (and some other farm products) consumed on-farm at below-market planned prices. The divergence between planned and market prices widened greatly during the second half of the 1980s, and worsened the downward bias in the SSB's figures for rural income. The understatement of rural income is estimated to have been at least Y70 per capita in 1989, equivalent to 12% of the Y602 average per capita rural income reported in that year. # The market price of grain declined with 1990's record harvest, and the understatement of rural income reportedly declined to Y53 per capita or about 8% of 1990 rural income. E By using the below-market planned price of grain in the calculations of the poverty lines, 4/ In 1989 the difference between the planned ('"contract") price (Y490/ton) and the mixed average procurement price (Y750/ton), which represents a lower bound to the market value of grain, was Y260/ton. About 232 million tons of grain (expressed in milled equivalent), or about two-thirds of 1988 total grain production, was consumed on-farm in that year. The product (Y60 billion) of the price difference and the volume of grain consumed on-farm, divided by the rural population of 832 million, comes out to Y72 per capita, representing a lower bound to the understatement of average rural per capita income in 1989. 5/ Officially, rural per capita income is reported to have been Y630 in 1990 (SSB, 1991). The SSB Rural Survey Office, however, has unofficially reported that 1990 average rural per capita income would have been Y683 had grain (and other farm products) consumed on-farm been valued at prevailing market prices. The SSB plans, beginning with 1991 survey data, to record income by valuing on-farm consumption at the volume-weighted average of planned and market prices. 23 Table 2.1: Distribution of AnnuaL Per Capita Income. Poverty Lines. and Incidence of Absolute Povertv Unit 1978 1981 1984 .1985 1989 1990 ANNUAL PER CAPITA INCOME Urban <240 X 2.1 I 240-300 X 5.5 1 .7 11.1 4.8 300-420 X 31.8 10.5 7.4 420-600 X 42.3 38.9 24.8 600-720 X 11.9 22.7 19.5 5.0 720-840 X 12.8 16.0 6.9 5.7 840-960 X 11.0 9.3 7.0 960-1080 X 7.0 10.8 8.8 1080-1200 X 4.1 11.2 8.8 1200-1320 X 6.5 13.4 2.6 10.2 9.8 1320-1440 X 8.6 9.2 1440-1560 X 7.5 8.0 1560-1680 X 3.9 6.0 7.4 1680-1800 X 4.4 5.8 >1800 X 15.8 22.3 Average Y 500 660 749 1388 1523 Rural 100 X 33.3 4.7 0.8 1.0 0.6 0., 100-150 X 31.7 14.9 3.8 3.4 1.3 0.9 150-200 X 17.6 16.0 9.4 7.9 2.8 2.2 200-300 X 15.0 34.8 29.2 25.6 10.9 9.5 300-400 % 14.4 24.5 24.0 15.6 14.4 400-500 X 2.4 5.0 14.1 15.8 15.6 15.1 >500 X 3.2 18.2 22.3 53.2 57.4 Average Y 134 223 355 398 602 630 Rural Gini Coefficient 0.21 0.24 0.26 0.26 0.31 0.31 POVERTY LINES Urban expenditure on food Y 108 117 132 190 197 food's budget share X 0.63 0.62 0.61 0.62 0.61 poverty Line Y 171 190 215 304 321 Rural expenditure on food Y 74 105 112 120 165 173 food's budget share X 0.75 0.66 0.63 0.63 0.63 0.63 poverty line Y 98 158 179 190 262 275 INCIDENCE OF ABSOLUTE POVERTY Urban million people 3.9 0.8 0.9 0.9 1.3 X of urban population 1.9 0.3 0.4 0.3 0.4 Rural million people 260.5 194.3 88.7 96.4 102.5 96.8 X of rural population 33.0 24.3 11.0 11.9 12.3 11.5 Total million people 198.2 89.5 97.3 103.4 98.2 X of total population 19.8 8.6 9.2 9.2 8.6 Source: SS8 (1991a) for urban and rural income distribution data and Gini coefficients. Annex 1 to this report for details of the estimates of poverty lines and incidence of absolute poverty. 24 this report's poverty estimates roughly correct for the understatement of rural income. Details of the calculation of the poverty lines and the incidence of absolute poverty for all years 1978-90 are reported in Annex 1. A second important feature of the SSB rural income distribution data is that the survey sample was completely redrawn and greatly expanded (from 31,375 households in 1984 to 66,642 households in 1985) in 1985. The 1985 sample, which was left almost completely unchanged until 1990, almost certainly extended the SSB survey into more remote and hence poorer areas than did previous samples. Comparison of the 1990 poverty estimates with the 1985 figures, rather than the 1984 figures, therefore yields a more accurate assessment of the trends in poverty in the second half of the 1980s. C 2.04 Poverty Lines. This report's urban and rural poverty lines have been calculated as the annualized cost of the subsistence food basket, inflated by the reciprocal of food's budget share for low income households (from the SSB survey data) to approximate the cost of nonfood subsistence goods. The food basket comprises fixed quantities of grain (which supplies 90% of food energy), vegetable oil, vegetables, pork and eggs, and provides 2150 calories per day. The content of the food basket is held constant, but its annualized cost increases in accordance with annual changes in planned and market prices for these foods. As shown in the middle section of Table 2.1, the estimated urban poverty line increases from Y171 in 1981 to Y321 in 1990, or by 88%. Over the same period, the rural poverty line increases 74%, from Y158 in 1981 to Y275 in 1990. The rural poverty line jumps by 61% in the three years 1978-81, and nearly triples over the years 1978-90. Consistent with strong real income growth, the estimated urban poverty line declines steadily from 34% of average per capita urban income in 1981 to 21% in 1990. The decline would have been somewhat less pronounced in the absence of urban consumer food subsidies (para 1.02). Similarly, rapid real income growth in the countryside during 1978-84 was associated with a decline in the estimated rural poverty line from 73% of average per capita rural income in 1978 to 50% in 1984. With the stagnation of rural incomes during 1985-88, the decline slows and in 1988 the rural poverty line is 42% of average rural income. With the sharp increases in rural grain prices and the 7% decline in real rural income in 1989, the estimated poverty line increases to about 44% of average rural income in that year. This report's estimated 1985 rural poverty line of Y190 is bracketed by the State Council's 1985 cutoffs of Y150 and Y200 for extreme poverty. The 1988 poverty line of Y231 is about 10% less than the SSB's unofficial estimate of the rural absolute poverty line in that year of Y260. 7 6/ It is quite likely that the complete redrawing and doubling of the SSB sample explains at least all of the modest increase in the number of rural poor -- from 89 million in 1984 to 96 million in 1985 -- noted in this report. 7/ The SSB rural poverty line for 1988 (Yano and Liu, 1990) provides for a diet of 2400 calories per day, more than 10% above the 2150 calorie line used in this paper. 25 2.05 E'overty Estimates. The estimates of the incidence of absolute poverty, shown at the bottom of Table 2.1, indicate that the number of poor declined from about 270 million in 1978 to just less than 100 million in 1985, and then stagnated between 86 and 103 million during 1985-90. E As a sensitivity test, the incidence of poverty has been estimated with the urban poverty line inflated by 20% and the rural poverty line inflated by 10% (see Annex 1). With the inflated poverty line, estimated urban poverty remains at extremely modest levels -- fewer than 2% of registered urban households -- in all years 1981-90. Inflating the rural poverty line by 10% does not significantly alter the observed trend in rural poverty of a sharp and uninterrupted decline in poverty during 1978-85 followed by stagnation during 1985-90. Reflecting the bunching of a large share of the rural population around the poverty line, however, the 10% increase in the poverty line corresponds to much larger increases -- of about 20% in 1978 and between 25% to 37% during 1980-90 -- in the estimated number of rural poor. 2.06 The poverty estimates were also tested for sensitivity to the assumptions made regarding the price of grain (para 2.03). For this purpose, an alternative rural "procurement price" poverty line has been constructed by valuing gra,in at the official mixed average procurement price and then directly deducting the difference between subsistence grain consumption valued at the mixed average procurement price and at the estimated plan price. Though it too corrects for the understatement of the SSB rural income data, the alternative "procurement price" poverty line exceeds the base case "plan price" poverty line by an increasing margin during 1985-89. The "procurement price" poverty line corroborates the same tremendous reduction in rural poverty during 1978-85, but indicates a temporary resurgence in rural poverty in 1989 to about 1983 levels. The sensitivity of the poverty estimates to these and other changes in assumptions are detailed in Annex 1. 2.07 Poverty Gap. The "headcount index" of poverty reported in the previous paragraph indicates the number of people in poverty, but gives equal weight to individuals with income just below the poverty line and those with income half that of the poverty line. Differences in the depth of poverty can be measured by the "poverty gap," or the additional income that would be necessary to bring a poor person up to the poverty line. Summed across the poor population and normalized against the poverty line and population size, the "poverty gap index" can be used to assess the relative depth of poverty among subpopulations, across time, or internationally. The Chinese rural poverty gap index -- 2.1% in 1985 and 2.5% in 1990 (para i) -- is very small by international standards, showing that the poor are concentrated at income levels just below the poverty line. 21 This suggests that the challenge of 8/ The distribution of urban income in 1978, and hence the incidence of urban poverty in that year, is not known. Extending back in time the sharp decline in urban poverty observed for the early-1980s, it appears likely that the number of urban poor was between 5 to 10 million in 1978. If so, total absolute poverty in 1978 would have been between 265 to 270 million. 9/ By conaparison, the aggregate poverty gap index for 86 developing (continued...) 26 helping the remaining rural poor cross the poverty line is somewhat less daunting than that faced in other poor countries. However, the increase in the poverty gap index during the second half of the 1980s indicates a modest deepening of poverty over a period when the headcount index was stable. The value of the Foster-Greer-Thorbecke index, which increased from 0.77 in 1985 to 0.94 in 1990, also indicates a modest increase in the severity of poverty during the second half of the 1980s. 10 2.08 Official Estimates. Comparing this report's estimates of the incidence of absolute poverty with those of the Chinese government must be done with care, for the government itself generates more than one set of estimates and the estimates are not all based on transparent criteria. The Chinese press reported that the number of absolute poor in the late-1970s was about 200 million (Lardv, 1982) -- somewhat less than this report's 1978 estimate of 270 million. This report's estimate for 1985, of 96 million rural poor, agrees closely with the State Council's official figure of 102 million rural poor in that year. However, the State Council's figures for 1988 and 1989, 27 million and 39 million rural poor respectively, are far below this report's estimates. The implausibly low official rural poverty figures for the late-1980s, which are based upon the 1985 poverty line of Y200 unadjusted for inflation, highlight the need to establish an independent and objective poverty monitoring and assessment system at the national and provincial levels. iL- Urban poverty estimates appear to be generated only by the Ministry of Civil Affairs system. Their 1989 estimate of 17.5 million urban poor far exceeds the estimate in this report, but is clearly a measure of relative, rather than absolute, poverty. -Urban Poverty 2.09 For official urban residents, the strategy for poverty alleviation (para 1.02) must be judged a major success. In 1990, average per capita income among the poorest 5% of urban residents was Y689, or more than double the urban absolute poverty line of Y321 and greater than the per capita income of 65% of rural residents. Less than 1% of the urban population, or about 1 9/(.. .continued) countries was about 9.2% during the late-1980s (Ravallion, 1991). 10/ The Foster-Greer-Thorbecke index, which weights the poverty gaps of the poor by the relative size of those poverty gaps, is a preferred measure of the severity of poverty. 1l/ Recently released unofficial SSB estimates show the number of rural poor declined sharply from 124.1 million in 1985 to 84.1 million in 1990. The SSB has reported the poverty lines used in their analysis -- Y206 in 1985 and Y268 in 1990 -- but has not released detailed documentation of the methods and data used in their calculations (para 2.04). In contrast to the 60% increase in the rural retail price index during 1985-90, the SSB's 1990 poverty line is just 30% greater than their 1985 line -- suggesting a strongly optimistic bias to the SSB poverty estimates. By comparison, this report's 1990 poverty line is 45% greater than its 1985 poverty line. 27 million people, had income levels falling below the estimated absolute poverty line each year from 1983 to 1990. Superior income levels, complemented by annual consumer food subsidies of at least Y200 per urban recipient (World Bank, 1990d), leave the registered urban population much better nourished than their rural counterparts. SSB household income and expenditure surveys show, for example, that urban residents among the bottom 5% of the 1988 urban income distribution consumed 15% more vegetable oil, 27% more meat, and 117% more eggs than did the average rural inhabitant. It is therefore not surprising that anthropometric surveys have shown that less than 3% of urban children under 5 years suffered even mild malnutrition (defined as either moderate stunting or wasting) -- a rate similar to those achieved in advanced developed countries and several fold less than observed for rural children in China (UNICEF). 2.10 Relative Poverty. In contrast to their practice for rural areas, the Chinese use a relative measure to estimate urban poverty, with each province defining its own poverty line. Iv Using those lines, in 1989 the Ministry of Civil Affairs (MCA) estimated that about 17.5 million urban residents (or 5.9% of the urban population) were poor and assisted some 4.9 million of them, primarily with small, one-time grants. only 340,000 people qualified for regular monthly subsidies that year. Urban poverty is found disproportionately in county towns. Cities, which held about 63% of the urban population in 1989, have more revenue sources and greater fiscal autonomy than towns. That economic strength reflects itself in more and higher paying jobs and much higher levels of infrastructure investment than enjoyed by towns. Average city worker incomes run about 6% higher than town incomes and typically the lowest income decile city residents in a province will have better access to such amenities as electricity, piped water in the house, private toilets and hospital services than do the highest income decile town residents. 13/ 2.11 Geoaraphic Distribution. Provincial average urban per capita incomes vary substantially, ranging from Y777 to Y1202 in cities and Y638 to Y1050 in towns in 1987 (the most recent year for which data are available). Urban poverty displays a geographical distribution distinct from that of rural 12/ A recent article by Ahmad and Wang in The World Bank Economic Review found much higher levels of urban poverty and a different pattern of rural poverty than derived in this paper. Their 1985 urban poverty line was placed at an income level equal to approximately 50% of mean urban income and thereafter adjusted by the urban cost-of-living index. The 1985 rural line was taken as approximately 50% of the much lower mean rural income, with the result that in 1985 an urban resident with an urban poverty line income had a real income some 68% higher than a rural resident with a rural poverty line income. This paper holds rural and urban residents to the same real poverty standard, a more valid approach when attempts are made to compare poverty across sectors. 13/ The income figures are derived from SSB (1988) and the social services results derived from an SSB published fertility survey (SSB (nd)). 28 poverty. The six provinces where the poorest 5% of city dwellers averaged Y400 per capita or less in 1987 are contiguous and extend in a belt from Xinjiang to Jilin. Town poor are more generally distributed, but concentrated in provinces north of the Yangtze, with, in the south, only Guizhou having very poor towns (See Annex 1, Table 6 for details). Average incomes also vary substantially within provinces--in a sample of three provinces available for 1987, the average income in the poorest town in a province was as low as 52% of that in the wealthiest town. Lv These patterns of urban income and poverty reflect (i) national and provincial decisions about granting city status to urban places and urban residency status to individuals and (ii) opportunities for collective enterprise development and the distribution of investment in state owned enterprises. 2.12 Chanaes over Time. The lowest income urban people shared fully in the economic gains of the early to mid-1980s. In fact, from 1981 through 1987 the bottom income decile realized a 9.2% average annual increase in real income, greater than that for any other income decile. During the retrenchment of 1988 and 1989, however, the lowest income groups failed to keep pace with either higher income groups or inflation, the bottom decile losing 8.2% in real income over those two years. Those losses were recovered during 1990, when the real income of the poor grew 8.7%, approximately the average for all income groups that year (SSB, 1991a). Despite the recovery in 1990, low income urban groups remain vulnerable to poor urban economic performance. 2.13 Poverty. Employment and Dependency. The main characteristic differentiating a poor urban family from a non-poor family is the ratio of dependents to workers. Analysis of the urban household survey from 1990 shows that families in the bottom 3% of per capita income (well below the Chinese urban relative poverty line but above the absolute line used in this paper) had an average of only 34% of family members employed, while families in the next 4% (on average, just at the Chinese urban poverty line) averaged 43% employment and the next 7% had a 48.9% employment rate (SSB, 1991a). Average wages of employed members increase with per capita income, but much more slowly than dependency falls. In 1989, a worker from a family in the bottom 5% of per capita income earned Y1,845 per year, against Y2,165 for a worker in the 30% income group, only a 17% difference. Per capita family income differences derive primarily from the higher dependency ratios in the lower income group, and those ratios, not worker income levels, drive poverty in urban China. An analysis of household data shows that the relatively high dependency ratios in the poorest families stem both from proportionately fewer working age family members and from the fact that fewer working age members have jobs outside the home. 151 Those not working outside have lower 14/ These differences are calculated after adjusting for differences in average education and experience of the labor force. Actual differences are even larger. The provinces and low-high differences are: Liaoning 80%, Guangdong 59% and Shandong 52%. Calculated from (SSB (nd)). 15/ Calculated from a subsample of the 1988 SSB urban household survey in the (continued...) 29 education levels than the population as a whole, with many not having finished more than three years of formal education. 2.14 Urban Labor Market Reform. The impetus for urban labor market reform could markedly change the nature of urban poverty. Proposed reforms promise to significantly improve the productivity of urban enterprise, but would do so by shifting responsibility for surplus labor and for other social welfare functions from enterprises to specialized government agencies. The key to urban enterprise reform is maintaining enterprise staff at levels that maximize the long run profitability of the enterprise and compensating employees in a way that more closely approximates their contribution to enterprise income (World Bank, 1992a). Properly implemented, these changes would substantially increase productivity, increase both enterprise and government net revenues, and help maintain a level of economic vitality that would create new jobs to replace those of redundant workers and provide for new entrants to the labor force (paras 3.12 to 3.14). In the short run, however, substantial dislocation would result, with a jump in unemployment and shifts in income that would lower some incomes through reduction in unit-based subsidies. While the current employment system limits unemployment to the young and, usually, unmarried, proposed reforms would lead to unemployment among currently employed older people, those more likely to have high dependency burdens. Evidence from the recent austerity period suggests that layoffs would occur first among women. Such a pattern of discrimination would reduce the probability of all family income earners simultaneously losing their jobs and spread the burden of unemployment more widely across families, but create a new source of inefficiency in labor allocation. 2.15 In addition to absorbing surplus labor, enterprises have traditionally been responsible for providing employees with various consumption goods and a range of services, including housing, income supplements for poor families, health care, and, in some cases, education. Linking these services, especially housing, to place of employment strongly inhibits labor mobility (World Bank, 1992c). Direct enterprise provision of, or subsidies for, services and commodities creates substantial gaps between apparent and actual returns to labor that have to date nullified attempts to link income to performance. Either monetizing benefits or, where appropriate, shifting service delivery to specialized government agencies should result in superior targeting of social services at reduced social cost. The current system results in relatively strong enterprises -- those whose workers already enjoy above average wages and bonuses -- providing much higher "social welfare" income supplements than do the weak enterprises whose workers are most at risk. This system has the perverse result that unit-based subsidy payments are highly regressive, with the top 10% of the urban income distribution receiving more than three times as much per capita in enterprise subsidies as those in the bottom 10% (SSB, 1990c). Considerable scope exists, therefore, for reducing total social welfare costs without impairing 15/(.. .continued) possession of the Economics Institute, Chinese Academy of Social Sciences (Riskin). 30 the welfare status of the urban population by shifting welfare functions frozm enterprises to specialized government agencies. 2.16 The urban labor market reform agenda described above would in the short run increase urban poverty as measured by earned income. At the same time, it would generate savings more than sufficient to compensate those whose loss of income placed them at risk of poverty. To deal effectively with increased numbers of poor, and new, broader responsibilities for guaranteeing urban social welfare, the MCA system would need a modest expansion in personnel and a substantially increased budget. Continued MCA reliance on neighborhood committees to make initial assessments of need and monitor status would be key to the effective handling of the additional work load. Needed budgetary resources could be generated by reallocating from enterprises to MCA funds earmarked for social welfare. L1 2.17 The Floating Population. China's floating population, those people living on either a long-term or seasonal basis outside of the place where their residence records are maintained, number as many as 70 million people (Li, 1991). Members of the floating population live in both urban and rural areas and work in a wide variety of occupations. Their jobs are marked by an absence of tenure and their residence rights are tenuous. In times of economic or social pressure, local governments often force these people back to their original homes. When considering the place of these people in urban poverty it is useful to divide them between short term, often seasonal, migrants and long term migrants. Both groups share a vulnerability unknown among the registered population, but the short term migrants are at much less risk than the long term. 2.18 Short term migrants are most often rural people who travel during the agricultural slack season in search of unskilled or semi-skilled employment, and are estimated to number up to 40 million in any year. Interviews in sending villages suggest that most of these people find employment in other rural areas or in county towns. They are characteristically healthy adult males between 18 and 45 travelling alone or with fellow villagers. 17' Although some depart home without a clear idea 16/ Increased funding for MCA-provided unemployment benefits could be derived from several sources. Enterprise social welfare funds now designated for "families in difficulty" could be shifted to MCA control, which would provide at least Y215 million. Some portion of non-wage related subsidies now given by enterprises to employees could also be channelled through MCA. Such subsidies totalled Y21.5 billion in 1988 (an additional Y33 billion was paid in subsidies that the government considered wage related) (SSB 1990c). In addition, reduced government subsidies to the existing urban consumer food system, which now benefit all urban residents rich or poor and cost more than Y50 billion annually, could provide funds for transfers targeted to the small proportion of the registered urban population in need of assistance. 17/ Nearly 80% of short term migrants were adult males in a 1986 sample (continued...) 31 about where work is to be found, many have established patterns of association with particular receiving areas. As this brief profile makes clear, the age, gender, and health status of these people leaves them among the least vulnerable of any rural residents. Furthermore, because their migration is short term, family and village support systems remain intact and they are assured of sustenance if illness or unemployment leaves them facing a subsistence crisis. This group does not typically contribute to urban poverty. 2.19 Long term migrants, numbering approximately 30 million people according to the 1990 census, have a substantially different profile. More women -- many young and unmarried -- enter the long term migrant stream, as do families with children. 181 Also, urban areas, most often county towns and small cities, contribute to this migrant stream. Most long term migrants appear to move to areas where jobs have already been identified or are thought likely because of the past success of other villagers or relatives. Limited evidence on the welfare of this group comes from a 1986 survey, in which almost two-thirds of those interviewed claimed a marked improvement in income and welfare from the move (Shanghai Statistical Bureau). The 1989 economic slowdown, however, demonstrated the vulnerability of long term migrants. Many were forced out of their jobs and back to their native place. Sichuan alone had 1.3 million return from outside the province. Newspaper articles from that time make it clear that reabsorbing the returnees was difficult and associated with considerable hardship for those returning and their families. The home areas were already plagued by underemployment and the long separation from family and village had created the presumption that the migrants were no longer valid claimants on local welfare. 2.20 Unfortunately, the urban social service system responds poorly to the needs of short or long term migrants. Access to important services such as education and medical care is very limited and expensive. The urban safety net for needy migrants now consists of no more than a ticket back to their area of origin. As unregistered migrants increase in number and duration of stay, the pressure on urban resources of all types--housing, transport, medical care, welfare, and schools--will continue to grow. An important test of urban reform will be the delinking of those services from formal residence status and work unit connections. A failure to broaden access to urban resources will encourage the development of barrios holding a marginalized community of unregistered migrants. Simply sending ill or indigent migrants 17/(... continued) survey of sending villages (Yu). Information in that survey corresponds closely to the results from field interviews in 1991. Female migrants face special risks, and in particular are vulnerable to sexual crime. Reports are not uncommon of young women lured from their villages with promises of good jobs elsewhere, then forced into marriage and virtual bondage. 18/ Results in Yu indicate that the proportion of women jumps by about 25% among the long term migrants from the rural sending areas. State Council Census Office shows 47% of all long term migrants to be women. 32 avay becomes a less tenable urban response as migrant residence periods extend in years. The NCA must rethink their approach to the problems of these nigrants and other arms of government must work to reduce the vulnerability of these people by easing access to social services. -Rural Poverty 2.21 Rural Income and Poverty. Trends in rural income and poverty during 1978-90 are illustrated in Figure 2.1 and summarized below: Rural Income and Poverty Poverty Line Rural Income ---------------- Poverty (Y per capita) Share of Incidence -------------- Nominal --------- Year Real Nominal Yuan Income (%) million ------------------------------------------------------- 1978 134 134 -98 73.4% 260 1980 180 191 134 70.0% 218 1981 202 223 158 70.7% 194 1982 241 270 167 61.8% 140 1983 274 310 175 56.5% 123 1984 311 355 179 50.4% 89 1985 324 398 190 47.8% 96 1986 325 424 199 47.0% 97 1987 335 463 210 45.4% 91 1988 336 545 231 42.4% 86 1989 313 602 262 43.6% 103 1990 319 630 275 43.7% 97 Average Annual Growth Rate (%) 1978-90 7.5% 13.8% 9.0% -4.2% -7.9% 1978-84 15.1% 17.7% 10.6% -6.1% -16.4% 1984-90 0.4% 10.0% 7.4% -2.3% 1.4% ------------------------------------------------------- Source: SSB (1991a) and Table 1.2 for nominal and real real income. Annex 1 for poverty lines and incidence. 33 FIgure 2,.1: Rural Income and Poverty Cyuan per capita and milliTon people) 700 600 a & 500 - 400 - 4-, - 300 200- 00 1978 1980 1981 1982 1983 1984 1985 1986 1987 1988 1989 1990 Year 0 Rural Poverty Line + Nominal Income R Real Income A Poverty Incidence Source: SSB (1991a) for nominal income and Table 1.2 for real income. Annex 1 for poverty lines and incidence. The dramatic two-thirds decline in the incidence of rural poverty, from 260 million in 1978 to about 90 million in 1984, was driven by the rapid growth of rural real incomes, averaging as much as 15% per year. 91 Nominal rural income increased at an average annual rate of nearly 18% during these years, far above the 11% average annual increase in the poverty line. As a result, 19/ The 15% annual growth rate was derived by deflating nominal per capita income growth by the consumer price index. That index is widely thought to underestimate true inflation and an alternative estimate, based partially on adjustments to the consumer price index and partially on a reweighting of income sources, would have income growth at slightly above 11% per year over the period (Travers. 1984). The somewhat lower income growth rate does not alter the analysis. 34 the estimated absolute poverty line declined from 73% of nominal per capita rural income in 1978 to 50% by 1984. Real income growth was, by comparison, minimal during 1984-90, and during that period the poverty line remained roughly constant as a share of nominal rural per capita income -- ranging between 42% and 48%. The incidence of rural poverty correspondingly stagnated between 86 and 103 million, or from 10.4% to 12.3% of the rural population. 2.22 Income Ineaualitv. Most observers agree that the early-1950s national land reform and the subsequent agricultural collectivization program, which culminated with the implementation of the commune system in 1958, sharply reduced differences in land and other asset ownership as a source of income inequality at the local level, but also conclude that land reform and the commune system did little to reduce interregional rural disparity (e.g., Blecher 1976, Khan 1977, Roll 1976, Schran 1969, Selden 1985, and Vermeer 1982). Lardy (1978) has argued that the concentration of modern industrial inputs in regions with the greatest potential for agricultural growth may have actually intensified rural inequalities during the 1960s and 1970s. Furthermore, the World Bank (1983) has noted that tight controls on intrarural and rural-to-urban migration in China constrained an important income equalizing force, and concluded that in 1979 "interprovincial differences in agricultural output per capita were of major distributional significance." 2.23 There is conflicting evidence regarding trends in inequality during the burst of rural income growth from 1978 to 1984, but the subsequent period of income stagnation was almost certainly accompanied by increasing inequality. The SSB's unofficial Gini coefficients for rural income increased steadily from 0.21 in 1978 to 0.26 in 1984, indicating a substantial increase in inequality during these years. However, the large proportion of the rural population concentrated in the lowest income intervals in the 1978 and 1980-83 distribution data severely limits the accuracy of Gini coefficients estimated from the SSB rural income distribution data for those years. I Using the published SSB rural income distribution data, for example, Selden estimated Gini coefficients which instead decline from 0.28 in 1978 to 0.22 in 1983. 2 Furthermore, an examination of the SSB's provincially disaggregated average 20/ Furthermore, the SSB rural household sample size expanded rapidly over the period 1978-1985. By 1985, when the sample size exceeded 66,000 households, observers agree that it was well designed. One improvement in the urban sample as it also grew over those same years was its coverage of a more diverse population (capturing, in low income groups, retired people and the unemployed, and among high income people those deriving their income from private enterprise). If the rural sample behaved in the same way, measures of inequality would tend to grow even if the sample means accurately estimated the population mean. 21/ Based on more detailed province-specific rural income distribution data, Nee found a slight increase in inequality in Fujian during 1975-85 and a more pronounced decline in income inequality in Anhui during 1978-85. Similarly, income data in Hunan Statistical Bureau show a slight decrease in inequality in that province from 1978 to 1980, with a return to the 1978 level by 1984. 35 per capital rural income (reported in Table 2.2) identifies a negative, statistically significant relationship between 1980 average per capita rural income and provincial income growth rates during 1980-84. That is, on a provinciaL basis, income gaps narrowed during the period. 2.24 The SSB's Gini coefficients for 1985-90, by comparison, have been estimated from very large data sets (averaging more than 66,000 households annually) disaggregated by Y10 income intervals. These Gini coefficients accurately document a significant increase in rural income disparity during the second half of the 1980s. This finding is supported by the provincially disaggregated income results, which show a highly significant positive correlation between 1985 average per capita rural income and provincial income growth rates during 1985-89. Adjusting for provincial population, the SSB provincial rural income data document a narrowing of interprovincial income differentiLals during 1980-84 followed by an even greater widening of those differentials during 1985-89. D Despite the evident widening of income differentials during the second half of the decade, overall rural inequality in China remained fairly modest by developing country standards throughout the 19809. 2 2.25 The mechanism likely responsible for these trends in income disparity has been identified by Rozelle in an analysis of Jiangsu provincial data. Rozelle disaggregated changes in income by source and was able to demonstrate that in Jiangsu income increases deriving from agricultural activities reduced inequality. On the other hand, income increases generated by non-farm activity, particularly rural enterprises, increased inequality. Nationally, rural income growth in the 1978-1984 period was driven primarily by increased agricultural prices, output, and marketing. Agricultural growth fell substantially after 1984 (para 1.08), while rural enterprise and other non-farm income sources appear to have generated such growth as there was. Rozelle's analysis fits well with this pattern of growth and the observed changes in the Gini coefficient. 2.26 National average measures of inequality mask substantial differencBes among provinces in the degree of intraprovincial income inequality. Provincial Gini coefficients for rural income, estimated from data for mean income and share of households in each of 12 income intervals, ranged from 0.209 in Shanghai to 0.356 in Xinjiang in 1989 (Table 2.2). The highest Gini coefficients are found in those provinces relying mainly on rainfed agriculture, particularly the belt from Heilongjiang through Xinjiang. The observed pattern of intraprovincial inequality helps explain the 22/ Population-weighted correlation coefficients between provincial average per capita rural income growth rates during 1980-89, 1980-84 and 1985-89 and income levels in 1980, 1980 and 1985 are 0.25, -0.11 and 0.56 respectively. 23/ The SSB's unofficial Gini coefficients for China's rural income distribution are slightly less than Gini coefficients for rural income in other Asian developing countries such as India, Pakistan, Sri Lanka and Pakistan (World Bank, 1983). 36 prevalence of poverty in some provinces, especially the Northeast, where relatively high average income levels are associated with greater inequality. 2.27 The provincial Gini coefficients are uninformative on the issue of whether every area within a province has roughly the same average income and distribution, or whether the Gini is generated by a pattern of distribution in which average incomes (and possibly their distribution) also vary widely among localities. The distinction is important in designing interventions to ameliorate poverty. In the former case, the effort must be very broad based and will probably be oriented on individual families. In the latter case, a regional development focus on a relatively small number of places would be more appropriate. A review of the range of county average rural incomes, summarized below, shows very sharp differences in those incomes among counties within each of the several provinces for which such data are available: County Level Averaae Rural Per Capita Income Province Year Highest Lowest Ratio Gansu 1988 1251 196 6.38 1984 773 88 8.78 Ningxia 1989 896 194 4.62 1985 532 156 3.41 Anhui 1989 1189 351 3.39 Guangxi 1988 720 240 3.00 Hunan 1986 1029 259 3.97 Source: Gansu (Ningxia, Anhui, Guanaxi and Hunan) Statistical Bureau. This simple evidence helps validate the area-based development strategy chosen in the Chinese poverty amelioration effort. 2.28 Reaional Concentration of Poverty. Consistent with the post-1984 pattern of slower growth of rural income in provinces with lower rural income levels (para 2.24), the SSB 1989 provincial rural income distribution data show that most of China's remaining rural poverty is concentrated in the northwestern, southwestern, northeastern and northern regions of the country. Provincial incidence of rural poverty in 1989, estimated by applying the 1989 national poverty line (Y262) to the SSB 1989 data, is reported in Table 2.2 (discussed in detail in Annex 1) and illustrated in Map 1 -- "1989 Incidence of Rural Poverty (SSB Data)" -- to this report. The data indicate that in 1989 the rate of absolute poverty among rural households ranged between 17% to 34% in the northwestern provinces of Inner Mongolia, Shanxi, Shaanxi, Ningxia, Table 2.2: Provincial Indicators of Social Development RuraL Income 1989 Incidence of HeaLth (yuan per capita and % change) Rural Poverty --------------------------------------- Rural ------------------- Life With Access ReaL Gini Share of Share of IlLiteracy Expectancy Birth to Safe Water Increase ----- Provincial TotaL Poor --------- a t Birth Rate (X 1987 Rural 1980 1984 1985 1989 1980-89 1989 HousehoLds PopuLation 1990 1981 1990 Population) National Average 191.3 355.3 397.6 601.5 75X 0.31 11.41 100.0X 22.3X 67.8 21.1 60.2X North BeiJing 290.5 664.1 775 1 1230.6 136X 0.224 0.24 0.0% 11." X 72.0 13.0 90.4X Tianjin 277.9 504.6 564.6 1020.3 1041 0.240 0.41 0.0% 11.71 71.1 15.6 98.01 Hebei 175.8 345.0 385.2 589.4 871 0.312 13.01 7.1X 21.91 70.7 20.5 81.41 Henan 160.8 301.3 329.4 457.1 581 0.272 16.51 12.71 22.9X 69.8 24.9 73.31 Shandong 194.3 404.2 408.1 630.6 811 0.279 6.8X 5.0X 23.11 70.2 18.2 71.91 Northeast Liaoning 273.0 477.4 467.8 740.2 51X 0.313 8.01 1.91 11.6% 70.8 16.3 83.01 Jitin 236.3 486.8 413.7 624.0 471 0.321 12.21 1.91 14.31 69.0 19.5 72.51 Hei Longjiang 205.4 443.2 397.8 535.2 451 0.332 18.31 3.6% 15.01 68.3 18.1 77.5X Northwest Inrner. Mongolti a 181.3 336.1 360.4 477.5 471 0.332 23.51 3.61 21.91 67.0 21.2 42.71 Shanxi 155.8 350.5 358.2 513.9 841 0.313 17.41 4.11 16.51 67.9 22.5 73.01 Shaanxi 142.5 262.5 295.3 433.7 701 0.281 20.31 5.81 25.31 65.2 23.5 25.51 Ningxia 178.1 313.2 321.2 521.9 631 0.319 18.91 0.71 32.21 65.9 24.3 56.31 Gansu 153.3 221.1 255.2 365.9 331 0.272 34.21 6.71 39.81 65.1 20.7 26.91 Qinghai 264.2 343.0 457.5 0.328 23.71 0.81 40.61 61.6 24.3 44. 7 Xinjiang 198.0 362.7 394.3 545.6 531 0.356 18.7X 1.61 19.31 61.3 26.4 44.31 Yangtze River Shanghai 397.4 785.1 805.9 1380.0 931 0.209 O.01 0.01 13.41 73.0 10.3 81.21 Jiangsu 217.9 447.9 492.6 875.7 1241 0.302 3.41 1.91 22.6% 69.6 20.5 57.01 Zhejiang 219.2 446.4 548.6 1010.7 1571 0.278 2.01 0.81 23.21 69.7 15.3 67.81 Anhui 184.8 323.0 369.4 515.8 551 0.227 7.71 3.96 34.41 69.3 24.5 69.91 eianxi 180.9 334.4 377.3 558.6 721 0.226 5.1 1.61 23.81 66.3 24661.91 Hubei 169.9 392.3 421.2 571.8 871 0.245 6.0X 2.6% 22.21 65.8 21.6 42.41 Hunan 219.7 348.2 395.3 558.3 421 0.235 6.21 3.51 17.11 64.5 23.9 86.41 South Fujian 171.7 344.9 396.5 697.3 1261 0.231 1.81 0.51 22.81 68.6 24.4 62.21 Gaingdong 274.4 425.3 495.3 955.0 941 0.274 0.91 0.51 15.21 68.6 22.3 65.91 Hairnan 674.3 0.273 3.31 0.21 21.31 24.9 Southwet Guangxi 173.7 267.2 303.0 483.0 551 0.283 15.41 6.11 16.51 68.3 20.2 43.61 Sichuan 187.9 286.8 315.1 494.1 461 0.258 11.21 11.21 21.41 64.3 19.1 40.71 Guizhou 161.5 260.7 287.8 430.3 481 0.257 17.8X 5.41 36.41 61.9 23.1 25.51 Yuman 150.1 310.4 338.3 477.9 77X 0.302 19.01 6.51 37.51 61.5 23.6 30.01 Tibet 353.0 397.3 0.309 67.61 24.0 Sources: SSB (1991a) for rural income illiteracy and birth rates, Annex 1 for incidence of rural poverty, Banister for life expectancy, and DUTCEF for access to safe water. 