DISCUSSION PAPER NO. 1423 90030 Any Guarantees? China’s Rural Minimum Living Standard Guarantee Program Jennifer Golan, Terry Sicular and Nithin Umapathi August 2014 Any Guarantees? China’s Rural Minimum Living Standard Guarantee Program Jennifer Golan Terry Sicular Nithin Umapathi The University of Manchester The University of Western The World Bank Ontario August 2014 We are grateful to Luo Chuliang, Wang Dewen, Philip O’Keefe, Song Jin, and Reena Badiani for their suggestions and input. 1 Abstract This paper examines China’s rural minimum living standard guarantee (dibao) program, one of the largest targeted transfer schemes in the world. Using household survey data matched with published administrative data, we provide background on the patterns of inequality and poverty in rural China, describe the dibao program, estimate the program’s impact on poverty, and carry out targeting analysis. We find that the program provides sufficient income to poor beneficiaries but does not substantially reduce the overall level of poverty, in part because the number of beneficiaries is small relative to the number of poor. Conventional targeting analysis reveals rather large inclusionary and exclusionary targeting errors; propensity score targeting analysis yields smaller but still large targeting errors. Simulations of possible reforms to the dibao program indicate that expanding coverage can potentially yield greater poverty reduction than increasing transfer amounts. In addition, replacing locally diverse dibao lines with a nationally uniform dibao threshold could in theory reduce poverty. The potential gains in poverty reduction, however, depend on the effectiveness of targeting. JEL Classification: I38, O15 Keywords: Rural poverty, cash transfers, targeting, China 2 Contents I. Introduction ...................................................................................................................................... 4 II. Background on China’s rural dibao program .................................................................................... 7 III. Data ................................................................................................................................................. 11 IV. Patterns of income inequality and poverty in rural China, 2007-09 ............................................... 15 V. Patterns of dibao participation, thresholds and transfers.............................................................. 17 VI. Impact of dibao transfers on incomes and poverty ........................................................................ 20 VII. Conventional analysis of dibao targeting........................................................................................ 22 VIII. Correlates of dibao participation and propensity score analysis of dibao targeting...................... 24 IX. Policy simulations: Expand Coverage versus Increase Transfer Amounts .................................... 27 X. Policy simulation: Nationally uniform transfer and threshold....................................................... 30 XI. Conclusions...................................................................................................................................... 33 XII. References....................................................................................................................................... 36 XIII. Figures ............................................................................................................................................ 39 XIV. Tables .............................................................................................................................................. 43 3 I. Introduction China’s economic reforms have brought substantial growth in rural incomes, but have been accompanied by a substantial weakening of public goods provision and the social safety net in rural areas. Since the late 1990s China’s central government has pursued a multi-pronged effort to rebuild rural social programs. Relevant measures have included the new rural cooperative medical system, the expansion of universal, free nine-year education in rural areas, and the minimum living standard guarantee or dibao program (Lin and Wong 2012, World Bank 2009). The last of these—the rural dibao program—is the focus of this study. The stated aim of the rural dibao program is to provide income transfers to households with income per capita below an income threshold. The transfers are intended to bring the recipients’ incomes up to the threshold. The threshold and transfer amounts are determined locally in light of local conditions. The government’s adoption of this approach to poverty alleviation was motivated by the changing structure of poverty in rural China. During the 1980s and 1990s the overall incidence of poverty in rural China declined substantially, poverty became more dispersed geographically, and transitory poverty emerged as an important issue (World Bank 2009, World Bank Social Protection Group 2010). In contrast to China’s earlier “poor area” poverty alleviation programs, which targeted localities and communities, the dibao program targets households and individuals wherever they reside and provides transfers based on income shortfalls. Thus, it is well suited to the new environment. Experiments with dibao programs began in the 1990s, and China’s rural dibao program was adopted nationwide in 2007. By 2010 its coverage exceeded 50 million people, making it one of the largest social relief programs in the world. Program expenditures are also substantial, in 2011 equivalent to 0.14% of GDP and 0.6% of total government expenditures. Despite its size, little is known about the program’s performance. Several reports have provided insightful descriptive analyses and preliminary evaluations of the program’s successes and challenges (World Bank Social Protection Group 2010, World Bank 2011; Luo and Sicular 2013). To our knowledge, there has been no systematic analysis of the rural dibao program’s 4 benefit incidence and impact on poverty reduction since 2007, when the program was rolled out nationwide. The literature on poverty program evaluation in developing countries is extensive, as is the debate regarding appropriate methodologies (see, for example, Deaton 2010 and Ravallion 2008). A central focus of this literature is how to address empirical issues that arise due to selection bias and due to the behavioral responses of program participants. These concerns are relevant to China’s dibao program, but are not the focus of this paper. In view of the lack of basic information about China’s rural dibao program, our goal is description and basic analyses using well-known empirical methods that can inform policy. Our work follows in the footsteps of recent analyses of China’s urban dibao program (Chen, Ravallion and Wang 2006; Gao, Garfinkel and Zhai 2009; Wang 2007; Ravallion 2008), with some differences in approach reflecting differences between the urban and rural programs as well as data availability. Our analysis makes use of household-level data from the China Household Income Project (CHIP) surveys, matched with administrative data on the dibao program from the Ministry of Civil Affairs (MOCA), for the years 2007-2009. Our central finding is that in practice China’s rural dibao program provides substantial income benefits to program beneficiaries, bringing many low-income beneficiaries above the dibao income thresholds and also out of poverty. Nevertheless, due to limited coverage relative to the large total number of rural poor in China, as well as high exclusion and inclusion errors, its effect on poverty reduction has been small. The overall impact of the rural dibao program is thus less than expected given the program’s design and scale. Our findings suggest that although the dibao benefits are adequate, improvements are needed in coverage and targeting. These conclusions emerge both from a conventional targeting analysis using household incomes as the evaluation criterion and from an alternative, propensity score approach. In settings such as rural China where measurement of household income is difficult, administrators of conditional transfer programs often rely on observable correlates of income to determine eligibility. Even in China’s urban areas, where income is more likely to be in the form of salaries and wages and so easier to observe, measurement errors can arise (Chen, Ravallion and Wang 2006). In their evaluation of the urban dibao program, Chen, Ravallion and 5 Wang (2006) suggest use of a propensity score approach that evaluates the program’s performance based on the sorts of income proxies that are likely used by local officials carrying out the program. We adopt this approach to analyze China’s rural dibao program. Although the propensity score analysis reduces the magnitude of exclusion and inclusion errors, the targeting errors remain large. Our findings raise questions about whether changes in the rural dibao program might increase its impact on poverty reduction. The government has, in fact, further expanded the dibao program since 2009. We therefore carry out simulations that explore the impact of increasing the dibao budget from its observed level in the 2009 CHIP data by (a) expanding the number of beneficiaries without changing the transfer amounts, and (b) doubling the transfer amounts without increasing the number of beneficiaries. These simulations assume that, aside from changes in the transfer amounts and number of beneficiaries, other aspects of the program are unchanged. The results indicate that expanding coverage has the potential to yield greater reductions in poverty than increasing transfer amounts. In actual practice, the dibao thresholds and transfer amounts are set locally at the county level and are correlated with local fiscal capacity. Consequently, poor counties tend to have lower dibao thresholds and transfers than do rich counties, with implications for targeting and the poverty impact of the program. We construct several simulations to investigate the impact of adopting a uniform nationwide dibao threshold combined and a uniform nationwide transfer amount. The results of these simulations indicate that adopting uniform transfer amounts in the context of the existing system would likely have little poverty impact. A uniform transfer would be beneficial only if inclusionary targeting error is reduced. Shifting to a nationally uniform eligibility threshold has the potential to substantially reduce poverty, but again depending on targeting performance. We begin in the next section with an overview of the rural dibao program and discussion of some relevant literature. Section III describes the data. Section IV provides background on overall trends in rural inequality and poverty in China. Section V describes patterns of dibao participation, thresholds, and transfers in the data. Section VI examines whether dibao transfers bring recipient households above the dibao thresholds and out of poverty. Section VII 6 analyzes the targeting effectiveness of the program using conventional targeting analysis. Section VIII examines the characteristics of dibao and nondibao households and reports the results of probit analyses that identify the characteristics associated with program participation. In this section we also discuss the results of a propensity score analysis of dibao targeting. Sections IX and X discuss the policy simulations. We conclude with a recap of our major findings and implications for policy and future research. II. Background on China’s rural dibao program China’s rural dibao program is modeled after the urban dibao program, which began i n the early 1990s on an experimental basis in some cities. In 1999 the State Council implemented the urban dibao program in all cities nationwide. Participation in recent years has stabilized at about 22 to 23 million urban individuals (Chen, Ravallion and Wang 2006, O’Keefe 2004, Ministry of Civil Affairs 2011). Experiments with rural dibao began in the 1990s, mainly in more developed areas. By the early 2000s rural dibao programs were fairly widespread, but they relied on local funding and, due to differences in local fiscal capacity, varied across counties in terms of the level of support and criteria for eligibility. In 2004 the central government called for the rural dibao program to expand and began to provide funding for the program in poor areas; by the end of 2006 roughly 80 percent of the provinces and counties in China had adopted some form of rural dibao program (Ministry of Civil Affairs 2007, World Bank Social Protection Group 2010, Xu and Zhang 2010). In early 2007 the central government announced that the rural dibao program was to be implemented nationwide in all counties and with central subsidies (Xinhua 2007a, 2007b; World Bank Social Protection Group 2010; Xu and Zhang 2010). Under this new initiative, the program would become more standardized and would absorb or complement several pre- existing programs that had provided subsidies for poor households such as the five-guarantee (wubao) program and the subsidy program for destitute households ( tekun jiuzhu). Although 7 central funding of the program increased, the minimum income thresholds and subsidy amounts continued to be set locally at the county level in light of local fiscal capacity. Official statistics indicate that the rural dibao program grew quickly after 2006 (Table 1). In 2007, the first year of nationwide implementation, the rural dibao program provided transfers to 35.7 million rural individuals (4.9% of the rural population) and accounted for three-quarters of the rural recipients of social relief, followed in a far second place by the five-guarantee program, which covered 5 million recipients (Department of Social, Science and Technology Statistics of the National Bureau of Statistics 2008, p. 330; National Bureau of Statistics 2009, pp. 89, 939). By 2010-11 program participation had leveled off at about 53 million individuals, equivalent to 8% of the rural population. This is more than double the size of the urban dibao program (23 million), and it far outnumbers the sum total of participants in all other rural poverty relief programs (17.9 million in 2010; does not include disaster relief) (Ministry of Civil Affairs 2011; National Bureau of Statistics 2011). Spending on the program has grown apace (Table 1). According to official statistics, in 2007 total spending on the rural dibao program was 23 billion yuan, with an average transfer amount of 1,210 yuan per recipient per year. In 2011 total spending on the rural dibao program was 67 billion yuan or, on average, 1,258 yuan per recipient per year, an amount equivalent to more than half of the official poverty line in that year (2,300 yuan). In view of the diversity of China’s rural economy and the difficulty of measuring income for rural households, it is not surprising that the program’s implementatio n has varied among localities and evolved over time. Local variation and flexibility was explicitly built into the central dibao policy regulations (Poverty Alleviation Office of the State Council 2010). Reports based on fieldwork provide insights into how the program has worked on the ground. According to reports based on fieldwork from the World Bank (World Bank Social Protection Group 2010, World Bank 2011), variation exists in the extent to which applications are open versus by invitation of local officials. In practice village leaders often identify potential beneficiaries and invite them to apply. Village committees, which include village leaders and other community members, play a central role in identifying and screening potential beneficiaries. Members of village committees live in close proximity to and have local 8 knowledge of potential beneficiary households. Applications or nominations for dibao benefits are submitted to the township government and forwarded to the county Department of Civil Affairs. Decisions are made by township and county officials, who review the documentary evidence submitted by households and villages, and who sometimes visit the households to check on, or to collect additional, information. The names of applicants are, in principle, made public in the villages and are subject to community review and feedback. National policy permits, and local officials in practice make use of, a range of information to evaluate eligibility. This might include information about household income, assets, and housing conditions, as well as the presence of household members who are able or unable to work, or of illness or disability (Poverty Alleviation Office of the State Council 2010; World Bank Social Protection Group 2010; World Bank 2011). In principle the dibao program tops up the income of recipients to the level of the local dibao threshold. The amount of the dibao benefit, then, should depend on the level of the dibao threshold and the level of a household’s per capita income. As will be discussed in more detail later, dibao thresholds vary substantially among provinces and counties. Practices regarding how to determine the amount of the benefit also vary. In some areas local officials estimate the gap between the household’s income and the local dibao threshold and decide on the benefit accordingly. Due to difficulties accurately measuring income, most localities use other approaches. The 2007 national policy allowed local officials to classify households in tiers according to their apparent level of poverty and to set fixed benefit amounts associated with each tier. This tier-classification approach appears to have been widely used (World Bank Social Protection Group 2010). Several reports have noted that although the flexible design of the dibao implementation policy has advantages, it gives officials at the county, township and village levels considerable discretionary power. The program does not appear to have well-functioning checks and balances, in part because of limited resources at the local level for administration of the program. These characteristics of the program create the potential for irregularities (World Bank 2011). In the Chinese-language media reports of dibao irregularities are numerous, so much so that they have been classified into standard categories: giving dibao on the basis of 9 connections or personal relationships (guanxi bao, renqing bao), cheating (pian bao), and mistakes (cuo bao). An internet search using Baidu yielded many reports of irregularities in multiple localities, including a widely discussed case of dibao corruption in Fang County, Hubei, as well as cases in Shaanxi, Shandong and Guangxi. Problems with the dibao program are of concern to China’s central leadership and policy circles. In 2012 He Guoqiang, a member of the Politburo Standing Committee and Secretary of the Central Commission for Discipline Inspection, made a speech about the problem of corruption in China that explicitly mentioned corruption in the dibao program, which he referred to using the phrase “a tide of unhealthy practices in urban and rural dibao ( chengxiang dibao zhongde buzheng zhi feng)” (Zhu Wurong 2012). He outlined major reasons for these problems: “first, local village and township cadres don’t do their jobs, they don’t go out to the villages and meet with the people, don’t really understand and grasp which are the households in difficulty; second, dibao work is not sufficiently transparent and open; and third, a few village and township cadres are selfish and looking out for their own benefit, and they give dibao benefits to relatives, friends, or even themselves.” The Ministry of Civil Affairs has openly acknowledged the existence of such irregularities and called for improvements in dibao work. A recent news report published comments by the Minister of Civil Affairs regarding the findings of an internal review of the dibao program. The Minister reported that the review found cases of cheating, mistakes, and awards based on connections, but concluded that the overall incidence of such problems is relatively small. The internal review estimated that the rate of incorrect/mistaken dibao benefits was 4% (Xinhuanet 2013). The basis of this estimate is not explained. To address problems in dibao implementation, in early 2013 the Ministry of Civil Affairs announced some new policies that were to be adopted nationwide. The new policies include (1) allowing households to apply for dibao benefits directly to the county Department of Civil Affairs rather than having to go through