38 Gansu, Qinghai and Xinjiang, between 11% to 19% in the southwestern provinces of Guangxi, Sichuan, Guizhou and Yunnan, and between 12% to 18% in the northern provinces of Hebei and Henan and the northeastern provinces of Jilin and Heilongjiang. These 15 provinces, with less than 50% of China's rural population, accounted for nearly 80% of rural poverty (about 80 million poor). The SSB provincial income data also indicate that China's worst absolute poverty -- those 4.6% of rural households with average income levels of less than Y200 per capita -- is concentrated in these same provinces. - 2.29 With more than 34% of rural households falling below the national poverty line, Gansu had the greatest rate of absolute poverty in 1989. The rate of poverty was less than 24%, by comparison, in all other provinces. As a region, the northwestern provinces had the highest rate of poverty, averaging nearly 23% of rural households, and together these provinces accounted for 23% of all rural poverty. The rate of poverty in the two northern provinces of Hebei and Henan averaged 15% and, reflecting their large share of national rural population, together they accounted for about 20% of rural poverty -- just slightly less than all seven northwestern provinces. Corresponding to the province's large rural population and greater than average rate of poverty, the rural poor in Henan are estimated to have numbered about 13 million in 1989 -- more (in absolute terms) than in any other province. The rate of poverty in the two northeastern provinces of Jilin and Heilongjiang averaged about 16% but, because of their lesser share of rural population, they accounted for less than 6% of rural poverty. The rate of poverty in the southwestern provinces (excluding Tibet) averaged about 14% and together they accounted for about 29% of rural poverty. The figures also indicate modest levels of poverty, of between 5% to 8% of rural households, in Shandong, Liaoning, and the interior provinces of Anhui, Jiangxi, Hubei, and Hunan. Together these six provinces accounted for another 19% of the national total. The incidence of rural poverty in Beijing, Tianjin and Shanghai and the coastal provinces (except Liaoning and Shandong) was minimal in 1989. 2.30 The provincial disaggregation of poverty must be interpreted with caution since rural income in any one year can be greatly influenced by local variations in weather. In particular, rural income levels in the northeastern provinces of Jilin and Heilongjiang and the northwestern provinces of Qinghai and Inner Mongolia were far below trend in 1989 -- nominal rural income in 1989 was less than the 1988 level in all four provinces -- clearly indicating that the 1989 provincial poverty estimates overstate the depth of poverty in these provinces. Furthermore, the provincial estimates do not correct for price differences between provinces. Preliminary estimates of provincial 24/ There is a strong positive correlation between provincial percentage of rural households with average income levels less than Y262 per capita and provincial percentage of rural households with average income levels of less than Y200 per capita. Sichuan -- where the poverty rate of 11.2% is very close to the national rate of 11.4%, but only 3% of the household incomes fall below Y200, in contrast to the national rate of 4.6% -- appears to be the only significant exception to this rule. 39 incidence of poverty, calculated on the basis of province-specific poverty lines (Annex 1), confirm the regional concentration of poverty in western and northern China, but indicate somewhat lower rates of poverty in northwestern, northeastern and northern China and slightly increased rates in the central provinces and Guangxi. 2.31 An Alternative View: MOA Rural Income Data. Ministry of Agriculture (MOA) county-average rural per capita income data for 1985 and 1989, summarized by province and region in Table 2.3 and illustrated in Map 2 -- "1989 Incidence of Rural Poverty (MOA Data)" -- to this report, evidence a significantly dissimilar regionalization of poverty during the second half of the 1980s. For each province, the MOA data specify the number of counties (and the total rural population of those counties) with average per capita rural income levels of less than Y200 and Y250 in 1989 and less than Y150 and Y200 in 1985 (hereafter, "lower-income counties"1 ). Since the data do not include intracounty distribution of income, they can be used to derive only very rough estimates of the number of poor. Nevertheless, the NOA data have played a key role in the government's assessment of the location of poverty and the targeting of poverty alleviation funding. The following review of the MOA data therefore clarifies the government's perception of the location of absolute poverty and evidences the real need to upgrade China's national and provincial poverty assessment and monitoring system. 2.32 In contrast to the 1989 SSB provincial rural income distribution data, the 1989 MOA data suggest a much greater rate of poverty in southwestern China and substantially lower rates in northwestern and northern China. Specifically, the MOA data show that 19%, 10% and 2% (1.0:0.5:0.1) respectively of the rural population in southwestern, northwestern and northern China resided in lower-income counties (less than Y250). Reflecting the large population base and greater number of lower-income counties, the rural population of the lower-income counties of southwestern China comprised 60% of the total population residing in all lowez-income counties. The rural population of the lower-income counties of northwestern and northern China accounted for another 17% and 8% of the rural population residing in lower- income counties. The SSB data, by comparison, indicate that (i) 14%, 23% and 12% (1.0:1.6:0.8) of the rural households in southwestern, northwestern and northern China had per capita income less than the national poverty line of Y262 and (ii) the poor of these three regions represented 29%, 23% and 25% respectively of the national total. >} 25/ The poor performance of agriculture in Jilin and Heilongjiang in 1989 (para 2.30) explains much of the difference between this report's estimates (of above average incidence) and the government's perception (of very modest levels) of poverty in these provinces. The MOA statistics indicate that less than 2% of the rural population of these two provinces resided in lower-income counties (less than Y250) in 1989, confirming the perception of modest poverty in Jilin and Heilongjiang. It is possible, however, that the MOA data might instead be consistent with an unusual pattern of extensive household and village poverty within counties with average income levels exceeding the poverty line (and therefore not identified as lower-income in the MOA data). Table 2.3: MOA County Per Capita Income Data 1985 1989 gricultural employment figures do not adequately distinguish underemployment in the sector. The figures are widely recognized to treat all rural workers not employed in rural off-farm pursuits as being fully employed in agriculture, and thereby mask mounting underemployment of the agricultural working aige population. Accounting for this increasing underemployment, available evidence (presented in paras 3.39 and 3.40) indicates that agricultural labor requirements stagnated or at most increased by a small margin during the second half of the 1980s. Assuming that agricultural employment did not increase at all during 1985-90, the balance between labor supply and demand would have deteriorated significantly during the 1980s -- holding agricultural employment constant at the 1985 level (304 million), total employment appears to have actually declined from 82.7% of the working age population in 1978 to 77.1% in 1990. 3.09 Urban Emplovment. In 1978, the state, collective and private sectors provided 78.3%, 21.6% and 0.2% respectively of total employment in the urban economy. Due to difference in their employment growth rates, their shares of labor absorption during 1978-90 -- 55.5%, 28.8% and 12.6% respectively -- differ significantly from their 1978 employment shares. ~ Prior to the reforms of the urban employment system initiated in 1980, the state sector was required to provide life-time employment to the number of workers specified by administrative quota. 5J Adherence to the employment quota was at least in principle mandatory, and state-owned enterprises (SOEs) were restricted to the employees assigned to them by the government. An official employment system reform directive issued in 1980 allowed workers and, to a lesser extent, the SOEs greater latitude in seeking employment in ' Jointly-operated (state, collective, private and other) enterprises accounted for the remaining 3.1% of labor absorption in the urban economy during 1978-90. it Reform of the urban employment system is discussed in detail in World Bank (1992a). 54 and recruiting employees from job fairs, exchange centers for skilled workers and, in some cities, the public domain. The labor contracting system was also first tested in a number of cities in that year. While these and subsequent reforms introduced greater flexibility in sourcing workers, the Ministry of Labor (MOL) and the local Labor Bureaus have retained control of overall employment growth in the state sector. In conformance with MOL labor requirement plans, derived from comprehensive labor and enterprise planning, employment growth in the state sector remained relatively stable throughout 1978-90. The use of temporary workers in the SOEs, which is not covered by the labor plan, has expanded in recent years. F Since they are neither guaranteed permanent employment nor eligible for the wide range of benefits offered to permanent staff, temporary workers provide SOE managers with greater flexibility and considerable cost savings. In order to contain unemployment within planned targets, however, urban Labor Bureau authorities continue to pressure the SOEs to restrict their use of temporary rural workers in favor of employment of permanent staff with urban registration. 3.10 The employment growth rate of the urban-based collective sector exceeded that of the state sector during 1978-90, though it declined sharply from 7.2% during 1978-85 to only 1.3% during 1985-90 and provided no new jobs during 1988-90. Collectively-owned enterprises (COEs) are more market oriented in their employment practices, with most of their workers recruited from the job fairs organized by local Employment Bureaus. 3.11 Employment in the urban-based private sector increased at an average annual rate of 37.3% during 1978-90, and increased as a share of total urban employment from 0.2% in 1978 to 4.5% in 1990. Prior to 1981, the existence of urban privately-owned enterprises (POEs) was actively discouraged by government sanctions and policy. As a result, private sector employment declined from 9.0 million in 1953 to less than 0.2 million in 1978. These restrictions weakened in the late-1970s and were formally lifted in 1981, and POE employment rebounded at an average annual rate of 46% during 1978-88. POE employment growth faltered during the austerity period of 1988-90, and averaged 8.3% annually during 1985-90. Despite the strong growth since 1978, total private sector employment amounted to only 6.7 million in 1990 -- just three-quarters the 1953 level of 9.0 million. -Expected Labor Market Trends 3.12 Given the expected sharp decline in the growth rate of the active working age population during the 1990s (para 3.06), even the vezy modest growth of employment expected during the decade will be sufficient to significantly reduce the current surplus of rural labor. The active working age population will increase by about 76 million during the 1990s, at an average annual rate of only 1.3%. The conservative estimates discussed below suggest that total employment growth will amount to at least 83 million during F Official urban employment figures often exclude those temporary and seasonal rural workers not recruited through official channels, and therefore understate actual labor absorption in the urban economy during the 1980s. 55 the decade, averaging about 1.4% annually. 7/ Contrary to MOL and MOA fears of significant increases in underemployment during the 1990s, employment growth is therefore expected to exceed the expansion of the active working age population by between 5 to 10 million. F Assuming (i) employment growth in the urban economy equals the relatively depressed rate of 1985-90, (ii) no growth in agricultural employment, and (iii) modest growth (considerably less than the average annual rate of 1985-90) in rural nonagricultural employment, total employment is expected to increase from 81.4% of the 1990 total working age population to 82.6% in 2000. 3.13 Urban Employment during the 1990s. Annual employment growth in the urban economy during the 1990s can be expected to at least equal the 1985-90 average rate of 2.8% (Table 3.1). Depressed by the sharp decline in the employment growth rates of the collective and private sectors, the 1985-90 growth rate is significantly less than that achieved during 1978-85 and the overall rate for 1978-90. But even at this lower rate, the urban economy would provide an additional 48 million jobs and account for about 60% of expected labor absorption during the 1990s -- nearly double its share of total labor absorption during 1978-90. Employment growth within the state sector is projected to increase at about the same rate as the urban economy as a whole, and would account for 70% of total labor absorption in the urban economy. Prior to the reforms of the 1980s, the ranking system of state enterprises gave managers considerable incentive to maximize employment, and was partly responsible for the current overstaffing in the SOEs. Recent enterprise and labor reforms have limited this pro-employment bias. The contract responsibility system, most importantly, now holds managers responsible for both profitability and the total wage bill. Managers now have a stronger incentive to limit employment in order to maximize the per employee distribution of allowed wage and bonuses increase. Despite these reforms, the MOL employment plans have guided a fairly steady increase in SOE employment throughout the 1980s. It is certainly possible that enterprise reform and restructuring during the 1990s may depress employment growth within the state sector below the 1985-90 rate, but such a decline would likely be at least Y' GNP is expected to increase at an average annual rate of between 6% to 7.5% during 1990-2000 (World Bank, 1992b). On the basis of the average employment-GNP elasticity for 1978-90 of 0.33, that level of GNP growth would be consistent with an employment growth rate of between 2.0% to 2.5% -- much greater than the growth rate of 1.4% projected in this report. F MOA estimates, for example, suggest that expected productivity gains in agriculturet during the 1990s will push underemployment of the rural working age population from about 100 million at present to between 160 to 200 million in 2000 ("E'ewer Jobs for Rural Labourers" in China Daily, June 14, 1991). Similarly, the State Council's Rural Research Center believes that surplus labor in thLe rural economy will increase from 100 million at present to between 146 to 183 million ("Surplus Workers in the Spotlight" in China Daily, June 21, 1991). However, both sets of estimates appear to grossly overstate expected growth in the active working age population (see footnote 3 to this chapter). 56 partially offset by a corresponding reform-induced surge in COE and POE employment. 3.14 The employment growth rate within the collective sector was more than twice that of the state sector during 1978-85, but declined sharply during 1985-90 to a level less than half that of the state sector. Employment growth within the collective sector is projected to increase at the modest 1985-90 annual rate of 1.3%, and would account for 10% of expected labor absorption in the urban economy. The rapid growth of employment in the collective sector during 1978-85 suggests that actual COE employment growth during the 1990s may exceed the projection level. The very strong growth of employment in the private sector during 1978-88, and the fact that POE employment levels have still not recovered to the levels of the early-1950s (para 3.11), suggest considerable potential for continued POE employment growth during the 1990s. Projected to increase at the 1985-90 growth rate of 8.3%, private sector employment would account for 17% of expected labor absorption in the urban economy. Should China continue its program of enterprise reform and restructuring during the 1990s, it is likely that existing barriers to growth and discriminatory practices against the COEs and POEs will be further eroded. If so, employment growth with the collective and private sectors would probably exceed the projection levels. COEs and POEs are heavily engaged in the service sector, and the currently modest levels of employment within the service sector are indicative of the employment growth potential within the collective and private sectors. The service sector's share of total employment in China (16% in 1985) is much less than in other populous developing nations such as Indonesia (30% in 1982), Philippines (37% in 1985), and Brazil (47% in 1982). 9' 3.15 Rural Employment during the 1990s. Employment growth in the rural economy during the 1990s is expected to fall sharply from levels achieved during the 1980s. Assuming no growth in agricultural employment and modest growth in rural nonagricultural employment, the employment growth rate in the rural economy is projected to decline to only 0.8% during the 1990s, less than one-third the 1985-90 average annual growth rate of 2.5% (Table 3.1). At this low rate of growth, the rural economy would provide an additional 35 million jobs and account for 42% of expected labor absorption during the 1990s. In contrast to official agricultural employment figures, which indicate a steady growth of 1.9% annually during 1985-90, no employment growth in agriculture is projected to occur during the 1990s (para 3.41). This conservative estimate, which holds agricultural employment in 2000 constant at the 1990 level of 333 million, assumes that labor productivity gains will fully balance expected 2/ The service sector's share of total employment also varies widely across provinces within China. Service employment is greatest in the three municipalities and the developed coastal provinces, accounting for 40%, 29%, 29% and 21% respectively of total employment in Beijing, Shanghai, Tianjin and Guangdong. Service sector employment is particularly limited, by comparison, in the southwestern provinces, where it accounts for only 11% to 12% of total employment in Guangxi, Sichuan, Guizhou and Yunnan (Jefferson and Rawski, 1991). 57 growth of agriculture sector output during the 1990s. L' Assuming constant agricultural employment, labor absorption in the rural economy will be entirely dependent upon nonagricultural employment growth. Based on MOA estimates, the expected expansion of rural nonagricultural production will provide an additional 35 million rural jobs by 2000. The MOA estimate appears to be conservative, since it implies an average annual growth rate of nonagricultural employment of 3.5%, much less than the relatively sluggish growth of 1985-90 and less than half that during 1978-90. -Labor Mobility and Poverty Alleviation 3.16 Overview. The impact of the projected gradual reduction of surplus labor during the 1990s on the employment and income prospects of the poor will depend largely on the location of job creation and the ability of the poor to migrate to emerging employment opportunities outside their home village and township. Available evidence suggests that the expansion of nonagricultural employment in rural enterprise (RE) was very modest in poor counties and, in particular, poor townships and villages during 1978-90. Since this pattern of unequal development is expected to continue, it is most likely that the poor will be able to take direct advantage of employment in developing RE in the 1990s primarily through rural-to-rural and rural-to-urban out-migration. As discussed below, however, intercounty and interprovincial out-migration from the poor areas has heretofore been fairly limited. 3.17 Rural-to-Rural Out-Migration. The majority of new employment opportunities in RE appear to have been filled locally during the 1980s. A joint 1985 World Bank - Chinese Academy of Social Sciences (CASS) study, for example, found that workers from other counties comprised only between 3% to 15% of RE employment in the four counties surveyed. In three of the four counties, between 93% to 97% of the RE work force were sourced from within the same county and between 80% to 94% from within the same township and village (Byrd and Gelb). Similarly, an independent 1986 CASS survey of 222 villages in 11 provinces found that of the 37% of the rural work force engaged in nonagricultural pursuits, 62% were employed in RE within their home township, at least another 8% were so employed within their own county, about 12% were employed in RE outside their own county, and another 13% were employed in urban areas outside their own county (Yu). However, for the poor western provinces, these national averages overstate both the share of rural workers able to find off-farm work and, in particular, the share leaving their home township to work in RE. In Ningxia, for example, the 1986 CASS survey shows that only 28% of the rural work force were engaged in nonagricultural pursuits. Of these, 79% were employed in RE within their home township, at least another 3% were so employed within their own county, only 4% were employed in RE outside their own county, and another 13% were employed in urban areas outside their own county. .01 In the absence of agricultural labor productivity growth, agricultural employment would be expected to increase at an average annual rate of 1.0% during the 1990s (para 3.41). 58 3.18 The 1986 CASS survey also found that 75% of those employed in nonagricultural pursuits were male and that 57% were employed on a seasonal basis. Seasonal employment is even more dominant in the western provinces, and accounted for 80% of nonagricultural employment in the villages surveyed in Ningxia. Very few of those employed either within or outside agriculture were well educated in 1986. Nationally, about two-thirds of those employed in nonagricultural pursuits were either illiterate, semi-illiterate, or had only primary school education. Another 25% had a middle school education, and less than 10% had a high school or better education. These rates of educational attainment are nearly the same as those of the entire surveyed rural work force. Educational attainment in the three western provinces surveyed (Ningxia, Qinghai and Guangxi) was even less impressive -- 74% of those employed in off-farm pursuits (and 85% of the surveyed rural work force) were either illiterate, semi-illiterate, or had only primary school education, and only 5% had a high school or better education. Surprisingly, educational attainment was not a characteristic of those able to find nonagricultural employment outside their home township -- 72% of those rural workers employed in nonagricultural pursuits outside their home township were either illiterate, semi-illiterate, or had only primary school education, 6% more than among the general surveyed rural work force and 12% more than those employed in RH within their home township. The 1986 CASS survey also shows that nearly two-thirds of those employed in nonagricultural pursuits outside their home townships found those jobs on their own. More than 80% of rural workers in western provinces employed in nonagricultural pursuits outside their home townships found those jobs on their own, or nearly twice the proportion of 44% observed for the four eastern provinces surveyed (Shanghai, Jiangsu, Zhejiang and Fujian). 3.19 Rural-to-Urban Labor Mobility. Tight government control, through the household registration system and the urban grain ration system, has limited rural-to-urban migration to something considerably less than its potential. These controls were first relaxed in the late-1970s for millions of "rusticated" young adults who had been sent to rural and border areas during the Cultural Revolution period (1966-76). The return of these young people, which was completed by the mid-1980s, led to a temporary surge in the urban working age population and unemployment. With the reduction of urban unemployment levels to politically acceptable levels by 1982-83, the government eased controls and allowed millions of rural inhabitants to move temporarily or permanently to towns or cities. In 1983, the government allowed temporary migration from villages to townships provided the migrants retained their village household registration. In 1984, permanent migration from villages to urban areas was approved, provided the migrants supplied their own food grain. In 1986, grain sales at negotiated prices to rural laborers working in urban areas were legalized. 3.20 Available data are not sufficient to allow accurate quantification of total rural-to-urban migration during the 1980s. Official figures indicate that the urban population increased by 4.8% annually during 1978-90 (para 1.18). Natural increase, that is more births than deaths, is known to have accounted for about 1% of this growth. The establishment of new cities and urban towns and rural-to-urban migration account for the remainder of the 59 increase, but it iB still impossible to disentangle their individual contributions. Several surveys during the 1980s have shown that, after the return of the rusticated youth, long-term rural-to-urban migration was not a major factor in urban population growth. The 1987 One Percent Sample Census, for example, shows that long-term rural-to-urban migration during 1982-87, including those who had been in their present residence for six months or longer (either with or without a change in permanent residence registration), accounted for about 4.2% of 1987 total urban population (or about 12 million urban inhabitants in that year). However, since no census has counted the floating population who had been away from their registration location for less than 6 months, statistics for the temporary or seasonal movement of labor are particularly weak. The significance of the floating population in interregional labor flows is underscored by the 1986 CASS survey, which shows that temporary and seasonal migration (of less than six months duration) accounted for more than half of rural laborers employed in nonagricultural work outside their home township (Yu). While there remains much uncertainty about the size and composition of the floating population, perhaps as many as 50 million rural "temporaries" were working or seeking work in urban areas by end-1989. However, rural-to-urban migration is believed to have temporarily declined in 1989 and 1990 when provincial governments vigorously sent millions of former rural residents -- particularly the temporaries -- back to the countryside. 3.21 Auamentina Labor Mobility. Obstacles to rural labor mobility include administrative and punitive economic measures adopted by local governments, lack of information regarding outside job opportunities, and social and cultural traditions. The 1985 World Bank - CASS study found that local governments are inclined to restrain the inflow of external labor in order to maximize local employment and incomes. The ability of local governments to restrict the inflow of labor is largely determined by the structure of RE ownership -- in counties where local government dominates RE ownership, the government is able to dictate both sourcing of labor and wage rates. Obstacles to the free flow of labor are so great in some instances that "labor costs in some rural areas have skyrocketed and even surpassed the salary level in local state enterprises" (Mena). The World Bank - CASS survey report concluded that "the low level of development of interregional labor markets is harmful to the development" of RE, and cited numerous instances where high labor costs had forced REs to prematurely purchase labor saving equipment, switch out of labor intensive activities, or cease production altogether. 3.22 Despite these obstacles, organized and unorganized labor migration did gain some momentum during the 1980s. Several of the western provinces have established active programs to find jobs for their surplus rural labor in the more developed areas of their own provinces and in the coastal provinces. Sichuan is the nation's leading exporter of rural laborers and, in 1989, provincial labor export agencies reportedly arranged employment for some 600,000 rural workers outside the province's border. These exported workers comprised about 5% of the province's surplus rural working age population in that year. A much greater number of rural laborers are known to have found employment in other provinces without the assistance of government 60 authorities. In Gansu, active efforts to organize labor exports from resource poor areas began in 1986. From 1986 to 1990, more than 5.3 million workers were recruited to work outside their home township and reportedly earned a total of Y3.2 billion in wage income. More than 70% of the organized migrant laborers found employment within Gansu. A small number of those provided employment within the province were voluntarily resettled from poor upland areas to newly irrigated areas within the province (paras 3.52 to 3.54). Overall, about two-thirds of the recruited laborers originated from extremely poor counties. The number of organized migrant laborers during the last three years was constant at 1.2 million and, in 1990, about 37% of the migrants were recruited from poor upland areas. Available evidence suggests that a much greater number of rural laborers found outside employment without the assistance of government authorities. However, since only a limited number of migrant laborers were employed outside the province on a long term basis, there has only been a modest impact on the total labor supply within the province. 3.23 Realizing the potential for augmenting economic growth and efficiency while reducing income disparity and poverty, the MOA and MOL have recently joined together to initiate an experimental program to foster rural employment and labor mobility. Under the direction of the State Council, staff of the two ministries decided in May, 1991 to launch 22 rural labor market and mobility experimental projects. Building upon the successful experience with labor export management in Linquan County in Anhui (OdQ), nine of the experimental projects are now underway. The overall objective of the program is to facilitate the flow (or "export") of labor from over-populated and resource poor areas to rural areas of the developed coastal provinces. In the experimental projects, Labor Export Management Offices (LEMOs) from the labor surplus regions seek to identify potential employment opportunities in labor deficit areas. Many of these LEMOs have established representative offices in major cities throughout the nation. When a labor importer is identified, the LEMO negotiates a one to three year contract detailing the level of educational attainment or other qualifications the candidate employees must meet, pay scales, housing conditions, medical and accident insurance, and any on-the-job training to be provided. The LEMO then recruits appropriate candidates from its labor surplus area and arranges for any necessary predeparture training. B. Rural Enterprise -Overview 3.24 Development Trends. Rural enterprises (RE) expanded at a remarkable rate during the 1980s and now play a substantial role in China's economy. -' In the late-1970s, some 1.5 million REs provided employment to about 30 million workers -- less than 7% of the nation's total work force -- ill The development of RE in China is discussed in detail in gdy and Bvrd and Lin. 61 and accounted for about 7% of national gross output value. The gross value of RE output increased in real terms at an average annual rate of 22.5% during 1979-90, surpassing the gross value of agricultural output for the first time in 1987. By 1990, RE accounted for more than 22% of national gross output value and about 30% of national industrial output value. The MOA recently reported that 34% of the country's coal, 63% of its nylon, 60% of its garments, 80% of its farm machinery and equipment, and more than 90% of its bricks are now produced by REs. 3.25 As summarized in Table 3.2, total RE employment increased by more than 60 million during 1978-90, growing at an average annual rate of 10.4%. At present, 18.5 million REs provide full or part time employment to 92.6 million people, or about 16% of the nation's total work force and 22% of the rural working age population. 2 Employment growth in rural industry accounted for about two-thirds of the total increase in RE employment. With the recognition of individual and private enterprises by the central government beginning in 1984, employment in these below-village units expanded by 34.4 million during 1984-90. Having increased at an average annual rate of 25%, below-village RE employment growth accounted for 85% of total RE employment growth, of 40.5 million, during 1984-90. The figures also indicate that the slowdown in RE during 1988-90 was associated with a 3% decline in total employment and a 10% decline in net profits. However, below-village RE employment remained constant and total RE nominal tax payments to the central government continued to increase during these recent years. W 3.26 RE Employment Growth Durina the 1990s. The 8th FYP targets REs to grow at an annual rate of 11% and to generate a total of 18 million new jobs by 1995. The MOA expects that real average annual growth of RE output value will be maintained at 11% or more throughout the decade, and that total employment generation will be at least 35 million by 2000. The expected RE employment growth rate during the 1990s, of 3.3% annually, appears moderate in '2 Perhaps because they classify those engaged mainly in agriculture and only secondarily in rural industry as rural enterprise employees, these employment figures modestly exceed the rural nonagricultural employment data reported in Table 1.2 for the years 1985-90. The difference between the two employment series is discussed in detail in Odv. A definitional change introduced in 1984 is known to have resulted in the inclusion of RE employees previously excluded from the employment series. Adjusting for this definitional change reduces the average annual RE employment growth rate during 1978-90 from 10.4% to 9.5%. 3/ Preliminary figures indicate that RE employment rebounded to 93 million in 1991 while gross output value, tax payments and net profits increased by the greater margins of 18%, 10% and 6% respectively ("Employment Rises in Rural Enterprises" in China Daily, December 30, 1991). 62 Table 3.2: Rural Enternrise Develooment. 1978-90 Financial Indicators Employment ----------------------------- -------------------------- Gross Tax Payments Number of Township Below- Output to Central Net Enterprises Total & Village Village Value Government Profits Year --------------- million --------------- ------- billion yuan -------- 1978/79 1.5 28.3 28.3 na 54.8 2.2 8.8 1984 6.1 52.1 39.8 12.3 171.0 7.9 12.9 1985 12.2 69.8 43.3 26.5 272.8 10.9 17.1 1986 15.2 79.4 45.4 34.0 354.1 13.7 16.1 1987 17.5 88.1 47.2 40.9 476.4 16.8 18.8 1988 18.9 95.5 48.9 46.5 649.6 23.7 25.9 1989 18.7 93.7 47.2 46.5 742.8 27.3 24.0 1990 18.5 92.6 45.9 46.7 846.1 27.6 23.3 Source: SSB (1991a). comparison with the 10.4% annual growth achieved during 1978-90. -W In light of the decline in employment and profitability during 1988-90, however, the planned growth rate may be realistic. If these planned rates of growth are realized, RE employment growth could provide jobs for fully half of the expected increase, of 70 million, in the participatory working age population during the 1990s (Table 3.1). However, the patterns of RE development during the 1980s discussed below suggest that RE development in the poor areas will remain fairly limited in the 1990s. Maximizing the participation of the poor in whatever RE employment growth obtains during the 1990s will therefore depend in large part on facilitating seasonal and long term migration of the poor from their home counties to RE employment opportunities in other counties and provinces. -National Trends 3.27 Background. Government policy has played a major role in the evolution of RE in China. In the early-1950s, rural industry was mainly staffed by farmers holding part-time jobs in the commercialized handicraft industry. Many of these workshops were very small scale, either because their technology or the organization of their work offered no economies of scale. During the commune era (1958-83), these workshops and individual handicraft workers were organized into a large number of commune, brigade and team enterprises. Prior to the late-1970s, the commune system ensured that W' The expected rate of RE employment growth also appears moderate in light of the average employment-output elasticity of 0.46 during 1978-90. Expected RE output and employment growth during the 1990s would be consistent with a decline in the employment-output elasticity to 0.30. 63 handicraft labor was a very low cost input to rural enterprises, since most compensation remained in the form of work points. 2' The egalitarian ideology of the time required that enterprises be collectively owned and, consequently, enterprise scale was in general not too small. The minimum scale was several employees -- enterprises that might work best as single proprietors simply were not allowed to function during the commune era. A variety of activities were profitably undertaken in this environment, including the production of building materials and some farm inputs, subcontracting of various sorts, and some simple processing of agricultural outputs. Overall, however, industrial and sideline production was subordinate to agricultural production, and RE development was quite modest during the commune era. 3.28 The 1980s. In addition to stimulating agricultural productivity, the widespread adoption of the PRS in the early-1980s altered the operation of the commune-run enterprise system. With the abolition of the work point system, nonagricultural labor began to be compensated with a wage at least equal to the marginal return to labor in agriculture. Individual unit accounts were now being kept, so enterprise losses became more visible at the very time when farmers' resistance to subsidizing administrative overhead was increasing. SOE monopoly and monopsony powers were eroded during this period, and opportunities in the urban economy were becoming more accessible to REs. Response to these liberalizations varied considerably and the pace of RE development diverged quite sharply among provinces. Collectively-run enterprises began to face a much sterner profitability test to justify their existence, and this test was failed much more frequently in the poor provinces. As shown in Table 3.3, growth of RE employment in most poor provinces of northwestern, northern and southwestern China lagged behind the national average during 1980-83. Furthermore, RE employment grew from a much smaller base in these provinces -- only 6.6% and 3.7% respectively of the rural working age population in northwestern and southwestern China were employed in RE in 1980, significantly less than the national average of 9.4%. As the adoption of PRS accelerated, seven of these provinces -- Inner Mongolia, Shaanxi, Ningxia, Xinjiang, Hebei, Guangxi and Sichuan -- experienced declines of between -1% to -18% in RE employment. In contrast, based on their superior achievements in primary and secondary education, well developed transport networks, and historically strong commercial and kinship ties with major urban areas, RE employment in the majority of the coastal provinces and the three municipalities increased by between 7% to more than 60%. .5S Under the work point system, any enterprise able to generate returns to labor in excess of the marginal product of labor made economic sense. Most rural areas had excess labor and the marginal product of labor was very low. 64 Table 3.3: Provincial Rural Enterprise Emoloyment. 1980-90 RE EmpLoyment (1000 empLoyees) Percent of Rural Change in RE Employment Working Age ------------------- 1990 Population 1980-83 1980-90 Annual --------------- EmpLoyed in RE ------- Growth Rate (X) Town & Below- ---------------- TotaL ---------------- 1980 1983 Village Village Total 1980 1990 (t) Total Share Total 29997 32346 45925 46722 92647 9.4 22.1 7.8 11.9 8.9 North Beijing 314 506 885 203 1088 19.1 59.1 61.1 13.2 11.9 Tianjin 302 427 742 145 887 20.0 51.9 41.4 11.4 10.0 Hebei 1886 1712 2259 4153 6412 10.7 27.2 -9.2 13.0 9.8 Henan 1603 1721 2531 6288 8819 6.4 25.8 7.4 18.6 14.9 Shandong 3343 4026 5393 4046 9439 12.6 28.2 20.4 10.9 8.4 Northeast Liaoning 1117 1280 1798 1253 3051 16.5 36.7 14.6 10.6 8.3 Jilin 378 441 567 985 1552 11.2 25.3 16.7 15.2 8.5 Heilongjiang 485 565 585 1037 1622 11.1 30.1 16.5 12.8 10.5 Northwest inner Mongolia 265 237 296 680 976 6.0 18.1 -10.6 13.9 11.7 Shanxi 883 970 1260 1144 2404 12.6 27.1 9.9 10.5 8.0 Shaanxi 628 608 923 1613 2536 7.0 21.1 -3.2 15.0 11.6 Ningxia 51 43 79 156 235 5.1 16.7 -15.7 16.5 12.5 Gansu 226 244 506 788 1294 3.9 15.8 8.0 19.1 14.9 Qinghai 38 51 62 84 146 3.7 10.8 34.2 14.4 11.4 Xinjiang 194 159 189 275 464 7.5 15.9 -18.0 9.1 7.8 Yangtze River Shanghai 699 905 1449 66 1515 24.8 60.6 29.5 8.0 9.4 Jiangsu 3886 4590 6729 2232 8961 17.2 32.2 18.1 8.7 6.5 Zhejiang 2349 2631 3524 1431 4955 15.7 24.3 12.0 7.7 4.5 Anhui 864 983 1691 2932 4623 5.2 20.1 13.8 18.3 14.5 Jiangxi 838 800 979 1348 2327 9.0 16.7 -4.5 10.8 6.4 Hubei 1527 1540 2022 1868 3890 10.2 21.7 0.9 9.8 7.9 Hunan 1994 1603 2097 2086 4183 10.0 16.1 -19.6 7.7 4.9 South Fujian 1174 1263 1397 1394 2791 16.1 27.6 7.6 9.0 5.5 Guangdong \a 2008 2153 3444 3384 6828 10.2 26.7 7.2 13.0 10.1 Southwest Guangxi 593 513 510 1481 1991 4.6 11.3 -13.5 12.9 9.5 Sichuan 1730 1715 3032 4026 7058 4.6 14.4 -0.9 15.1 12.0 Guizhou 201 228 253 816 1069 2.2 7.6 13.4 18.2 13.4 Yunnan 421 432 700 785 1485 3.5 9.0 2.6 13.4 9.9 Tibet 21 23 44 4.9 Source: SSS (1991c). \1 1990 figures include Hainan. 