the village and township levels, (2) requiring that county-level officials visit and check at least 30% of applications, (3) instituting a filing and auditing system for close relatives of local officials and village leaders involved in dibao implementation, (4) establishing and improving systems for community feedback, and (5) 10 establishing a systematic mechanism for checking information on dibao applications against information in other departments, e.g., vehicle registration data and savings account information (Xinhuanet 2013). These sorts of reports reveal divergence between policies and implementation. Although it is difficult to know exactly the extent of such divergence, the reports raise questions about the rural dibao program’s performance, targeting, and impact on poverty. III. Data For our analysis we use two types of data. First, we use rural household survey data for the years 2007, 2008 and 2009 collected by the China Household Income Project (CHIP) in conjunction with the Rural Urban Migration in China (RUMiC) project. Hereafter we will refer to these as the CHIP data. During the years covered by the CHIP data the rural dibao program expanded rapidly nationwide. As of 2009, coverage was about 90% of the program’s level at full implementation of 53-54 million, which was attained after 2010. Second, we use administrative data published by the Ministry of Civil Affairs (MOCA) on rural dibao thresholds, transfers and expenditures. The MOCA data are available at the county level. We use the MOCA data for counties covered in the CHIP survey to create a matched dataset. There are 82 counties covered in the CHIP rural survey and for 77 we are able to match county-level information from MOCA. The CHIP rural survey sample is a panel of about 8000 rural households containing 30,000 individuals in nine provinces (Hebei, Jiangsu, Zhejiang, Anhui, Henan, Hubei, Guangdong, Chongqing and Sichuan). These nine provinces cover nearly half of China’s total population and span China’s eastern, central and western regions. Table 2 shows the sample size for each year and gives information on the panel aspect of the dataset. Ninety-eight percent of households and ninety-three percent of individuals in the sample are present in the dataset for all three years. In this paper we do not exploit the panel aspect of the dataset, but we plan to do so in future work. A detailed description of the CHIP sample can be found in Li, Sato and Sicular (2013). Here we highlight key features relevant to our analysis. The CHIP sample is a subset of the 11 National Bureau of Statistics (NBS) annual rural household survey sample, which covers 68,000 households in all 31 provinces. Like the larger NBS rural sample from which it is drawn, the CHIP sample is representative at the provincial level. CHIP’s provincial sample sizes are not proportional to the provincial populations. For this reason, and also because of the deliberate selection of provinces covered by CHIP so as to represent China’s three major regions (eastern, central, western), for most analyses we use two-level weights reflecting the provincial and regional populations. Weights are constructed using population statistics from China’s annual 1% population sample surveys (NBS, various years). The nine provinces in the 2007-09 CHIP sample exclude the Northeast and China’s autonomous regions in the Northwest and Southwest. These autonomous regions contain relatively high concentrations of the poor, which may explain in part why the CHIP dataset has lower poverty rates than the full NBS sample. Based on the 2009 official poverty line and the full NBS national rural household survey data for 2009, China’s poverty rate was 4.7%; using the same poverty threshold and (weighted) CHIP rural data, the poverty rate is 3.2%.1 The nine provinces covered in the CHIP sample also have lower concentrations of dibao participants than is the case nationwide according to the official data. In 2009 the nine provinces covered by the CHIP rural sample contained 47% of China’s rural population but only 38% of China’s rural dibao recipients.2 Nevertheless, the mean values of key variables such as income are similar to those in the full NBS sample (Table 2; Li, Sato and Sicular 2013). Thus, with careful interpretation in light of sample coverage, the CHIP data provide a reasonable approximation of the situation in much of China. The CHIP dataset contains detailed information on incomes, consumption, household composition and demographics, and many other (but not all) variables collected by the NBS as 1 These estimates were kindly provided by Luo Chuliang. Note that these poverty rates are calculated using the 2009 official poverty line, which is lower than the 2011 official poverty line that we use to calculate estimates reported in the next section. 2 Population data from NBS (various years). Provincial and national rural dibao data are for the month of December, 2009, and are published on the Ministry of Civil Affairs website. Note that in December 2008 the nine provinces contained 36% of rural dibao recipients in China. See http://files.mca.gov.cn/cws/201001/20100128094132409.htm and http://cws.mca.gov.cn/accessory/200905/1243323064255.htm, accessed December 31, 2012. 12 part of its annual rural household survey. Additional information about the households was collected using an independent questionnaire designed by the researchers associated with the CHIP and RUMiC. The dataset contains matching community-level data gathered through a village survey. The availability of rich information at the individual, household and village levels provides a unique resource for our analysis. The income data were collected using a diary method. Although the diary method reduces recall error, the income data contain some unknown degree of measurement error. Error could arise due to difficulties keeping track of the complex and diverse income sources in rural China, which include farming, nonagricultural self-employment, formal wage employment, and informal or casual jobs, and which generate incomes both in cash and in kind. Error could also arise due differences in the ability and willingness of respondents to record accurate data in the diaries. The CHIP datasets contain information on household participation in the dibao and wubao programs. Participation is self-reported. In our analyses we treat households that indicated participation in either the dibao or wubao programs as dibao households and their members as dibao participants, because the distinction between the two programs is not always clear at the local level and because during the time frame of our analysis the wubao program was to some extent being absorbed by the dibao program (World Bank Social Protection Group 2010). Table 2 shows the number of dibao (including wubao) households and individuals in the CHIP datasets. The numbers of dibao households and individuals increase markedly over the three years, reflecting the expansion of the program during this time frame. The numbers of dibao households and individuals are adequate for analysis at the national level, but with disaggregation the numbers quickly become too small. Consequently, our analysis is carried out primarily at the national level. In order to evaluate the dibao program’s targeting performance and poverty impacts, we need to estimate the “ex ante” or counterfactual level of income that households would have had in the absence of the dibao transfers. Here we estimate ex ante income as equal to reported or “ex post” income minus the amount of dibao transfers received by the household. 13 This approach assumes that households that the dibao transfers do not change household behavior. It is widely recognized that households that receive transfers are likely to alter their behavior, for example, by reducing effort to earn income. If this is the case for rural dibao recipient households, our estimates of ex ante income will understate the true counterfactual income that households would have had in the absence of the transfer. Consequently, our estimates of ex ante income are likely to be too low, thus exaggerating the difference between ex post and ex ante incomes and leading to overstatement of the impact of the dibao program on incomes and on poverty. As shall be seen in later sections, we find that despite this possible overstatement, the impact of the dibao program on poverty rates is relatively small. The CHIP household survey data contain ex post incomes, but unfortunately they do not contain information on the amounts of dibao transfers received by the households. 3 Information about dibao transfers is, however, available at the village and county levels. The CHIP village-level data contain information for 2008 and 2009 on the number of dibao and wubao households within the village and on the average dibao transfer per recipient within the village. Also, MOCA publishes county-level data on rural dibao participation and expenditures, which can be used to calculate county average dibao expenditures per recipient.4 It is possible that county expenditures include some categories of government spending on the dibao program other than the dibao transfers to households; as discussed later, however, the county average dibao expenditures are quite similar to the village average transfers. We use the local village average dibao transfers and county average dibao expenditures amounts as proxies for household level dibao transfers. In this way we obtain two estimates of ex ante income for dibao households: one is equal to ex post household income per capita 3 The data contain information on the total transfer income received by the households, including both private and public transfers, but without any breakdown of the total transfer income by source or type of transfer. We found no correlation between total transfers received by households and their dibao participation. 