65 3.29 Government legitimization of small scale private enterprises in 1984 initiated a five year period of rapid RE growth. La This growth spurt was associated with a surge in employment in below-village RE, particularly in the poor provinces of northwestern, northern and southwestern China. In contrast to the early-1980s, the rate of RE employment growth was significantly greater in the poor provinces of northwestern, northern and southwestern China during 1983-90 than in the coastal provinces and the three municipalities. Though still considerably less than the national average of 22.1%, the share of the rural working age population employed in RE in northwestern and southwestern China increased to 17.9% and 10.6% respectively by 1990. Below-village RE appears to have flourished because so many economic activities, especially in transport and commerce, share with agriculture the attribute of being best managed on a small scale or by individuals. -RE Developrment in the Poor Areas 3.30 The expansion of RE during the 1980s, and the strong performance of RE in the poor provinces of northwestern, northern and southwestern China during 1981-90 in particular, hides tremendous variation in experience within provinces and obscures the impact of RE development on the poor. In Guangxi, for example, 11% of all 1988 RE income was generated in just two of the province's 105 counties. Similarly, in Yunnan, Kunming City REs accounted for 29% of 1990 provincial RE income. RE development in the more remote and poorer counties, townships and villages of these provinces, by comparison, has been quite limited (Working caper 4). In general, RE performance has been highly correlated with county and township rural income levels. This is not surprising, since rural income levels and RE performance are in the main the result of the same set of forces -- the natural resource endowment of an area, the abilities of the labor force, the adequacy of local infrastructure, and proximity to market. Most poor areas do not share these attributes and have experienced quite limited RE development. Available evidence confirms that the share of the rural working age population employed in RE is closely related to county average rural per capita income -- less than 4% of the rural working age population was employed in RE in 120 counties with rural income levels averaging Y410 per capita in 1989, while 56% of the rural working age population was so employed in 254 counties with income levels averaging Y750 per capita (SSB, 1991b). Byrd and Gelb have also shown that the limited RE development typical of poor areas does little to soak up surplus labor or improve workers' incomes. 3.31 I:n addition to limited resource endowments and disadvantageous location, available evidence suggests that local government policies have contributed to the failure of RE development in many poor counties, townships and villages. Most importantly, the struggle to mobilize revenue forces local governments in the poor areas to extract excessive funds from local enterprisesi regardless of their financial performance. Byrd and Gelb document an extreme example of such "fiscal predation" in a very poor township in l1/ The spurt was partly statistical artifact, since private enterprises had existed but been disguised before 1984. 66 Shangrao County, Jiangxi in 1986 -- payments to local government actually forced township enterprises to suffer net losses in that year. It is also likely that, in poor counties and townships, local government policy favors collectively-owned enterprises over small scale private enterprise despite the probability that the latter can more efficiently exploit poorly developed local markets. Local governments find small scale enterprises more difficult to tax and, because they can often compete effectively with larger scale collectively-owned enterprises, may actually consider such enterprises to be a threat to a proven and vital source of revenue. Unfortunately, local government often responds to the challenge of small scale private enterprises by attempting to restrict them. 3.32 Government sponsored initiatives intended to augment the development and poverty alleviation impact of RE in the poor areas include the introduction of appropriate technologies, the subcontracting of simple manufacturing or production activities to poor households and villages, and the provision of subsidized credit to enterprises which agree to employ a specified minimum number of poor. The recently enacted Regulations on Rural Collective Enterprises, which seek to limit local government's excessive extractions from enterprises and protect the rights and interests of both private and collective enterprise, are also of particular importance to creating a favorable environment for poor area RE development. These initiatives have been successful in many instances, and should be continued and expanded where market-determined financial viability can be documented. However, the potential for poor area RE development in the 1990s must not be overestimated -- the development of poor area RE is only one of several possible means of reducing poverty and misplaced efforts to force RE development may lead to inefficient use of scarce investment funds. Most of the subsidized credit made available for poor area RE development, for example, is now directed to larger scale county and township owned enterprises. These enterprises are frequently less efficient than smaller scale private units, and actual adherence to targets for employment of the poor is frequently questioned. C. Agriculture and Natural Resources 3.33 Sustaining Agricultural Growth in the 1990s. China's rural reform program (para 1.07), initiated in 1978, featured: -rural decollectivization, including reestablishment of nearly 200 million family farms through the adoption of the PRS system during 1979-81 and the subsequent elimination of more than 50,000 communes, whose administrative functions were reassumed by township and village government; -agricultural marketing and price reforms which, with the exception of grains and a few other commodities, shifted a large and growing share of agricultural marketing from the state commercial system to free markets. This was accompanied by substantial increases in the planned procurement prices of grain and other commodities subject 67 to continued administrative control and a loosening of the state monopoly on agricultural foreign trade. Exports of live animals, meats and aquatic products, and a wide variety of processed agricultural products boomed; and, -sharp increases in the availability and use of modern farm inputs, including a near tripling of the application of chemical fertilizer, from 8.8 to 25.9 million tons during 1978-90, a four- fold increase in the number of pedestrian tractors, from 1.4 to 7.0 million during 1978-90, and the widespread adoption of hybrid and improved varieties of rice, wheat, corn and other crops. The outcome of these reforms was predictable, but nonetheless remarkable. The real gross value of agricultural output more than doubled during 1978-90, and was accompanied by tremendous diversification of the structure of production and increased productivity. Sustaining agricultural growth in the 1990s will depend critically upon completing the rural reform agenda, including the further decontrol of agricultural input and output markets, and increasing investment in public infrastructure. In particular, incomplete liberalization of the land market and uncertainty about land use tenure rights, constraints on labor mobility, continued planned allocation of material inputs such as fertilizer, and distorted prices and markets for outputs still subject to administrative controls, perpetuate considerable inefficiency in the agriculture sector and limit farmers' on-farm investment. These essential reforms, plus increased private and public investment in the sector, are key to continued agricultural growth and a prerequisite to further reductions of poverty. 3.34 Agricultural Growth and Povertv Alleviation. Broad participation in strong agricultural growth played a key role in bringing about the tremendous reduction in poverty during 1978-84. The subsequent slow down of agricultural growth during 1985-90 coincided with the stagnation of poverty levels. These trends would appear to suggest that the modest agricultural growth of 1985-90 was not sufficient to achieve further reductions in poverty, and imply that a resumption of vigorous growth would now lead to further reductions of poverty. However, the more likely explanation of these trends is that the easy reductions of poverty through agricultural growth had been exhausted by 1985, and that even continued rapid growth during 1985-90 would not have necessarily achieved further poverty reductions. Most of the poor in 1978 resided in less remote and less hilly areas, where rapid productivity gains through increased use of modern farm inputs and adoption of "green revolution" technologies were more feasible, and so were able to more fully participate in the agricultural growth of 1978-84. By 1985, however, China's remaining absolute poor were mostly concentrated in more remote and less productive rainfed uplands. While some increases in productivity were achieved, available evidence shows that modest agricultural growth was largely offset by population growth in these remote upland areas throughout the 1980s. This strongly suggests that sustained nationwide agricultural growth in the 1990s will be essential to preventing an erosion of past successes in reducing poverty, but will not be sufficient to achieving further reductions in most upland areas. Instead, since most of the rural poor will continue to reside 68 in remote mountainous areas during the 1990s (para 3.04), improving agricultural productivity in the poor areas will be an essential component of any program to increase -- or at least maintain -- current levels of income. 3.35 Natural Resources. It is not coincidental that most of China's poor reside in areas of severe rural environmental degradation -- these poor have no alternative but to extract their meager livelihood on some of the nation's least productive but most ecologically fragile lands. The poor are both the perpetrators of upland environmental destruction, abetted by sometimes adverse government policy, and the victims of that destruction in that it further diminishes agricultural potential. Particularly in the poor areas, continued population growth has reduced the per capita natural resource base supporting agricultural production. Most importantly, arable land per agricultural laborer has fallen by nearly half from 0.57 hectares in the 1950s to 0.29 hectares today. The plains areas already had little uncultivated land in the 1950s, but farmers there have been able to more than overcome land pressure through intensified use, supported by increased irrigation and greater application of chemical fertilizers and other modern inputs, and through exit to off-farm employment opportunities. Farmers in the poorer hill and mountain lands, where irrigation is more difficult and modern inputs hard to find, have by contrast been forced to extend production to ever steeper and more fragile slopes. 3.36 Population growth has also exacerbated pressure on pasture land and forest areas. The pasture lands of northwestern China have been subjected to increased overgrazing, resulting in serious pasture degradation. During the 1960s and 1970s, furthermore, the government's program of local self- sufficiency in grain production mandated conversion of pasture land to grain crops and contributed to the deterioration in those areas. Population growth has also spurred encroachment of forested watershed areas in southwestern China, and is frequently reported in the Chinese press as the cause of increased siltation and flooding of major rivers, including the Yangtze. Insecurity of forest tenure, which encourages peasants to extract immediately whatever forest products they control and to minimize their investment in forest land, and rapidly growing demand for wood in industry and construction and as fuel in rural areas, also augment forest and watershed destruction. The government has been experimenting with a variety of management systems and tenure policies, but has yet to identify those that will induce widespread peasant investment in forest lands. 3.37 Strategv for the 1990s. Essential to increasing income levels in the poor areas, agricultural development programs must be tailored to the special characteristics and potentials of the upland natural resource base and address the contradiction between the immediate survival needs of upland farmers and environmental protection. The limits to the effectiveness and efficiency of assisting the poor through agricultural development schemes must also be recognized. Financial and economic rates of return must be given greater emphasis as key selection criteria for poor area agricultural development projects. Excessive attention to production-oriented projects, furthermore, can lead to an underemphasis of alternative investments. Most importantly, greater attention must be directed to better agricultural 69 extension and training and to farmers' marketing systems in the poor areas. In southwest China, where losses of 25% or more of poor households' grain reserves due to rats and other pests are common, provision of basic household grain storage containers and other pest management treatments represents another example of an inexpensive yet high return investment. The following sections review past and expected trends in agricultural labor requirements and explore options for agricultural development in the loess plateau in northwestern China. -Employment in Agriculture 3.38 Overview. The official figures summarized below show that the number of rural inhabitants engaged in agriculture increased from 275 million in 1978 to 324 million in 1989, or by about 50 million: Agricultural Emplovment: Million Workers 1978 1982 1986 1989 Total 275 301 305 324 Crop Cultivation 257 282 254 na Other Farm Activities 18 19 51 na Source: SSB (1991c). Note: Other farm activities include animal husbandry, forestry, fisheries, and nonindustrial sideline production. Employment growth averaged 1.5% annually during this period, and was greatest during 1978-82 and 1986-89 when it averaged 2.4% and 2.0% respectively. Average annual employment growth, however, was only 0.3% during 1982-86. The decomposition of the figures into employment in crop cultivation and other farm activities suggests that the slowdown of employment growth during 1982-86 was consistent with the restructuring of agriculture in the early-1980s. Adoption of the responsibility system and the breakup of the commune system encouraged labor-shedding from crop cultivation during the period and a surge in employment in animal husbandry, forestry, fisheries, and nonindustrial sideline production. W0 As a result, crop cultivation's share of agricultural employment declined from 93% in 1978 to 83% in 1986. 3.39 Labor Reguirements in Agriculture. An examination of labor requirements in agriculture during 1978-89, however, suggests that these official figures mask increasing underemployment among the growing residual of the rural working age population without access to off-farm employment 7 Employment in crop cultivation reportedly declined by 27 million during 1982-84. Employment in animal husbandry and nonindustrial sideline production, on the other hand, increased by 14 and 13 million respectively during the same period. 70 opportunities. A means of estimating agricultural employment favored by Chinese researchers is to apply average labor norms to major farm activities. Using this approach, estimated labor requirements for major farm activities in 1978, 1984, 1989 and 2000 are reported in Table 3.4. Lv In contrast to the official agricultural employment figures reported above, these estimates suggest that agricultural labor requirements plunged more than one-third during 1978-84 and then declined marginally (by about 2%) during 1984-89. A close examination of the estimates, on the other hand, shows that decreased labor norms account for more than all of the decline in labor requirements -- holding labor norms constant at 1984 levels suggests that agricultural labor requirements instead increased by about one-quarter during 1978-89. As shown in the bottom row of Table 3.4, adopting the 1984 labor norms for each year 1978, 1984, and 1989 results in increases in agricultural labor requirements of 8% and 13% in 1978-84 and 1984-89 respectively. 3.40 The sharp reduction in labor norms during 1978-84 corresponds both to increased transparency of the widespread underemployment previously obscured by the commune work point system as well as a true increase in productivity brought about by the adoption of the PRS system. Agricultural mechanization explains little if any of the decline in work norms during 1978- 84, since area ploughed and sown by machine declined during the period by about 20%. Increased mechanization is consistent with declining labor norms during 1984-89, on the other hand, as area ploughed, sown and harvested by machine increased by between 20% to more than 50%. On balance, this evidence suggests that increased labor requirements for greater cultivation of labor intensive cash crops and expanded output of animal and aquatic products were mostly offset by improved efficiency and productivity. At most, agricultural labor requirements increased by only 10% during 1978-90 -- or about half the increase indicated by official figures (see Table 3.1). 3.41 Agricultural Labor Recuirements in 2000. Based on an extension of the World Bank's agricultural sector simulation model for China to 2000 (World Bank, 1991b) and assuming 1989 labor norms, agricultural labor requirements would increase by a total of 10% during the 1990s (Table 3.4). 191 Most of this increase would occur in or near the more developed rural areas, where demand for labor intensive cash crops and animal and aquatic products is greatest. Labor-substituting mechanization is also most likely to occur in the developed areas, and overall it is expected that at least half of the projected increase in labor requirements will be offset through improved efficiency and productivity. At the national level, agricultural employment would therefore increase at most by about 5% during the 1990s -- the L These estimates exclude labor requirements for a number of lesser crop cultivation, animal husbandry and forestry activities for which labor norms are not available. Nevertheless, they do include most significant farm activities and are useful as indicators of trends in total agricultural labor requirements. L9 Assuming 225 workdays per workyear, the increase would be equivalent to an increase of about 25 million workyears of agricultural employment. Table 3.4: Estimated AAricultural Labor Requirements. 1978. 1984. 1989 and 2000 1978 1984 1989 2000 Scale of Labor Labor Scale of Labor Labor ScaLe of Labor Labor ScaLe of Labor Labor Operation Norm Requirement Operation Norm Requirement Operation Norm Requirement Operation Norm \a Requirement million workdays million million workdays miLlion miLLion workdays million million workdays miLlion Activity mu per mu workdays mu per mu workdays mu per mu workdays mu per mu workdays Rice 516.3 38.1 19672 497.7 21.2 10551 490.5 19.6 9599 453.0 19.6 8866 Wheat b 437.7 30.7 13439 443.6 14.8 6566 447.6 12.0 5371 459.7 12.0 5516 Corn _ 299.4 31.1 9312 278.0 16.7 4643 305.3 14.1 4305 496.1 14.1 6995 Soybeans 107.2 22.2 2379 109.3 10.1 1104 120.9 8.1 979 126.4 8.1 1024 Tubers 176.9 31.1 5503 134.8 16.7 2252 136.5 14.1 1924 135.2 14.1 1906 Other Grains 271.2 31.1 8436 229.8 16.7 3837 182.3 14.1 2571 Cotton 73.0 60.5 4416 103.8 40.9 4247 78.1 37.8 2950 79.6 37.8 3010 Peanuts 26.5 35.7 947 36.3 25.6 930 44.2 24.6 1087 50.4 24.6 1239 Rapeseed 39.0 30.4 1185 51.2 21.4 1096 74.9 16.7 1251 85.4 16.7 1426 Fiber Crops 11.3 54.3 612 7.3 40.5 295 8.5 40.5 342 8.6 40.5 349 Sugarcane 8.2 65.2 537 10.9 49.8 544 14.4 49.0 705 21.5 49.0 1053 Sugar Beet 5.0 11.3 56 7.5 25.4 191 8.5 14.1 120 12.7 14.1 179 Tobacco 11.8 83.9 987 13.5 54.9 739 27.0 54.9 1481 27.5 54.9 1511 Other Cash Crops 41.9 83.9 3513 58.7 54.9 3225 59.4 54.9 3258 87.4 54.9 4800 VegetabLes 50.0 83.9 4192 64.8 54.9 3558 94.4 54.9 5180 90.9 54.9 4990 Tea 15.7 57.4 902 16.2 48.4 782 16.0 41.4 661 16.2 41.4 672 Fruit Orchards 24.9 115.6 2873 33.3 84.5 2814 80.6 50.0 4029 88.5 50.0 4424 Pigs (million head) 301.3 24.9 7502 306.8 17.4 5338 352.8 17.4 6139 487.8 17.4 8489 Aquatic Products 435 48.9 45.6 2230 57.6 35.0 2014 85.3 35.0 2984 TotaL 86896 54941 53968 59433 Total (2 1984 Labor norm) 50709 54941 62136 68285 Source: Agricultural Publishing House (1990) and (for 1978 Labor norms) SSB (1991c). Scale of operation in 2000 from the MAT-1 model (World Bank, 1991b) updated and extended to 2000. \a 1989 Labor norms. \_ Includes "other grains" in 2000. \E Estimated on the basis of 1978 output of aquatic products. 72 equivalent of between 10 to 15 million workyears -- and might not increase at all in the less developed areas. Sluggish employment growth in agriculture will be compensated by greater growth rates in the other sectors of the economy. Even assuming no agricultural employment growth during the 1990s, total employment growth should modestly exceed the expansion of the active working age population (para 3.12). Nevertheless, while some of the poor should be able to find off-farm employment within or outside the poor areas, it is most likely that the majority of China's absolute poor will remain trapped in subsistence agriculture in remote mountainous areas for the 1990s and beyond. Programs to improve agricultural productivity, when they meet the basic requirement of economic efficiency, therefore represent a key component of the limited number of options for increasing income levels in these upland areas. The potential for increasing productivity in one such region, the loess plateau of northwest China, is discussed in the section below. -The Loess Plateau 3.42 Soils and Climate. The loess plateau is an ancient land form which extends over some 624,000 km2 of northwestern China, covering much of Gansu, Ningxia and Shaanxi. The soils, formed from accumulated wind-borne deposits, are of uncommon depth -- large tracts are between 50 to 100 meters deep. Mean annual rainfall varies from less than 200 mm to more than 800 mm, with most areas receiving about 400 mm. About 75% of the annual rainfall occurs in the four month period June to September, and frequently comes as sporadic heavy falls. The brief season of effective rainfall fortunately coincides with the period of highest mean air temperature, and provides the most favorable period for plant growth. The efficiency of rainfall use is reduced, however, by limited infiltration on steep slopes and surface crusting. 3.43 Land Degradation and Rural Poverty. Expanding and exploitative agriculture, in combination with geologic erosion of these naturally fragile soils, has progressively removed the loess plateau's vegetative cover (Chen). Since unprotected loess is easily eroded, the removal of the vegetative cover has by now resulted in the advanced environmental destruction of most of the plateau. Streams, which have in many areas cut through to the underlying sandstone, have heavily dissected the plateau, leaving a landscape of steeply sloped hills of relatively uniform height. It is estimated that, of the 530,000 km2 of the loess plateau situated along the middle reaches of the Yellow River, more than 450,000 km2 (85%) are severely eroded "badlands" (Cook). Erosion products are transported out of the area by the silt-laden Yellow River, leading to heavy siltation of downstream dams and irrigation systems and dangerous increases in the height of the river bed itself. 3.44 Extensive environmental degradation is both an important cause and result of low productivity and severe poverty in the rainfed agricultural areas of the loess plateau. Agricultural productivity and incomes were diminished during the 1960s and 1970s when, as part of the national policy of achieving local self-sufficiency in grain production, upland farmers were encouraged to switch from pastoralism to extensive cultivation of grain. Unfortunately, extending grain cultivation to steeply sloped land exacerbated 73 land degradation and ultimately led to reductions in carrying capacity. ' Many of the 81 million rural inhabitants of the loess plateau continue to suffer fronm extreme poverty, and the effort to improve their well-being receives central and provincial government attention and support. The per capita farm input, farm output, and income figures shown below for three poor counties in the loess plateau are indicative of the low agricultural productivity and severe poverty in the area: Rural Per Capita Farm Inputs, Farm Outputs and Income: Three Poor Loess Plateau Counties, 1988 Gansu Ningxia National Average Tongwei Zhangiiachuan Jincyuan CuLtivated Area (mu) 1.7 5.0 (304%) 2.5 (151%) 3.4 (202%) Irrigated Area (mu) 0.8 0.1 (16%) 0.1 (16%) 0.5 (67%) MuLtipLe Cropping Index (100%=100) 151.6 109.4 (72%) 102.4 (68%) 86.1 (57%) Grain Yield (tons/ha) 3.6 0.9 (25%) 1.4 (39%) 1.6 (44%) Fertilizer Use (kg of nutrients) 24.7 13.0 (53%) 8.3 (34%) 10.9 (44%) Electric Power (kWh) 82.1 7.2 (9%) 19.3 (24%) 21.1 (26%) Grain Production (kg) 454.4 225.4 (50%) 193.2 (43%) 250.8 (55%) Meat Production (kg) 25.3 15.3 (61%) 4.6 (18%) 10.5 (42%) Rural SociaL Output Value (Y) 1445.3 378.6 (26%) 679.5 (47%) 369.4 (26%) Income (Y) 544.9 243.9 (45%) 224.0 (41%) 169.0 (31%) Source: SSB (1991a) for national average, and SSB (1991b, 1989a and 1989b) for poor counties. Note: Figures in parentheses are percent of national average. Tongwei and Zhangjiachuan are in Dingxi and Tianshui prefectures in Gansu. Jingyuan is in Guyuan Prefecture in Ningxia. 3.45 Cultivated land per rural inhabitant in the three poor counties ranges between 150% to more than 300% of the national average. However, irrigated land is quite limited (only 16% of the national average in Tongwei and Zhangjiachuan counties), there is very little if any multiple cropping, and grain yields are less than half the national average (only one-quarter of the national average in Tongwei). Fertilizer use per capita is only one-third to one-half the national average, and only one-quarter to one-third of the national average when measured as fertilizer application per unit of sown area. Per capita grain production is roughly one-half the national average. Figures for earlier years indicate extreme variability and, with the exception ' Per capita grain production in the Dingxi Region, which occupies much of the loess plateau in central-eastern Gansu, was about 20% less in the early- 1980s than in the early-1950s (World Bank, 1988). Dingxi Prefecture in Gansu and Guyuan Prefecture in Ningxia, two well known areas of extreme poverty in the loess plateau, were both net exporters of grain until the late-1960s. 74 of Zhangjiachuan, no increase in per capita grain production during the 1980s. IL' Per capita rural social output value in the three counties is only one-quarter to one-half the national average, and average rural per capita income is 30% to 45% of the national average. The hardships imposed by low and variable crop production and depressed income levels are made considerably worse over much of the loess plateau by inadequate access to safe drinking water (fluorosis affects much of the population) and insufficient fuel supplies for cooking and heating. 3.46 Land Rehabilitation Programs. Erosion control work on the loess plateau since the 1950s has included the construction of wide terraces on hillsides, soil dams in gullies, and large-scale afforestation. ' The pace of erosion control was accelerated in the 1970s with the adoption of a comprehensive regional plan to construct hundreds of soil dams in each of the counties along the middle reaches of the Yellow River. However, implementation of these soil conservation measures was insufficient to offset growing population pressure and the policy induced switch to extensive grain cultivation (para 3.44). Beginning in the late-1970s, the government responded to the limited effectiveness of the previous erosion control measures with intensive research and development programs to simultaneously improve erosion control and agricultural production. More than 300 research agencies and stations are now developing and testing a number of soil erosion control measures and improved agricultural production techniques. A land use and social data base has been established, and comprehensive land rehabilitation programs are being implemented in numerous small watersheds. 3.47 The erosion control strategy adopted in land rehabilitation project watersheds is to replace the current low-yield extensive cropping system with an integrated system of intensive cropping of lowlands and terraces combined with fodder, fuelwood and tree crop production on steeply sloped land. Under the new integrated system, unterraced hills with slopes greater than 250 are utilized exclusively for fodder and tree crop production. Since much of the steeply sloped land is currently sown to low-yield grain crops, the conversion of these lands to grass, shrub or tree cover eliminates a portion of village and household grain production. In most project watersheds, it is expected that this decrease will be more than fully offset by increased yields on existing and newly created flatland and wide terraces. The package of erosion control measures are: (i) construction of narrow terraces, cut into currently cultivated hillsides with slopes of up to 250. With associated changes in production practices and input use, this labor intensive action effectively doubles crop yields and reduces erosion; IL' In Tongwei, per capita grain production was 237, 271 and 195 kg respectively in 1980, 1985 and 1987. In Zhangjiachuan, grain production was 153, 178 and 185 kg in those same years and, in Jingyuan, 262, 214 and 181 kg. 9 Rehabilitation of the rainfed agricultural areas of the loess plateau is reviewed in Hua and Mou, World Bank (1988), and Doolette and Piazza. 75 (ii) establishment of pasture plants, planted by hand as seed on unterraced hillsides with slopes of 250 to 350. Pasture species replace crop cultivation and uncontrolled grazing and are often harvested for fuel; (iii) establishment of trees and woody shrubs, planted in small bench terraces on hillsides with slopes of more than 350. The tree and shrub cover protects the soil and augments generally limited local supplies of fuel wood; (iv) construction of soil dams and dam land, including large soil dams near the outlet of each watershed and numerous small soil dams situated along many of the small feeder gullies. Soil dams stop or retard the movement of erosion products down and out of the watershed, and the impounded soil soon forms fertile flatland; and, (v) improved agricultural technologies, including better chemical and organic fertilizer strategies, testing and introduction of improved varieties, and new cultural practices for improving water use efficiency. 3.48 Evaluation and Issues. Variants of this rehabilitation package have been implemented as part of well-orchestrated comprehensive watershed development plans in a UNDP and WFP assisted project in Mizhi County in Shaanxi, in a World Bank assisted project in Dingxi County in Gansu, and at the village level in 200 micro-watersheds throughout the loess plateau. - Most of these projects and experiments are now well advanced, and impact assessments confirm that the rehabilitation package improves agricultural productivity and profitability while sharply reducing soil erosion and siltation. The Mizhi County project has helped increase project area grain and horticultural crop yields and output of animal products, diversify production and increase income levels, greatly reduce soil losses, and served as a model for other WFP assisted repeater projects. The Dingxi County project staff also report sharp increases in agricultural production and income levels, and claim that the completion of three large soil dams has all but completely eliminated the exit of erosion products from the watershed. Data for the large number of village level pilot operations similarly indicate that grain yields on newly terraced land generally increase by at least 100%, large areas of sloped land have been returned to pasture and shrub cover, and soil erosion has been curtailed by 50% to 90%. 2J With UNDP and then WFP assistance, beginning in 1980 and 1985 respectively, the "Improvement of Land Use of the Loess Plateau" project (Number 2744) rehabilitated 18,500 ha of uplands in Mizhi County, Shaanxi. With World Bank assistance, the Guanchuan River Basin subcomponent of the Gansu Provincial Development Project (Credit 1793-CHA) began rehabilitation of 76,000 ha in 1987. The components and performance of the rehabilitation programs adopted in some 200 village level integrated pilot programs, undertaken by a variety of Chinese research agencies in the loess plateau over the last 30 years, are reviewed by Bi. 76 3.49 The government iB supporting the widespread implementation of the rehabilitation strategy through recently adopted regulations which stipulate the cessation of crop cultivation on all unterraced loess plateau lands with slopes exceeding 250. According to the new regulations, unterraced land exceeding 250 slope must instead be terraced or converted back to pasture, trees or shrubs. The regulations are to be supported during the Eighth Five Year Plan, furthermore, by the expansion of government financial support for terracing and establishment of pasture, trees and shrubs through the Chinese Food-for-Work Program (para 5.19). Past experience in the Mizhi, Dingxi and village level pilot operations suggest that additional policy initiatives and other measures could greatly expedite the popularization and efficiency of the rehabilitation strategy. Most importantly, means must be found to reduce the investment costs of the rehabilitation package when implemented on a project basis -- it is simply too expensive in its current form to be adopted on other than a pilot area basis. 2' Labor is the principal cost of most of the rehabilitation package measures, and cost benefit analysis indicates that investment in these measures is attractive only at extremely low wage rates of Y3 per workday or less. The analysis shows that the rehabilitation package is financially and economically viable, but only in the poorest parts of the loess plateau and only when the physical work of the rehabilitation measures is undertaken by the poor themselves. & The Pood-for-Work Program is an excellent means of minimizing the costs of terracing and the establishment of pasture, shrubs and trees, and can be used to provide the absolute poor with employment opportunities. Significant cost savings can also be achieved by cutting back on the "over-engineering" associated with some rehabilitation measures. Trees and shrubs, for example, can be planted on steeply sloped lands with only minimal or no terracing. 3.50 It is also essential that benefits to farmers choosing to participate in the rehabilitation program be maximized. Most importantly, provided they adequately maintain vegetative cover and meet soil conservation standards, households must be guaranteed full land use rights and entitled to inheritable transfer of newly terraced lands and sloped lands converted to pasture, trees or shrubs. Localized experiments have shown that farmers' incentive to invest their labor in the construction of terraces and the planting of pasture, trees and shrubs is greatly enhanced when land tenure is 2' Investment costs in the Mizhi project exceeded US$800 per ha and in the Dingxi project are about US$300 per ha. At an average cost of US$500 per ha, implementing the rehabilitation package over a 100,000 km2 stretch of the loess plateau situation along the Middle Reaches of the Yellow River would require an investment of US$5 billion. 9' The cost benefit analysis of an earlier study similarly showed that the rate of return to these rehabilitation measures is critically determined by the marginal value of labor. Doubling the economic value of labor, from YO.6 per workday to Y1.2, diminished the estimated economic rate of return to individual rehabilitation measures of between 13% (for construction of soil dams) and 40% (for planting woody shrubs on steeply sloped land) down to between 5% and 23% (World Bank, 1988). 77 assured. ' Measures to enhance marketing of nongrain products are also essential to realizing favorable returns to investments. Improving access to markets for fruits, vegetables, and especially animal products increases cash incomes and sustains enthusiasm for diversification into these and other nongrain crops well suited to the area's agronomic potential. The collapse of the market for wool produced in the loess plateau area in recent years is a particularly serious problem, since it undermines farmers' interest in maintaining pasture on steeply sloped land. At the currently depressed price of wool, loess plateau farmers may receive a greater profit from growing potatoes on steeply sloped land than from pasture-based wool production. 3.51 It is also essential that the full yield potential of terraced land be realized through increased supplies of fertilizer and other modern inputs, greater testing and promotion of improved varieties, and better extension of improved cultural practices. Inadequate availability of chemical fertilizer, in particular, continues to be a serious impediment to achieving economic yield potential on terraced land in the Mizhi and Dingxi project areas. Benefits from newly established pasture could be significantly enhanced by testing and promoting new species -- especially grasses -- and improving management practices. L' Overgrazing on unimproved pasture "commons" (that is, pasture areas not yet subject to the PRS system) is extensive in the loess plateau, and the control or elimination of grazing through the partitioning of remaining pasture commons to individual households has considerable potential for increasing productivity and reversing environmental degradation. 2' 3.52 The Voluntary Resettlement Program. Even with successful adoption of the rehabilitation program, however, the rural population would exceed the carrying capacity of the land throughout much of the lowest rainfall sections of the loess plateau. Based on detailed assessments of carrying capacity and estimated potential for absorption of excess labor in off-farm rural enterprise, the government initiated a program in the late-1970s to provide opportunities for voluntary resettlement to more than 10% of the rural population of the worst affected areas. The resettlement program calls for relocating 600,000 upland farmers of the Dingxi Region of Gansu and 300,000 farmers of the Xihaigu Region of Ningxia to newly developed irrigated land by 2' China Daily (February 13, 1990), "Achievements Made in Erosion Control." ZZ Deficiencies in pasture management include early cutting of alfalfa (Dingxi) and uncontrolled grazing after the second cutting of alfalfa (Mizhi). W Hassall and Associates (1988) report that uncontrolled grazing on unimproved common pasture land in the loess plateau results in suboptimal resource utilization and increased soil erosion. Unimproved common pasture represents about 20% and 70% respectively of total pasture area in Hassall and Associates project survey sites in Shishe and Tuqiao townships in Qingyang County, Gansu. 78 2000. 2' Under the supervision of the Three-Xi Commission, four-fifths of the settlers are to move short distances (200 km or less) to newly irrigated lands within the Dingxi and Xihaigu regions, and the other fifth are to relocate from the Dingxi Region to newly irrigated areas in Gansu's Hexi Corridor some 500 km to 1000 km to the northwest. 3.53 By the end of 1990, the resettlement program had successfully arranged the relocation of 320,000 upland farmers to newly irrigated areas, including about 170,000 within and from the Dingxi Region and 150,000 within the Xihaigu Region. 2' Resettlement has been and will continue to be on a strictly voluntary basis. In practice, the number of households wishing to resettle exceeds available slots, and selection has been competitive. Preference is given to very poor families with sufficient labor power to open up new land, and to families from areas lacking access to safe drinking water, deficient in fuel for cooking and heating, or suffering extreme soil erosion problems. Settler families retain use rights to their rainfed lands for the first three years after relocation, and are allowed to reverse their decision and return to those lands at any time during this initial period. Settlers receive assistance with transport of family members and their belongings, first year housing, farm inputs and machinery. Electrification, agricultural extension services, education, health and other social services are typically established in settler townships and villages during the second and third year after relocation. 3.54 Very few if any of those settling within the Dingxi and Xihaigu regions have so far chosen to return to their rainfed lands. Between 5% to 10% of those moving to the Hexi Corridor were unable to adjust to their circumstances, however, and have returned to the Dingxi Region. Those settlers returning from the Hexi Corridor reportedly had excessively high expectations of immediate improvements in their income levels and standards of living. L-/ Random sample surveys of 752 settler households in 1986 and 337 settler households in 1990, conducted by the Population Research Institute of the Ningxia Academy of Social Sciences (Yan, 1990), clearly document the success of the resettlement program in the Xihaigu Region. The surveys show that the average settler enjoys a three fold increase in income -- from about Y100 to more than Y300 -- within the first three years of relocation, and that very few settler households seriously regretted having made the move. The most common complaint voiced by those surveyed has been the lack of primary '2 The Dingxi Region comprises 20 counties in 6 prefectures in central and eastern Gansu. The Xihaigu Region, which is adjacent to the Dingxi Region, includes 8 counties in Guyuan and Yinnan prefectures in southern Ningxia. Gansu's resettlement program is reviewed in detail in World Bank (1988). 3' China Daily (June 21, 1991), "Relocation Project Aids Poor Peasants." 31/ Perhaps not surprisingly, since the intensively irrigated Hexi Corridor is well known as a highly productive agricultural area and enjoys average per capita rural income levels in excess of the national average and more than double the provincial average. 79 schools during the first years after relocation. Provincial authorities confirm that in some cases inadequate funding has delayed the planned establishment of social services, but note that schools, health centers and other services are generally adequate in areas settled for more than three years. Gansu provincial authorities have also experienced difficulty interesting minority people -- who comprise a disproportionate share of Gansu's absolute poor -- in resettlement. Recent efforts to offer resettlement to entire minority villages, instead of integrating small numbers of minority households into Han communities, have proven more successful. 3.55 Resettlement and Irrigation DeveloDment. The resettlement program depends critically on the development of newly irrigated land to receive the settler population, and more than one-third of the required 180,000 ha of irrigated land has now been established. Most of the irrigated land has been or will be developed through gravity or low-lift systems with at least minimally acceptable rates of return (World Bank, 1988). In order to complete resettlement targets, however, a number of high-lift irrigation schemes have been undertaken and several more are being planned. One of the largest of the high-lift schemes, Jingtai Phase II, was completed in 1990. Situated in the Dingxi Region, the Jingtai Phase II project employs 28 pump stations to lift water an average of 460 meters from the Yellow River to irrigate 33,000 ha of previously uncultivated land. At full development, the project will provide 150,000 settlers from the Dingxi Region with standards of living modest in comparison with the national average, but well above what they were in the rainfed areas. The 460 meter lift requires a great deal of energy -- 500 million Kwh annually -- and farm income and the project's economic rate of return (ERR) are critically determined by the value of electricity. Settlers pay the equivalent of only YO.01 per kWh and, at this heavily subsidized rate, are able to quickly increase their income above the absolute poverty line. By comparison, the costs of hydroelectric and thermal power generation in Gansu are at least YO.05 to YO.08 per kWh, and the 1990 negotiated price for electricity was at least YO.15. Valuing electricity at YO.05 per kWh, the estimated ERR for Jingtai Phase II is only 9%, and at YO.11 per kWh the estimated ERR turns negative. The government's difficult decision to reduce poverty through heavy recurrent subsidies for high-lift irrigation is discussed in Box 3.1. 80 Box 3.1: Jingtai Phase II -- Laudable Objectives. QuestionabLe Economics CoWpleted in 1990, the Jingtai Phase 11 project supplies 33,000 ha of previously uncultivated land with irrigation water from the YeLLow River. With an average lift of 460 meters, the centraL conveyance system comprises 28 pumping stations and about 100 km of trunk canal. Total investment costs, including 340 km of secondary canals, land leveling and field preparation, and first year resettlement costs, amounted to more than Y350 miLlion. At full deveLopment, a total of 150,000 settlers from the Dingxi Region will plant this land to wheat, corn and other fodder crops, sugar beet, fruits, and vegetables, and diversify into animal husbandry, forestry, and agroprocessing. The project expands on the earlier Jingtai Phase I project, which irrigates 20,000 ha of nearby land and provides settlers with income levels which are the envy of farmers in the rainfed areas of the Dingxi Region. The Jingtai Phase II project requires 500 million kWh of electricity annually for the 460 meter lift from the Yellow River, and farm incomes and the project's ERR are critically determined by the value of that power. Farm incomes, which are expected to increase to at the least the provinciaL average within five years of settlement, are insulated from the high energy costs of lifting irrigation water by massive operating subsidies. Through water fees, settlers pay the equivalent of YO.01 per kWh -- only a fraction of the estimated economic costs of hydroelectric and thermal power generation in Gansu of at least YO.05 to YO.08 per kWh. Farm incomes would be cut by about one-third were water fees increased to levels corresponding to power generation costs of YO.05 per kWh, and would be at or below the absolute poverty line at the higher rate of YO.08 per kWh. Valuing electricity at YO.05 per kWh, the project's estimated ERR is onLy 9% (less than the opportunity cost of capital). At YO.08 per kWh the ERR declines to less than 6%. Costs of production, furthermore, may significantLy understate the economic value of energy. The 1990 negotiated price for electricity in Gansu, which may better represent the economic value of power in northwest China, was at least YO.15 per kWh. The project's estimated ERR turns negative at YO.11 per kWh, and at YO.15 per kWh, the costs of field crop production at full development would exceed output value by Y20 million annually. Assuming power generation costs of YO.05 to YO.08 per kWh, the annuaL fiscal costs of subsidized water fees amount to between Y20 to Y35 million. At the higher negotiated price of electricity, the subsidy value is Y70 million. The subsidy burden is shared by the central and provinciaL government, primarily as foregone revenue. The government's difficult decision to undertake the Jingtai Phase 11 project was motivated by the laudable objective of poverty alleviation. The full extent of the trade-off between poverty alleviation and economic efficiency, however, was obscured by the government's (i) estimate of the financial cost of hydroelectric power generation in Gansu of YO.02 per kwh and (ii) expenditures on relief programs in the Dingxi Region, which are expected to decrease by perhaps Y10 million after resettLement to the Jingtai Phase 1I project area has been completed. The government has undertaken three other major high-lift irrigation schemes in Gansu, with similarly unattractive rates of return and heavy recurrent subsidies, and has plans to start at least one more such project in the near future. World Bank (1988) presents an economic cost benefit analysis of the Jingtai Phase 11 project (Annex A) and discusses the costs of electric power generation in Gansu (Annex 8 Appendix 3). 81 4. SOCIAL SERVICES IN THE POOR AREAS 4.01 The ability of the poor to take advantage of off-farm employment, new agricu:Ltural technologies, and other income generating possibilities depends very critically on their being adequately nourished, educated and healthy. These and other aspects of human capital development delivered in part through social services have been identified as crucial to successful poverty alleviation efforts elsewhere (World Bank, 1990a). This chapter identifies a program of government intervention to reduce the problems of illiteracy, illness, and malnutrition that now hinder poor area development in China. That program would be based on an existing series of experiments in education (para 4.17) and health (paras 4.22 and 4.31) that demonstrate low cost, culturally appropriate measures to increase school attendance and health care. Nutrition in the most vulnerable families would be improved by returning to the level of the early 1980s the real resources of the Ministry of Civil Affairs devoted to rural food supplements. A. Financing Social Services in the Poor Areas 4.02 Expenditure on Social Services. In contrast to the substantial real increases in total government (all levels) expenditure for education, health and rural relief since the late-1970s, support for these services in the poor areas has remained roughly constant. Total expenditure on education and health increased in real terms by 180% and 220% respectively during 1978- 88, and together increased from 4.7% of GNP in 1978 to 5.7% in 1988. Real expenditure on rural relief -- including assistance to areas affected by natural disasters or extreme poverty -- increased by the lesser margin of 15% during 1980-89. However, most of the increase in funding for these services has benefitted the well-off urban population and middle and upper income rural inhabitants -- very little of the increase has reached the absolute poor. Central ancd provincial government support for education and health in China's poorest townships and villages is at present extremely limited. Relief assistance, provided almost exclusively in the form of small quantities of low-cost or free grain, does reach most extremely poor rural households, but is only sufficient to avoid outbreaks of severe malnutrition. 4.03 In the absence of meaningful levels of funding from higher levels of government, support for social services in the poor areas must be mobilized from the limited local revenue base and user fees. Although recurrent expenditure on social services claims an enormous share of local government expenditure in the poor areas -- education alone now claims 30% to 60% of government recurrent expenditure in most poor counties -- total government expenditure is simply insufficient to establish and maintain adequate social services in the poorest townships and villages. With their very meager cash incomes, furthermore, the poor often can not afford to make use of even the very limited education and health services which do exist in their vicinity. 82 4.04 Impact. Inadequate support for social services in the poor areas has an obvious negative impact on the well-being of the poor. Although their total numbers are greatly reduced, China's remaining absolute poor are as illiterate, unhealthy and malnourished today as they were in the late-1970s. At least 50% of the boys in many of the poorest towns and villages and, particularly in some minority areas, nearly 100% of the girls do not attend school and will not achieve literacy. Infant and maternal mortality in poor counties is at least 50% to 100% greater than the national average, and is much greater yet in the poorest townships and villages. Incidence of several infectious and endemic diseases, including tuberculosis and iodine deficiency disorders, is concentrated in poor and remote areas. Roughly 50% of children in households at or below the absolute poverty line are mildly malnourished (stunted), and iron, vitamin A, and other micronutrient deficiencies remain a severe problem among the poor. As many as 90% of poor children suffer chronic helminthic infection. 4.05 Fundina Reauirements. Now that the incidence of absolute poverty has been reduced to less than 10% of total population, the government is in the position to address the most critical education, health, and relief needs of the poor at modest and acceptable costs. Annualized incremental funding requirements for improving education and health services and status in the poor areas, through the specific programs outlined in this chapter, are estimated to be about Y2 billion. Through these programs it should be possible to universalize six-year primary education for 15 million poor children and to reduce infant, child and maternal morbidity and mortality in the poor areas by at least one-third. Failure to adopt such programs will preclude realization of the government's goals of universal nine-year primary education and "Health-For-All in 2000." Incremental funding for the programs, which must be provided by the central and provincial governments, could be partially defrayed through efficiency gains in planned expenditure on education during the 1990s. Even at full cost, however, incremental funding requirements would represent only about 0.6% of current total government planned budgetary expenditure. B. Education 4.06 Overview. Children in many poor villages are as likely to be illiterate today as were children in those same villages ten years ago and, throughout much of the country, poor children are now less likely than before to attend secondary school. This low educational attainment has a tremendous social cost, including larger, less healthy families, lower adoption rates of new agricultural technology, and diminished access to alternative off-farm employment. Poor performance comes despite the fact that expenditure on education now consumes between 30% to 60% of government recurrent expenditure in most poor counties -- even this large share is insufficient in counties lacking adequate revenues. Government funding for education did increase substantially in real terms during the 1980s, but most or all of this increase has benefitted urban and better-off rural areas. Given the limited prospects for increased revenue mobilization in the poor areas and the inability of many poor families to pay even current school fees, strengthening primary education 83 in poor rural areas will require substantial new central and provincial government budgetary funding. It is estimated that achieving universal six- year primary education in the poor areas would require at least Y1 billion annually in incremental funding during the 1990s -- additional funding would be required for universal nine-year education or to significantly increase provision of secondary education in the poor areas. -The Poor's Access to Education 4.07 Official figures indicate steady improvements to near universal primary education at present -- national enrollment rates for primary school- age children increased from only 50% in 1952 to 95% by 1975 and 97% in 1989. Over the same period, illiteracy reportedly decreased from about 80% to 27%. Though real improvements have certainly occurred, official figures overstate those improvements and mask the poor's relatively limited access to education. Actual attendance rates are significantly less than the official enrollment rates suggest -- World Bank (1991a) reports that in Guizhou, for example, "about 30% of the children do enroll, and maintain their enrollment, but do not attend classes." Since literacy is often measured by reported years of education, not direct testing, these non-attenders are often listed as literate on survey forms. The official national and provincial averages, furthermore, include the performance of relatively well off rural areas, county towns, and cities. Available evidence from poor areas indicates much less favorable levels of educational attainment. Even if they begin school, few poor children stay long enough to gain lasting literacy. Lavelv (1990) has shown that, in provinces situated along the upper reaches of the Yangtze River, non-attenders and illiterates are heavily concentrated in more remote rural areas. In those areas, few children appear to continue beyond third grade and attainment levels off at a troublingly high rate of illiteracy. Similarly, in Guangxi, only 60% of those starting grade one reportedly continue to grade five. This is particularly unfortunate, since completion of primary school (normally 5 years in rural areas) appears to be key to obtaining marked benefits from education in China. As a 1985 study in Shaanxi shows, fewer years of formal education often leave the student illiterate: Years of Schooling Literacy Rate 0 2.9 1 10.7 2 17.9 3 38.2 4 57.6 5 72.4 6 87.2 _______________________ Source: Xu (1991). 4.08 Women's Access. Female students have increased as a share of total student enrollment in primary, secondary and tertiary schooling from 33%, 24% and 24% respectively in 1952 to 46%, 42% and 34% in 1989 (Croll, 1990 and SSB, 1990a). Despite this improvement, more than 80% of the 4.8 million school age 84 children who dropped out of school in 1990 were girls, mostly from rural and remote mountainous areas and from minority groups, and there are still more than twice as many illiterate women as men. 1 The family failure to support girls through school happens in part because in Han and most other ethnic groups females marry out of the natal family very early in their working lives and are typically not expected to make financial contributions thereafter. In other cases, schools fail to meet cultural rather than financial norms. For example, Hechi Prefecture, Guangxi, has all-expenses-paid special minority schools in Yao areas, yet these fail to attract representative numbers of girl students. In this case, the absence of Yao women teachers is believed to perpetuate girls' low attendance rates. Successful strategies to fully incorporate females in poor area education must recognize and accommodate the diverse financial and cultural pressures on poor families. ' -Returns to Education 4.09 Education affects personal choices about a range of development- related issues. Female education, in particular, is known to lead to later marriage and lower total fertility. For Chinese women, age of marriage increases slightly with primary education and markedly with a secondary education. Total marriage duration fertility falls nearly 20% with a primary education and an additional 3% with a secondary education (Lavely, 1990). Female education also results in much higher child survival rates. A study in Shaanxi showed the following relationship between mother's education level and infant mortality: Education Level Infant Mortality Rate (per 1,000) No Schooling 72.7 Some Primary 72.1 Completed Primary 44.6 Completed Junior High 35.1 Senior High or better 14.6 Source: Xu (1991). Y' China Daily (December 6, 1991), "Basic Education to Be Secured for Children." v Of course, cultural pressures are not unique to poor families or to females. An abiding problem in Chinese primary education is the language of instruction. China has many local languages and teaching is much more efficient in a language familiar to the child. However, mastery of the national standard language is needed for a child to advance beyond primary school. Educators in some poor areas, especially minority areas, complain that they spend most of their effort in primary school trying to help students master a language that 80% of them will not use once they leave school. Yet to do otherwise is to condemn the children to no more than a primary education and very limited access to written material. 85 Achievement of universal primary education among females promises to aid government in reaching a broad array of population and health policy goals. 4.10 Private Costs and Returns. Financial factors conditioning parent's decisions about sending children to school include the cash costs of schooling, any income from child labor which would be foregone if the child attends school, and the child's long-term potential contribution to the household economy. For the very poor, the costs of sending a child to school represent a very high proportion of total cash income and, conversely, even a small contribution from child labor is welcome. World Bank (1991a) reports that the direct private costs of primary education were Y41 per student-year in rural Shaanxi and Y39 per student-year in rural Guizhou in 1988. For the average farm family in Shaanxi and Guizhou those costs were not excessively burdensome!, as total family income approached Y2,000 and cash income about Y1,250 per annum. But the poorest 8% of families, with total income of less than Y1,200 yuan and cash incomes ranging downward from about Y500, would find these costs difficult to finance. Foregone child labor was also found to represent a major cost to poor rural households choosing to educate their children and to play a major role in the dropout problem in poor areas of these two provinces. School quality in poor areas is relatively low, furthermore, lowering potential returns to education. 4.11 The evidence is ambiguous on private financial returns to education in China. Analysis of Chinese data shows that an extra year of education for a rural head of household results in a 10% increase in family sideline income. However, in field crop production, the extra education results in no income increase (Jamison and van der Gaaa, 1987). This anomalous result likely derives from close state control of field crop cropping patterns, input use and output markets. Social returns may be much higher than private if the state effectively taxes away or otherwise captures the benefits of increased productivity. The experience in other Asian, African and Latin American countries shows an average 26% social return to an additional year of primary education (Psacharopoulos, 1985). Since parent's choices about investment in their children's education are conditioned by private, not social, returns, the pattern of returns shown above will result in investment based on possible employment outside of grain farming. Those prospects are best in wealthier areas, where greater income levels already induce parents to invest in primary education. -Education Financing and Services 4.12 Financing. As summarized below, total nominal public expenditure on education increased from Y8.3 billion in 1978 to Y39.8 billion in 1988, or by about 180% in real terms: 86 Expenditure on Education --------------------------------------------------- Total As a Share of: Government ---------------------- Budgetary Extra- Total Budgetary Expenditure GNP Total Budgetary Budgetary Expenditure GNP ----------------- ------- Y billion ------- -------- % --------- --- Y billion --- 1978 8.3 7.1 1.2 6.4% 2.3% 111.1 358.8 1985 27.0 25.5 1.5 13.8% 3.2% 184.5 855.8 1888 39.8 36.4 3.4 13.4% 2.8% 270.7 1398.4 Source: SSB (1990a). Note: All figures are current yuan. Total budgetary expenditure on education increased by 200% in real terms over this period, expanding from 6.4% of total government budgetary expenditure in 1978 to 13.4% in 1988. Extra-budgetary expenditure on education increased by about 65% in real terms over this period, and remained roughly constant as a share of total government extra-budgetary expenditure during 1982-87. The 1985 reform of education finance transferred much of the responsibility for mobilizing financing of secondary and higher education, and control of the allocation of those funds, from the central government to provincial and local authorities. Primary education, on the other hand, has always been the responsibility of the provinces and primary schools continue to be administered by county governments. Overall, real expenditure per primary school child nearly doubled during the 1980s, from Y23 in 1980 to Y45 in 1987 (constant 1980 yuan; World Bank, 1991a). 4.13 Available evidence suggests that the real increases in education expenditure during the 1980s have primarily benefitted urban and better-off rural areas (World Bank, 1991a). In urban areas, schools benefit from a much greater local revenue base and from categorical grants from the provincial or national government. Rural primary schools, however, must rely almost entirely on local community support. Much of the funding increase for primary education during the 1980. has been directed to (i) increasing the number of teachers -- during the last ten years the national average student-teacher ratio declined from 28:1 to 18:1 in urban areas and from about 35:1 to 30:1 in rural areas -- and, in rural areas, (ii) improving allowances for minban teachers (hired directly by local authorities) or upgrading them to conoban status (officially employed by the government and enjoying superior salaries, retirement pensions, food subsidies and "urban residency"). While increasing the number and compensation of teachers has almost certainly served to improve the overall quality of rural primary education, it appears to have primarily benefitted better-off rural areas. 4.14 Consistent with growing differences in local capacity for revenue mobilization, disparities in local level support for rural primary education are substantial and increasing. Counties and townships with a strong county, township, and village enterprise base draw much of their funding from that 87 source. But such enterprises are rare in poor areas and where found tend to have low profit rates. In poor areas, the authorities are instead forced to rely on the inherently difficult vehicle of direct taxation to garner funds from a population with relatively little ability to pay. Y Local revenue generation in poor regions is in general very limited, and there appears to have been little if any real increase in the provision of education in poor communities during the 1980s. Central, provincial and county governments are known to play a very limited role in equalizing primary education funding and opportunities, and at present none of the limited categorical grants from these higher levels of government reach primary schools in poor villages. In Yunnan, for example, central and provincial contributions to primary education rose only 2% during 1985-90 while total costs increased by 82% (Working paper 4). World Bank (1991a) details substantial inequalities in education financing at all levels of government due to the dependence on local revenue sources, and reports that the lack of a viable revenue base at the township and village level makes it difficult to decentralize public and education finance to that level in poor areas. That report notes that many poor counties, faced with decreasing fiscal revenues in 1989 and 1990, significantly reduced their allocations for education. 4.15 Services. Field studies have documented that the inadequate funding of primary education in many poor rural areas has resulted in schools which are in poor physical condition, are inadequately staffed, and are deficient in books and other minimum necessities. More than two-thirds of primary schools in poor areas are incomplete (often including only first and second grades) and, more often than not, comprise only one or two small rooms. In poor counties, complete rural primary schools (at least grades one through five) are in most cases located several hours away (by foot) from the poorest villages, and are therefore inaccessible to the majority of poor children. Urban primary schools are staffed by conoban teachers and enjoy student- teacher ratios of 14:1 or less. Poor area rural primary schools, by comparison, are staffed primarily by poorly paid minban teachers (who often take on othLer employment to supplement their income) and suffer student- teacher ratios of 60:1 or more (World Bank, 1991a). In sum, the underfunding of rural schools in poor areas condemns them to mediocrity and the poor to a bleak, ill-educated future. -Intervention Strategies 4.16 Overview. Improving primary education in poor rural areas will require measures to (i) assure that all poor children have the means to go to school, (ii) assure that the curriculum meets local needs, (iii) increase the number and improve the quality and motivation of their teachers, and (iv) increase school rooms and equipment. Achieving these three tasks is essential to realizing the government's goal of nine-year universal primary education v In such an environment, user fees play an important role in funding social services, particularly education and health. When the poorer families in a poor village have no more than Y200 a year in cash income, school costs of Y40 per child can be prohibitive. 88 and, given the limited prospects for increased revenue mobilization in the poor areas, will require increased central and provincial government budgetary funding for provision of primary education in poor rural communities. The costs associated with improving primary education in poor rural areas, which are estimated below, can be met through a combination of incremental transfers to the poor areas earmarked for this purpose and efficiency gains in planned education expenditure during the 1990s. These efficiency gains could be achieved over time through the gradual redirection of planned education expenditure from over-staffed and over-equipped urban schools to poor rural schools and careful control of the costs of secondary education -- particularly higher education institutions and the vocational schools. ' However, even assuming no efficiency gains, the annual incremental funding requirement of Yl billion for the proposed poor area education program would still represent less than 0.3% of total government planned budgetary expenditure. 4.17 Direct Costs of Attendance. The financial costs of increasing poor children's access to primary education depend on household income, the direct costs of attendance, perceived benefits, and foregone income. A number of recent experimental programs provide evidence on the question of how much of a subsidy parents need in order to decide to send their children to school. World Bank (1991a), for example, cites experiments in two primary schools in Guizhou's Taijiang County, where halving the costs of textbooks and school fees resulted in an increase in female enrollment from 20% of the student population to 50%. Similarly, in Guizhou's Jingping County, provision of free schooling for girls resulted in 91% of school-age girls enrolling in 1987 -- much greater than the 1988 provincial enrollment average for girls of 77%. Providing between Y30 to Y50 per female student to cover school fees and books, the Zigen Fund's experimental program in 13 poor Miao minority villages in Guizhou's Leishan County has reportedly increased female attendance in first and second grades from nearly zero to almost 100% 1 5 4.18 Such experiments provide the basis for developing culturally acceptable, minimum cost national and provincial programs to improve primary education among the rural poor. An expanded program might target the roughly 15 million primary school aged children now living in absolute poverty. At the current average cost of about Y40 per child, full scholarships could be given to all in the poorest group for Y600 million annually. In practice, interventions would be more varied than simply giving scholarships and may in some instances benefit all children in a particular school. An additional Y50 million fund should also be established to provide full scholarships for secondary education for a limited number of the most promising poor area children. V World Bank (1991a) discusses the potential for efficiency gains in education during the 1990s, and concludes that "the Chinese Government should be able to meet the recurrent costs of universal nine-year education without increasing education's share of the total government budget." 5/ Information provided to mission by Zigen Fund officials. 89 4.19 Teachers and School Rooms. An additional Y350 million in funds would be needed to hire an additional 350,000 minban teachers for the poor areas, complete more village level schools through the construction of more classrooms, and purchase some minimal classroom equipment. Teachers are a critical input to the education process, and field studies have shown the importance of the morale and status of the village teacher to school attendance and education quality. Hiring an additional 350,000 minban teachers would reduce the average student-teacher ratio in the poor areas to 25:1. Providing all current and the newly hired minban teachers with the current government-stipulated minimum minban salary would require about Y200 million annually. a The additional Y150 million would be sufficient to construct (or rebuild) and equip about 100,000 classrooms annually, and would be adequate to complete or upgrade most poor area village schools within a period of four years. 7' 4.20 Total Costs. The total program would be about Yl billion per year, or about 2.5% of current education expenditure. An initial period of experimentation, covering 10% of the target population and running for two years, would require a total of no more than Y200 million yuan. During this start-up period, these funds would be used to support an array of experiments across the poor areas of China designed to further identify the costs and most efficient means of achieving universal primary education. The experiments would be accompanied by the development of better indicators of achievement (not just enrollment or attendance), and carefully structured experiments using a variety of subsidies or other incentives will test response by ethnic group, gender, household income level and school quality. s' F' A Ministry of Finance regulation mandates that provinces provide a minimum cash subsidy of Y540 per year for all minban teachers, and the national average compensation from that source is now about Y650. Minban teachers may also be paid in cash or in kind by the village in which they work (usually their home village). In addition, village teachers may continue to derive income from farming (teachers, and all other village residents, received land use rights during implementation of the PRS) and other economic pursuits. Y Assuming classrooms would be constructed through China's existing Food- for-Work program (para 5.18), or some other low cost method. Y In some villages, the goal will not be to subsidize all village children or to completely cover all costs of participating children. The need for case-by-case assessment of family circumstance in some villages lends itself to the use of the MCA system (discussed in section D below), not the educational system, to allocate funds. MCA already evaluates poor families for other pioverty-related subsidies, so this effort would require little additional staff time. Hence an appropriate element of the experiments would be to test the effectiveness of directly linking the educational subsidy to relief grain amounts, further lowering administrative work. 90 C. Health 4.21 Overview. Health services are the weakest component of the rural safety net. This weakness is directly responsible for much of the excessive morbidity and mortality in the poor areas. Access to basic preventive and curative health services in the poor areas, which was always quite limited, stagnated during the 1980s due to a relative reduction in public financing for health and the privatization of most health services in the rural areas. While the privatization of health services may have been successful in the better-off rural areas, it has failed in the poorest areas. Given extremely limited financial resources in poor villages, the government's current program to establish locally financed rural medical insurance cooperatives is not an adequate means of improving health services and status in the poor areas. Instead, supplementary central and provincial government funding of a limited array of inexpensive medical interventions should be included as a fundamental component of the poverty alleviation program. Programs to increase the availability and quality of health staff, facilities, equipment and other inputs in poor townships and villages should be supplemented with continued efforts to increase access to health services, safe drinking water and improved sanitation. -Maternal and Child Health 4.22 Infant and Maternal Mortality. A 1989 MOPH survey of health services and status in 300 poor counties (MOPH, 1991), undertaken with the technical assistance and financial support of UNICEF and UNFPA, documents the dismal state of maternal and child health (MCH) in the poor areas. The survey found that (first year) infant mortality (IMR) averaged 68 per 1000 in the 300 MOPH counties -- about 50% greater than the national average -- and exceeded 100 per 1000 in 38 of the counties. More than 80% of infant deaths occurred at home or on route to medical facilities and, among infants who died, less than 50% were seen by any health worker in the 24 hours preceding death. Pneumonia (24.4%), neonatal asphyxia (16.4%), prematurity (14.3%), diarrhea (8.4%) and neonatal tetanus (8.8%) together accounted for more than 70% of first year IMR. Only one-third of the women in the MOPH counties received any antenatal or postpartum care, and only 36% of deliveries met basic hygienic standards. Official statistics (All China Women's Federation, 1991), by comparison, indicate that the national average hygienic delivery rate reached 97% in 1988. 2' Maternal mortality averaged 202 per 100,000 in the MOPH survey counties, or more than twice the national average of 95 per 100,000, and exceeded 300 per 100,000 in some poor and remote areas (MOPH, 1990). Two- thirds of maternal mortality occurs at home or on route to medical facilities, and 45% of those who died were not seen by any health worker in the 24 hours preceding death. Postpartum hemorrhage and infections account for 55% of maternal death. In a number of field projects in China, these causes of 2' The national figure probably corresponds to the distribution, but not the actual use, of sterile delivery packets. The 300 county MOPH survey, by comparison, used a stricter definition of hygienic delivery requiring presence of a trained attendant. 91 infant and maternal death have been sharply reduced through simple and relatively inexpensive improvements in pre-natal care, maternal nutrition and delivery practices, care of the newborn, and home or health center primary treatments (for simple prevention and case management of pneumonia and diarrhea). 4.23 Upon closer examination, it appears likely that even the MOPH survey may significantly understate infant mortality in the poorest areas. Rural per capita income in the MOPH counties averaged between Y113 in Ningxia to Y565 in Jilin -W and average IMR ranged between a low of 38 per 1000 in Zhejiang to a high of 108 per 1000 in Guizhou. Surprisingly, there is no statistically significant relationship between income and IMR -- rural per capita income explains only 3% of the variation in IMR. Detailed data for the 20 counties in Yunnan and the 16 counties in Guangxi included in the MOPH survey, moreover, evidence a weak positive relationship between county average rural income levels and IMR. The absence of the expected strong negative relationship between rural income levels and IMR is most likely due to some combination of (i) the underrepresentation of the poorest villages in reported county-wide averages and (ii) the underreporting of neonatal mortality in those very poor villages included in the averages. (Other confounding factors include (iii) the possible underreporting of income in counties seeking inclusion in the survey and (iv) relatively high income in areas, such as remote grassland counties, with geographically-determined limited access to health services.) The latter hypothesis is supported by a statistically significant positive relationship between county-average hygienic delivery rate and neonatal mortality observed for the 16 MOPH counties in Guangxi. W' The counterintuitive relationship between hygienic delivery and neonatal mortality strongly suggests that poor villages lacking in MCH services are not inclined, or are inadequately staffed, to fully measure and report all neonatal mortality. The supposition that the MOPH survey data may significantly understate infant and child mortality in the poorest areas is also strongly supported by poor area village studies (Croll, 1991) -- the majority of poor households interviewed had experienced at least one child death and many households had lost two or more children. 4.24 Nutritional Status. Recent large-scale anthropometric surveys evidence high levels of childhood malnutrition, as indicated by below average height-for-age and weight-for-age, in the poor areas and document a strong negative correlation between malnutrition and income and parental literacy.fL' Lot Average rural per capita income in the MOPH counties -- Y374 -- was nearly 40% less than the 1989 national average of Y602, but exceeds this paper's estimated 1989 poverty line of Y262 by more than 40%. Li' County average hygienic delivery rate explains 29% of the variation in IMR (significant at the 5% level). L2 The physical growth rates of children under age five are determined more by their health and nutrition than by their genetic potential. Height and (continued...) 92 Height-for-age data for 10,000 rural preschool children, from annual surveys undertaken in 18 poor areas of seven provinces during 1986-89 (Child Nutrition Surveillance Working Team, 1991), show that the incidence of malnutrition averaged 38% in areas with per capita income levels of between Y200 to Y500 -- more than double the range of malnutrition, of 10% to 20%, observed for a broader sampling of rural preschool children in 1985 (UNICEF, 1989). The incidence of malnutrition ranged between 19% to 78% across the survey's 18 surveillance points, and was greatest among minority children -- exceeding 60% in the survey's surveillance points in Miao, Tujia and Buyi minority areas of Guizhou and Hunan. In their analysis of the survey data, Chinese researchers found malnutrition to be negatively correlated with both income and parental literacy. The poor area data show that malnutrition was greatest among children ages 2 and 3, and the results of a companion food intake survey led the Chinese researchers to conclude that the lack of access to nutrient-dense weaning foods in the poor areas was the principal contributory factor. 4.25 Anthropometric data for 91,000 children under age six, from the 9 province 1987 Survey on the Situation of Children in China (SSB, 1989), provide even more convincing evidence of the high incidence of malnutrition among poor children and the strong negative correlation between malnutrition and income and parental eduction. Researchers of the Chinese Academy of Preventive Medicine and the Beijing Institute of Pediatrics (Chinese Academy of Preventive Medicine, 1991) disaggregated the 1987 data by income group, and found that more than 45% of the children of rural households with per capita income levels less than Y250 were malnourished. As shown below, this level was nearly 10% greater than the rate for children of the 47% of rural households with per capita income levels between Y250 and Y500, and about double the rate for children of the 36% of rural households with per capita income greater than Y500: Rural Incidence of Incidence of of Malnutrition Household Average Malnutrition Educational Maternal Paternal Per Capita Income Urban Rural Status Education Education less than Y250 30.0% 45.6% Illiterate 48.0% 52.2% Y250 to Y500 22.7% 37.6% Primary 40.5% 45.6% Y500 to Y750 12.9% 26.9% Middle 33.1% 36.3% Y750 to Y1000 9.4% 23.1% High School 26.9% 32.5% more than Y1000 >9.4% >21.0% University 8.2% 18.2% _______________________________ Source: Chinese Academy of Preventive Medicine, 1991. LV(... continued) weight are therefore commonly used indicators of health and nutritional status of young children. A child is deemed to be (or to have been) at least mildly malnourished if his or her height-for-age is less than 90% (equivalent to at least two standard deviations below) of WHO international standards. 93 Similarly, 30% of the children of urban households with per capita income levels less than Y250 were malnourished, or more than double the rate for children of the 21% urban households with per capita income levels between Y500 and Y750, and more than triple the rate for children of the 67% of urban households with per capita income greater than Y750. Disaggregating the rural survey data by parental education, the researchers also found that half of the children of illiterate parents were malnourished. As shown above, the incidence of malnutrition declined to about one-third for the children whose parents had attended middle school, and to less than one-fifth for the very few (0.1% of the sample) rural children whose parents had attended the university. 4.26 Immunization. Initiated in 1978, China's "Planned Immunization Program" has intensified immunization efforts and contributed to a rapid decline in the 1980s of the diseases preventable by means of immunization, particularly measles, pertussis, diphtheria and polio. By 1987, at least 12 provinces had achieved the Program's target of 85% immunization of children under age one for these four diseases. The MOPH recently announced that this target had been achieved for all provinces by the end of 1989. 123 The 1987 Survey on the Situation of Children in China (SSB, 1989), however, indicates that immunization rates differ sharply by region. The Survey shows that more than 60% of urban children under age 6 have been immunized for all four diseases, or about twice the rate of 33% for rural children. B Only between 10% to 30% of the children in the mountainous areas of southwest China had been immunized for all four diseases. Coverage statistics for 1987-88 show that inadequate funding, difficulties with transport and maintenance of the cold chain, and resistance to immunization constrained coverage to well below the 85% target in remote and minority counties of the northwestern and southwestern provinces of Gansu, Qinghai, Xinjiang, Guangxi, Guizhou, and Tibet (UNICEF, 1989). However, a recent UNICEF sponsored survey documents much greater extension of the Program in recent years -- nearly 100% of the 283 counties surveyed in 1991 had achieved 85% coverage of their under age one population. -Disease Morbidity 4.27 Index of Infectious Diseases. Available figures suggest a nationwide decline in disease morbidity during the 1980s. The index of infectious disease morbidity, which is the simple sum of the reported rates of 35 infectious diseases, reportedly fell from 875 per 100,000 in 1985 to 339 per 100,000 by 1989. 