4 MOCA publishes county-level dibao data on a monthly basis. In our analyses for 2008 and 2009, we use year-end (December) values of the MOCA county-level dibao participation and expenditure levels to calculate monthly dibao expenditures per recipient. To obtain annual dibao expenditures, we multiply the December amounts by twelve. These estimates therefore capture the level of transfers per capita attained by the end of the calendar year. Since the MOCA county-level data are not available for 2007, for 2007 we use the January 2008 county-level data, multiplied by twelve. We compared the January versus December values of the MOCA dibao variables for later years (December 2008 versus January 2009, and December 2009 versus January 2010) and did not find systematic differences. 14 minus the village average dibao transfer, and the other is equal to ex post income per capita minus the county average dibao expenditure.5 This approach effectively assumes an egalitarian distribution within villages or within counties of dibao benefits among dibao recipients.6 The dibao participation rates in the CHIP rural survey are lower than the aggregate rates implied by official data.7 To some extent this reflects the selection of provinces in the CHIP sample, but the discrepancy remains even for the nine CHIP provinces (to be discussed in more detail below). The reason why the CHIP sample has lower dibao participation rates than the official data is not clear. It is possible that dibao households are under-sampled in the CHIP survey. Under-sampling of poor households—which are presumably more likely to be dibao recipients—is a known feature of the NBS household survey samples from which the CHIP samples are drawn. It is also possible that some dibao households do not report their dibao participation. Households may not be aware that the transfers they received were from the dibao program, or they may not want to disclose their participation in the program. A third possibility is that the official numbers overstate true participation rates. It is widely accepted that local-level governments in China massage the statistics that they report to higher levels so as to appear to comply with central government policy targets and in order to obscure local irregularities in program implementation (Hvistendahl 2013). IV. Patterns of income inequality and poverty in rural China, 2007-09 During the period 2007-2009 inequality increased and poverty decreased in rural China. Table 3 shows estimates of several measures of inequality calculated using household net income per capita as reported in the CHIP data with population weights. For all measures, inequality increased between 2007 and 2009, with the overall increase ranging from 6 to 19%, depending 5 In the few cases of missing village-level (county-level) data we use county-level (village-level) information to impute missing values. 6 In fact, most villages and counties contain multiple dibao households. In future work we may explore whether different assumptions about the distribution of transfers yields different conclusions; however, even with the assumption of egalitarian distribution of transfers within villages or counties, we find that the dibao program is quite successful in reducing poverty among recipient households, and the modest overall impact of the program on poverty is due to insufficient coverage, rather than insufficient transfers to covered households. 7 Gao, Garfinkel and Zhai (2009) find that in the CHIP urban data (for 2002) the rate of dibao participation is also lower than the officially reported rate. 15 on the measure. The increase is smaller for the Gini coefficient than for the Mean Log Deviation (MLD) index, the Theil index, and the dispersion ratios, which place more weight on the tails of the distribution. The decile dispersion ratio, for example, increased by 19%, and the quintile dispersion ratio by 13%. For purposes of comparison, Table 3 gives estimates of inequality published by the NBS. The NBS’s estimates of the rural Gini coefficient are higher than ours by 6 to 8%, and the NBS’s quintile dispersion ratios are also higher, by about 20%. The discrepancy between the NBS and our estimates of the Gini is not surprising given the provincial coverage of the CHIP dataset; however, the discrepancy between the NBS and our estimates of the quintile dispersion ratio is larger than expected. Regardless, none of the estimates of inequality in Table 3 is overly high. All estimates of the Gini coefficient, for example, are below 0.40, indicating a moderately low degree of inequality in rural China. Inequality increases over time for both the CHIP and official estimates. From 2007 to 2009 the NBS’s rural Gini coefficient increased by about 3%, as compared to 6% for CHIP, and the NBS’s quintile dispersion ratio increased by 9%, as compared to 13% for CHIP. Figure 1 shows the growth incidence curve, a plot of annual income growth (in constant prices) between 2007 and 2009 for each percentile group in the income distribution, arranged in order from the poorest to the richest decile. This figure is constructed using the CHIP data. Figure 1 reveals that from 2007 to 2009 the poorest percentiles experienced negative income growth. At the third percentile income growth becomes positive; at the seventh percentile it reaches 5% per year. As one moves further up the income distribution, the rate of income growth rises above 10%. For most percentiles in the top 40% of the income distribution, income growth is close to or exceeds 10% per year. Overall, Figure 1 shows that during this period, incomes of poorer groups lagged behind those of middle- and high-income groups, a pattern consistent with rising inequality as reported in Table 3. For estimates of absolute poverty, we use three different poverty lines. First, we use China’s official poverty line as of 2011 (adjusted back to 2007, 2008 and 2009 using the national rural consumer price index). We use the 2011 official poverty line rather than the contemporaneous official poverty lines because before China made a large upward adjustment 16 to the official poverty line in 2011, before which time the official poverty line was widely regarded as too low (World Bank 2009). We also use the $1.25 and $2 per person per day international poverty thresholds based on purchasing power parity (PPP) income. We note that the $1.25 poverty line is not much different from the 2011 official poverty line. Finally, we use two relative poverty lines that are equal to 50% and 60% of median income in each year. Table 4 shows these poverty lines in current prices and explains their construction. Table 5 shows our estimates of poverty incidence calculated using the CHIP data and the poverty lines in Table 4. For all three absolute poverty lines, poverty incidence declined substantially from 2007 to 2009. For the official poverty line, for example, the poverty rate in 2009 was 25% lower than in 2007. Although absolute poverty declined, relative poverty increased. For both of our relative poverty lines, poverty incidence increased by more than 10% from 2007 to 2009. The different direction of change in absolute and relative poverty rates reflects that although the absolute level of income of the poor grew, their income growth was slower than that of higher income groups (as evident in Figure 1). The poverty gap is a measure of the amount of funding that would be needed eliminate poverty if transfers could be perfectly targeted to individuals below the poverty line, and in amounts exactly equal to their income shortfalls below the poverty line. Table 6 gives estimates of the poverty gap calculated for the three absolute poverty lines. In all cases the poverty gap declined between 2007 and 2009. For example, measured using the official poverty line, the poverty gap declined from 61 trillion yuan in 2007 to 59 in 2008 and 56 in 2009. In real terms, this was equivalent to a decline of 10% decline in 2008 and of an additional 5% in 2009. V. Patterns of dibao participation, thresholds and transfers The levels of inequality and poverty outlined in the last section provide a context for evaluating the rural dibao program. In this section, using the CHIP household data combined with MOCA statistics, we describe the patterns of dibao thresholds, transfers, and participation, with some comparisons to poverty lines and poverty incidence. 17 Consistent with national dibao policies, our data show substantial expansion of the dibao program since 2007. The mean dibao threshold, calculated using MOCA county-level data for all provinces, increased from 1,064 yuan per capita in 2007 to 1,428 yuan per capita in 2009 (Table 7). The mean dibao transfer per capita also increased (Table 7). Dibao transfers were, on average, somewhat lower than China’s official poverty lines at the time (785 yuan in 2007, 1,067 yuan in 2008, and 1,196 yuan in 2009), and also lower than the 2011 official poverty line that we use in our analysis (Table 4). Table 7 also shows the average dibao thresholds for the nine provinces covered in the CHIP sample; these are similar to the national averages. According to official policy, the dibao thresholds are set locally and so can vary across counties. The MOCA county-level data indeed show substantial variation in thresholds. Figure 2 is a graph of the distribution of county dibao thresholds in current prices for the CHIP sample counties in each of the three sample years. In 2007 and 2008 the county dibao thresholds ranged from less than 500 yuan per capita per year to more than 3,000 yuan. In 2009 the lowest thresholds had risen above 500 yuan, and the highest to more than 4,000 yuan. Figures 3a and 3b show the distributions of dibao transfer amounts in the CHIP sample counties for 2008 and 2009. The distributions based on the county-level averages from MOCA data and on the village-level averages from CHIP are similar, although variation is wider at the village level (to be expected because averaging at the county level eliminates variation within counties). As is the case for the thresholds, variation in the dibao transfers is substantial. In 2009, for example, county average dibao transfers ranged from less than 500 to more than 3,000 yuan per capita. Dibao participation increased along with dibao thresholds and transfer amounts. Calculated using the CHIP data, the rate of participation in the rural dibao program increased from 1.9% in 2007 to 3.0% in 2009 (Table 8). Dibao participation rates in the CHIP data are lower than national participation rates implied by the MOCA statistics, which increased from 5.0% of the rural population in 2007 to 6.9% in 2009. Possible