12' Though China's disease reporting system improved 3 China Daily (April 1, 1991), "Nation Seeks to Curb Diseases." 'L Immunization rates from the 1987 survey, since they are for all children under age 6, are much lower than Program targets and reported rates for children under age one. WS/ See "Health Care Grows Fast in Five Year" in China Daily, December 6, 1990. 94 considerably during the 1980s, the accuracy and representativeness of these index figures are considered somewhat questionable since (i) patients treated at the village level only, accounting for the majority of morbidity for some diseases, may not be reported, (ii) diagnoses are not always reliable, and (iii) they exclude a number of important diseases such as tuberculosis and pneumonia. 4.28 The detailed data for 13 poor counties in Guangxi shown below indicate a similarly steep decline in infectious disease morbidity, from 2323 per 100,000 in 1980 to 509 per 100,000 in 1989, but also illustrate some of the problems in interpreting trends in these index figures. Seven of the more common diseases make up over 98% of all the disease reports. Influenza comprises 44% of 1980 disease reports, and the tremendous decline in influenza morbidity, from 1021 to 107 per 100,000, accounts for half of the total decline in morbidity during 1980-89. However, there is no control program for influenza in China and the decline in influenza morbidity most likely reflects a change in case definitions. Influenza occurs at all levels of economic development, is usually not severe, and is easily confused with other conditions. The substantial declines in two of the water and food borne diseases -- dysentery (morbidity data for which exclude common acute diarrhea) and hepatitis A -- should be interpreted with caution, since the reliability of reports for these diseases may be low and single year figures may not accurately reflect longer term trends, but may represent real improvements in rural water supply and sanitation in some areas. The reported declines in measles, pertussis and malaria, on the other hand, most likely represent real achievements of the immunization and malaria control programs in rural areas. The decline in morbidity due to these three diseases accounts for more than 30% of the total decline in reported infectious disease morbidity during 1980- 89. Infectious Disease Morbidity in 13 Poor Counties of GuanQxi Cases Per 100,000 Population Disease 1980 1989 % Change Influenza 1021 107 -90% Dysentery 553 268 -52% Viral Hepatitis 82 48 -42% Typhoid/Paratyphoid 33 38 +15% Measles 366 24 -93% Pertussis 96 6 -94% Malaria 151 11 -93% All others 18 8 -56% Total 2323 509 -88% Source: Guangxi Bureau of Public Health. 4.29 Tuberculosis. Morbidity data for diseases not included in the index of infectious diseases, such as tuberculosis, present a less sanguine 95 picture of health status in the 1980s -- particularly in the poor areas. MOPH national surveys of tuberculosis rates do show that the prevalence of infectious cases declined by about 2.6% per annum during the 1980s, from 187 per 100,000 in 1979 to 156 per 100,000 in 1984-85 and then to 141 per 100,000 in 1990. Lu However, modern outpatient treatment for tuberculosis is highly effective and should produce annual rates of reduction of at least 5% to 10%. Failure to achieve such reductions in the 1980s is directly attributable to failure of the tuberculosis control program -- particularly in poor and remote areas. Just as new highly effective methods of short-course intensive chemotherapy were being successfully introduced elsewhere in the world, the health sector financial reforms of the early-1980s sharply reduced the budgetary support available for tuberculosis treatment in China. The full burden of financing for examination and treatment now falls directly on the patient, and the cost of treatment -- about Y150 -- is too high for uninsured poor families. This has led to failure to treat and to ineffective partial treatments, resulting in large numbers of chronic infectious cases in the poor areas. It is estimated that half of all current infectious tuberculosis cases have received partial treatment, and that many of these have developed resistant forms of the disease (World Bank, 1991c). Tuberculosis rates correlate closely with poverty, and there is little prospect for controlling the disease in the worst affected areas without provision of (i) free short- course chemotherapy, particularly for highly infectious patients, and (ii) adequate operational budgets for preventive services. 4.30 Endemic Diseases and Micronutrient Deficiencies. Endemic diseases and micronutrient deficiencies, including schistosomiasis and iodine deficiency disorders (IDD), are often highly localized and frequently concentrated in poor areas. While uncommon nationally, they impose a serious burden on the more than 60 million people residing in seriously affected communities without effective control programs. Specific control measures can prevent or at least control all of these diseases, and rates of disease depend directly on the activity of local control programs. The national schistosomiasis program has a history of more than 30 years of successful control work and, by the end of the 1970s, had eradicated the disease from much of its original range. However, decreased funding and attention to schistosomiasis in the 19809 led to the spread of the disease to formerly controlled areas and increased numbers of acute infections. A major effort is now required to protect the 44 million rural inhabitants still at risk. Considerable progress has been achieved in controlling IDD -- iodized salt has greatly reduced the rates of goiter and cretinism in many areas -- but more than 20 million people in remote poor areas are still beyond the reach of the salt iodinization program (Li Jianaun, 1989). 4.31 Other micronutrient deficiencies, including iron and vitamin A, are also prevalent in poor areas. The 1987 Survey on the Situation of Children in China (SSB,, 1989) showed that 6.2% of the rural children surveyed suffered severe anemia (hematocrit less than 9 g per mm), or nearly three times the W6' "Nationwide Random Survey for the Epidemiology of Tuberculosis in 1984- 85" and provisional results of the "1990 Nationwide Tuberculosis Survey." 96 rate for urban children of 2.3%. Due primarily to inadequate intake of iron in the diet and losses due to hookworm infestation, incidence of severe anemia was greatest in rural Guangdong (18.2%) and upland rural areas of Yunnan (11.3%). The 1987 Survey also documents that mild anemia (hematocrit less than 11 g per mm) is common in both poor and better off areas. Incidence of moderate anemia averaged between 38% among urban children to 50% for rural children, and exceeded 80% in rural Guangdong. A recent study of vitamin A supplementation on childhood diarrhea and respiratory disease in 7 low income villages in upland Hebei (Chena Lie and Chang Yina, 1990) indicates considerable vitamin A deficiency in poor areas and confirms the cost effectiveness of vitamin A supplementation. The study found biochemical evidence of vitamin A deficiency -- between 20% to 45% of village children had serum vitamin A levels of less than 20 ug per dl -- and achieved sharply reduced rates of diarrheal and respiratory disease through vitamin A supplementation. Pending confirmation of these results through larger-scale surveys, no firm conclusions can yet be drawn regarding the potential for reducing morbidity and mortality through vitamin A supplementation. 4.32 Water Supply and Sanitation. Between one-quarter to one-third of the rural population still have critical water supply needs -- the water supply for 77 million people has excessive fluoride levels, 100 million drink saline or alkaline water, 130 million use untreated surface water contaminated by domestic, industrial or agro-chemical wastes, and 43 million live in water- scarce areas. Proper sanitation, furthermore, is almost totally lacking in rural areas. The use of "night soils" (human wastes) as fertilizer provides an opportunity for contamination of soil and water supplies with fecal pathogens, and is a major source of related infection in the warmer more humid areas of central and southern China. The scarcity of safe water supplies and poor sanitation are worst in the poor and remote areas (para 2.36), and contribute significantly to increased incidence of diarrheal disease, fluorosis, helminthic infestation, and other health problems in those areas. A recent investigation has shown that between 60% to 80% of rural children, and more than 90% of the rural children of poor remote areas, suffer from roundworm, pinworm and whipworm. W0 While improved access to safe water supply has reduced morbidity and saved time and energy spent on water collection in many parts of rural China, the expansion of safe water supply in the poorest areas has been relatively limited. Greater government support, in combination with low-cost alternatives to piped water supply, are now needed to extend development of safe water supply in the poorest areas. -Health Financing and Services 4.33 Financing Disease Prevention Programs. Health sector financial reform and the dismantling of commune-run health care services have led to concerns that the effectiveness of disease prevention programs in rural areas may have declined during the 1980s. The adoption of cost recovery as the foundation of health care financing did lead to a relative decline in public resource flows to the health sector, from 28% of total recurrent expenditure 17 China Daily (June 4, 1991), "Health Plan Targets Parasites." 97 on health in 1979 to 19% in 1987, and was associated with a relative decline in recurrent expenditure on disease prevention programs. Lw Most significant disease prevention programs are provided by the public institutions specializing in disease prevention -- the Anti-Epidemic (AES) and MCH stations, most importantly -- and public expenditure on these programs declined from about 5.5% of total recurrent expenditure on health in 1979 to about 3.2% in 1987. As a result, public expenditure on disease prevention programs now represents a very small and declining share of GNP (less than 0.1% in 1987). In contrast to this relative decline, however, real public recurrent expenditure for disease prevention programs increased by 50% during these years, from Y540 million in 1979 to Y870 million in 1987 (expressed in constant 1980 yuan). L'9 Furthermore, greater absolute levels of expenditure on disease prevention programs have been matched by a steady increase in the number of AES and MCH stations and intensified childhood immunization efforts since 1978 (para 4.26). This real increase in expenditure and infrastructure for disease prevention programs notwithstanding, there is concern that strong pressure on all health institutions, including those devoted primarily to prevention, to earn income from alternative services (e.g. inspection of factory products, liver clinics, ultrasonic examination in pre-natal care, etc.) diverts staff time from adequate support for preventive programs in the countryside. 4.34 Rural Health Insurance and Financing. There is also concern that the privatization of health services in the late-1970s and the subsequent elimination of the commune system may have reduced the effectiveness of rural health services more generally. In 1975, cooperative health insurance schemes and subsidized health services provided up to 85% of the rural population with basic preventive and curative services at low charge to the patient, and protected some individuals and families from some of the cost of more expensive higher level medical care (Chao Limin et. al.. 1982). The commune work point system supported the health work of large numbers of "barefoot doctors" and brigade (now village) "health aides" and put accessible basic services in most villages (Kan Xuecui, 1990). Beginning in the late-1970s, however, most cooperative health insurance schemes were discontinued. Coverage slipped to 58% in 1981, and only 20% of the rural population now have "limited communal insurance coverage." The bulk of the rural population, including members of the "floating population" and those in poor areas, now lt Health sector expenditures and finance are discussed in detail in World Bank (1991e). 19/ Real recurrent expenditure on health (all sources) is estimated to have increased by 160% during 1979-87, from Y10.7 billion to Y27.7 billion (expressed in constant 1980 yuan). Public expenditure on disease prevention programs, which represents the bulk of all expenditure for disease prevention, has been approximated as public recurrent expenditure for the AES and MCH stations (World Bank, 1990e, Annex Tables 9.1 and 9.4). Public recurrent expenditure for the AES and MCH stations increased in real terms in all provinces during 1979-87. 98 have virtually no coverage. (Most registered urban dwellers, by comparison, are well covered by government or labor insurance systems.) 4.35 The impact of the privatization of health services and the discontinuation of health insurance on the effectiveness and availability of health services in the poorest areas is, however, uncertain. Barefoot doctors and brigade health aides, recruited from the local population and also comprising young people sent to the countryside from urban areas, had highly variable levels of basic education and medical training, the latter often lasting only a few months. Particularly in the poor areas, their effectiveness in delivering health services must be suspect. More significantly, risk sharing was at the level of the brigade during the 1970s, and there was a great deal of variation between brigades in the ability to fund health services -- the poorest areas never had health insurance. Plans to reestablish cooperative health insurance schemes -- including targets of 50% coverage of the rural population by 1995 and 70% coverage by 2000 -- figure prominently in the government's strategy to achieve its "Health-For-All in 2000" targets. ' These plans do include the poor areas, but just as was the case in the 1970s, the limited financial resources of poor villages will severely constrain the effectiveness of locally funded insurance schemes. 4.36 MOPH financial support for rural doctors declined 45% in real terms during 1979-87, from Y36 million to Y20 million (in constant 1980 yuan). As a result, most of China's village doctors (upgraded barefoot doctors who have passed a qualifying exam) now rely on patient registration fees and a fixed mark up on drugs (prescribed for curative services) for their compensation. Government assistance and patient fees are so limited that, in the poor areas, village doctors spend most of their time engaged in the more remunerative activities of agricultural field work and animal husbandry, and are often not available in an emergency. Financial support for village doctors' preventive work may come from fixed payments by the village, township, or (rarely) county government, the collection of fees at the village level for preventive services, and patient paid "contracts" for full immunization or maternity care. But in many poor areas, village doctors are expected to do preventive work without specific financial support. 4.37 The financial threat of illness to a poor family comes both from lost income or services and the cost of medical care, when it is available. The Ministry of Civil Affairs has identified medical expenses as the leading cause of indebtedness among poor families. Poor area village studies (Croll, 1991) show that the costs of medical care, especially hospitalization and drugs, impedes the poor's access to curative health care. The practice of requiring a deposit prior to hospital admission (Y50 for a township hospital) is the most frequently reported deterrent to hospital consultation. Hospitals are consulted only as a last resort, and are now frequently associated with debt and death by poor villagers. g' China Daily, "Grim Diagnosis on the State of Rural Health Care" (July 13, 1991) and "Rural Areas Plan Group Health Care" (July 10, 1991). 99 4.38 Health Infrastructure and Personnel. National statistics indicate an unbroken increase in overall health infrastructure and services during 1978-89. The total number of health institutions increased each year, from 170,000 in 1979 to 207,000 1989, as did the number of personnel and doctors in these institutions. The number of AES and MCH Stations steadily increased, from 2989 and 2459 respectively in 1978 to 3591 and 2796 in 1989. aowever, most of this expansion occurred at the county and higher levels, and the national aggregate figures obscure evidence of a concurrent erosion of some health infrastructure and services at the township and village level. The number of township health centers and hospitals, which play the key role of supervising the work of village health workers and providing first level referral for common conditions, declined by about 15% from a peak of 55,500 in the early-1980s to 47,400 in the mid-1980s. With the greater emphasis on cost recovery adopted during the 1980s, some township hospitals were able to prosper and expand their curative services. Most, however, appear to have lost in competition with the more advanced county level facilities, which are preferred by individuals with the ability to pay and who bypass the township level. This has led to a loss of high quality staff, equipment and facilities, and further declines in numbers of patients. In Liaoning, a relatively wealthy province, one third of township hospitals are reported to be bankrupt. 2i' In poor areas, the condition of township health centers depends critically on the support offered by township and county governments to carry out their preventive work. In some of the poorest areas, including 273 poor townships in Guangxi (accounting for 20% of the Region's townships), township health centers are simply absent. In villages with average and above average income levels, there is generally adequate financial support to maintain basic health services. In the poorest areas, however, much of the entire infrastructure of village level facilities may be absent or, if present, lacking in even basic equipment and supplies. At present, about 14% of China's villages lack health stations. 2' 4.39 Available figures for rural health care personnel during 1975-88, summarized below, confirm a significant erosion of health care personnel at the village level after 1975: 21/ China Daily (April 18, 1991), "Hospitals Suffer From Lack of Funds." The article also reports that a survey in Hunan determined that only 46% of rural patients were treated by local village and township facilities. The remainder went directly to the county level hospital. 2a China Daily (January 30, 1991), "State Calls for Health Scheme in Rural Areas." 100 Salaried Commune \_ Brigade Rural Health Health Barefoot Rural Health Care Workers Care Workers Doctors \b Midwives Aides Year ----------------------- million workers -------------------------- 1975 1.56 0.62 3.28 1977 1.28 1978 1.32 0.96 1.67 0.74 3.11 1981 1.58 1.40 0.59 2.01 1984 1.70 1.01 1.25 0.52 1.16 1986 1.73 0.88 1.28 0.51 1988 0.87 1.25 0.47 Source: SSB (1991 and 1990b) and World Bank (1990e and 1984). \a Reclassified as township workers after 1984. \b For 1985-88, the sum of "rural doctors" and "village aides." The figures indicate that the number of salaried rural health care workers steadily increased from 1.3 million in 1977 to more than 1.7 million in 1986. However, this increase in staffing was apparent only at the county level. By comparison, the number of commune (now township) health care workers remained roughly constant during 1978-88 while the number of village level barefoot doctors and midwives declined from 2.2 million in 1975 to 1.7 million in 1988. The decline in the number of brigade health aides, from 3.3 million in 1975 to 1.2 million in 1984, was most pronounced. The total number of village level health care workers (including all part-time barefoot doctors, midwives, and brigade health aides) declined from 5.5 million in 1975 to 4.0 million in 1981, to 2.9 million in 1984, and appears to have continued to decline thereafter. While much of this decline is explained by the "retirement" of very poorly trained part-time workers, a considerable portion must also correspond to the exit of health care workers during the privatization of health services and other rural reforms initiated in the late-1970s. 2' 4.40 Trends in Health Status During the 1980s. There is little convincing evidence of a significant change in health services or status in the poor areas during the 1980s -- the health status of the absolute poor appears to be at least as miserable at present as it was in late-1970s. Available evidence does document a significant erosion of township and village level health infrastructure and personnel after 1975 (paras 4.38 and 4.39). However, health services were at best extremely limited in the poorest areas, so it is unlikely that the decline in cooperative health insurance schemes and free medical services in the late-1970s, or the dismantling of the commune system in the early-1980s, could have had any significant impact on the health of the absolute poor. Instead, some disease prevention programs crucial to the health status of the poor are know to have expanded during the 1980s. 2' Official MOPH figures indicate an even greater decline from 5.5 million village level health care workers in 1978 to 2.2 million in 1984 to 1.7 million in 1990 (MOPH, 1991). 101 Real funding for disease prevention programs (para 4.33) and the number of AES and MCH stations (para 4.38) increased steadily during the period. Official figures (P,Qricultural Publishing House, 1991) also show that improvements to rural water supply reduced the number of rural inhabitants lacking access to safe water supply by almost 40 million during 1985-89, including more than 6 million no longer exposed to excessive fluoride levels. While recognizing the limitation,s of available disease morbidity data, the declines reported for measles, pertussis, malaria, IDD, viral hepatitis and dysentery in the poor areas during the 1980s (paras 4.27 and 4.28) are almost certainly the result of improved immunization, malaria and IDD control, and rural water supply and sanitation. These gains may have been partially offset, however, by a resurgence of other diseases, such as tuberculosis and schistosomiasis, in the poorest areas (paras 4.29 and 4.30). -Costs of Improving Health Services and Status in the Poor Areas 4.41 Overview. The government's "Health-For-All in 2000" program calls for the reestablishment of the rural cooperative medical insurance system in all areas of China (para 4.33) and an increase in the training of students from rural areas in secondary medical schools. W' While these measures may serve to improve health services in better-off rural areas, they will certainly not be sufficient to upgrade health services and status to acceptable levels in the poor areas. Instead, equity and efficiency require that the central and provincial governments increase their assistance to the poor areas in support of a limited set of health services directed at the principal causes of morbidity and mortality. A health program for the poor areas, which comprises both (i) the establishment and operation of the basic infrastructure of the health system and (ii) ten cost-effective public health programs, is summarized below and detailed in Ouinley (1992). 4.42 Basic Infrastructure. China's well established "three tier network" of rural health services comprises: at the village level, a clinic with one or more health workers who undertake preventive and simple curative activities; at the township level, a health center or small hospital which serves the local population directly, provides referral services for patients with more severe disease, and supervises the preventive activities of the village level; and, several institutions at the county level, usually including a county general hospital, a county health school, and AES and MCH stations, which train staff and supervise the activities of lower levels and serve the urban population and referrals from lower levels. The three tier network exists throughout most of China, though wide variations in the numbers and capacities of the health institutions leave most poor villages and townships at a considerable disadvantage (paras 4.36 and 4.37). Extending the W4 "Better Health Care for Farmers" (China Daily, February 13, 1991) reports that very few health workers (most of whom were raised and educated in urban areas) have been willing to practice medicine in rural areas. MOPH hopes that students from rural areas, having had only limited exposure to urban life, will be more inclined to remain and serve in rural areas after graduation. 102 village and township tiers of the rural health service network to minimally acceptable levels in a typical poor county would require: personnel and training -- at least one "village doctor" and one village health aide per 1,000 population, one township assistant doctor and one township health technician per 1400 and 3400 population respectively, three years of training for new village and assistant doctors, one year of training for new health aides and technicians, and between three months to one year to upgrade the skills of existing doctors, aides and technicians; buildings and maintenance -- construction or renovation of 600 m2 township health centers and annual maintenance; and, 2' drugs and equipment -- essential drugs for provision of a variety of financially accessible services (but excluding the drugs required for the public health interventions discussed in paras 4.41 to 4.43) and, per UNICEF guidelines, minimum equipment requirements for the simplest village and township facilities. Including personnel and training, buildings and maintenance, and essential drugs and equipment, the minimum annualized cost of completing the village and township tiers of the rural health service network in the poor areas would cost about Y6 per capita: Costs of ComDletina Health Services in a Poor County: Basic Infrastructure Unit Cost/ Total Annual Cost Item Number Salary Cost Per Capita ------------------------------------------------------------------__--------- Personnel and Training: -village doctors 400 Y500 Y200,000 Y0.50 -village health aides 400 Y400 Y160,000 YO.40 -township assistant doctors 280 Y1500 Y420,000 Y1.05 -township health technicians 120 Y1200 Y144,000 YO.36 -training of new staff YllO,000 YO.28 -upgrading existing staff Y34,000 YO.08 subtotal Y1.068.000 Y2.67 Buildings and Maintenance: -construction and renovation Y12650 Y278,000 Y0.69 -maintenance and utilities Y130,000 YO.33 subtotal Y408.000 Y1.02 Drugs and Equipment: -essential drugs Y840,000 Y2.10 -equipment Y84,000 YO.21 subtotal Y924,000 Y2.31 Total Y2,400,000 Y6.00 _________________________ Source: ouinley (1992). Note: Typical county with 400,000 residents, 22 townships and 300 villages. W Provided there is a safe place to store equipment and supplies, there is no need for a special building for basic health services at the village level. 103 4.43 Public Health Programs. Basic health infrastructure is essential to the provision of health services, but does not by itself assure a substantial impact on morbidity and mortality. For most preventive services, and for some curative services, special public health programs are needed to insure that available health infrastructure is oriented towards the most cost- effective health activities. In the poor areas, ten public health programs targeted at maternal and child health and selected infectious diseases will be the most efEicient means of reducing excess morbidity and mortality. 4.44 Each of the health programs comprises three general costs -- in- service training, supervision, and health education -- as well as specific costs unique to each program. Regular in-service training is essential to a steady improvement in preventive and curative health programs in townships and villages. At least two weeks training per year is needed for all village and township doctors, aides and technicians. A limited number of staff of the county MCH and AES stations, hospital, health school and health bureau, dealing with high priority diseases or populations, would also be provided two weeks of in-service training. This training would be delivered at the township, county and higher levels and, including teaching materials and payments to trainers, would cost between Y50 to Y300 per week or about YO.42 per capita. Adequate levels of supervision are also essential to the success of specific preventive and curative health program -- particularly new programs. County level supervisors should visit each township on a regular basis. At an average cost of Y16 per visit, each of four supervisors could visit each township once per month at a total cost of about YO.04 per capita per year. 2' Health education is another vital part of the public health programs. The most important health education occurs during routine contacts between the village or township health staff and patients or families. The major cost of health education is therefore the required staff time, and the staff ratios presented above (para 4.40) should be adequate for this purpose. Other costs of health education include posters for the village and township clinics, teaching aids for discussions with families, radio messages, slide shows, and pamphlets. The average annual cost of such health education materials is estimated to be about YO.5 per capita. 4.45 Principal content and costs of the ten public health programs are: Maternal and Neonatal Health -- Interventions to reduce morbidity and mortality take place during antenatal and post-partum visits and during labor and delivery. Pre-existing conditions or complications during pregnancy, which can be treated directly or which indicate high risk, should be identified during antenatal visits. Post-natal visits during the first month of life make possible early identification of problems such as late bleeding and neonatal illnesses (especially pneumonia), and re-enforcement of health education messages, such as for breast feeding. w' It may be possible, for example, to cover all MCH related programs in one supervision visit. Similarly, respiratory disease and diarrheal disease programs can be supervised jointly, or even combined with supervision of the MCH programls. 104 Using improved techniques, labor and delivery would be at home for low risk women and at the township level for high risk women. = Incremental program costs, of YO.46 per capita, are primarily for labor and delivery. (Antenatal and post-partum visits require time, but very little material costs.) s' Expanded Program on Immunization (EPI) -- This program is operational in nearly all counties. A series of immunizations, usually requiring four to six contacts before the age of twelve months, protects against six common childhood diseases (measles, polio, pertussis, diphtheria, tetanus, and childhood tuberculosis). Village immunization days are held monthly in most counties, and registration, gathering, and immunization of children is done by village and township EPI staff. Vaccine is provided by the county level in cold boxes, delivered on a routine schedule to each township. The county EPI program has freezers and refrigerators for vaccine, and usually either its own or a health bureau car for delivery of vaccine. (YO.06 per capita) Neonatal Tetanus Control -- Tetanus in the newborn occurs when the umbilical stump is contaminated with material containing tetanus spores, often around the time of delivery. Even with advanced medical care, infections are usually fatal between three and 28 days after birth. This problem is very closely associated with poverty, and the highest rates of neonatal tetanus are generally found in the poorest areas. In addition to continued efforts to prevent contamination of the cord after delivery, neonatal tetanus can be prevented by immunizing the mother against tetanus -- maternal antibodies will protect the newborn during the susceptible period. (YO.02 per capita) Hepatitis B Immunization -- Infection with hepatitis B can result in a sometimes severe clinical hepatitis. More importantly, infected infants, who usually do not develop clinical hepatitis, often become chronic carriers of hepatitis B infection. Worldwide, an estimated 25% of chronic hepatitis B carriers eventually die from primary hepatic cancer or cirrhosis, usually before age 60. In China, about 10% of the population are chronic carriers. Without intervention, infections of ~' Including better antisepsis to avoid maternal sepsis; cord and placental delivery techniques, manual massage, and possibly oxytocin to reduce hemorrhage; monitoring fetal heartsounds and partographs to identify fetal distress and obstruction for earlier referral; newborn resuscitation techniques to reduce asphyxia; prevention of cold injury of newborns; improved cord care to prevent tetanus; and immediate breastfeeding to reduce newborn infections. Emergency referral for mothers or infants is difficult in poor rural areas. However, significant improvements can be made with planning, even in the worst situations. & In very poor villages, a significant proportion of births are attended by relatives or informal untrained birth attendants. Free deliveries would greatly encourage poor households to use trained attendants. 105 newborns and young children will maintain this high level of carriers indefinitely. Immunization of all newborns (with special focus on those whose mothers are hepatitis B carriers) can reduce the production of new carriers by over 90%. This prevents both the long term sequelae of chronic infections and acute infections at any age. (YO.65 per capita) Tuberculosis -- Control of tuberculosis requires that a high percentage of patients found and brought under treatment are cured. The best means for achieving this is through supervised intermittent short course chemotherapy. Under this system, diagnosis, registration, and evaluation is the responsibility of the county tuberculosis service, usually located in the AES station. Supervision of patients taking medication is usually the responsibility of the village doctor. Passive case finding through routine medical contacts is sufficient to identify a high proportion of cases. (YO.10 per capita) Acute Respiratory Infections (ARI) -- Pneumonia is the number one killer of children during the first five years of life. Pneumonia can develop very rapidly in small children, and parents may fail to recognize the signs of serious illness. Particularly in poor areas, reluctance to spend money on what is mistakenly believed to be a simple cough toc often leads to failure to seek medical attention. ARI control programs focus on several steps. Parents are taught to bring children with respiratory symptoms to village health workers for evaluation. Village (and township) health workers are taught to divide cases into diagnostic groups. Common colds receive only symptomatic care. Simple pneumonia may be treated on an outpatient basis. Severe disease is admitted for hospitalization. (YO.96 per capita) Diarrheal Disease Control -- Diarrheal deaths are due to dehydration and other complications. Use of simple diagnostic criteria by families, village health workers and township doctors, and appropriate treatment with home fluids, feeding, oral rehydration or antibiotics as appropriate, can greatly reduce diarrhea associated morbidity and mortality. (YO.76 per capita) Iodine Deficiency -- More than 20 million Chinese, residing in poor and remote areas with low levels of iodine in local water and food supplies and no effective IDD control program, are still at high risk for endemic goiter, cretinism, and other IDD. Endemic goiter is the most visible problem and can affect a high percentage of the population. Cretinism, which results in moderate to severe mental retardation and physical deformities, may affect 1-10% of children born in severely affected areas. There is now considerable evidence that milder iodine deficiency produces a general reduction in the intellectual development of children, and control of mild iodine deficiency has taken a high priority among public health interventions. The mainstay of China's IDD control program -- the promotion of iodized salt -- has sharply reduced the incidence of IDD for more than 300 million Chinese residing in 106 iodine deficient areas. 3' Unfortunately, the IDD control program does not yet reach many extremely poor and remote counties. In some instances, furthermore, the poor can not afford to purchase iodized salt or may not understand its potential benefits. The IDD control program must be extended to the remaining at risk population and, in areas where it exists but is ineffective, appropriate interventions -- including removing non-iodized salt from the market, subsidizing iodized salt, increasing the concentration of iodine in the little salt that is purchased, and using oral or injectable iodized oil (especially for reproductive age women and for young children) -- should be undertaken. (YO.15 per capita) Iron Deficiency Anemia -- In 1987, the Shanghai Medical University initiated an intervention trial in seven counties of several provinces for childhood anemia. They found that 47% of children had at least mild anemia, with the highest rates in the 6-24 month age group. Moderate and severe cases were given iron supplements and other treatment. A variety of control strategies, including health education, prophylaxis, and supplementation, were tested on other children. All interventions were found to be fairly effective, and the health education alternative the most cost-effective. (YO.02 per capita) Vitamin A Deficiency -- The greatest levels of vitamin A deficiency, as well as the most vitamin A associated morbidity and mortality, occur in younger children. Vitamin A status can be improved through supplementation, fortification, and improved dietary intake, and is most important for children under three. All vitamin A capsule programmes should be accompanied by health education for the parents on the correct feeding of infants and children. Where available foods do not contain sufficient vitamin A, local production of such foods should be considered. (YO.02 per capita) The total cost for the ten programs would be about Y3.2 per capita. Other possible cost-effective public health programs include (i) control of other infectious and endemic diseases, such as malaria, sexually transmitted diseases, and helminthic infections, (ii) growth monitoring, which is a sensitive means of identifying malnourished children needing special assistance, (iii) improved family planning services, (iv) health and nutrition education, particularly in the areas of infant feeding practices and improved child nutrition, and (v) increased supply of safe water supply and improved sanitation. 4.46 Total Costs. The health program proposed for the poor areas includes both basic infrastructure and specific public health programs. This is consistent with government policy, which calls for the establishment of basic health infrastructure in all communities, and takes advantage of the W However, recent evaluations using more sensitive measures have discovered that many areas previously thought to be under control do in fact have significant rates of moderate daficiency. 107 strong complementarity between the two. At an average per capita cost of Y10, the total program cost would be about Y1 billion per year, or about 2% of total expenditure on health at present. This lower bound estimate excludes most county level costs, and is several fold less than actual levels of expenditure in most of rural China. It is expected that the proposed health program would reduce infant, child and maternal morbidity and mortality by at least one-third. The program should be accompanied by the careful monitoring of health status in the poor areas, and program content refined on the basis of demonstrated cost-effectiveness. Finally, it is essential that the funds not be used to extend more advanced medical services to the county urban population. D. Relief -Introduction 4.47 The current government program to guarantee subsistence income for all citizens was put in place in the mid-1950s and has changed little since then. The program goal has been the simple one of helping citizens avoid starvation. The responsibility for that work has been given to the family, the work place, and the state, in that order. If the resources of one level are inadequate to meet the problem, the next level takes action. The natal family is obliged to provide its members with adequate consumption, even when those members live in separate households. That obligation derives from Confucian tradition and was incorporated in the Marriage Laws of 1952 and 1980. The government invokes the law primarily to force adult children to support elderly, non-working parents. 4.48 Urban Programs. Urban families lacking the means to maintain minimal levels of subsistence are expected to turn first to their employers for assistance. Urban work places often have social welfare funds to be used to subsidize poor families, and provincial governments set a level of per capita household income below which urban families are eligible for work place support. The urban work place determines the degree of need and the subsidy to be provided, but has no specific obligation to meet the provincial standard. Actual assistance reflects the unit's economic strength as much as family need, and SSB urban income surveys document a perverse "U"* shaped distribution of such assistance, with per capita payments greater in absolute terms to the top two income deciles than the lowest two (SSB, 1988). 4.49 If, after enterprise assistance, an urban family still has serious economic difficulties they can apply through their neighborhood committee for help from the Ministry of Civil Affairs (MCA) system. If the neighborhood committee approves the application they forward it to the municipal relief committee for consideration. MCA welfare programs target the old, infirm, widowed, handicapped, and orphans, and support can be in the form of a one- time grant or a monthly allowance. The MCA estimates that in 1989 about 17.5 million urban people qualified for relief payments. However, the MCA budget allocation for urban poverty, of Y220 million annually, was only sufficient to provide 4.8 million urban residents with short term grants and another 0.2 108 million with monthly allowances in that year. Furthermore, only households holding residence permits for the city in question are eligible to apply for aid. Needy urban households lacking residence permits are told to rely on their original towns or villages, and if necessary are given train or bus tickets to return to that place. 4.50 Rural Programs. The national land reform program implemented during the early-1950s provided all rural families with assets allowing self- employment. During the commune era, all able-bodied rural residents were assigned jobs on commune lands or in sideline industry and, given the relatively egalitarian workpoint system then in use, this work usually provided a subsistence income. For those who could not work, many work teams used a distribution system based on a minimum per capita grain ration and team members lacking sufficient work points were permitted to go into debt to the team to obtain that ration (Dixon, 1981). In cases where team and commune production was inadequate, a localized sharing of poverty and malnutrition resulted, as did the need to turn to the state for assistance. 