reasons for discrepancies between the CHIP and official dibao statistics include those discussed earlier. These dibao 18 participation rates are also substantially lower than poverty rates calculated using the CHIP data (Table 5). Geographic variation in dibao participation rates is considerable (Table 8). In 2009 dibao participation rates (calculated using the CHIP data) ranged from less than 1% in Hebei and Zhejiang provinces to 5 or 6% in Guangdong and Chongqing. Variation in participation rates is also evident in the official data. Such variation reflects differences across locations in dibao thresholds, financing and implementation, as well as differences in incomes and thus eligibility. The fact that dibao thresholds vary, and that they tend to be lower in poorer than richer counties, raises the question of whether dibao participation rates are in fact higher for lower income groups. Using the CHIP data, we calculate dibao participation rates by ex ante income decile for 2007, 2008, and 2009, shown in Figure 4. The blue lines represent the distribution based on estimates of ex ante income that subtract village average dibao transfers per capita, and the red lines are based on estimates that subtract county average dibao expenditures per capita. Village-level data are not available for 2007; for 2008 and 2009 the two estimates yield similar patterns of participation rates across the income distribution. Figure 4 reveals that, in general, dibao participation rates are higher for poorer income groups. In all three years the participation rates are highest for individuals in the poorest decile of the income distribution. Dibao participation drops sharply for the second poorest decile, and thereafter tends to decline further as one moves to higher income groups. In all years, however, less than 10% of individuals in the poorest decile are dibao participants. Moreover, in all years dibao participation is evident for all income deciles, including the very richest. With expansion of the dibao program over time, the pattern of participation has shifted more towards poorer income groups (Figure 4). Between 2007 and 2009 participation rates increased for most income groups, with relatively large increases for the bottom deciles. Participation rates, however, also rose for middle deciles. For the richest four deciles, changes in participation rates were small and remained below 2% in all three years. Figure 4 reveals that even though poorer groups are more likely to participate in the dibao program, participation by middle-income and richer deciles is nontrivial. This pattern suggests leakage in targeting, which we explore later. 19 VI. Impact of dibao transfers on incomes and poverty Do dibao transfers provide a minimum income guarantee, that is, do they bring household incomes up to the level of local dibao thresholds? Do they reduce rural poverty, and if so, to what extent? Here we provide answers to these questions through comparisons of ex ante and ex post incomes. As explained earlier, our estimates of ex ante income are equal to reported income minus the amount of the dibao transfer, which implicitly assumes that the receipt of dibao transfers does not change household behavior. Our estimates of the impact of the dibao program on incomes and poverty are therefore probably overstated. Did the rural dibao program provide a minimum income guarantee? In order to answer this question, we compare ex ante and ex post incomes of individuals whose incomes were below the local (county) dibao threshold. Table 9 gives the percentages of individuals in the CHIP sample with ex ante and ex post incomes below the local dibao thresholds in each of the three years. The first three rows classify individuals using ex post incomes; the second three rows using ex ante incomes calculated using village average transfers; and the bottom three rows using ex ante incomes calculated using county average transfers. The first column shows the percentages of all individuals in the CHIP sample, including both beneficiaries and non-beneficiaries, whose incomes were below the dibao thresholds. The percentage of individuals whose ex post income was below the dibao thresholds increased over time from 2.4% in 2007 to 2.6% in 2008 and further to 3.8% in 2009. This increase is somewhat surprising given the dramatic expansion of dibao participation and transfers during these years; however, dibao thresholds were also raised. Examination of ex ante incomes reveals that eligibility rates also increased: from 2007 to 2009 the share of individuals in the CHIP sample with ex ante incomes (calculated using county average transfers) below the local dibao thresholds rose from 2.5% to 4.1%. Did the dibao program provide a minimum income guarantee? In all three years the percentage of dibao recipients with ex ante incomes below the dibao thresholds exceeded the percentage with ex post incomes below the thresholds. For example, in 2009 more than 12% of 20 dibao recipients had ex ante income below the dibao thresholds, and only 5.7% had ex post income below the dibao thresholds. In other words, the dibao transfers raised more than half of dibao recipients who started out below the dibao threshold above the threshold. We conclude that the rural dibao program was reasonably successful in providing an income guarantee for dibao recipients whose pre-transfer income was below their local dibao threshold. Of course, these numbers ignore non-recipients whose incomes were below the dibao thresholds. About 90% of individuals with income below the threshold did not receive dibao transfers. For these individuals, the dibao program did not provide a minimum income guarantee. The lack of guarantee to this group reflects a substantial exclusionary error in targeting, which we discuss in the next section. Did the dibao program reduce poverty? We answer this question by comparing poverty incidence and the poverty gap calculated using ex ante versus ex post incomes. As shown in Table 10, which reports estimates of poverty incidence calculated using our three absolute poverty lines, in all cases poverty incidence was higher for ex ante incomes than for ex post incomes. This is consistent with a poverty-reducing impact of the dibao program. In all cases, however, the difference in ex ante versus ex post poverty incidence is smaller than half a percentage point. In other words, the dibao program apparently had a negligible impact on poverty incidence. Table 11 shows estimates of the poverty gap calculated using ex ante incomes and ex post incomes. As expected, the poverty gap calculated using ex ante is larger than that calculated using ex post incomes, which include the dibao transfers. In 2007 and 2008 the ex ante poverty gap was 2-3% larger than the ex post poverty gap, and in 2009 it was 6.5% larger. Again, however, the difference is modest, especially when compared to total dibao expenditures. According to the official data, in 2007 total dibao expenditures were equivalent to 18% of the ex ante poverty gap; by 2009 total dibao expenditures had risen to 64% of the ex ante poverty gap. The reduction in the poverty gap per yuan dibao expenditure was therefore fairly small. In 2007 each yuan of dibao expenditures was associated with a reduction in the poverty 21 gap of 0.13 yuan. In 2009 each yuan of dibao expenditures was associated with a reduction in the poverty gap of 0.10 yuan. Dibao participation in the CHIP sample is lower than that reported in official statistics, and it may be more appropriate to evaluate the program’s poverty impact using the level of dibao expenditures implied in the CHIP data. We calculate total dibao expenditures implied by the CHIP data as equal to the weighted sum of county level transfers times the number of dibao recipients within each county (see note to Table 11). 8 By this calculation, total dibao expenditures are substantially lower than the official numbers. In 2009, for example, they are only 36% of the official total. Even using these lower estimates of total dibao expenditures, the poverty impact of the dibao program remains modest. In 2009, for example, dibao expenditures implied by the the CHIP data were equivalent to 26% of the ex ante poverty gap, but the poverty gap calculated using ex post incomes was only 6.5% lower than that calculated using ex ante incomes. Each yuan of dibao expenditures was associated with a reduction in the poverty gap of only 0.24 yuan. These discrepancies between dibao expenditures and poverty reduction suggest leakages in targeting. VII. Conventional analysis of dibao targeting What is the extent of inclusionary targeting error, that is, to what extent do dibao benefits go to individuals with ex ante incomes above the dibao thresholds? The dibao program’s stated goal is to assist households with incomes below the dibao thresholds, so inclusionary targeting error is a relevant criterion for evaluation of the program. What is the extent of exclusionary error, that is, to what extent are individuals with ex ante incomes below the dibao thresholds excluded from the program? The dibao program does not claim to cover all individuals with incomes below the dibao threshold, so exclusionary error may not measure the success of the 8 For dibao recipients who live in counties for which MOCA county-level transfer data are missing, we use the village average transfers from CHIP. 22 dibao program in meeting its own objectives. Nevertheless, analysis of the program’s exclusionary targeting error is informative. Table 12 contains estimates of inclusionary and exclusionary targeting error of the dibao program calculated using local dibao thresholds as the targeting criterion. Targeting errors have declined over the three years. For example, based on estimates using the county average dibao expenditures, from 2007 to 2009 inclusionary error declined from 94% to 86%, and exclusionary error from 94% to 89%. Despite these improvements, the overwhelming majority of dibao beneficiaries had ex ante incomes higher than the local dibao thresholds. Moreover, the dibao program reached only a small proportion (11% or less) of individuals with ex ante incomes below the dibao thresholds. In all years, then, it appears that the vast majority of eligible individuals did not benefit from