4.51 The distribution system ended with the dissolution of the communes in the early-1980s. Land was redistributed on a per capita basis, providing both the able-bodied and the non-working indigent with a productive asset. Most villages retained responsibility at that point primarily for maintaining the well-being of the elderly or young orphans without labor power or family support. Other families impoverished by death, illness, disability or other problems were also forced to turn to the village government for assistance, but in most poor areas the village has nothing to offer and serves only as a crucial intermediary with the MCA system in applying for state funds. The MCA system now runs two major programs to aid rural subsistence crisis families. The first assists families after natural disasters and the second helps the needy under more normal circumstances. 4.52 Urban and Rural: A Contrast. A key difference between the urban and rural income maintenance programs comes in assumptions about state responsibility for insuring employment. Employment serves as an effective guarantee against subsistence failure and assures minimum real levels of income, and the government has accepted as its obligation the provision of adequately remunerative employment to registered urban residents (para 1.02). The government does not assume this responsibility for the rural population. To meet its obligation to the urban population, the government has limited rural-to-urban migration and, in times of economic stress, has expelled large numbers of unregistered workers from urban places, effectively shifting the burden of unemployment to rural areas. Indeed, in the 1962-65 period even some registered urban residents were forced back to the countryside. In addition, the government helps protect real urban income levels through grain and other subsidies (para 1.02), while letting rural areas absorb annual production fluctuations. The most vivid example of this occurred during the famine of 1960-61, when compulsory grain sales allowed the government to keep urban per capita consumption within 2% of the normal level while rural consumption fell 25% (Lardy, 1985). 109 4.53 The employment obligation reflects itself in another vulnerable segment of the population -- the disabled. The urban MCA system runs some 4,700 social welfare enterprises that hire handicapped people unable to find employment in regular enterprises, with another 37,000 organized by local governments and taking advantage of concessional tax rates. MCA has no corresponding program in rural areas, where the number of disabled far exceed the urban total. One example of this can be found in Guangxi. The provincial government estimates that of 500,000 employable handicapped people, some 150,000 reside in urban areas and another 350,000 in rural areas. Social welfare enterprises hire two-thirds of the urban employable handicapped, but make no attempt to reach the 350,000 rural employable handicapped. -Rural Relief 4.54 Disaster Relief. Rural relief assistance for natural disasters can be requested by village officials after drought, flood, hail, pests or other causes cause crop losses exceeding 30 percent. Depending on the area damaged in the disaster, township, county, prefecture, or even provincial MCA officials will join with local administrative leaders to inventory losses. The MCA response to the loss depends on the time of year and magnitude of the disaster. If time permits, a short duration replacement crop will be planted, possibly with an MCA subsidy for inputs such as seed and fertilizer. If the land cannot be immediately replanted, village leaders and MCA staff will assess each family's grain stocks and probable deficit and grant or sell (through the Grain Bureau) relief grain to make up the difference. Very little protection against the 10S of assets such as equipment or housing is provided. Compensation for a destroyed house, for example, averaged only 100 yuan in recent years, while the per room average construction cost of rural housing now exceeds Y600. 4.55 The objective of MCA disaster relief is to assure subsistence, not compensate for losses. In 1988, for example, total grain losses in qualifying natural disasters were estimated at over 50 million tons. MCA official estimates were that over 211 million people were affected by these losses and that their consumption shortfall would be over 15 million tons of grain. Of the affected group, only 68 million people were given aid -- approximately the number estimated to have crop losses exceeding 60 percent -- and a total of over 5 million tons of grain distributed. Farmers are forced to rely on their own grain stocks, market purchases, catch crops, and reduced consumption to make up the difference. Families who lose a house to natural disaster often live for years in makeshift huts while they save enough to rebuild. 4.56 MCA was proscribed from diverting disaster relief funds to any other purpose until 1985, when they were authorized to invest excess funds, not to exceed 30% of the disaster relief budget, in enterprises that would employ the poor. In the six years 1985-1990, over Y1.9 billion (accounting for a consistent 30% of the roughly Y900 million annual disaster relief budget) has gone to "self-relief through development" enterprises discussed in Box 4.1. The experiment with self-relief through development was terminated in 1991, when the Ministry of Finance called a halt to the use of disaster relief funds for investment. 110 Box 4.1: "Self-Relief Through Development" Entervrises Growth of MCA supported "self-relief through development" enterprises was rapid after 1985. In the six years 1985-90, over Y1.9 billion was invested in county, township, village, and privately controlled enterprises, with the condition that poor people comprise a majority of employees. A total of over 40,000 enterprises have been funded, ranging in size from 7 to over 400 employees, and are situated in nearly all counties. Employment in 1990 amounted to more than 1.1 million people and monthly wages ranged from Y70 to Y150. Profits from the enterprises are returned to the MCA system for further relief or development work. Those profits have been substantial -- in the five years ending 1989 the enterprises reportedly generated Y1.5 billion in profits -- with over 60% of the enterprises profitable, another 35% breaking even, and 5% incurring losses. Predictably, the most profitable MCA enterprises appear to be in the wealthier areas of China where greater availability of well developed markets, technical skills, and an educated work force predispose enterprises to success. These conditions are less likely to be found in poor areas. The Yunnan Bureau of Civil Affairs, for example, had invested Y86 million in 741 enterprises by end-1990 (additional investments by other parties were made in the 411 of the enterprises which are joint ventures or wholly owned by counties or townships), but reports cumulative profits of only Y5 million through end-1990 -- a rate far lower than the national average. The Yunnan Bureau of Civil Affairs requires that at least 70% of the workers come from families deemed poor and, in order share the opportunity to earn cash incomes more widely, requires that workers be rotated every year or two. Key technical workers have permanent jobs (and are rarely drawn from the ranks of the poor), but the forced turnover of unskilled labor may contribute to the limited profitability of the Yunnan self-relief through development enterprises. As an alternative to investing in enterprises, some provincial Civil Affair Bureaus have used part of their social welfare funds to provide agricultural input loans to poor farmers. Guangnan County in Yunnan, for example, has over Y700,000 in such loans. The Guangnan loan fund was created in part because Agricultural Bank of China regulations constrain lending to farmers with outstanding bank debt. Many of the poor have existing bank debt dating back several years and are therefore unable to secure the credit necessary to adopt modern agricultural technologies requiring purchased inputs. Guangnan county reports that loan repayment rates have been limited, perhaps because MCA's tradition of serving as a source of grant funds may impede their attempts to gain repayment. 111 4.57 Income Maintenance: "Five-Guarantee" Households. The system described above deals with intermittent, large-scale disaster relief needs. But even in normal times every village has families in which age, illness, or disability has left too few members earning an adequate income. If these families cannot be sustained through contributions by relatives, they must rely upon village resources. The village obligation is akin to that of employers in the urban sector (para 4.48) -- support is provided only if funds are available. But since 1956, the one category of neighbor that villagers have been obligated to provide assistance to are the "five-guarantee" households. A five-guarantee household has "three lacks" -- they lack a source of financial support, a source of livelihood, and labor power -- and are in principle guaranteed food, clothing, housing, medical care, and burial (in the case of orphans, education is also guaranteed). In practice only young orphans and the elderly qualify for the five guarantees. During the commune era, poor communes rarely had welfare funds adequate for more than their five-guarantee households. A 1980 survey of eight low and medium income teams in communes in Anhui, Henan, Shandong, for example, found that only one team had a welfare fund (Wiens, 1985). of the four poorer teams, only one distributed any cash as part of the normal income distribution and none were capable of offering anything more than a modest grain ration to indigents. In this respect poor villages now differ little from those of ten years ago. 4.58 In 1990, 3.65 million rural inhabitants qualified for the five- guarantee program, and of these 3.36 million received aid. Of program participants, nearly 80% are elderly and some 450,000 are in institutional care, usually in one of over 37,300 rural retirement homes. 30' Areas without retirement homes (and few poor counties have them) depend on village cadre and the kindness of neighbors to monitor five-guarantee households. More than half of total annual expenditures on the five-guarantee program reportedly goes for the 450,000 people in rural retirement homes. Assistance to all other beneficiaries is by comparison quite modest, averaging only a little more than Y40 a year per recipient. Many five-guarantee households, therefore, are provided only one (food) or two (food and clothing) of the five guarantees. Medical costs are reportedly particularly difficult for the system to cover, even in wealthier regions. 4.59 Other Income Maintenance Programs. Rural households ineligible for the five-guarantee program but needing assistance are identified through provincially determined poverty lines and the judgement of local cadre. On this basis, MCA estimates that the number of poor in need of relief increased from about 66 million people in 1978 to about 90 million in 1988.3L' These 30' One third of the counties in China have retirement homes, and eight provinces (Jilin, Heilongjiang, Shanghai, Shandong, Ningxia, Beijing, Tianjin, and Liaoning) have them in every township. 32/ The MCA rural poverty line is not known, but the increase in the number of poor during 1978-89 must correspond to a substantial increase in the poverty line in excess of the increase in subsistence expenditure over the same period. The MCA poverty figures are therefore not useful indicators of the trend in absolute poverty during the reform period. 112 people have been identified as suffering from persistent poverty, not simply natural disaster, but limited central government and local support has constrained relief to a modest share of the total. In 1978 about 30 million people in over 6 million households received payments totalling Y205 million in central government relief as well as 20 kg of grain and Y5.8 cash per capita from their collectives. LI In 1988, 35 million people in 8 million rural households shared Y156 million in central government funds and Y293 million in collective funds. Comparisons of average support levels are problematic, since 1978 payments were a combination of workpoints, grain and cash. In broad terms, it appears that the 1978 level of support was approximately Y18 per capita and the 1988 level was Y13 (current yuan) per capita. Rl MCA staff recognize that the 1988 payments failed to meet the needs of the recipients and that many needy households received no support at all. 4.60 In response to village difficulty in funding subsistence needs of its members, the MCA has promoted the development of alternative, voluntary, forms of mutual assistance. The two most common are the grain bank associations and rotating fund committees. The former, said to be more common in poorer areas, use contributions to a collective grain bank to loan grain to members in difficult times. The latter does much the same, but uses cash contributions and loans rather than grain. Said to be peasant organized and managed, with technical assistance from the MCA, there are now more than 130,000 grain bank associations and rotating fund committees nationwide, with over 2 million members. In Yunnan there are 2,741 associations with an average of 180 households as members. The average family contribution to the grain banks is 15 kg of grain and about Y15 in cash. The 298 rotating fund committees in Yunnan average about Y285,000 in assets and are said to operate only in wealthier areas of the province. -Financing Rural Relief: An Expanded Role For the Ministry of Civil Affairs 4.61 Available official figures for the number of rural inhabitants receiving poverty relief assistance and total financial support for disaster and poverty relief programs during 1978-89 are summarized in Table 4.1. The number of rural recipients of five-guarantee and other income maintenance assistance appears to have peaked in 1980 at nearly 50 million, and ranged between 37 to 43 million people in all other years. Beneficiaries of the five-guarantee program have on average accounted for about one-tenth of the total number of rural poverty relief recipients. 3V These are incomplete returns that do not include Shandong, Yunnan, Ningxia and Tibet. 33/ Using YO.8 as the value of a full day of work points in the average team and the 162 million work days credited to the needy. 113 Table 4.1: Rural Disaster and Poverty Relief -- Numbers Assisted and Fundina Disaster and Poverty Relief Funding Number Assisted \a --------------------------------------------------------------------- ----------------------- Five-Guarantee Other Poverty Relief \b Five- Other ------------------- ----------------------------- Guarantee \k Total Disaster Central CentraL ----------------------- ReLief CoLLective Gov't Collective Gov't SubtotaL TotaL Year --- million people --- ------------------------- million yuan ---------------------------- 1978 na 30.2 na 419.4 na 25.9 330.0 \c 204.8 534.8 na 1980 2.9 46.4 49.3 448.0 154.5 20.3 164.7 225.0 389.7 1012.5 1983 3.0 35.0 38.0 640.4 297.0 31.5 155.6 236.9 392.5 1361.4 1984 3.0 38.0 41.0 690.6 347.4 36.3 170.2 259.1 429.3 1503.6 1985 3.0 38.0 41.0 958.5 443.1 37.2 146.5 177.4 323.9 1762.7 1986 2.9 39.9 42.8 1070.3 na 42.4 172.8 171.0 343.8 na 1987 2.9 37.0- 39.9 987.2 461.6 44.3 167.9 158.3 326.2 1819.3 1988 2.5 34.5 37.0 1040.5 645.7 49.7 292.7 156.0 448.7 2184.6 1989 2.9 34.5 37.3 1227.1 602.5 53.2 167.2 163.8 331.0 2213.8 Source: SSB (1990b). \a Excludes beneficiaries of disaster reLief. \b Other rural income maintenance programs. \q Staff estimate. 4.62 Funds for disaster relief, which are managed by central MCA and distributed as needed to the provinces (except for six provinces that use a contract system for disaster relief), nearly tripled in nominal terms from Y420 million in 1978 to Y1.23 billion in 1989. In real terms, expenditure on disaster relief increased by about 50% over the period. Furthermore, rural household income levels and their private grain stocks both increased substantially through the first half of the 1980s, providing farm families a greater cushion against crop losses and leaving many ineligible for assistance or eligible only for reduced levels of assistance during the second half of the decade. 4.63 Funding for the five-guarantee and other rural poverty relief programs remained roughly constant in real terms during the 1980s, having increased in nominal terms from Y565 million in 1980 to Y987 million in 1989. As shown in Table 4.1, however, the increase in collective support for the five-guarantee program (from Y155 million in 1980 to Y603 million in 1989) accounts for nearly all the increase in nominal rural poverty relief funding during the 1980s. As a result, the five-guarantee program increased its share of total rural poverty relief funding from less than one-third in 1980 to two- thirds in 1989. Since the bulk of five-guarantee program funding is directed to the elderly in higher income rural areas (para 4.57), this trend suggests that assistance to extremely poor children and adults through the five- guarantee program declined in real terms during the 1980s. This decline was 114 not offset by other programs, furthermore, since real funding for all other rural relief programs declined by more than 50% during the 1980s. An examination of 1989 rural poverty relief expenditure by province indicates that provincial poverty relief expenditure is positively correlated with provincial incidence of poverty. 34 4.64 Available evidence suggests that the MCA system functions effectively to prevent, or at least tightly control, outbreaks of severe malnutrition among the rural population. Although exact figures are considered highly confidential, the MCA system arranges for the distribution of an estimated 2 million tons of free relief grain annually -- enough to satisfy the minimal subsistence requirements of 30 to 40 million absolute poor. L5 This grain is provided in small quantities on an individual basis to households lacking sufficient grain (or the means to purchase grain) to last to the next harvest. It might also be argued that the decline in real support for rural poverty relief programs (other than the five-guarantee program) has been consistent with the tremendous reduction in the number of rural absolute poor during the 1980s. However, the number of recipients of such assistance has remained roughly constant during the 1980s and is now still significantly less than the total number in need (para 4.61). An expansion of central government funding for rural relief programs would allow for an increase in the number of recipients and a modest improvement in subsistence standards. As shown in Table 4.1, such funding is quite modest at present (Y164 million in 1989) and has declined more than 50% in real terms during the 1980s. The MCA system is also well suited to playing a role in meeting the health and educational needs of the poorest. Unlike any other government program, MCA has staff in every township whose concern is to assess individual and family well-being. 34 More specifically, provincial share of total rural poverty relief expenditure (SSB, 1990b) is positively correlated with provincial share of total poverty (Table 2.2). (The analysis excludes collective support for the five-guarantee program, for which no provincial breakdown is available.) This positive relationship is not perfect, however. The higher income central and coastal provinces of Shandong, Jiangsu, Zhejiang, Jiangxi, and Guangdong do expend a somewhat disproportionately large share of the total while the poorer northern, northwestern and southwestern provinces of Henan, Shaanxi, Gansu, Guangxi, Guizhou, and Yunnan expend a somewhat disproportionately small share. & The relief grain is provided to the poor through the Ministry of Commerce's Grain Bureaux system. 115 5. POVERTY ALLEVIATION POLICIES. PROGRAMS. AND INSTITUTIONS 5.01 Poverty Alleviation Programs of the Earlv-1980s. With the initiation of the rural reform program in the late-1970s, the government's concerns about chronic poverty and widening regional inequalities were translated into new development assistance policies and programs focussed on resource-poor areas. It was recognized, in particular, that certain groups, by reason of geography primarily, were not likely to share in the increased income accruing to household-managed farming. Severe drought and resultant hardships in Gansu and Ningxia in 1981-82 dramatized the plight of millions in resource-poor parts of China held back by conditions beyond their control. Studies of mountainous regions in South China carried out by the State Council's Research Center for Rural Development (RCRD) in 1983-84 provided additional evidence that the rural reform program alone would not be sufficient to overcome continued low productivity in poorly endowed areas. 5.02 As early as 1980, the Ministry of Finance (MOF) created a special fund, the Development Fund for Underdeveloped Areas, to provide grant monies to resource-poor provinces in northwest and southwest China. In response to the downturn in agriculture in Gansu and Ningxia, the State Council in 1982 authorized establishment of the Sanxi Agricultural Construction Fund, another grant fund administered by MOF, to provide development aid to 39 poor counties in Gansu and 8 in Ningxia. On the recommendation of the RCRD, the State Planning Commission in 1984 set up a Food-for-Work Program designed to use surplus grain and cotton stocks to pay for infrastructure construction in financially depressed areas. In 1985, the MCA was authorized to allocate 30% of its welfare funds to finance start-up costs of rural enterprises employing low income applicants. These several programs differed from traditional disaster relief efforts and standard income maintenance schemes to assist the persistently poor (paras 4.54 to 4.60); they were envisioned as long-term interventions geared to strengthening infrastructure and increasing efficiency of resource use as the base for sustained improvements in agriculture. A. Poverty Alleviation During the 7th Five Year Plan (1986-90) -Strateoy 5.03 Income-generating Poverty Alleviation Programs. China's 7th Five Year Plan (1986-90) sought to expand prospects for the newly prosperous while providing hope for those not fully able to participate in overall economic growth. Its strategy for rural economic growth was grounded in the 1985 directive which replaced the state's monopoly on purchase and supply with a dual state and free market system of pricing and commodity exchange. Expanding marketing networks, freeing up prices, encouraging diversification and commercialization of agriculture were emphasized as the means of extending the strong agricultural growth of 1978-84. At the same time, the Plan 116 recognized that general economic growth was a necessary but not sufficient condition to overcoming poverty in some areas of the country, and that to achieve stated growth with equity goals, specific anti-poverty measures must be incorporated into regional economic planning. The new, explicit poverty reduction approach was based on policy recommendations prepared by RCRD in 1984. It emphasized (i) development assistance -- developing income- generating activities based on local resources and local organizations -- as the key to sustainable poverty reduction, (ii) continued provision of direct relief aid and social services, but with the burden of financing shifting more from central to local levels, (iii) establishing preferential tax, pricing and other measures to stimulate poor area development, and (iv) targeting assistance funds to poor areas based on a standard set of income criteria for determining need. -Institutions 5.04 Dozens of line ministries and agencies play a role in poverty alleviation in China. Some, like MCA, the Ministry of Commerce's Grain Bureau system, and SPC, have a tradition of involvement in relief and income maintenance activities. Others, such as ABC, PBC, and MOF, initiated grant and special subsidized credit programs in the early-1980s. Many government agencies support development efforts which touch poor and non-poor alike -- the Ministry of Water Resources, MOA, MOPH, and SEdC, for example. Other government agencies, including SSB with its household surveys and also CASS, have devoted resources to charting the progress of the poor. l' 5.05 Leadinq Group for the Economic Development of Poor Areas (LGEDPA). To better orchestrate this increasingly complex poverty reduction program and in order to institutionalize the new anti-poverty initiatives of the 7th Five Year Plan, the State Council in 1986 authorized establishment of a task force and executive agency structure to be replicated at all administrative levels. At the central level is an inter-ministerial task force called the Leading Group for the Economic Development of Poor Areas (LGEDPA). The new anti- poverty executive agency, the Poor Area Development Office (PADO), reports directly to the State Council via LGEDPA, though its logistical support and most of its staff are from MOA. Central PADO is a small (25 member) organization consisting of five divisions: Planning (5 staff), Research (6 staff), Policy (4 staff), Sanxi Area Affairs (4 staff), and Support Services (6 staff). From its own budget, the State Council funds another four anti- poverty units which operate within the Leading Group orbit: the China Development Foundation for Poor Areas, the Cadre Training Center, the Training Center Office, and the Economic Development Service Center. The latter two, while operationally independent, are required to provide routine reports to PADO as well as to the Leading Group. Provinces, prefectures, and counties have all established Leading Groups and PADOs after the central model, and many townships have a 'designated person" to handle anti-poverty work. I/ Hinton (1991) lists (and provides either detailed evaluations or summary facts for) 39 Chinese institutions involved in poverty reduction work. 117 -Defining ithe Poor 5.06 Nationally Designated Poor Counties. Employing the methodology previously adopted by the RCRD, PADO in 1986 used MOA county-level rural income data to develop a roster of 328 poor counties eligible for development assistance. v (The PADO roster was subsequently increased to 331 counties with the addition of three nationally designated poor counties in Hainan.) A county was identified as poor if its 1985 average per capita income fell below assumed poverty lines of Y300, Y200, or Y150, depending on other locational or political factors. Average per capita grain production of less than 200 kg was also adopted as a second key indicator of poverty. While a large proportion of individuals within those counties must have been poor, this approach failed to capture the poor in otherwise well-off counties. 3 Based solely on 1985 income data, furthermore, the method did not adequately correct for single year aberrations in county rural income levels. It has already been observed, finally, that the regionalization of poverty evidenced by the MOA county-level rural income data (which are the basis of PADO's list of poor counties) is significantly different from that indicated by SSB's provincial rural income distribution data (paras 2.28 to 2.30). £ 5.07 The PADO review added about 100 counties to RCRD's original list of 225 nationally designated poor counties, and this list remained unchanged throughout the 7th Plan period. These counties, representing a total of 22 provinces and 30% of the total number of counties in China, were grouped into the following three categories: (a) 273 counties in 18 mountainous regions where rural net per capita incomes in 1985 were less than Y150 (all included on the list), less than Y200 (if old revolutionary base areas or minority areas), or less than Y300 (special cases). These counties are eligible for subsidized credit from ABC's Poverty Reduction Fund and other funds and grants; 2/ Based on three-year (1981-83) average township level rural per capita income data, RCRD identified 50 million absolute poor residing in about 7,000 townships, situated in 225 counties in 11 mountainous regions. While a preferred means of monitoring poverty, the RCRD township level methodology was not an operationally viable mechanism for targeting China's poverty alleviation programs. In 1986 LGEDPA switched to the county level for the purposes of poverty alleviation programs and monitoring. 3/ The PADO figure of 102 million poor in 1985 assumes that the poor comprise 60% of the rural population of counties with average per capita income of less than Y150, 50% of the rural population of counties with average per capita income of Y150 to Y200, and 15% of the rural population of counties with average per capita income of Y200 to Y300 (or about 23, 43 and 37 million poor respectively of a total rural population of 39, 85 and 250 million). 4/ PADO reserved, and in some instances exercised, the right to reject the MOA county-level rural income data when those data were believed to be inaccurate. 118 (b) 27 counties in Inner Mongolia, Gansu, Hebei, Sichuan, Qinghai and Xinjiang with rural net per capita incomes of less than Y300 (if pastoral) and less than Y200 (semi-pastoral). These counties are eligible for subsidized credit from ABC's Pastoral Areas Fund; and, (c) 28 counties in central Gansu and southern Ningxia which have low incomes (not specified), low precipitation (less than 400 mm/year), and a high degree of environmental degradation. Such counties receive special grants through the Sanxi Development Fund. The PADO determination of poor counties, and the subsequent allocation of poverty alleviation funding, was necessarily political. Counties with favorable political credentials and strong supporters -- most notably the old revolutionary base areas -- were able to force their inclusion to the roster of poor counties despite having per capita income levels twice the level deemed to represent subsistence. The inclusion of politically favored counties forced the exclusion of many counties with per capita income levels well below the poverty line. Yunnan, for example, was forced to adopt a special cutoff of Y120, rather than the nationally mandated level of Y150, in order to reduce its number of poor counties. The list does not include any counties of Tibet since (i) no county had reported average per capita rural income levels below the poverty line and (ii), a political special case, the Autonomous Region receives separate earmarked development funds. 5.08 Provincially Designated Poor Counties. In addition to the national roster, there are 368 provincially designated poor counties, eligible for provincial funding, which are selected on the basis of poverty lines determined by the province. Province-specific poverty lines used during the 7th Plan period ranged from Y150 in Yunnan to Y400 in Jiangsu. In addition to drawing up a roster of poor counties, most provinces have identified townships in otherwise well-off counties as eligible for special assistance. As evidenced by the wide difference in qualifying income levels (Y150 to Y400), the provincial rosters include a number of counties which are relatively poor by provincial standards but which would not qualify as absolutely poor by the national minimum standard. That provinces created their own special programs testifies to a concern with relative poverty and, in many instances, to provincial budgetary strength. On the other hand, the provincial roster does in some cases compensate for the incorrect exclusion of counties with average per capita income levels below the national minimum standard. The Yunnan provincial roster, for example, includes all the counties which should have received central government support for absolute poverty alleviation (that is, 15 counties with average per capita income of between Y120 to Y150). 5.09 Obiectivity and Eouity. A provincial breakdown of the PADO listing of 699 nationally and provincially designated poor counties is shown in Table 5.1. Roughly 40% of all the counties of the northwestern and southwestern provinces are included in the PADO listing, and together these 393 counties comprise more than half the total number of poor counties. Jiangxi (not part of western China), Guangxi and Shaanxi have the greatest proportion of their counties on the poor county list -- 67%, 58% and 57% respectively. The number Table 5.1: PADO (1985) and MOA (1989) Poor Counties and SSB (1989) Poverty Incidence by Province PADO (1985) SSS (1989) Central Provincial Total MOA (1989) PADO/MOA PADO/MOA Rural Poverty Number --- -------- ------------- ------------- ------------- OverLaps Difference incidence of Share Share Share Share -------- ---------- ------------- Counties Number (%)\a Number (%)\a Number (%)\a Number (%)\a Number Number (%)\§ National 2104 331 16% 368 17% 699 33% 187 9% 144 556 11.4% North Beijing 8 0.2% Tianj'in 5 0.4% Hebei 139 14 10% 35 25% 49 35% 11 8% 9 40 13.0% Henan 117 15 13% 9 8% 24 21% 4 3% 2 22 16.5% Shandong 99 9 9% 5 5% 14 14% 14 6.8% Northeast Liaoning 44 3 7% 8 18% 11 25% 4 9% 2 9 8.0% JiLin 41 11 27% 11 27% 11 12.2% Heilongjiang 69 8 12% 8 12% 2 3% 2 6 18.3% Northwest Inner.MongoLia 84 16 19% 24 29% 40 48% 13 15% 12 25 23.5% Shanxi 100 14 14% 21 21% 35 35% 9 9% 4 31 17.4% Shaanxi 93 34 37% 12 13% 46 49% 12 13% 12 34 20.3% Ningxia 18 8 44% 8 44% 7 39% 7 1 18.9% Gansu 75 31 41% 12 16% 43 57% 4 5% 3 40 34.2% Qinqlhai 39 10 26% 10 26% 20 51% 19 23.7% - Xinjiang 85 17 20% 13 15% 30 35% 1 1% 1 26 18.7% Yangtze River Shanghai 9 0.0X JIan9su 64 3.4% Zhejiang 67 3 4% 3 4% 3 2.0% Anhui 72 9 13% 8 11% 17 24% 17 7.7% Jiangxi 84 17 20% 39 46% 56 67% 54 5.0% Hubei 71 13 18% 24 34% 37 52% 6 8% 6 31 6.0% Hunan 95 8 8% 20 21% 28 29% 9 9% 9 29 6.2% Sotuth Fujian 64 14 22% 2 3% 16 25% 16 1.8% Guangdong 77 4 5% 20 26% 24 31% 30 0.9% Hainan 17 3 18% 5 29% 8 47% 8 3.3% Southwest Guangxi 83 23 28% 25 30% 48 58% 11 13% 10 38 15.4% Sichuan 181 21 12% 30 17% 51 28% 17 9% 13 33 11.2% Guizhou 81 19 23% 12 15% 31 38% 27 31% 20 11 17.8% Yurnan 123 26 21% 15 12% 41 33% 49 40% 32 9 19.0% .--------... Sources: LGEDPA (1989), MOA (1991) and TabLes 2.2 and 2.3 of this report. \a Share of nationaL or provincial total number of counties. 15 Share of national or provinciaL total number of poor rural households. 120 of poor counties in the Yangtze River and south China provinces, 141 and 47 respectively, is also considerable, and together they account for about 27% of the total number of poor counties. Virtually all of the counties on the list are typified by a combination of geographical, ethnic, and political correlates including mountainous terrain, limited natural resources for agricultural production, a large share of minority peoples (40% of the 331 national poor counties are special minority autonomous counties), and status as a revolutionary base. 5.10 PADO's list of nationally and provincially designated poor counties is compared with MOA's 1989 list of 187 "lower-income" counties (see Table 2.3) and SSB 1989 poverty incidence in Table 5.1. This comparison shows that the PADO listing: -includes most, but not all, counties with a large number of absolute poor; and, -includes a number of relatively, but not absolutely, poor counties. A cross check of county names ("overlap" in Table 5.1) shows that most (nearly 80%) of the MOA lower-income counties are included as either national or provincial poor counties. However, 40 counties reported by MOA to have had average rural per capita income levels of less than Y250 in 1989 are not included in the PADO list. Two-thirds of these 40 counties are in the four southwestern provinces, and nearly half in Yunnan alone. The PADO list of 699 poor counties includes 556 counties not included in the MOA list. Most of the 556 counties are situated in provinces with a corresponding number of counties with MOA-reported average rural income of less than Y400, suggesting a reasonably close correspondence between the 1985 PADO list and 1989 MOA rural income figures. E However, the number of poor counties in Jiangxi, Fujian, Guangdong and Hainan on the PADO list (103 counties) greatly exceeds the number of counties with MOA-reported average rural income of less than Y400 (38 counties) in these provinces. The difference presumably reflects central government support for old revolutionary bases, particularly in Jiangxi and Fujian, and provincial efforts to address relative poverty in Guangdong and Hainan. A number of poor counties have prospered since 1985 and, by 1990, 36 of the PADO poor counties had achieved average rural per capita income levels exceeding national average rural income of Y602 per capita. 5.11 The PADO list also evidences a significantly different regionalization of poverty than does the SSB 1989 provincial rural income distribution figures. Normalizing for the difference in the proportion of poor counties (33%) and the SSB incidence of poverty (11%), in particular, the PADO list suggests a surprisingly high incidence of poverty in Jiangxi, Hubei, Hunan, Fujian, Guangdong and Hainan. In sum, the PADO list correctly includes most of China's poorest counties, but also includes many counties with average 5/ A cross check of county names is not possible for counties with MOA- reported average rural income of Y250 to Y400. 121 incomes well above the subsistence level. It also ignores the millions of individual poor households not residing in otherwise poor counties. The political realities which made the PADO list an effective basis for initiating a national anti-poverty program should not now restrict broader efforts to better monitor and ameliorate household poverty wherever found. B. Development Assistance Programs -Domestic Programs 5.12 PADO-Administered Income-generating Proarams. Central PADO administers a large program of development assistance (grants, credit and standard loans) provided by MOF and through the national banking system to the 331 national poor counties. Central PADO's responsibilities are to determine overall program guidelines and to supervise lending and auditing procedures. PADO also takes an active role in determining provincial allocation levels for MOF's grant assistance. The banks, with reference to PADO indicative guidelines and relying primarily on their provincial branches, determine annual allocations for poverty alleviation as part of the normal credit planning process. Proposed funding levels are worked out with the provincial planning ccmmissions in accordance with project lending programs submitted by the poor counties, and the resulting draft credit plans are then sent to the bank's central office for approval. Project-based lending is a relatively recent approach; before 1985, poverty alleviation funds were distributed on an "average allocation basis", a formula based on an actual headcount of poor at various income levels. However, although allocation of poverty alleviation funding by province is not known, available information suggests that it still roughly accords with the SSB 1989 incidence of poverty -- Yunnan, for example, with about 7% of China's poor, receives about 6% of ABC's annual anti-poverty funding. 5.13 As shown below, central government assistance to poor areas consists of- eight programs -- several of which pre-date establishment of the PADO system -- and amounted to Y4.05 billion annually by 1988: Central Government Poor Area Assistance Programs Funding Rate (%) Program Y billion Start Channel (in 1989) 1. Development Grant 0.8 1980- MOF grant 2. Sanxi Development Fund 0.2 1983- MOF grant 3. Revolutionary Bases, Minority and Remote Areas 1.0 1984- PBC 3.9% 4. Economic Development Fund 0.3 1984- ABC standard rate 5. Poverty Reduction Fund 1.0 1986- ABC 2.4% 6. Pastoral Areas Fund 0.05 1988- ABC 2.4% 7. County-run Enterprises 0.4 1988- PBC 3.9% 8. County-run Enterprises 0.3 1988- ICBC 20% discount 122 These programs provide subsidized credit (about 75%) or grant (25%) funds to county, township and village, and below-village enterprises in the 328 national poor counties. (The 370 poor counties on the provincial list are also eligible for some funding from MOF's interest-free fund for underdeveloped areas.) ABC plays a major role in the poverty alleviation program, providing a third of the Y4 billion total from its own sources and managing two-thirds of the Y1.4 billion in credit provided by the People's Bank of China (PBC). Most of the Y4 billion is used to finance development of "productive" infrastructure -- such as irrigation works through the Sanxi Fund, but excluding "nonproductive" infrastructure such as household water supply and rural roads -- and the development of county enterprises engaged in plantation and livestock production, agro-processing and mining. Remaining funds are extended as loans to township and village enterprises and to households directly, primarily through ABC. Originally, individual household loans were common. However, these proved administratively costly and difficult, leading ABC to direct its lending instead to "economic entities" (most commonly, township economic committees and county, township and village enterprises), which in turn provide poor households with (i) credit in the form of fertilizers, planting materials and other inputs, (ii) a guarantee to purchase raw materials, or (iii) the opportunity for wage work. Projects chosen are expected to promote one or more of the State Council's "four ones" goals for poor areas: (i) one mu of high-and-stable-yield arable land per household member, (ii) the sale of one animal per year, (iii) one forestry or fruit tree venture per household, and (iv) off-farm employment of one laborer per household in village or town enterprise. 5.14 Provincially Administered Programs. Twenty three provinces have their own prtgrams of poverty reduction assistance covering a total of 370 counties. Similar to the central government's programs, these provincial programs provide a combination of subsidized credit and grants to provincially designated poor counties and townships. A complex array of agencies is involved in program and project development for these counties, beginning with the provincial planning commissions which set the current year allocations by county, and including bureaus of finance, agriculture, rural industry, forestry, ABC, and, particularly in the northwest and southwest, the minority commissions. Total provincial government support to the 370 provincially designated poor counties is believed to roughly equal central government support to the 328 nationally designated poor counties. 5.15 Local Government Operations. County planning and economic committees, PADOs and local ABC (or other bank) offices are the key players in the design and preparation of poverty alleviation projects. Project identification and review and approval of feasibility studies are generally handled jointly by the local PADO and the local bank office, often with the assistance of appropriate technical agencies. E If the proposed project I/ In the case of rural industry projects, PADO and ABC also require that feasibility studies be carried out by the borrowing enterprise itself. Reportedly, the final approved projects are selected from a large pool of (continued...) 