the program. By comparison, for China’s urban dibao program Chen, Ravallion and Wang (2006) report an inclusionary error of 43% and an exclusionary error of 71%. Although based on data for earlier years, their estimates suggest that the targeting performance of China’s urban dibao program is markedly better than that of the rural dibao program. Weaker performance of the rural dibao program is not overly surprising given the uneven capacity and resources of local governments in rural China, as well as the difficulty of measuring rural incomes. The targeting performance of the rural dibao program can also be evaluated relative to the poverty line so as to ascertain the extent to which the program benefited the poor versus nonpoor. Table 13 shows the shares of the poor and nonpoor who received dibao benefits. These shares are calculated using our three poverty lines and ex ante incomes. In all cases, less than 10% of the poor received dibao transfers. A higher proportion of the poor than nonpoor, however, were dibao recipients. For example, based on the official poverty line, the percentage of the poor receiving dibao benefits in 2009 was 8%, versus less than 3% of the nonpoor. Also, the proportion of the poor who received dibao benefits increased over time. For example, based on the official poverty line, the share of the poor receiving dibao benefits increased from 4.7% in 2007 to 8.0% in 2009. How well does the dibao program target poor households? Table 14 shows the inclusion and exclusion errors calculated using ex ante incomes in relation to the official 23 poverty line. The inclusion error is between 64 and 75%, depending on the estimate and year. That is, between 64 and 75% of dibao recipients were not poor. The exclusion error is between 92 and 95%, indicating that the overwhelming majority of the poor did not benefit from the dibao program. VIII. Correlates of dibao participation and propensity score analysis of dibao targeting Our conventional analysis of dibao targeting implicitly assumes that the local officials who implement the program select program beneficiaries based on current year household incomes, that their information about those incomes is the same as that collected by the NBS and reported in the CHIP survey, and that these income data are accurate. As discussed by Chen, Ravallion and Wang (2006), these assumptions may not be correct. Local officials who implement the dibao program do not have access to detailed income data such as that collected by the NBS, and even if they did, the NBS income data contain some unknown degree of measurement error. In reality, local officials are likely to select beneficiaries based on some measure of permanent income rather than current income, and based on observable correlates of income. Indeed, China’s national rural dibao policies allow for such practices, and local regulations explicitly mention alternative criteria for identifying recipients. In view of these considerations, Chen, Ravallion and Wang (2006) propose an approach based on the idea that local officials select beneficiaries on the basis of a latent income variable that is correlated with ex ante income as measured in the survey as well as other characteristics and an error term. Local officials select households whose latent income is below the dibao threshold as beneficiaries. Targeting analysis can then be carried out on the basis of latent household incomes (Ravallion 2008). The first step is to estimate a probit regression with dibao participation as the dependent variable and with ex ante income and other relevant attributes as measured in the survey data as the independent variables. The other attributes are chosen based on local implementation practices and include household characteristics such as demographic composition, health of household members, and human and physical capital or assets. Second, the results of the probit model are used to predict a conditional probability of program 24 assignment (the propensity score). The estimated coefficients from the probit regression provide weights placed on the different characteristics; these correspond to the implicit weights assigned by program administrators when deciding on beneficiaries. Third, a cutoff is determined based on the observed coverage rate. Beneficiaries are selected by counting off households ranked from highest to lowest propensity score until the cutoff is reached. The selected households are used to calculate the targeting errors. Here we carry out such an analysis using the CHIP survey data. We note that in this analysis, including the probit regressions and calculations of propensity scores and targeting errors, we use households as the unit of analysis. Tables 15, 16 and 17 contain descriptive statistics (unweighted) for attributes associated with rural dibao implementation. Comparison of dibao and nondibao households reveals differences in mean incomes as well as other attributes, although not all differences are statistically significant. Both ex post and ex ante incomes of dibao households are, on average, lower than those of nondibao households. A smaller share of the income of dibao households is from wage employment, and in 2007 and 2008 (but not 2009) dibao households are less likely to have a member engaged in migrant work than nondibao households. Household size is smaller for dibao households, and they contain markedly higher shares of elderly members and of bad health and disability. In 2007, for example, 20% of dibao households contained a family member over the age of 60, 41% contained a member in bad health and 35% contained a member with a disability, as compared to 10%, 14% and 12%, respectively, for nondibao households. Differences also exist in ownership of physical assets. Housing conditions, as measured by whether housing is multi-storey and the presence of piped water and flush toilets, are poorer for dibao households. Ownership of durable goods such as household appliances and motorized vehicles is lower. Tables 15-17 also reveal that the communities in which dibao households live are somewhat different from those of nondibao households. A higher share of dibao households live in villages that are located in mountainous areas, do not have a paved road, are distant 25 from the nearest township government, and experienced some sort of natural disaster in the survey year. Probit regressions reveal that many of the characteristics in Tables 15-17 are statistically significant predictors of dibao status. Table 18 reports the estimated marginal effects of the probit regressions. Specification 1 uses ex ante income calculated using village average dibao transfers, and specification 2 uses ex ante income calculated using county average dibao expenditures.9 In all years the probability of receiving dibao benefits has a significant, negative association with household income. The marginal effects imply that a 1% increase in ex ante income reduces the probability of receiving dibao by 0.7 to 1.0 percentage points. Other characteristics that are consistently significant in most years and specifications are: household size (negative), bad health (positive), disability (positive), the share of wages in income (negative), share of income from non-agricultural business (negative), and absence of a major appliance (positive). The estimated coefficients change somewhat across the years. Notably, more variables are significant in 2009 than in the earlier two years. For example, the share of elderly becomes significant (positive) in 2009, indicating that selection criteria may have changed to emphasize households with elderly family members. The presence of a migrant worker (positive), marriages (negative), deaths (positive), and cultivated land area (negative) also become significant in 2009. These changes may reflect the refinement of, or adaptation in, the criteria used by local officials to decide on eligibility for the program, or perhaps smaller standard errors in our estimates due to the larger number of dibao households in the 2009 sample than in 2007 and 2008. Also, the expansion of the dibao program during this time period may have allowed the widening of eligibility criteria to include more characteristics. Tables 19 and 20 show the results of the propensity score analysis of targeting performance. In 2007 17% of households selected as eligible based on propensity scores 9 We also estimated the probits using ex post income because of concerns about measurement error in calculation of ex ante income using village and county average dibao transfers. We found that the estimated coefficients on ex post income were in fact smaller and the standard errors relatively bigger than those on our estimates of ex ante income. We concluded that our estimates of ex ante income, despite their possible weakness, are useful for this analysis. 26 received dibao benefits, as compared to 6% for the conventional targeting analysis based on the dibao thresholds (see Table 12). In 2008 20% of eligible households and in 2009 17% of eligible households received dibao benefits according to the propensity score approach, as compared to 7% and 11% using the conventional approach. Thus dibao coverage for households classified as eligible using latent income as the selection criterion is substantially higher than that implied by conventional targeting analysis. The inclusionary error for the propensity score method is also lower than that for conventional targeting analysis. 