123 amount is over Y100,000, the project is reviewed by the prefectural PADO; projects over Y500,000 are typically referred to the province. Most projects, particularly in crop and livestock development, are small and therefore handled by the county alone. Project appraisal involves an assessment of the resource base in the project area and the project's potential to increase local revenues, absorb surplus labor, and serve as a model easily replicable elsewhere. Project implementation and loan collection are the responsibility of the bank branch and relevant technical agencies. The county PADO handles monitoring and assists the participating bank and the State Audit Agency in preparation of annual audit reports. PADOs at the provincial and prefectural levels have general oversight functions, but are not involved in project operations except to review and approve projects exceeding specified funding ceilings. 5.16 PADO and ABC rely on village leaders to determine which households are eligible for project participation. In the case of low-interest loans to enterprises, ABC's regulations stipulate that 70% of enterprise employees must be members of poor households. For greater efficiency in funds supervision, loan agreements are made between a borrowing enterprise or a township economic committee and the ABC, not between household and ABC directly. The enterprise or township committee in turn signs contracts with participating households; at the household level loans are extended in the form of input supplies rather than cash. Project implementation is the job of one or several technical agencies, depending on the nature of the project. For example, a chestnut plantation in Guangnan County, Yunnan, financed through PADO and ABC, involves the county personnel, statistics, and forestry science bureaus as key implementing agencies. Loan collection is the responsibility of ABC which also participates in audits of project funds along with PADO and the planning and economic committee. The system of overlapping responsibilities and multiagency inputs is complex, but typical of Chinese operations. 5.17 Ministry and Agency Special Proiects. Responding to State Council directives, 27 state agencies have each established a special development assistance relationship with a single poor area on the PADO list. In most cases, the designated area is a prefecture or one of list's 18 mountainous regions which include several counties in two or three provinces (such as the Wuling Mountains area of Hubei, Hunan and Guizhou). Typically, the ministry or agency works on a matching fund basis with the local PADOs to develop projects which take advantage of its particular resources or expertise. A successful example of this sort of special relationship is the work of the State Science and Technology Commission (STC) in the Dabie Mountains, an area bordering Anhui, Hubei and Hunan provinces. Working with a small anti-poverty budget (Y7.2 million annually) over the past four years, central STC has channeled most funds to projects which train local farmers in technical and business skills and set up demonstration sites to promote improved crop varieties and new techniques in livestock and silkworm raising. Local STC 6/( ..continued) proposals. 124 offices coordinate their efforts with PADOs and attempt to draw upon the full array of local technical expertise available through the extension system, industries, water and power agencies and the like. STC has now extended its special projects development efforts to the Jinggangshan area of Jiangxi and Guizhou's karst region. 5.18 MCA Enterprise Develo=ment Assistance. Beginning in 1985, MCA received MOF authorization to redirect 30% of MCA's welfare fund to investments in RE development for the poor. In the six year period 1985-1990, over Y1.9 billion has been allocated to support establishment and expansion of some 40,000 county, township, village, and privately-managed enterprises. The program is targeted to poor families, not poor areas, and all but one county in China have at least one such enterprise. According to MCA guidelines, enterprise employees are to be drawn from the ranks of needy families. In 1990, over 1.1 million people were so employed at monthly wages ranging from Y70 to Y150. Despite the apparent success of the enterprise development program, which is discussed in Box 4.1, a rapid fall in the real value of welfare funds led MOF to prohibit further diversion of relief funds to the program since 1990. 5.19 Food-for-Work Program. China's Food-for-Work Program, initiated in the winter of 1984, was designed to use surplus commodity stocks as in-kind payment for construction work on water supply system and road building projects. Managed by the Western Regional Office of the State Planning Commission, the Program is funded on a provincial government matching grant basis. The first cycle of the Program, during 1984-86, aimed at building roads and water conservancy projects. The central government committed Y2.7 billion of grain, cotton, and cotton cloth for the three year period. The central government committed the lesser amount of Y600 million to the second cycle, during 1987-89. The second cycle included mostly road building and drinking water projects. Project workers were issued coupons redeemed at designated stores in exchange for low and medium quality industrial goods. The central government committed Y1.5 billion worth of goods for the third cycle during 1990-92. The third cycle includes water conservancy and other agricultural capital construction, road building and drinking water projects. Including matching provincial government grants, total grant commitment through 1992 amounts to about Y10 billion. 5.20 Program funds are distributed in accordance with population figures for the nationally and provincially designated poor counties. On this basis, Sichuan received Y380 million in the first cycle and Heilongjiang and Jilin less than Y30 million each. Sichuan and Guizhou have been major beneficiaries of the Program, receiving to date about 10% and 7% respectively of the funds. Jiangsu, Shanghai, Tianjin, and Beijing were excluded from the Program from the beginning and Guangdong graduated after the first cycle. Wages, paid in coupons redeemable in kind, are set by the province, generally in the range of Y1.5 to Y3.0 per day. These wage rates are below those for alternative employment, and laborers hired for Food-for-Work projects are often from low- income disadvantaged groups. However, Food-for-Work project contracts are frequently arranged on a piece-work basis in which the actual wage rate will often exceed Y3 per day. In such cases, availability of employment 125 opportunities is not restricted solely to the poor. Through 1989, Food-for- Work projects have resulted in the construction of over 110,000 km of roads and water systems benefitting more than 13 million people. If carefully targeted, managed, and monitored, Food-for-Work has much to offer as a way to provide additional employment and build rural infrastructure in China's poor areas. 5.21 Preferential Policies. Since 1985, the central and provincial governments have adopted a complement of preferential policies to augment poor area economic development. The central government has authorized major tax exemptions for the 328 national poor counties in the agriculture, energy, and transport sectors. It has also lowered the bank reserve requirement in those counties and increased provision of fertilizer, steel, and vehicles at fixed state prices. Measures instituted by provincial governments include reduction or exemption from mandatory grain quotas, special tax breaks for rural enterprises, price cuts on essential medicines, and reduction of collateral requirements for low income loan applicants. Poor counties themselves frequently seek exceptions to standard regulations. Wenshan County (Yunnan), for example, requested a budgetary variance to allow provincial anti-poverty funds to be used to cover the county's required contribution to construction of water supply systems. In the education sector, provinces offer an array of fee exemptions and special incentives to poor counties to promote increased enrollments. In many provinces, minority students, especially females, are allowed fee exemptions and special consideration in such matters as age-in- grade (new school entrants may be as old as 15) and conduct (girls may bring younger siblings to school) as a means of encouraging school attendance. -International Assistance 5.22 Foreign Assistance for Poverty Alleviation. Over the past decade, a host of outside donors -- multinational institutions, bilateral aid agencies, and NGOs -- have provided grant and loan assistance to support China's long-term development process. Total concessional aid amounted to US$2 billion in 1988 and $2.2 billion in 1989, about 80% in the form of loans. The two largest donors in 1989 were Japan, which contributed 38% of all assistance, and IDA which accounted for 27%. Total World Bank lending to China over the period 1981-91 amounted to US$10.6 billion, with IBRD's contribution $5.7 billion and $4.9 billion from IDA. Other agencies within the UN system provided another US$1.0 billion over the same period. Bilateral assistance totalled US$5.8 billion during 1985-90; Japan alone, as China's largest bilateral donor, contributed US$3.5 billion. Many NGOs have been active in China for a number of years. Though their numbers and volume of assistance have been increasing, NGO assistance still represents less than 1% of total assistance (UNDP). The enormous infusion of foreign assistance funds has been directed mainly to large-scale infrastructure and industrial development projects aimed to support broad-based economic growth. Direct intervention for poverty reduction has been only a secondary theme. Until the mid-1980s, it was generally held that China had done very well in meeting its 126 basic needs goals and that sustained overall economic growth was the best means of alleviating poverty. - 5.23 IFAD and WFP. Two exceptions to the growth-oriented approach have been the IFAD and WFP programs which are explicitly directed to poverty concerns. Since its lending program to China began in 1984, IFAD has provided US$136 million in support of six projects specifically targeted at poor households. IFAD's emphasis has been on (i) improving the production base of the poor through reduction of soil salinity, reclamation of wasteland, and rehabilitation of pastureland, and (ii) providing a package of appropriate technologies -- in fodder production, fish farming and the like -- easily replicable in other parts of China. Projects have focussed on small, well- defined geographical areas -- counties or prefectures within a single province rather than multi-province efforts -- where poor households are identified by income levels. Overall objectives are to increase employment opportunities and incomes of low income households, increase food production, and improve nutritional and general living standards. 5.24 Like IFAD, WFP targets its assistance -- all in the form of grant aid -- specifically to low income households in resource-poor counties. WFP's focus is less on the "pockets" of poverty found in all parts of China than on provinces with the lowest rural income levels, including Tibet, Qinghai, Gansu, Shaanxi, Guizhou, Yunnan and Guangxi. WFP projects tend to have a single sub-sectoral focus, such as irrigation rehabilitation in Tibet, erosion control in Guizhou, watershed development in Shaanxi, and land improvement in the Wuling Mountains in Guizhou. Between 1985 and 1990, WFP approved 50 development projects and two emergency operations totalling US$585 million. Assistance takes the form of direct food aid, provided as a grant to the government, which is used primarily as an incentive to recruit local farmers for labor-intensive work to repair irrigation facilities, build roads, construct terraces, plant protective forests and the like. Government matching funds cover materials, equipment, training, wage subsidies and a variety of management costs. 5.25 UNDP. In its Third Country Programme, with resources of $189 million over the period 1991-95, UNDP targets poverty alleviation as one of its five major themes. UNDP mainly works with research, training and other institutions and the municipal governments in poor and remote areas. The largest project in the Programme has a strong poverty reduction focus. Covering the semi-arid and arid areas of Northwest China, the project provides research and training funds to institutions working in five areas critical to improving agricultural productivity -- desertification control, rainfed farming, mechanization, education, and diversification. Other poverty reduction focussed assistance includes a project to train officials and managers in poor and remote areas in project preparation (para 5.30) and sponsorship of an investment promotion forum for five poor provinces in 7/ UNDP (1992) provides a comprehensive overview of international assistance for China's long-term development process. Hinton (1991) also lists and evaluates outside donors supporting poverty reduction work. 127 Northwest China. UNDP has also established the UN Interagency Working Group on Poverty Alleviation in China to discuss current issues, exchange information and consider the possibility of collaboration on joint projects. 5.26 UNICEF. The overall framework for the current China-UNICEF cooperative programme is the government's National Programme of Action (NPA) which itself is based on the World Summit for Children Declaration. Targeted to disadvantaged areas, the cooperative programme concentrates UNICEF's resources in order to have the greatest impact on improving education and reducing mortality and malnutrition. Sectoral assistance is directed to provinces and counties where education and health status indicators suggest the need is greatest. The maternal and child health and family planning project, for example, is targeted to 300 counties with high IMR, high birth rates and low per capita income levels. The project, which is funded in part by UNFPA and technically supported by WHO, supports training and development of services at the township and village levels and leadership capacity at the county level. Multi-sectoral projects are also directed to disadvantaged provinces and counties. With the objective of improving the well-being of children and women, 20 such counties have received support in at least three of the following areas: health, education, water and sanitation, income generation opportunities for women, and vocational and literacy training for women. C. 8th Five Year Plan (1991-95): Reaching the Chronic Absolute Poor -Strateqv and Funding 5.27 The Government's commitment to sustained efforts to eradicate poverty figures prominently in the 8th Five Year Plan (1991-95). The Plan reconfirms and extends LGEDPA's central role as the coordinating body responsible for poverty measurement and research, project planning and monitoring, and management not only of domestic funding but of bilateral and multilateral assistance for poverty reduction. As the basic premise for anti- poverty interventions, the Plan calls for maintaining steady economic growth through policies supporting gradual structural reforms, expansion of RE, technology imports and international trade more generally, and diversification of agriculture. During the Plan period, the Government projects an annual growth rate of agriculture of 3.5% and of industry, 6.8%. Within this framework of overall growth, the poverty reduction strategy has three essential thrusts: (a) a continued emphasis on developing the productive capacity of poor area agriculture through subsidized loans; (b) better targeting of assistance to reach the poorest of the poor; and, (c) improved management and coordination of anti-poverty activities. 128 5.28 Rural Development. The 8th Five Year Plan continues to stress the importance of the "self-help" approach taken during the previous plan period: namely, the creation of new income-earning opportunities in poor areas based on best use of indigenous natural and human resources. Most of the poor are farmers, and the anti-poverty agenda outlined in the Plan focusses on agriculture. First, the Plan supports continued efforts to diversify agriculture, particularly to initiate new commercial ventures in livestock, forestry, fruit trees and plantation crops. At the same time, the Government continues to voice concern about maintaining stable grain production in the poor areas, and the Plan explicitly states that where appropriate conditions exist, poor areas should strive to achieve grain self-sufficiency. Second, the Plan emphasizes the need to improve input supply, marketing, and technical advisory services to help poor farmers expand productivity and find new avenues for sale of goods. Third, the Plan calls for development of rural infrastructure -- roads, power, water supply -- as a means to raise productivity and improve overall living standards in disadvantaged areas. Fourth, the Plan endorses developing off-farm employment opportunities to absorb surplus labor, and envisions movement of labor to new agro-industries or mining enterprises in small or medium towns within the same county. Finally, the Plan calls for integration of these production oriented programs with improved education, health and family planning services in the poor areas. 5.29 Assistance for poor area development will continue to be provided in the form of subsidized credit, grants, and Food-for-Work. The goal, as outlined in the Plan, is to first move from the subsistence level ("wenbao") to a relatively comfortable standard of living ("xiaokang"). An additional 187 counties have been designated "national poor counties" in the 8th Five Year Plan. All of the newly designated counties had average per capita rural incomes levels of less than Y300 in 1989, and most were previously provincially designated poor counties. None of the original 698 counties have been graduated from the roster of poor counties, though such is clearly warranted (para 5.10). To be used exclusively for programs in the newly designated poor counties, annual PADO-administered grant and credit assistance will increase by Y500 million. The provincial governments, however, are to supply all the incremental PADO funds. Annual central government funding for the third cycle of the Food-for-Work Program is slated to increase from YSOO million to Y1 billion (para 5.19). The incremental funds, which are to be provided as grain and industrial goods, will be used to support terracing and other measures to reduce soil erosion. -Monitoring and Targeting 5.30 Monitoring. The Government recognizes the importance of strengthening its poverty measurement methodology (paras 5.06 to 5.11) as a first step in improving its ability to assess and monitor poverty at the national and provincial levels. However, little action and no formal decisions have as yet been taken. Key priorities are to (i) establish an independent and objective monitoring system and (ii) update the poverty line and establish corresponding province-specific lines. Fortunately, the SSB has both the necessary data and skills to undertake an update of national and 129 provincial poverty lines and initiate more objective monitoring of the incidence of poverty. This national and provincial level analysis should be supplemented with additional detailed county and village level studies undertaken by PADO research staff, SSB, and other research agencies. 5.31 Targeting Disadvantaged Groups. The extension and strengthening of assistance to the poorest of the poor residing in the worst physical environments is to be supported under the 8th Plan by individual programs tailored to the special needs of minorities, communities residing in remote and often high altitude areas, and other disadvantaged groups. The 8th Plan is explicit in its concern for the progress of China's minorities. It emphasizes the need to expand income-earning opportunities among this clearly disadvantaged group, most of whom live in inaccessible, mountainous parts of China's border provinces. The Plan calls for five major development initiatives: (a) strengthening crop, forestry, and livestock production. Specific mention is made of raising the grain self-sufficiency rate, improving pastureland and solving water supply problems in major livestock areas, establishing special production bases -- for example, sugar cane in Yunnan and Guangxi, grain in Inner Mongolia and Ningxia -- promoting tropical crops production in Yunnan, Guangxi and Hainan, and developing river valleys in eastern Tibet; (b) developing transport and communications services. Priority is given to construction of rail lines, including completion of the Nankun (Nanning to Kunming) railway now underway; (c) promoting mineral and mining development. This includes developing coal and natural gas fields in Inner Mongolia, Shaanxi, Ningxia, Gansu, Xinjiang, and Guizhou, oil in Xinjiang, and phosphate deposits in Yunnan and Guizhou; (d) improving special products processing and overseas and border trade in such products; and, (e) raising the levels of scientific knowledge, education, literacy, health, and physical fitness in minority areas. Special attention is given to achieving universal elementary level education, promoting adult and vocational education, and increasing medical supplies and clinics. This is an appropriate set of initiatives, targeted at a group clearly in need and combining an emphasis on income-generating activities with a concern for improving the access to and quality of social and basic services. Translating initiatives into effective programs and projects will require (i) sensitivity to the distinct social structures and special requirements of China's minority poor, (ii) adequate funding, (iii) careful financial analysis to determine project feasibility, and (iv) innovative thinking regarding financing mechanisms, technology application, and implementation and monitoring responsibilities. 130 -Institutional Issues 5.32 Central Operations. In addition to acting as the chief advocate of the absolute poor and of continued expenditure on poverty reduction activities, the central PADO coordinates poverty measurement and research, project planning and monitoring, and management of domestic funding and international assistance for poverty reduction. However, the central PADO's ability to complete these tasks is hampered by its "temporary" status and severe staffing constraints. At present, central PADO simply does not have the staff resources (para 5.05) to at once mount research studies of national import, determine funds allocation, and advise on project operations. The government should recognize that sustaining the effectiveness of poverty reduction efforts during the 1990s and beyond will require a permanent and much stronger institutional structure. In addition to formalizing central PADO's permanent status, the Government should arrange for a significant increase in central PADO's staff. Most importantly, PADO's capacity to undertake applied poverty related research needs to be strengthened through an increase in staff numbers and qualifications. This research work should be supported through the introduction of a modest information and statistical division. Consistent with its mandate to coordinate all poverty alleviation programs, central PADO should also establish a division specifically responsible for the assessment and improvement of education, health and other social services in the poor areas. This expansion of staff numbers and skills should be complemented by a better definition of the interaction among the various units which are part of the LGEDPA orbit. It might be advisable, for instance, to make the Economic Service Center and the Training Center, with its associated Office, subunits within PADO rather than leaving them as they are now as somewhat ambiguous independent agencies. Both have critical functions to play in anti-poverty operations. The Service Center, for example, which enjoys an annual budget of Y200 million and staff of 65, has the potential to become a major conduit of appropriate technology tried and tested in poverty alleviation projects outside China. 5.33 Prolect Conceptualization and Implementation. PADO expects to continue its focus on planning and monitoring with actual project design handled jointly with ABC offices at the local level and implementation assigned to line agencies. UNDP recently launched a US$2.4 million training program to strengthen PADO staff capabilities in all aspects of the project cycle. Since further decentralization of decision making regarding anti- poverty funds is expected during the Plan period, the training program will concentrate on upgrading local PADO personnel and county/township officials. The Ford Foundation has a program introducing PADO leaders to poverty alleviation efforts in other countries, with the goal of improving policy formulation. 131 REFERENCES Ahmad, Ehtisham and Yan Wang: "Inequality and Poverty in China: Institutional Change and Public Policy, 1978 to 1988" in The World Bank Economic Review 5, no. 2 (May, 1991): 231-257 Anhui Statistical Bureau (1990): Anhui Statistical Yearbook. 1990 (Beijing: SSB, 1990 and earlier years) Ashton, Basil et al: "Famine in China, 1958-61" in Population and Development Review 10 (June, 1984): 241-271 Banister, Judith: "China: Mortality and Health under the Economic Reforms" (mimeo, 1990) Banister, Judith: "Demographic Aspects of Poverty" (Working Paper to this report) Bi, Mingjian: "Economic Analysis of the Loess Plateau Land Rehabilitation Program" (Working Paper to this report) Blecher, Marc: "Income Distribution in Small Rural Chinese Communities" in China Ouarterlv 68 (December 1976): 797-816 Byrd, William and Alan Gelb: "Township, Village, and Private Industry in China's Economic Reform" [WPS 406] (Washington, D.C.: World Bank, 1990) Byrd, William and Lin Qingsong eds: China's Rural Industry: Structure. Development and Reform (Oxford: Oxford University Press, 1990) Chen Yongzong, et al: "The Temporal and Spatial Variation of Erosion and Sediment Yield on the Loess Plateau" in Lucian M. Brush, et al (ed) Taming the Yellow River: Silt and Floods (The Netherlands: Kluwer Academic Publishers 1989) 275-288 Chinese Academy of Preventative Medicine: Coale, Ansley: Rapid Population Change in China. 1952-1982, (Washington, D.C.: National Academy Press, 1984) Croll, Elisabeth (forthcoming): From Heavan to Earth: Images and Experiences of Development in China Croll, Elisabeth (1990): "China - Women in Development Country Assessment Paper" (Washington, D.C.: World Bank, 1990) Denny, David: "Provincial Economic Differences: A Forty Year Perspective" (Mimeo, 1991) 132 Dixon, John: The Chinese Welfare System. 1949-79 (NY: Praeger Publishers, 1981) Doolette, John and Piazza, Alan: "Impact of Combining Agricultural Techniques and Soil Conservation Measures in China's Loess Plateau" in Proceedings of the 1988 International Conference on Dryland Farming (TX: Texas Agricultural Experiment Station, 1990) FAO: Food Balance Sheets (Rome: FAO, 1991) Gansu Statistical Bureau (1990): Gansu Statistical Yearbook, 1990 (Beijing: SSB, 1990 and earlier years) Gao Xiaomeng: "Grain Procurement and Distribution in China, 1950-89" (Mimeo, 1990) Guangxi Statistical Bureau (1990): Guangxi Statistical Yearbook, 1990 (Beijing: SSB, 1990) Harrell, Paula et al: "Poverty in Southwest China" (Working Paper to this report) Hinton, William: "UNICEF Poverty Alleviation Implementation Study" (Mimeo, 1991) Hubei Statistical Bureau: An Overview of Hubei's Economy 1978-1989 (Wuhan: Hubei Statistical Publishing House, 1990) Hunan Statistical Bureau: Hunan Statistical Yearbook, 1989 (Beijing: SSB, 1989) Jamison, Dean and Jacob van der Gaag (1987): "Education and Earnings in the People's Republic of China" in Economics of Education Review 6 (No. 2, 1987) Jefferson, Gary and Thomas Rawski: "China: Urban Employment and Wage Study" (Mimeo, 1991) (a background paper for World Bank 1992a) Khan, Azizur: "The Distribution of Income in Rural China" in Poverty and Landlessness in Rural Asia (Geneva: ILO, 1977) Lardy, Nicholas (1985): "State Intervention and Peasant Opportunities" in William Parish ed., Chinese Rural Development (NY: M.E. Sharpe Inc., 1985) Lardy, Nicholas (1982): "Food Consumption in the People's Republic of China" in Randolph Barker and Radha Sinha, ed., The Chinese AQricultural Economy (CO: Westview Press, 1982) 133 Lardy, Nicholas (1978): Economic Growth and Distribution in China (Cambridge: Cambridge Universtiy Press, 1978) Lavely, William, et al (1990): "The Rise in Female Education in China: National and Regional Returns" in China Ouarterly 121 (March 1990) 61-93 LGEDPA (1989): Outline of Economic Development in China's Poor Areas (Agricultural Publishing House: Beijing, 1989) Li, Mengbai, et al: The Effect On and Responses to the Floatinq Population in Manor Cities (Economic Daily Publishing House: Beijing, 1991) (in Chinese) Lyons, Thomas P.: "Interprovincial Disparities in China: Output and Consumption, 1952-1987" in Economic Development and Cultural Change (April, 1991): 471-506 Ministry of Agriculture (1991): Chinese Agricultural Yearbook. 1990 (Agricultural Publishing House: Beijing, 1991) Ministry cf Public Health (1991): Selected Edition on Health Statistics in China, 1978-90 (MOPH: Beijing, 1991) Nee, Victor: "Inequality and Market Transition in State Socialism" a paper presented at the 42nd Annual Meeting of the Association for Asian Studies (Mimeo, 1990) Ningxia Statistical Bureau (1990): Ningxia Statistical Yearbook. 1990 (Beijing: SSB, 1990 and earlier years) Ody, Anthony: "China: Rural Enterprise, Rural Industry, 1986-90" (Washington, D.C.: World Bank, 1991) Piazza, Alan: Food Consumption and Nutritional Status in the PRC (Boulder, CO: Westview Press, 1986) Psacharopoulos, George (1985): "Returns to Education: A Further International Update and Implications" in Journal of Human Resources XX (4) Quinley, John: "A Program to Improve Health Services in the Poor Areas" (Working Paper to this report) Rao, Bhanaji: "Poverty in Indonesia, 1970-80: Trends, Associated Characteristics and Research Issues" (Washington, D.C.: World Bank, 1984) Ravallion, Martin (1992): "Poverty Comparisons: a guide to concepts and methods" LSMS Working Paper No. 88 (Washington, D.C.: World Bank, 1992) 134 Ravallion, Martin (1991): "New Estimates of Poverty in the Developing World, 1985-89" (mimeo, 1991) Ravallion, Martin et al (1991): "Quantifying the Magnitude and Severity of Absolute Poverty in the Developing World in the mid-1980s" PRE Working Paper WPS 587 (Washington, D.C.: World Bank, February, 1991) Riskin, Carl et al: "Household Income and Its Distribution in China" (Mimeo, 1992) Roll, Charles: "Incentives and Motivation in China: The Reality of Rural Inequality" (CA: The Rand Corporation, 1976) Rozelle, Scott: "Decomposing China's Rural Income Inequality: Emerging Patterns in the 1980s" (Mimeo, 1992) Schran, Peter: The Development of Chinese Agriculture, 1950-1959 (IL: University of Illinois Press, 1969) Selden, Mark: "Income Inequality and the State" in William Parish ed., Chinese Rural Development (NY: M.E. Sharpe Inc., 1985) Shanghai Statistical Bureau: The Shanghai Floating Population, (Shanghai: Shanghai Statistical Bureau Publishing House, 1989) State Council Census Office, ed: Important Data From The Fourth Chinese Population Census, (Beijing: SSB, 1991) (in Chinese) SSB (1991a): Chinese Statistical Yearbook. 1991 (Beijing: SSB, 1991 and earlier years) SSB (1991b): An Overview of Chinese Rural Economic Statistics By County. 1989 (Beijing: SSB, 1991 and earlier years) SSB (1991c): Chinese Rural Statistical Yearbook. 1991 (Beijing: SSB, 1991 and earlier years) SSB (1990a): Chinese Price Statistical Yearbook. 1990 (Beijing: SSB, 1990 and earlier years) SSB (1990b): Chinese Provincial Statistical Compendium, 1949-89 (Beijing: SSB, 1990) (in Chinese) SSB (1990c): Chinese Urban Residents Household Income and Expenditure Survey Data: 1988 (Beijing: SSB, 1990) (in Chinese) SSB (1988): Chinese Urban Residents Household Income and Expenditure Survey Data: 1987 (Beijing: SSB, 1988) (in Chinese) 135 SSB (no date): "China In-depth Fertility Survey (Phase 2), recode Data Version" (a household data tape covering 4 provinces in the 1987 fertility survey) Taylor, Jeffrey: "Rural Employment Trends and the Legacy of Surplus Labour, 1978-86" in China Ouarterly 116 (December 1988): 736-766 Travers, Lee: "Post-1978 Rural Economic Policy and Peasant Income in China" in China ouarterly 98 (June 1984) Tsui, Kai Yuen: "China's Regional Inequality, 1952-1985" in Journal of Comparative Economics 15 (1991): 1-21 UNDP: Development Cooperation Report for 1990: The People's Republic of China (Beijing: UNDP, 1992) UNICEF: Children and Women of China (Beijing: People's Education Press, 1989) Vermeer, E. B.: "Income Differentials in Rural China" in China Ouarterly 89 (March, 1982): 1-33 Wiens, Thomas: "Poverty and Progress in the Huang and Huai River Basins" in William Parish ed., Chinese Rural Development (NY: M.E. Sharpe Inc., 1985) World Bank (1992a): Report No. 10266-CHA: "China: Reforming the Urban Employment and Wage System" (Washington, D.C.: World Bank, 1992) World Bank (1992b): China: Country Economic Memorandum: Planning and Reform in the 1990s (Washington, D.C.: World Bank, 1992) World Bank (1992c): China: Implementation Options for Urban Housina Reform (Washington, D.C.: World Bank, 1992) World Bank (1992d): "Price Reform in China" (Washington, D.C.: World Bank, 1992) [update with working paper citation] World Bank (1991a): Report No. 8657-CHA: China: Provincial Education Planninq and Finance Sector Study (Washington, D.C.: World Bank, 1991) World Bank (1991b): China: Options for Reform in the Grain Sector (Washington, D.C.: World Bank, 1991) World Bank (1991c): Report No. 9894-CHA: China: Infectious and Endemic Disease Control Project (Washington, D.C.: World Bank, 1991) World Bank (1990a): World Development Report 1990: Poverty (Oxford: Oxford University Press, 1990) World Bank (1990b): China: Macroeconomic Stability and Industrial Growth under Decentralized Socialism (Washington, D.C.: World Bank, 1990) 136 World Bank (1990c): China: Between Plan and Market (Washington, D.C.: World Bank, 1990) World Bank (1990d): "Consumer Food Subsidies" (Working Paper No. 2 for World Bank, 1991b) World Bank (1990e): "China: Long-Term Issues and Options in the Health Transition" (Washington, D.C.: World Bank, 1990) World Bank (1989): Report No. 7605-CHA: China: Revenue Mobilization and Tax Polic , 1989 World Bank (1988): China: Growth and Development in Gansu Province (Washington, D.C.: World Bank, 1988) World Bank (1984): China: The Health Sector (Washington, D.C.: World Bank, 1984) World Bank (1983): China: Socialist Economic Development (Washington, D.C.: World Bank, 1993) Yang Jingying and Liu Cunxin: "Research on the Standard for Measuring Poverty in the Rural Area In China" (Mimeo, 1990) Xu Yanchun: "The Effect of Maternal Education on the Infant Mortality Rate: Shaanxi Province, China, 1950-1985"1 Masters thesis, Univ. of Illinois, Chicago Circle, 1991 Yu Dechang, ed: National 100 Village Labor Power Survey Data: 1978-86 (Beijing: SSB, 1989) (in Chinese) Zhang Shaoshan: "Resettlement" in Rivers, Multiple Taraets, Policy (Nanjing: Hehai University Publishing House, 1990) 51-55 (in Chinese) Zhang Yiming: "Rural Poverty: Characteristics, Causes and Policies in Baise Prefecture" (Working Paper to this report) 137 INCIDENCE OF ABSOLUTE POVERTY 1. Introduction. The estimates of the incidence of absolute poverty presented in this report have been calculated on the basis of absolute poverty lines, which correspond to the estimated cost of a subsistence basket of food and nonfood goods in urban and rural areas, applied to urban and rural income distribution data from the SSB's annual household sample surveys. Unfortunately, the available SSB rural income distribution data substantially understate true income levels by valuing on-farm consumption of grain and other farm products at below-market planned prices. This report's estimates of absolute poverty crudely correct for the understatement of rural income by using the planned price of grain in the calculations of the rural poverty lines; in effect, the understatement of rural income has been offset by an equivalent decrease in the rural poverty line. Estimates of national incidence of absolute poverty, using national average urban and rural poverty lines, have been calculated for 1978 and each year 1980-89. Estimates of provincial rural poverty in 1989 have also been calculated by applying both (i) the national average rural poverty line and (ii) province-specific poverty lines to provincial rural per capita income distribution data. -National 2. Poverty Lines. Urban and rural poverty lines have been calculated as the annualized cost of a subsistence basket of food and nonfood goods. As shown in Table 3, the food component of the subsistence basket includes modest quantities of vegetable oil, vegetables, pork and eggs, plus a little more than 0.5 kg of grain per day. The nongrain foods supply 10% and grain the remaining 90% of the subsistence diet of 2150 calories per day. 11 The content of the food basket, which is held constant in all years, has been expressed in value terms using the official and estimated average annual prices shown in Table 3. For the urban poverty line, actual prices paid by the low income urban population for food can be imputed from SSB urban household survey data. Imputed prices paid by the low income urban population have been used to value urban grain for all years 1981-90. Actual prices for nongrain foods can not be imputed for all years, however, so official mixed average retail prices have been used to value urban nongrain foods. As shown in Table 5, official mixed average retail prices closely approximate available 1/ These same standards of subsistence food energy intake (2150 kcal/day) and composition (90% of food energy from grain) are used in quantitative assessments of poverty in other countries (Rao, for example, for Indonesia). Increasing the subsistence level of food energy intake, or decreasing grain's share of food energy intake, would increase the estimated poverty line and, hence, incidence of poverty in all years. Such changes in assumptions would alter this report's point estimates of the number of poor, but would not significantly change the trends in poverty observed for the period 1978-90 (see para 6). 138 Annex 1 imputed prices for nongrain foods. Unfortunately, it is not possible to calculate actual prices paid for food from published SSB rural household survey data. Mixed average procurement prices, which reasonably approximate prices faced by the rural population, have instead been used to value rural nongrain foods. The planned price of grain, estimated as a weighted average of grain-specific quota (1978-84) and contract (1985-89) prices, has been used to value rural grain (see "planned price" in Tables 3 and 4). As shown in Table 5, the estimated planned price of grain was considerably less than the mixed average procurement price in recent years. Using the estimated plan price for grain corrects, albeit imperfectly, for the downward bias in the official SSB rural income distribution data (para 1). ' 3. To approximate the cost of nonfood subsistence goods, annualized expenditure on food has been inflated by the reciprocal of food's estimated budget share for low income households. Food's budget share for the poorest 10% of urban households, shown in Table 5, has been calculated from SSB urban household income and expenditure survey data. For poor rural households, food's budget share has been approximated as the average share for all rural households (from SSB rural survey data) inflated by the ratio of food's budget share for poor urban households and all urban households. 3/ An alternative rural poverty line has also been calculated using the mixed average procurement price to value grain. The difference between subsistence grain consumption valued at the mixed average procurement price and at the estimated planned price is then directly deducted from the subsistence basket of food and nonfood goods. This second rural poverty line, identified as the "procurement price" line in Tables 3 and 4, exceeds the "planned price" line in recent years. Y The two rural poverty lines are nearly identical, however, for the period 1978-85. ' Lacking sufficient data, it is not possible to directly correct the published SSB rural income distribution figures. Available evidence suggests that valuing on-farm consumption of grain at planned prices resulted in an understatement of rural income of nearly Y60 billion in 1989, or, on average, roughly Y70 per rural inhabitant. Valuing rural grain at the estimated planned price in 1989 diminishes (i) estimated subsistence expenditure on grain by about Y50 and (ii) the rural absolute poverty line by Y75. I Food's budget share for poor rural households has been approximated in the poverty calculations as 75% in 1978 and 1980, 66% in 1981, and 63% for all years 1982-90. Holding food's budget share constant (at, say, 63%) in all years 1978-90 would result in a corresponding substantial increase in the estimated number of poor during 1978-81. Such a change in assumptions would therefore amplify this report's finding that the incidence of poverty declined sharply during 1978-84. 4 In contrast to the planned price line (see footnote 1 above), the procurement price line diminishes (i) estimated subsistence expenditure on grain and (ii) the rural absolute poverty line by equal amounts (e.g., by about Y50 in 1989). 139 Annex 1 4. Incidence of Poverty. The estimated urban and rural absolute poverty lineis have been applied to the official SSB distributions of urban and rural households by per capita income (reported in Tables 6 and 7) to derive the percentage of poor households shown in Table 4. 5/ Based on available information on household size by income group (household size is negatively correlated with per capita income), the distribution of urban and rural population by per capita income (Tables 6 and 7) has been estimated. On the basis of the breakdown of total population by urban and rural registration shown in the top row of Table 4, the number of absolute poor has been calculated by applying the poverty lines to the estimated distribution of population by per capita income. As a sensitivity test, and representing an upper bound for absolute poverty, the incidence of poverty has also been calculated with both of the rural poverty lines inflated by 10%. 5. Results and Sensitivity Analysis. Comparing this report's estimated poverty lines with the official retail price indices shown in Table 1, it is evident that the urban poverty line closely tracks the growth rate of urban retail prices during 1981-85, but does not keep pace with retail prices during 1985-90. An examination of the prices used to value the urban subsistence food basket (shown in Table 5) reveals that the relatively slow growth of the urban poverty line is due entirely to the very modest increase in the price paid by poor urban households for grain. Due to the sharp increases in rural prices for all foods and, to a lesser extent, the decrease in food's budget share, the growth rate of the rural poverty lines greatly exceeds that of rural retail prices during 1978-82. With the adjustments made to compensate for the downward bias in reported rural income (para's 2 and 3), however, the "planned" price poverty line does not keep pace with the increase in rural retail prices during 1987-89. The "procurement" price poverty line, on the other hand, very nearly matches the increase in rural retail prices through 1989. 