83% of dibao recipients were not eligible based on the propensity score in 2007, as compared to 94% in the conventional analysis; in 2008 and 2009 the propensity score inclusion errors are 80% and 83%, as compared to 92% and 86%, respectively, for the conventional approach (Tables 12 and 20). Thus dibao leakage to households classified as ineligible using latent income as the selection criterion is lower than that implied by conventional targeting analysis. All in all, the propensity score targeting analysis yields smaller targeting errors than conventional targeting analysis. These findings are consistent with a situation in which local officials rely on observable characteristics of the households to determine dibao eligibility, or where local officials’ perceptions of household income are not identical to CHIP income estimates. IX. Policy simulations: Expand Coverage versus Increase Transfer Amounts In order to examine options for improving the impact of the rural dibao program, we conduct some simple simulations using the 2009 CHIP data. A first set of simulations investigates how expansion of the rural dibao program would affect the level of poverty. The rural dibao program has in fact expanded since 2009. According to the official statistics (Table 1), in 2013 the total rural dibao budget was 2.4 times that in 2009. Most of the increase reflected higher transfer amounts rather than wider coverage: the average dibao transfer per recipient doubled, while the number of recipients increased only 13%. Our simulations provide some insights into the potential consequences of such a program expansion. We explore the impact of expanding the program in two ways: by 27 increasing the amount of dibao transfers going to existing dibao recipients, and by expanding coverage to more recipients while keeping the transfer amounts unchanged. The first approach should reduce poverty if most dibao beneficiaries are poor and if their transfer amounts are insufficient to bring them above the poverty line. The second should reduce poverty if exclusionary targeting error is substantial and transfer amounts are adequate. Based on findings reported in previous sections of this paper, we expect that the second approach will have a larger poverty impact than the first approach. This first set of simulations retains local variation in dibao eligibility thresholds and dibao transfer amounts as observed in the 2009 CHIP data. We use the average transfer in the household’s village of residence, as reported in the CHIP village-level survey, as the local transfer amount.10 For the simulations, which involve the selection of recipient households based on ex ante income, we estimate ex ante income by subtracting the local transfer from ex post income as reported by households in the CHIP household-level survey. We calculate poverty measures in relation to the official poverty line.11 All calculations are done using weights. We begin by constructing a baseline case that reflects observed incomes and dibao participation in the 2009 data. Baseline poverty levels are equal to those observed in the 2009 CHIP data, that is, they are the levels of poverty implied by ex post incomes in the data. The dibao budget for the baseline case is equal to the (weighted) sum of local dibao transfer amounts for all dibao recipients observed in the data.12 We refer to this as the “observed” baseline. The simulations require a decision about how much to expand the program. For simplicity, we use a target budget equal to the amount of money that would be spent if the program were expanded to cover all eligible individuals in 2009 who were not yet dibao beneficiaries. In other words, we calculate the cost of providing local dibao transfers to all non- 10 In cases where data for the village average dibao transfer are missing, we use the county average transfer. 11 As elsewhere in this paper, the poverty line for 2009 is derived from the new official poverty line of 2300 yuan announced in 2011. We adjust this back to 2009 using the rural consumer price index published by the NBS. 12 Note that the dibao budget here does not equal the official number reported by MOCA, reflecting in part the lower dibao participation rate in the CHIP sample than in the official statistics (see Table 11 and related text). Also, the dibao budget here differs a bit from the CHIP total dibao expenditures in Table 11, which is calculated using county average dibao expenditures. Here we use village average dibao transfers. 28 recipients whose per capita incomes were below the local dibao thresholds, and we add this cost to the baseline dibao budget. This yields a target budget equal to 2.54 times the baseline budget. Simulations of the impact of expanding the dibao program are shown in Table 21. For the simulation with expanded coverage, we present two scenarios. Simulation (a) assumes perfect targeting: all added dibao recipients have income below the dibao eligibility threshold in their location of residence, in other words, zero inclusionary targeting error among added recipients. Moreover, the target budget is just sufficient to ensure that in this simulation all eligible individuals receive dibao transfers, so exclusionary targeting error is also zero. Simulation (b) assumes no targeting: additional recipients are selected randomly from among all non-recipients. Thus, there will be both inclusionary and exclusionary targeting error among the added recipients. These two simulations can be interpreted as optimistic and pessimistic targeting scenarios for expansion of coverage. The poverty results for simulations (a) and (b) reveal that expanding coverage has the potential to substantially reduce poverty relative to the “observed” baseline, depending on how the additional recipients are selected. If we assume optimistically that the new recipients are selected using perfect targeting (simulation a), then the expansion reduces the poverty headcount by more than 5%, the poverty gap by 24%, and the squared poverty gap by 17%. If we assume random selection (simulation b), the expansion reduces poverty by at most 3%. Would increasing the transfer amount to baseline recipients be more effective than expanding coverage? Simulation (c) shows the results of increasing transfers without changing coverage. The poverty impact is modest: relative to the baseline poverty reduction at best 3%. This impact is considerably smaller than that of expanding coverage when new recipients are well targeted (simulation a), and similar to that of expanding coverage when new recipients are randomly targeted (simulation b). We conclude that expanding coverage is a better policy choice than increasing transfer amounts. Even if targeting is imperfect, so long as targeting practices do a better job than random selection, expanding coverage should yield greater poverty reduction than increasing transfer amounts. These simulations suggest that the large increase in the dibao budget 29 observed between 2009 and 2013 could potentially yield substantial reductions in poverty depth and intensity and a respectable reduction in the headcount if the additional funds were used mainly to expand coverage rather than increase transfers, and if selection of new recipients were well targeted. The actual expansion of the program from 2009 to 2013, however, mainly increased transfer amounts. X. Policy simulation: Nationally uniform transfer and threshold In China dibao thresholds, transfers and rates of coverage vary locally and tend to be higher in richer than in poorer counties. Consequently, households above the official poverty line in richer areas may be selected for dibao, while households below the poverty line in poorer areas may be left out. Moreover, richer areas have greater fiscal capacity than poorer areas and can provide larger dibao transfers (Ravallion, 2009). For these reasons some studies have recommended that China adopt a nationally uniform threshold and more equal transfer amounts (World Bank 2009). We investigate the impact of adopting a uniform transfer and uniform threshold with a second set of simulations. For simulations of a uniform transfer, we set the transfer equal to the average transfer amount in the relevant baseline simulation. For simulations of a uniform threshold, we set the threshold equal to the official poverty line. Individuals are classified as eligible for dibao if their ex ante income per capita is below the official poverty line.13 We begin by replacing the locally diverse transfers in the “observed” baseline with a uniform transfer equal to the average “observed” transfer of 666 yuan. The dibao thresholds, recipients and budget are identical to the “observed” baseline case. The outcome of this policy is shown as simulation (d) in Table 21. Compared to the “observed” baseline, the poverty headcount decreases very slightly, the poverty gap increases very slightly, and the squared poverty gap is unchanged. These results suggest that, in the absence of any other policy changes, adopting a uniform national transfer would yield minimal poverty gains. 13 At 2,098 Yuan, the official poverty line is higher than the average dibao threshold; however, in some counties the threshold exceeds this level. See Figure 2 and Table 7. 30 One reason why simulation (d) has such a small impact on poverty is that dibao transfers will only affect the level of poverty if the recipients are poor, but most of the recipients in this simulation are above the poverty line. The recipients are the “observed” recipients, so that targeting is the same as that in the 2009 CHIP sample, which has a 75% inclusionary error (Table 14). Consequently, for three quarters of the dibao recipients in simulation (d), the poverty impact of switching to a uniform transfer is zero. How would adopting a uniform transfer affect poverty if targeting was good? To answer this question, we construct a new baseline case with perfect targeting relative to the observed local dibao thresholds. For this “perfect dibao targeting” baseline we assume that individuals receive dibao transfers if and only if their ex ante incomes are below their local dibao thresholds. In other words, all individuals who are dibao eligible, and no one else, receive transfers. In this baseline the transfers are equal to the local transfer amounts. The dibao budget implied by this “perfect dibao targeting” baseline is 23.7 billion yuan. Poverty outcomes are shown in Table 22. We now carry out a simulation that replaces the locally varying dibao transfers in the “perfect dibao targeting” baseline with a uniform transfer. The transfer is equal to the average transfer (887 yuan) in the “perfect dibao targeting” baseline. The results of this simulation (e) are shown in the second line of Table 22. Now the impact of adopting a uniform transfer is more substantial. All three poverty measures decline, especially the poverty gap, which is reduced by 12%. This simulation demonstrates that adopting a uniform transfer can substantially reduce poverty if there is no inclusionary targeting error. We now turn to the case of a uniform