6. This report's estimates of the incidence of poverty, summarized in Table 1, indicate the near absence of absolute urban poverty during the 1980s. Inflating the urban poverty line by 20%, furthermore, estimated urban poverty remains at extremely modest levels -- fewer than 2% of registered urban households in all years 1981-89. F The "planned price" estimates for the 5/ For the purpose of these calculations, the distribution of households within income intervals is assumed to be linear. As shown in Table 6, the SSB urban income distribution data for 1984-89 aggregate several of the lower income intervals. Interpolating on the basis of prior year distributions, these data have been disaggregated into the same lower income intervals reported for 1981-83. F The (i) assumption of a linear distribution within income intervals and (ii) interpolation of the lower income interval data for 1984-90 (see footnote 4) impart a wide margin of error to the urban poverty estimates. It should only be concluded, therefore, that (i) there was a real decline in urban poverty during 1981-83 and (ii) the extremely low levels of urban poverty (continued...) 140 Annex 1 rural population indicate a sharp and uninterrupted decline in absolute poverty during 1978-84 followed by stagnant levels of poverty during 1985-90. The rural "procurement price" poverty line, on the other hand, indicateB no decline in rural poverty during 1985-88 and a sharp resurgence in 1989 to about 1983 levels. This difference in trend and absolute levels corresponds to the increasing divergence between the two rural poverty lines during 1985- 90 (para 3). Inflating the rural poverty lines by 10% does not significantly alter the observed trend in rural poverty during 1978-90. Reflecting the bunching of a large share of the rural population in the proximity of the poverty lines, however, the 10% increase in the poverty line corresponds to a large absolute increase -- of about 20% in 1978 and between 25% to 37% during 1980-89 -- in the estimated number of poor. Annex 1 Table 1: National Poverty Lines and Incidence 1978 1980 1981 1982 1983 1984 1985 1986 1987 1988 1989 1990 Poverty Line (Yuan): -Urban 171 169 178 190 215 226 247 289 304 321 -Rural (planned) 98 134 158 167 175 179 190 199 210 231 262 275 -Rural (procurement) 99 135 160 170 175 178 193 206 222 249 292 301 Poverty Une Index (1985 = 100): -Urban 80 79 83 88 100 105 115 134 141 149 -Rural (planned) 52 71 83 88 92 94 100 105 111 122 138 145 -Rural (procurement) 51 70 83 88 91 92 100 107 115 129 151 156 Retail Price Index: -Urban (1985=100) 74 81 84 85 87 89 100 107 117 142 164 165 -Rural (1985=100) 81 86 88 90 91 93 100 105 112 131 155 160 Procurement Price Index: -Grain (1985=100) 63 87 92 94 94 95 100 112 122 135 180 172 Absolute Poverty (% of households): -Urban 1.5 0.6 0.5 0.2 0.2 0.1 0.2 0.2 0.2 0.3 -Rural (planned) 32.8 26.8 23.4 16.3 14.0 10.0 10.7 11.2 10.0 9.5 11.4 10.7 -Rural (procurement) 33.0 27.2 24.3 17.4 14.1 9.9 11.1 12.5 12.1 11.9 14.7 13.3 Absolute Poverty (million people): -Urban 4 2 1 1 1 1 1 1 1 1 poverty line +20% 5 2 2 1 2 2 2 4 3 3 -Rural (planned) 260 218 194 140 123 89 96 97 91 86 103 97 poverty line +10X 306 272 255 186 163 119 131 133 124 113 128 121 -Rural (procurement) 262 221 202 148 123 88 100 108 109 106 132 121 poverty Line +10X 310 276 264 195 164 118 138 146 144 136 168 160 Source: Table 4 and, for price indices, SSB 1991. F( ... continued) observed for all years 1981-90 are invariant to a 20% increase in the poverty line. 141 Annex 1 -Provincial 7. Poverty Lines and Incidence. Provincial rural poverty in 1989 has been estimated by applying both (i) the national average "planned" poverty line and (ii) province-specific "planned" poverty lines to detailed provincial rural per capita income distribution data available for that year. 7' The province-specific poverty lines have been calculated by correcting the national average poverty line for estimated differences in provincial subsistence expenditure on grain. Provincial subsistence expenditure on grain has been estimated as the product of subsistence grain consumption (Table 3) and estimated province-specific weighted (in accordance with provincial structure of grain production) average contract procurement prices (para 2). Since the contract price for milled rice (Y610 per ton) exceeds the weighted average contract price for all grain (Y490 per ton), the province-specific poverty lines exceed the national poverty line (Y262 in 1989) for provinces which produce an above average share of rice (that is, those provinces in the middle and lower reaches of the Yangtze River and South China). Poverty lines for all other provinces (except Sichuan), on the other hand, are less than the national poverty line. As reported in Table 2, the province-specific poverty lines range from Y277 to Y284 in seven provinces where milled rice accounts for 80% or more of total grain production. The province-specific poverty lines are Y250 or less, by comparison, in eight provinces where milled rice accounts for 20% or less and corn (Y356 per ton) more than 50% of total grain production. 8. The provincial estimates confirm the concentration of rural poverty in northwestern, southwestern, northeastern and northern China. On the basis of the national poverty line of Y262 in 1989, the percentage of poor rural households ranges between 17% to 34% in each of the 7 northwestern provinces, between 11% to 19% in each of the 4 southwestern provinces, and between 12% to 18% in four provinces in northern and northeastern China. Modest levels of poverty, of between 5% to 8% of rural households, are observed for Shandong, Liaoning and the central provinces of Anhui, Jiangxi, Hubei, and Hunan. The province-specific poverty lines confirm this same regional concentration of poverty in western and northern China, but indicate slightly lower rates of poverty in northern, northeastern, and northwestern China and slightly increased rates in the central provinces. -Conclusion 9. Available official data have been used to quantify trends and absolute levels of poverty at the national and provincial levels. The observed trends, viz., the virtual absence of urban poverty throughout the 1980s and the sharp decline in rural poverty during 1978-84 followed by stagnation during 1985-90, are not sensitive to increases of 20% and 10% Z/ The income distribution data are percent of rural households within each of 12 income intervals (viz., less than Y100, 100-150, 150-200, 200-300, 300- 400, 400-500, 500-600, 600-800, 800-1000, 1000-1500, 1500-2000, and more than 2000) for each province. Mean income by interval is not known. 142 Annex 1 Annez 1 Table 2: Provincial Poverty Lines and Incidence Incidence of Poverty: Percent of Households _____________________ National Provincial Poverty Poverty Poverty Line Line Line Difference -------------------------------------------------------------__---- National Average 262 11.3 11.3 North Beijing 250 0.2 0.1 0.0 Tianjin 253 0.4 0.4 0.1 Hebei 249 13.0 11.7 1.3 Henan 254 16.5 15.2 1.4 Shandong 250 6.8 5.9 0.8 Northeast Liaoning 249 8.0 7.1 0.9 Jilin 244 12.2 10.7 1.5 Heilongjiang 259 18.3 17.9 0.4 Northwest Inner Mongolia 247 23.5 20.8 2.7 Shanxi 246 17.4 14.8 2.6 Shaanxi 253 20.3 18.4 1.9 Ningxia 259 18.9 18.4 0.5 Gansu 250 34.2 30.4 3.7 Qinghai 252 23.7 21.8 1.9 Xinjiang 254 18.7 17.6 1.1 Yangtze River Shanghai 274 0.0 0.0 0.0 Jiangsu 268 3.4 3.6 -0.2 Zhejiang 278 2.0 2.3 -0.4 Anhui 268 7.7 8.3 -0.5 Jiangxi 284 5.0 6.4 -1.3 Hubei 274 6.0 7.0 -0.9 Hunan 282 6.2 7.7 -1.5 South Fujian 277 1.8 2.1 -0.3 Guangdong 279 0.9 1.1 -0.2 Hainan 277 3.3 3.8 -0.5 Southwest Guangxi 278 15.4 18.1 -2.6 Sichuan 262 11.2 11.2 0.0 Guizhou 259 17.8 17.4 0.5 Yunnan 258 19.0 18.3 0.7 Source: See para's 7-8. 143 Annex 1 respectively in the estimated national urban and rural poverty lines. Similarly, the observed concentration of rural poverty in western, northeastern and northern China in 1989 is not significantly altered by the use of province-specific poverty lines which correct for differences in provincial subsistence expenditure on grain. However, the limitations of these preliminary estimates must be recognized. Most importantly, adjusting the rural poverty line downward to correct for the downward bias in the official SSB rural income distribution data widens the margin of error in this report's estimates of poverty. The "procurement price" rural poverty line, which is an alternative means of correcting for the bias in the SSB data, indicates a significant resurgence and greater absolute levels of rural poverty in 1985 and 1989 than does the "planned price" rural poverty line. Further refinements of national and provincial poverty lines, including the preparation of poverty lines based on the expenditure elasticity of demand for food energy and more in-depth county and village level poverty profiles, are essential to an improved understanding of the incidence of poverty and appropriate policy formulation. Annex I Table 3: Urban and Rural Subsistence Expenditure on Grain and Nongrain Foods, 1978 - 1990 (Yuan per capita per davl 1978 1980 1981 1982 Average Average Average Average Average Average Energy Volume Procurement Procurement Retail Procurement Retail Procurement Energy Kcal/ grams/ Price Expense Price Expense Price Expense Price Expense Price Expense Price Expense Kcal/kg Share day day yuan/kg yuan yuan/kg yuan yuan/kg yuan yuan/kg yuan yuan/kg yuan yuan/kg yuan ----------------------------------------------------------------__-----------__-------------------__-_-_--------------------------__---------___--__-------__----_ NONGRAIN FOODS Vegetable Oil 8840 0.060 129 14.6 1.75 0.025 2.64 0.039 1.97 0.029 2.82 0.041 1.97 0.029 2.77 0.040 Vegetables 262 0.020 43 164.1 0.08 0.013 0.10 0.017 0.13 0.022 0.11 0.018 0.14 0.023 0.11 0.017 Pork 3002 0.015 32 10.7 1.43 0.015 1.82 0.020 2.11 0.023 1.82 0.020 2.12 0.023 1.83 0.020 Eggs 1450 0.005 11 7.4 1.38 0.010 1.71 0.013 2.15 0.016 1.84 0.014 2.17 0.016 1.86 0.014 subtotal 0.10 215 0.065 0.088 0.089 0.093 0.091 0.091 UWAN: Grain 3580 0.90 1935 540.5 0.38 0.206 0.38 0.208 Total: 2150 0.295 0.298 RURAL: Grain: Planned 3580 0.90 1935 540.5 0.25 0.138 0.35 0.188 0.36 0.195 0.36 0.197 Total: 2150 0.202 0.276 0.288 0.289 Grain: Procurement 3580 0.90 1935 540.5 0.26 0.142 0.36 0.195 0.38 0.206 0.39 0.212 Total: 2150 0.207 0.283 0.299 0.303 Table 3: continued 1983 1984 1985 1986 Average Average Average Average Average Average Average Average Retail Procurement Retail Procurement Retail Procurement Retail Procurement Price Expense Price Expense Price Expense Price Expense Price Expense Price Expense Price Expense Price Expense yuan/kg yuan yuan/kg yuan yuan/kg yuan yuan/kg yuan yuan/kg yuan yuan/kg yuan yuan/kg yuan yuan/kg yuan ----------------------------------------------------------------__-----------__----------------------------------------_----------_-----------___-__-------__----____- NONGRAIN FOODS Vegetable Oil 1.93 0.028 2.65 0.039 2.07 0.030 2.62 0.038 2.18 0.032 2.85 0.042 2.28 0.033 2.85 0.042 Vegetables 0.16 0.026 0.11 0.018 0.18 0.029 0.12 0.019 0.26 0.042 0.20 0.032 0.28 0.046 0.23 0.037 Pork 2.19 0.023 1.82 0.020 2.28 0.025 1.86 0.020 2.75 0.030 2.25 0.024 3.02 0.032 2.35 0.025 Eggs 2.25 0.017 1.89 0.014 2.36 0.018 1.98 0.015 2.69 0.020 2.27 0.017 2.95 0.022 2.42 0.018 subtotal 0.094 0.090 0.101 0.092 0.123 0.115 . 0.134 0.122 URBAN: Grain 0.39 0.212 0.405 0.219 0.44 0.237 0.43 0.230 Total: 0.306 0.320 0.361 0.364 RlURAL: Grain: Planned 0.39 0.211 0.40 0.216 0.39 0.213 0.41 0.222 Total: 0.301 0.308 0.328 0.344 Grain: Procurement 0.39 0.212 0.40 0.213 0.42 0.225 0.47 0.252 Total: 0.303 0.306 0.340 0.374 - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -_ _ - - - - - - - - - -_ _ - - - - - - - - - - - - - - - - - - -_ - - - - -_ - - - - - - - -_ _ _ _ - -_ -_ - - - - - - - -_ - - - - - - - - - - - -_ _ _ _ _ _ - - - - - -_ _ - - - - - - - - - Table 3: continued 1987 1988 1989 1990 Average Average Average Average Average Average Average Average Retail Procurement Retail Procurement Retail Procurement Retail Procurement Price Expense Price Expense Price Expense Price Expense Price Expense Price Expense Price Expense Price Expense yuan/kg yuan yuan/kg yuan yuan/kg yuan yuan/kg yuan yuan/kg yuan yuan/kg yuan yuan/kg yuan yuan/kg yuan NONGRAIN FOODS Vegetable Oil 2.40 0.035 3.08 0.045 2.67 0.039 3.68 0.054 3.36 0.049 4.32 0.063 3.43 0.050 4.42 0.065 Vegetables 0.38 0.063 0.28 0.046 0.48 0.079 0.31 0.051 0.51 0.083 0.27 0.044 0.53 0.087 0.42 0.068 Pork 3.50 0.038 2.59 0.028 4.96 0.053 4.04 0.043 5.40 0.058 4.49 0.048 5.30 0.057 4.32 0.046 Eggs 3.53 0.026 3.04 0.023 4.36 0.032 3.66 0.027 4.87 0.036 4.17 0.031 5.03 0.037 4.36 0.032 subtotal 0.161 0.142 0.204 0.175 0.226 0.187 0.231 0.212 URBAN: Grain 0.462 0.250 0.50 0.271 0.545 0.295 0.571 0.308 Total: 0.411 0.474 0.521 0.540 RURAL: Grain: Planned 0.41 0.221 0.42 0.224 0.49 0.265 0.48 0.262 Total: 0.362 0.399 0.451 0.474 Grain: Procurement 0.51 0.275 0.56 0.305 0.75 0.405 0.72 0.387 Total: 0.416 0.479 0.592 0.599 Source: See text for composition of subsistence basket of food. Price data from SSB 1991 and Table S. -n Annea I Table 4: Urban and Rural Absolute Poverty Lines and Incidence of Poverty, 1978-90 1978 1980 1981 1982 Rural Rural Rural Total Rural Tot-l Urban Pln Procur Urban Pln Procur Urban Plan Procure Pln Procure Urban Plan Procure Plan Procure POPULATION: -million 172 790 191 790 202 799 1001 215 802 1017 EXPENDITURE (Yuenl: -Daily Food \i 0.20 0.21 0.28 0.28 0.30 0.29 0.30 0.29 0.30 0.30 0.29 0.30 0.29 0.30 -Annual Food 74 76 101 103 108 105 109 106 109 109 105 111 106 110 -Food's Budget Share 0.75 0.75 0.75 0.75 0.63 0.66 0.66 0.66 0.66 0.64 0.63 0.63 0.63 0.63 POVERTY LUNE: -Yuan \_ 98 99 134 135 171 158 160 161 162 169 167 170 16" 170 INCIDENCE OF POVERTY: -X households 32.8 33.0 26.8 27.2 1.5 23.4 24.3 19.0 19.7 0.6 16.3 17.4 13.0 13.8 -% population 33.0 33.2 27.6 28.0 1.9 24.3 21.3 19.8 20.6 0.9 17.4 18.5 13.9 14.8 -people (mill) 260.5 262.0 217.9 221.2 3.9 194.3 202.3 198.2 206.2 2.0 139.7 148.3 141.7 150.3 Table 4: continued 19B3 1984 1985 1986 Rural Total Rural Total Rural Total Rural Total Urban Plan Procure Plan Procure Urban Plan Procure Plan Procure Urbon Plan Procure Plan Procure Urban Plan Procure Plan Procure f.- POPULATION: -million 223 807 1030 240 803 1044 251 808 1059 264 811 1075 EXPENDITURE (Yc.n.: -Daily Food \A 0.31 0.30 0.30 0.30 0.30 0.32 0.31 0.31 0.31 0.31 0.36 0.33 0.34 0.33 0.34 0.36 0.34 0.37 0.35 0.37 -Annual Food 112 110 110 110 111 117 112 112 113 113 132 120 124 122 126 133 125 136 127 136 -Food's Budget Share 0.63 0.63 0.63 0.63 0.63 0.62 0.63 0.63 0.63 0.63 0.61 0.63 0.63 0.63 0.63 0.59 0.63 0.63 0.62 0.62 POVERTY UNE: -Yu.n \b 178 175 i17 175 176 190 179 178 181 181 215 190 193 196 198 226 199 206 206 211 INCIDENCE OF POVERTY: h households 0.5 14.0 14.1 11.1 11.2 0.2 10.0 9.9 7.7 7.7 0.2 10.7 11.1 8.2 8.6 0.1 11.2 12.5 8.5 9.5 --E populotion 0.6 15.2 15.3 12.0 12.1 0.3 11.0 10.9 8.6 8.5 0.4 11.9 12.4 9.2 9.6 0.2 11.9 13.3 9.0 10.1 -people (million) 1.4 122.5 123.4 123.9 124.7 0.8 88.7 87.9 89.5 88.8 0.9 96.4 100.2 97.3 101.1 0.5 96.5 108.0 97.0 108.5 Table 4: continued 1987 1988 1989 1990 Rural Total Rural Total Rural Total Rural Total Urban Plan Procure Plan Procure Urban Plan Procure Plan Procure Urban Plan Procure Plan Procure Urban Plan Procure Plan Procure POPULATION: -million 277 816 1093 287 824 1110 295 832 1127 302 841 1143 EXPENDITURE (Yuan): -Daily Food \a 0.41 0.36 0.42 0.37 0.42 0.47 0.40 0.48 0,41 0.48 0.52 0.45 0.59 0.47 0.58 0.54 0.47 0.60 0.49 0.58 -Annual Food 150 132 .152 136 152 173 146 175 151 175 190 165 216 170 211 197 173 218 179 213 -Food's Budget Share 0.61 0.63 0.63 0.63 0.63 0.60 0.63 0.63 0.62 0.62 0.62 0.63 0.63 0.63 0.63 0.61 0.63 0.63 0.63 0.63 POVERTY UNE: -Yuan \_ 247 210 222 219 228 289 231 249 246 259 304 262 292 273 295 321 275 301 287 307 INCIDENCE OF POVERTY. -- households 0.2 10.0 12.1 7.5 9.0 0.2 9.5 11.9 7.1 8.8 0.2 11.4 14.7 8.5 10.9 0.3 10.7 13.3 7.9 9.9 -C population 0.2 1131 13.4 8.3 10.0 0.2 10.4 12.9 7.8 9.6 0.3 12.3 15.9 9.2 11.8 0.4 11.5 14.3 8.6 10.7 -people (million) 0.6 90.5 109.0 91.2 109.6 0.7 85.6 106.4 86.3 107.1 0.9 102.5 131.9 103.4 132.8 1.3 96.8 120.6 98.2 122.0 Source: See text and Table 3. Poverty lines estimated aS subsistence expenditure on food inflated by the reciprocal of food's budget share to approximate the cost of nonfood subsistence goods. Incidence of poverty calculated by applying estimated poverty lines to official SSB urban and rural income distribution data (Tables 6 and 7). \¶ Expenditure on all food, from Table 3. \ Annual subsistence expenditure on food inflated by the reciprocal of food's budget share. .Ss Annex 1 Table 5: Price Parameters and Food Budget Shares. 1978-90 (Yuan/kg and %) 1978 1980 1981 1982 1983 1984 1985 1986 1987 1988 1989 1990 Urban Prices Grain Unit Value \a 0.381 0.384 0.392 0.405 0.439 0.425 0.462 0.501 0.545 0.571 Average RetaTl Price 0.337 0.340 0.351 0.358 0.383 0.413 0.442 0.489 0.557 0.528 Vegetable Oil Unit Value \a 1.798 1.906 2.023 2.209 2.522 Average Retail Price 1.967 1.969 1.930 2.071 2.179 2.278 2.396 2.672 3.365 3.426 Vegetables Unit Value \a 0.166 0.189 0.260 0.343 0.425 Average RetaTl Price 0.134 0.140 0.156 0.178 0.257 0.283 0.382 0.481 0.506 0.531 Pork Unit Value \a 2.283 2.374 3.068 3.616 5.030 Average RetaiL Price 2.108 2.120 2.186 2.282 2.750 3.022 3.501 4.956 5.402 5.298 Eggs Unit Value \a 2.289 2.333 2.600 3.281 3.769 Average RetaTI Price 2.152 2.168 2.248 2.364 2.694 2.948 3.531 4.362 4.867 5.029 Rural Prices: Average Procurement Prices \b Grain 0.263 0.361 0.382 0.392 0.393 0.395 0.416 0.466 0.509 0.564 0.750 0.716 Vegetable Oil 1.746 2.641 2.819 2.773 2.654 2.624 2.846 2.846 3.078 3.676 4.318 4.425 Vegetables 0.082 0.104 0.112 0.106 0.110 0.118 0.198 0.226 0.279 0.308 0.271 0.416 Pork 1.143 1.456 1.459 1.461 1.456 1.487 1.801 1.876 2.073 3.233 3.593 3.459 Eggs 1.378 1.714 1.840 1.864 1.886 1.976 2.266 2.424 3.042 3.658 4.168 4.355 Planned Procurement Prices for Grain Milled Rice 0.280 0.330 0.330 0.330 0.330 0.330 0.446 0.446 0.446 0.446 0.610 0.610 Wheat 0.272 0.314 0.314 0.314 0.314 0.314 0.424 0.424 0.424 0.454 0.486 0.486 Soy 0.401 0.690 0.690 0.690 0.690 0.690 0.690 0.690 0.690 0.690 0.690 0.690 00 Corn 0.176 0.214 0.214 0.214 0.214 0.214 0.289 0.336 0.336 0.336 0.356 0.356 Planned Procurement Shares (% & million tons) Quota 75% 28.29 26.89 28.21 30.32 27.86 Above Quota 25% 20.28 23.83 26.94 59.74 73.62 Weighted Average Planned Procurement Price \c Grain 0.255 0.348 U.361 0.365 0.390 0.399 0.393 0.410 0.408 0.415 0.490 0.485 Food's Budget Share (%I: Urban Poorest 10% 63.05 64.30 62.83 61.58 61.21 58.66 60.81 59.82 62.47 61.34 Average 56.66 58.65 59.20 57.97 53.31 52.43 53.71 51.36 54.50 54.24 Ratio 1.11 1.11 1.11 1.10 1.06 1.06 1.15 1.12 1.13 1.16 1.15 1.13 Rural Average 67.70 61.80 59.70 60.50 59.30 59.00 57.70 56.30 55.20 53.40 54.09 54.86 Rural Poor \d 75.34 68.77 66.43 66.33 62.94 62.67 66.25 62.99 62.50 62.20 62.00 62.04 Source: Average retail and average procurement prices, unit values and food's budget share from SSB (1991a). Planned procurement prices for grain from World Bank (1991). Planned procurement shares from Gao (1989). \a Unit value for poorest 10% of households imputed from SSB urban sample survey data. \b Average of planned, negotiated and free market prices for individual grains weighted by share of total grain production and the structure of procurement. \c Average planned procurement price weighted by share of total grain production and assuming a 25% price premium for above quota procurement in 1978 and a 50% premium in 1980-84. \d Average rural share times the ratio of the shares for poor and average urban households. 149 Annex 1 Annex 1 Table 6: Distribution of Urban Households and Porulation by Per Capita Income. 1981 - 1990 (X and #) Income Group (Current Yuan) 1981 1982 1983 1984 1985 1986 1987 1988 1989 1990 Distribution of Households: <240 2.05 0.92 0.61 j I I 240 -300 5.46 3.68 2.97 1.67 I 11.05 1 5.67 4.27 1 8.31 4.76 300 - 420 31.81 25.63 20.32 10.52 1 I 7.38 420 -600 42.29 45.40 46.56 38.89 24.78 14.81 11.70 600 - 720 11.90 14.20 16.42 22.67 19.53 16.17 13.36 7.61 4.97 720- 840 1 I 12.81 16.04 17.86 15.74 10.26 6.90 5.71 840 - 960 I 1 1 10.96 14.82 14.60 12.53 9.30 6.95 960 -1080 1 1 1 1 7.01 10.24 12.09 12.46 10.83 8.76 1080 -1200 1 4.11 7.40 8.49 11.91 11.19 8.75 1200 -1320 6.49 10.17 1 13.12 13.44 2.63 4.60 6.08 9.21 10.24 9.84 1320 -1440 1 3.05 4.35 7.25 8.61 9.17 1440 -1560 1 I 1 1.83 3.03 5.32 7.45 7.96 1560 -1680 1 3.89 1 1.24 2.04 3.82 6.01 7.43 1680 -1800 1 1 0.69 1.11 2.95 4.41 5.81 >1800 1 1 1 1.62 3.17 8.36 15.84 22.26 Number of Households: 8715 9020 9060 12500 24338 31126 32855 34945 35235 35660 Estimated Average Household Size: \a c240 5.60 5.88 5.49 1 240 -300 5.22 5.22 5.13 5.77 1 5.03 1 4.90 1 4.75 4.63 4.51 1 4.51 300 -420 4.66 4.62 4.52 4.84 420- 600 4.00 4.02 3.99 4.19 4.21 4.47 4.40 600 720 3.70 3.80 3.75 3.80 3.75 4.26 4.23 4.20 4.17 4.17 720 -840 1 3.71 3.64 4.09 4.09 4.09 4.08 4.08 840 -960 1 1 3.59 3.91 3.95 3.98 3.99 3.99 960 -1080 I 1 3.50 3.74 3.81 3.87 3.89 3.89 1080 -1200 1 1 I 3.46 3.57 3.67 3.75 3.80 3.80 1200 -1320 3.20 3.34 1 3.55 3.48 1 3.42 3.40 3.53 3.64 3.71 3.71 1320 -1440 1 1 1 3.23 3.40 3.53 3.62 3.62 1440 -1560 1 1 1 1 3.06 3.26 3.42 3.53 3.53 1560 -1680 1 3.22 1 2.89 3.12 3.30 3.44 3.44 1680 -1800 1 1 1 2.71 2.98 3.19 3.35 3.35 >1800 1 2.70 2.98 3.00 3.20 3.20 Estimated Population Distribution: <240 2.72 1.31 0.83 240 -300 6.75 4.65 3.77 2.39 14.21 1 6.96 5.16 10.13 5.74 1 300 -420 35.11 28.64 22.71 12.60 1 1 1 1 9.10 420 - 600 40.07 44.14 45.94 40.34 26.67 16.58 13.11 1 600 - 720 10.43 13.05 15.23 21.33 18.72 17.23 14.38 8.43 5.54 720 -840 1 1 11.77 14.93 18.27 16.39 11.06 7.53 6.37 840 -960 1 10.06 14.52 14.68 13.14 9.92 7.57 960 -1080 6.27 9.59 11.73 12.69 11.29 9.33 1080 -1200 1 1 3.64 6.61 7.94 11.78 11.39 9.10 1200 -1320 4.92 8.22 1 11.52 1 11.58 1 2.30 3.91 5.47 8.83 10.17 9.99 1320 -1440 1 1 2.46 3.76 6.74 8.34 9.08 1440 -1560 I 1 1 1 1.40 2.51 4.79 7.04 7.68 1560 -1680 1 1 1 1 3.20 1 0.90 1.62 3.32 5.53 6.99 1680 -1800 1 1 I I I 0.47 0.84 2.48 3.95 5.32 >1800 I 1 1 1 1 1.09 2.40 6.61 13.56 19.48 Source: Household distribution from SSB 1991 and population distribution from SSB 1988. \a Figures for 1986-89 are interpolated. 150 Annex 1 Annex 1 Table 7: Distribution of Rural Households and Population by Per Capita Income. 1978 - 1990 (% and #) Income Group (Current Yuan) 1978 1979 1980 1981 1982 1983 1984 1985 1986 1987 1988 1989 1990 Distribution of Households: < 100 33.3 19.3 9.8 4.7 2.7 1.4 0.8 1.0 1.1 0.9 0.5 0.6 0.5 100 - 150 31.7 24.2 24.7 14.9 8.1 6.2 3.8 3.4 3.2 2.4 1.5 1.3 0.9 150- 200 17.6 29.0 27.1 23.0 16.0 13.1 9.4 7.9 7.0 5.0 3.3 2.8 2.2 200 -300 15.0 20.4 25.3 34.8 37.0 32.9 29.2 25.6 21.8 17.5 13.5 10.9 9.5 300 - 400 | 5.0 8.6 14.4 20.8 22.9 24.5 24.0 21.7 21.3 17.5 15.6 14.4 400 -500 2.4 1.5 2.9 5.0 8.7 11.6 14.1 15.8 16.5 17.2 16.7 15.6 15.1 > 500 I 0.6 1.6 3.2 6.7 11.9 18.2 22.3 28.7 35.7 47.0 53.2 57.4 Number of Households: 6095 10282 15914 18529 22775 30427 31375 66642 66836 66912 67186 66906 66478 Estimated Average Household Size: 5 100 6.0 6.0 6.0 6.0 6.0 6.1 6.2 6.3 6.4 6.5 6.7 6.8 6.8 100 -150 6.0 6.0 6.0 6.0 6.0 6.0 6.0 6.1 6.1 6.1 6.2 6.3 6.3 i50 -200 6.0 6.0 6.0 6.0 6.0 6.0 6.0 6.1 6.1 6.1 6.2 6.3 6.3 200 -300 6.0 6.0 6.0 6.0 6.0 6.0 6.0 6.1 6.1 6.1 6.2 6.3 6.3 300 -400 1 5.0 5.2 5.4 5.5 5.6 5.6 5.7 5.7 5.8 6.0 6.1 6.1 400 500 4.4J 4.4 4.5 4.6 4.8 5.0 5.2 5.3 5.4 5.5 5.8 6.0 6.0 > 500 1 2.7 3.1 3.5 3.5 3.9 4.1 4.4 5.7 5.0 5.4 5.6 5.6 Estimated Population Distribution: \a < 100 33.5 19.6 10.1 4.9 2.9 1.5 0.9 1.1 1.2 1.1 0.6 0.7 0.5 100 - 150 31.9 24.6 25.4 15.5 8.6 6.7 4.2 3.8 3.4 2.6 1.6 1.4 1.0 150 -200 17.7 29.5 27.8 23.9 17.1 14.2 10.4 8.8 7.4 5.5 3.6 3.0 2.4 200 -300 15.1 20.7 26.0 36.2 39.5 35.6 32.2 28.4 23.0 19.3 14.6 11.7 10.3 300 - 400 X 4.2 7.7 13.5 20.3 23.1 25.2 24.9 21.4 22.3 18.3 16.3 15.1 400 -500 1.81 1.1 2.2 4.0 7.4 10.5 13.5 15.2 15.4 17.1 16.9 16.0 15.6 > 500 a 0.3 0.8 1.9 4.2 8.4 13.7 17.8 28.3 32.2 44.3 50.9 55.1 Source: Household distribution from SSB 1991 and population distribution from Guangxi Statistical Bureau 1989. \a Figures are for Guangxi for 1988 -- househoLd size for aLL other years has been interpolated. 151 Annex 1 Annex 1 Table 8: Averaae Per Canita Income of Town and City Families. 1987 (Yuan) PROVINCE BOTTOM 5% BOTTOM 10% OVERALL CHINA 4221505 446/564 788/946 Anhui 356/483 388/528 725/886 Beijing ---/594 ---/651 --/1102 Fujian 382/487 408/540 762/917 Gansu - - -/385 -- -/443 - -/818 Guangdong 380/459 419/502 830/939 Guangxi 401/471 444/520 829/907 Guizhou 343/440 360/485 692/835 Neilongiiang 310/414 330/502 665/803 Henan 322/442 357/491 663/854 Hebei 405/492 452/542 770/912 Hubei 404/494 466/547 807/917 Hunan 423/467 443/527 764/930 Jiangsu 629/584 650/638 931/1031 Jifangxi 390/414 425/456 758/777 JiLin 318/398 343/446 638/777 Liaoning 356/454 396/516 724/883 Inner MongoLia 331/400 355/442 679/818 Ningxia 394/376 464/439 836/800 Qinghai 448/388 473/456 865/891 Shaanxi 340/422 386/474 756/903 Shandong 509/519 552/569 901/967 Shanghai - -- /675 -- -/750 --/1202 Shanxi 329/416 362/460 663/805 Sichuan 395/523 442/573 837/970 tianjin ---/563 -- -/618 --/1048 Xinjiang 344/234 378/295 768/945 Yunnan 457/493 484/553 844/919 Zhejiang 608/603 650/653 1056/1055 Note: In 1987, Hainan province was stiLt part of Guangdong and is so treated here. The absence of Tibet reftects a Like absence in official standard of living reporting. Gansu is known to have poor towns and statistics would doubtless reflect that were they pubLished. The numbers given are from SSB (1988) adjusted for differences in cost of Living. 152 Annex 1 Appendix 1 APPED IX 1: RURAL INCOME REPORTING SYSTEMS 1. This Annex uses income data generated by the State Statistical Bureau (SSB) system in its analysis of poverty. An alternative data set, generated by the Ministry of Agriculture (MOA), is discussed in Chapter 2. The MOA data also measure rural household incomes, but results in a somewhat different picture of poverty than that obtained from the SSB data. This appendix discusses how the two data collection systems function and how the resulting data are used. It then briefly contrasts the resulting household income estimates, with particular attention to systematic differences between estimates from the two sources. 2. MOA Income Data. The MOA rural household data have a longer and more consistent history than do the SSB data. MOA has collected a variety of production and income data for agricultural areas for over forty years. During the commune era, data reporting requirements included average per capita distributed grain and cash, figures easily derived from the accounting needed to make those distributions in the first place. Other elements of household income typically recorded included household production for the market and the collective distribution of important consumption items such as fuel. The dissolution of the communes did not mark the end of household data collection efforts. Village accountants, reporting to the Economic Management Stations of the MOA, continue to record, by household, production of major crops and income earned from various non-crop economic activities. ! 3. Field interviews with village accountants reveal a wide range of quality in the recorded Income estimates. All accountants are required to report their findings quarterly through the system, which leads from village accountants through the administrative village to the township and thence to the county, provincial and national offices. Not all natural villages have accountants, but all administrative villages do have them. Each accountant maintains a record for each household, on which are recorded the required information on crop yields, aggregate input and output transactions, and income from other sources. At its simplest, the village accountant asks each household each quarter what the appropriate figure for each line of the report should be, noting down responses. In the most sophisticated form witnessed, the township runs sample surveys of households to double check reported production, purchase and sales figures, and uses those results as audit guides for household self reporting. Either way, household recall is the primary tool for data collection. Because these data flow through the local administrative system, they are subject to local political influence, something that has occasionally been a problem in the past. F The resulting figures are assembled at the Economic Management Central Station (Jingying Guanli Zhongzhan). 153 Annex 1 ADpendix 1 4. The MOA survey concentrates on major agricultural outputs (broadly defined, in the Chinese system, to include outputs from animal husbandry, forestry, fisheries, and household handicraft production in addition to crop cultivation), making no attempt to capture a wide variety of by-products and minor products. The value of the latter are captured if marketed and reported, but are otherwise ignored. In the same fashion, labor income from village or township enterprises will be accurately recorded, but private labor income will be recorded only to the extent it is accurately reported. 5. The MOA system has the virtue of complete coverage of all Chinese villages. However, as far as is known, village accounts do not include current residents with registration in other villages, nor do the home villages of out migrants record their income except insofar as it is remitted home. Because of this, incomes earned by long term migrants will not be captured by the MOA reporting system, although that of short term migrants is, in theory, recorded when they return to their home villages. 6. MOA data are used for a variety of purposes, including some that have a direct bearing on poverty alleviation programs. When the initial selection of counties for the special national poverty alleviation program was being made in 1986, counties were ranked in terms of their MOA-recorded income, and that income status was a major element in the selection process. The SSB itself uses MOA data to stratify by income the rural county universe before selecting counties to include in their rural income survey. And, for the many counties, townships and villages not covered by the SSB income survey, the MOA survey results remain the means by which they define their economic performance. The MOA data therefore continue to play an important role in China. 7. SSB Income Data. The SSB conducted rural household income surveys sporadically prior to 1966, and not at all between 1966 and 1978. In 1979 a survey was conducted retrospectively for 1978 and concurrently for 1979, using 6,095 households for the 1978 sample. The sample expanded rapidly until 1984, when it reached 31,375 households. In 1985, the sample van completely redrawn and doubled in 8ize to 66,642 households. That panel was left unchanged, except for some very minor additions, until the end of 1989, when there was a complete rotation for six provinces, with the remaining provinces rotating at the end of 1990. By the latter year the sample totalled 66,478 households in about 700 counties. Current intentions are to replace 25 percent of the sample annually beginning in late 1991, while maintaining the overall sample at about the present size. 8. The SSB household sample is drawn through a multistage process that includes every province, with larger provinces selecting a smaller proportion of counties for inclusion in the sample. Within a province, counties are stratified by average income, as determined by the MOA system, with selection intervals determined by population. Within counties, the village or township are variously the next stage of selection, again based on income and 154 Annex 1 A pendix 1 population. Finally, households are chosen after being ranked by MOA income. 2 9. The SSB survey technique requests that the household keep a daily diary of economic activity, with any family member seeking temporary outside employment carrying an individual diary. Village statistical assistants check frequently to see that the diaries are maintained, and collate information for the county statistical office monthly. The SSB attempts more thorough and timely coverage of income sources, In cash or in kind, than does NOA. In particular, byproducts such as crop stalks are recorded in the SSB approach and private and collective income sources are accorded equal weight in the collection effort. 10. The SSB panel suffered somewhat from aging during 1985-1989, but the new approach of continuous replacement will avoid that phenomenon in the future. By all assessments, the panel is well drawn, although it may underrepresent the poor. Successful participation in the survey demands literacy and numeracy, not qualities often found among the very poorest. Sample selection rules insist that literacy no be consider in household selection, and the village assistant has the task of filling in forms for illiterate households. However, county level SSB officials have mentioned that households with no literate members may in practice be excluded from the sample. 11. The SSB income returns are used to develop provincial average rural income comparisons used in national analysis. The figures are also used to derive provincial and national income distributions and to analyze changes in family composition, income by source, and a host of other economic issues. The information from the surveys moves through an independently funded SSB reporting system, bypassing the local administrative network. Lo' Many provinces have augmented the centrally-funded sampling effort by extending the surveys to every county within the province. The results from those additional counties are not reported through the usual system, but rather remain for local use and guidance. 12. A Comparison of the MOA and SSB Systems. The most immediate comparison between the two systems is in their coverage. Both systems will miss long term in or out migrants, but the MOA system is much more complete, without any danger of missing household because they are illiterate. The price for that completeness is a much lower quality of coverage. Not only does MOA ignore outputs such as crop byproducts, the danger is great that 2' See World Bank, "China: Statistical system in transition" Report No. 9557- CHA, June 5, 1991, for a detailed discussion of sampling techniques and history. 1 Although this alternative routing is important in maintaining independence, the personnel of the local statistical bureaus are part of county government and not entirely free of possible influence. The system does appear to be well designed and relatively free of administrative bias. 155 Annex 1 Appendix 1 private non-agricultural enterprise activities will be missed because of the way income is reported. 13. A test has been done to assess the bias in HOA income recording. For a small number of years both the MOA and SSB provincial average income levels are available and can be used to look for systematic reporting differences. In fact, such differences come out very strongly and show a consistent pattern of change over time. Below, by year, are the estimating equations for deriving MOA provincial average income estimates from SSB estimates: 1983: MOA = -87.9 + 1.04(SSB) 1984: MOA = -59.4 + 1.00(SSB) 1985: MOA = -26.9 + 0.94(SSB) 1986: MOA = - 4.3 + 0.91(SSB) Source: Derived from MOA (1991) and SSB (1991a). In every case the R2 is above 0.90 and all coefficients have very high t ratios. The MOA income data have always underestimated total income and the nature of that underestimation has changed over time. In 1983-1984 the underreporting was essentially invariant to income level, but by 1986 it showed a strong correlation with income level. Possible explanations include the rapidly growing importance of private enterprise, which peasants would be reluctant to fully report because of possible tax liabilities, and off-farm employment, which they would also see no virtue in reporting. Both of those income sources appear to be proportional to total income, which would yield the pattern of change shown above. 14. Considerations of quality and completeness make the SSB income estimates preferable to the MOA when they can be obtained, and has led to their use in this report. MOA data continue to have some value when used with care and the recognition of major sources of bias. Distributors of World Bank Publications ARCENTINA ELSALVADOR JAPAN SOUTH AFiCA. BOTSWANA Caoes Hirsch, SRL Fusades Eastern Book Service For aijgle titii Galeria Guemes Alam Dr. Manuel Enrique Araujo #3530 Hongo 3-Ceome, Bunkyo-ku 113 Oxford University Pm Florida 165.4th Floor-Ofc. 453/465 Edifico SISA, ler. Piso Tokyo Southern Africa 1333 Buenos Aires San Salvador 011 PO. Box 1141 KENYA Cape Town 8000 AUSTRALIA, PAPUA NEW GUINEA, FINLAND Africa Book Service (EA) Ltd. FIJI, SOLOMON ISLANDS, Alateeminen Kirjakauppa Quaran House. Mfanpno Set For subcr*tiesordeis: VANUATU, AND WESTERN SAMOA P.O. Box 128 P.O. Box 45245 Internationai Subscription Service D.A. Books & Journals SF-00l01 Helsinki 10 Nairobi P.O. Box 41095 648 Whitehorse Road Craighull Mitcham 3132 FRANCE KOREA. REPUBLIC OF Johartnesug 2024 Victoria World Bank Publications Pan Korea Book Corporaton 66. avenue d'I4na P.O. Box 101. Kwangwhamun SPAIN AUSTRIA 75116 Paris Seoul Mundi-Prensa libro, SA Geroid and Co. Castello 37 Graben 31 GERMANY MALAYSIA 29001 Madrid A-10l1 Wien UNO-Vertag University of Malaya Cooperative Poppelsdorfer Allee 55 Bookshop, Limited Librerla Internatcional AEDOS BANGLADESH D-5300 Bonn I P.O. Box 1127, JaLan Pantai Baru Conseil de Cent. 391 Micro Industries Development 59700 Kuala Lumpur 08009 Barcelona Assistance Society (MIDAS) GUATEMALA House5, Road 16 Librerias Piedra Santa MEXiCO SRI LANKA AND THE MALDIVES Dhanmondi R/Area Sa. CaUe 7-SS INFOTEC Lake House Bookshop Dhaka 1209 Zona I Apartado Postal 22-860 P.O. Box 244 Guatmla City 14060 flalpan, Mexico D.F. 100, Sir Chitimapalam A. Br*h offi= Gardiner hMwatha 156, Nur Ahumed Sarak HONG KONG, MACAO NETHERLANDS Coloinbo 2 Chittagong 4000 Asia 2000 Ltd. Di Undeboom/InOr-Publikaties 4648 Wyndham Street P.O. Box 202 SWEDEN 76. KDA Avenue Winning Centre 7480 AE Haaksberger For single titia Kulna 9100 2nd Floor Fritzes Fackboksforetaget Central Hong Kong NEW ZEALAND Regeringsgaten 12. Box 16356 BELGIUM EBSCO NZ Ltd. S-10327Stockholm Jein De Lannoy INDIA Private Mail Bag 99914 Av. du Roi 202 Alied Publishers Private Ltd. New Market For nrtet aofrs, 1060 Brussda 751 Mount Road Auckland Wennergren-Williams AB Madras- 600 002 Box30D04 CANADA NIGERIA 5-104 25 Stockhoim Le DiffHseur Brnch ofle University Press Limited CrP. 85, 1501B rue Ampere 15 J.N. Heredia Marg Three Crowns Building Jericho SWITZERLAND Boucherville, Qubec Ballard Estate Private Mail Bag S095 For single tilkl J4BSE6 Bombay -400038 iadan Libi-ae Payot 1,rue de Bourg CHiNA 13/14Asaf Al Road NORWAY C rI10o Lausa China Financial & Economic New Delhi -110 002 Narveses informauon Center Publihingr Houxe Book Department For svMcrtin orde: 8, DaFo Si Dong le 17C]%itarajan Avenue P.O. Box 612 Ettead Libairie Payot Beijing Calcutta - 700 072 N-0602 Oslo 6 Service des Abonnements Ceam postaLle 3312 COLOMBIA jayadeva Hoste Building PAKISTAN CH 1002 LAUs ne Infoenilace Ltda. 5th Main Road Gandhinagar Mirza Bwok Agency Apartado Aereo 34270 Bangalore-560009 65, Shahrah-e-Quaid-e-Azam TANZANIA Bo"h D.E. P.O. Box No 729 Oxford Univer.ty Pren 3-511129 Kachiguda Cross Road Lahone 54000 P.O. Box 5299 COTE DiVOIRE Hyderabad-500027 Maktaba Road Centre d'Edition et de Diffusion PERU Dar es Salaam Africaines (CEDA) Prarthana Flats, 2nd Floor Editorial Desrrolo SA 04 B.P. 541 Near Thakore Baug Navrangpura Apartado3824 THAILAND Abid jan 04 Plateau Ahmedabd -380 009 Lima I Centrei Department Store 306 Shorn Road CYPRUS Patiala House PHiUPPINES Bangkok Cyprus College Bookstore 16-A Ashok Marg internationl Book Center 6, Diogens Street Engomi Lucknow - 226 001 Fifth Floor, Fiipinas Life Building TRINIDAD & TOBAGO, ANTIGUA P.O. Box 2006 Ayala Avenue. Makatd BARBUDA, BARBADOS, Nicos Centmi Bazaar Road Metro Manila DOMINICA, GRENADA. GUYANA. 60 Bajal Napar JAMAICA, MONTSERRAT, ST. DENMARK Nagpur440010 POLAND KITS & NEVIS. ST. LUCiA. SamfundsLitteratur ORPAN ST. VINCENT & GRENADINES Rosenoerrs Ale6 11 INDONESIA PaLac Kultury i Nauld Sytematic Studie Unit DK-1970 FrederiksbergC Pt Indira Limited 00.901 Warzawa #9 Watts Stret Jl Sam Ratulangi 37 Curepe DOMINICAN REPUBUC P.O. Box 18 PORTUGAL Trinidad, West Indie Editora Taller, C por A. Jakarta Pusat Livrania Portuga Resturaci6rn e Isbela Catlika 309 Rua Do Carmo 70.74 UNITD KINCDOM Apartado de Correos 2190 Z-1 ISRAEL 1200 Lisbon Microinfo Ltd. Santo Domingo Yozmot ULteture Ltd. P.O. Box3 P.O. Box 56055 SAUDI ARABIA. QATAR Alto Hampshire GU34 2PG ECYPT, ARAB REPUBUC OF Tel Aviv 61560 larir Book Store Englad Al Ahram Irsrel P.O. Box 3196 Al G&Laa Street Riyadh 11471 VENEZUELA Cairo ITALY Librana del Este Licosa Comnmimionaia Sansoni SPA SINGAPORE, TAIWAN, Aptdo. 6037 TheMiddle astObsever Via Duca Di Calabria, 1/1 MYANMAR,BRUNEI Caracas 1060-A 41. Sherif Street Casella Postale 5S2 Information Publiaos Cair 50125 Frenze Private, Ltd. 02-061st Flt, Pei-Fu Industria Bldg. 24 New Industrial Road Singapore 1953 F A K~~~~~~~ ~ ~ ~~~~~~~~~~~~~~ ~~~~~~~ U S S IA R UI 5 S I A KAZAKHSTAN |?t _/ SQ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~~~~ / ~~~~~ HEILONG IANC7 ) 1000 0 00000;; e | X NOoL,A X WX~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~83 4 ,,,j,X,87K ,, g- <_/P_ = ~~MOGOLI KYR U, |''- / 9 1Ni : 'y,1'0>0;0s UZBEKISTA~NOrA;\) 0 ,/'0: 18E.A 711 1 > , R-@ 0 000; t 1 19891NCIDENCEOF RURs0LPOVEECYI 51 ; ; A A ; L V n *; 6 ;> S 6h ua , l~~~~~~~~~~~~~~~~~/~ L6 I PERCENT O U AL OUSE OLDS EELOW 1 K ; ; I~~~~~~~~~~~~~~~~~~~RE I E=l ~~A17 NISTAnNAL UNDES1d RUIN? i L ue O . d, d, ,. IEEORR\400 0 ; W0iA f f 0 0 SX f f () DX 1 SS~~~~~1 REP O \ ~~~~~~~~~~~~~~R U S S I A sXR U S S I A KAZAKHSTAN J' MONGOLIA % , 0 >,KYRG HYZSA Urumi '-r {~~~~~~~~~~~~~~~ \s_ _/ ~~~~~~~~~~~~~~~~~~~~~~~~~Sh.neg ,,,; JAPAN| D. RJA (~FCiHlSTAN 0 - tHloun{Hhht / OF, |KOREA 1 11 .! -''f'0V', ,,004 T-''Sa"';,< ,' b:''d':\'',,,' I~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~1-'' "REP.OF ~~ ~ /' '''' ~~ KOREA PNEAL ' 11.0A CHINA . ? F 1989 INCIDENCE OF RURAL POVERTY(MOA DATA) PERCENT OF RURAL POPULATION RESIDING IN PACIFIC LOWER-INCOME COUNTIES 4iI K- i~~~~~~~~~~~~ (~~~~~TAW N Xz N |___ z1 15% (D PROVNCE CAPITALS YUNAN -UNGX| 5- IO'% @ AONAL CAPITAL _l 100% a=FPROMNCE BOUNDARIES ____ 1 >INTERNATIONAL BOUNDARIES A S s~~~~~~LO~MI S o 200 IO 600 oo ~ _D .R r?lo 3t/PHILIPPINES - As o O EO' Y MO s LA A I I / K)~~~~~~~~ The World Bank Headquarters European Office Tokyo Office 1818 H Street, N.W. 66, avenue d'Iena Kokusai Building Washington, D.C. 20433, U.S.A. - 75116 Paris, France 1-1 Marunouchi 3-chome Chiyoda-ku, Tokyo 100, Japan Telephone: (202) 477-1234 Telephone: (1) 40.69.30.00 Facsimile: (202) 477-6391 Facsimile: (1) 40.69.30.66 Telephone: (3) 3214-5001 Telex: wui 64145wORLDBANK Telex: 640651 Facsimile: (3) 3214-3657 RCA 248423 WORLDBK Telex: 26838 Cable Address: IN7BAFRAD WASHINGTONDC - Do ' ..'~