threshold. For the uniform threshold we use the poverty line, which implies that everyone who is poor is eligible. Since the poverty line is higher on average than the local dibao thresholds, the number of eligible will be larger than the number of dibao eligible (incomes < dibao thresholds). Consequently, if we use a target budget based on dibao targeting, it will be insufficient to cover all eligible individuals. We must therefore make some assumption about how recipients are selected from among the poor. We use two alternative assumptions. The first is that recipients are selected based on distance from the poverty line, starting with the poorest (simulation f). The second is that 31 recipients are selected randomly from among the poor (simulation g). Both selection methods yield zero inclusionary and zero exclusionary targeting error; however, selection in simulation (g) ignores depth of poverty. In these simulations the target budget is that in the “perfect dibao targeting” baseline, and recipients receive the local transfers. Comparing the poverty outcomes to the “perfect dibao targeting” baseline tells us whether, in a world of perfect targeting, replacing local thresholds with a uniform threshold would reduce poverty. As reported in Table 22, simulation (f) reduces the poverty gap and squared poverty gap compared to the baseline, but the poverty headcount increases. Simulation (g) reduces the poverty headcount and poverty gap, but the squared poverty gap increases. This difference in outcomes between simulations (f) and (g) is not surprising, because the dibao recipients in (f) are on average in deeper poverty than the recipients in (g). We conclude that adopting a uniform national eligibility threshold has the potential to reduce poverty substantially compared to retaining local dibao thresholds, although the nature of the poverty impact will depend on how recipients are selected among the poor. Finally, what would be the impact of adopting both a uniform dibao transfer amount and a uniform threshold? Simulations (h) and (i) explore this policy option. In both (h) and (i) the uniform transfer is set equal to the average transfer in the “perfect dibao targeting” baseline. Simulation (h) selects recipients based on depth of poverty, while simulation (i) selects recipients randomly among the poor. Both these simulations yield substantial reductions in some, but not all, of the poverty measures. Simulation (h) yields the largest reductions in the poverty gap and squared gap. In these regards it is superior to only adopting a uniform transfer (e). The poverty headcount is higher, however, than in both the baseline and simulation (e). If dibao recipients are selected randomly among the poor (simulation i), then adopting both a uniform transfer and threshold can substantially reduce the poverty headcount and poverty gap compared to the baseline, and also compared to adopting a uniform transfer (e). The squared poverty gap, however, is higher. Overall, the simulations in Table 22 indicate that uniform transfer and/or uniform threshold policies have the potential to increase the dibao program’s effectiveness, but predicated on the assumption of perfect targeting in both the baseline and policy simulations. 32 We know that in fact the dibao program has substantial inclusionary targeting error, and simulation (d) demonstrates that in the presence of inclusionary targeting error at observed levels, the impact of a uniform transfer policy will be minimal. It is difficult to construct a simulation of the uniform threshold in the presence of targeting error. We speculate, however, that a uniform threshold would be more effective than a uniform transfer, because it would increase the share of dibao recipients from counties with lower local thresholds, which tend to have more poor. This policy reform would require fiscal measures that increase dibao funding resources in poorer counties. XI. Conclusions China’s rural dibao program, which was adopted nationwide starting in 2007, is now one of the largest targeted transfer schemes in the world. The program’s implementation and expansion in recent years have coincided with reductions in rural poverty in China. This raises the question of whether, or to what extent, the program has contributed to poverty reduction. Using household survey data matched with administrative data for 2007-2009, we have examined the relationship between China’s rural dibao program and rural poverty and conducted targeting analysis using conventional and propensity score approaches. We find that during these years the rural dibao program provided sufficient income to poor beneficiaries, but the poverty impact of the program overall was small. Although total dibao expenditures are fairly large relative to the poverty gap, the program did not substantially reduce the poverty gap. Conventional targeting analysis reveals large inclusionary and exclusionary targeting errors. Propensity score analysis of targeting reduces the targeting errors, which suggests that the program has been implemented in reference to an unobserved latent income variable rather than income as measured by the NBS in the CHIP survey. Nevertheless, even using propensity score-based targeting analysis, the targeting errors remain quite large. Our analysis indicates that during these years a central reason for the program’s modest poverty impact and high exclusionary targeting error was that the proportion of the population covered by the program was small. Since 2009 the rural dibao program has expanded rapidly in 33 terms of its overall the budget. Most of the budget increase, however, has been used to increase transfer amounts. The number of recipients has changed fairly little. Using simulations, we explore whether modifications in the dibao program would increase its impact on poverty. We first explore the impact of increasing the dibao budget (a) by increasing transfer amounts and (b) by expanding coverage. Our results indicate that expanding coverage is more effective than increasing transfer levels, although the extent to which it is more effective depends on targeting among the added dibao beneficiaries. We also use simulations to explore whether adopting a uniform transfer and uniform threshold will improve the program’s poverty impact. Both these policies have the potential to substantially reduce poverty, but the extent to which that potential is realized depends critically on targeting. For example, a uniform transfer to the recipients observed in the 2009 data yields minimal improvement in poverty by any measure. Overall, the simulations yield several broad lessons. First, expanding the program’s coverage has more potential to reduce poverty than increasing transfer amounts, even with imperfect targeting and local variation in thresholds and transfer amounts. We recommend that future increases in the dibao budget should mainly be used to expand coverage rather than increase transfer amounts. Second, in theory adopting uniform dibao transfers and thresholds can yield gains in poverty reduction, but in practice the gains may be limited due to imperfect targeting, especially by extent of inclusionary error. Our simulations demonstrate that a uniform transfer policy will have minimal poverty impact if the targeting error is similar to that in 2009. We speculate that a uniform threshold policy is more likely to have an impact in the presence of targeting error, as it would tend to reallocate the regional distribution of dibao transfers from richer to poorer counties. Our simulation analyses explore the consequences of expanding the dibao program or from modifying the levels of transfers and eligibility thresholds, holding targeting constant. We fix our targeting assumptions and then alter parameters of the program. It is possible, however, that targeting is endogenous and influenced by the parameters of the program. For example, small transfers may promote self-selection by poorer households into the program, thus improving targeting as evidenced from the Brazilian Bolsa Familia program (Bastagli, 2008) 34 which relies of self-reported income for targeting small cash transfers. This sort of interaction between program parameters and targeting strengthens the case for expanding coverage versus increasing transfer amounts. The program implementation may affect local government behavior and so the size of the dibao budget. For example if there is a large poverty impact tradeoff between higher transfers or higher coverage depending on targeting accuracy, increasing transfer amounts under high rates of inclusionary errors undermines program legitimacy, especially among the tax-paying constituency, which may in turn impact the local budget. In comparison, reduced exclusionary error by virtue of program expansion can potentially build further fiscal support. Our findings are influenced by limitations of our data. One limitation is the lack of household-level information on dibao transfers. In our analyses we have used village- (and county-) level data on dibao transfers to construct estimates of the household-level transfers on the assumption that within villages (or counties) the transfers are uniform and equal to the local average. If in fact dibao transfers vary within localities such that poorer households receive larger transfers than richer households, then the program’s poverty impact may be larger than that implied by our estimates. Another data limitation is potential bias in the dibao participation rates. Dibao participation in the CHIP sample is considerably lower than in the official statistics published by the MOCA and NBS. Of course, this could be because the official statistics are biased. If, however, the discrepancy is due to understatement in the CHIP survey of dibao participation rates, then our findings will understate the program’s poverty impact. 35 XII. References Bastagli, Francesca, 2008, “The design, implementation and impact of conditional cash transfers: An evaluation of Brazil’s Bolsa Familia”, PhD thesis, London School of Economics Chen, Shaohua, Martin Ravallion, and Youjuan Wang, 2006, “ Di Bao: A Guaranteed Minimum Income in China’s Cities?” World Bank Policy Research Working Paper WPS 3805. Deaton, Angus, 2010, “Instruments, Randomization and Learning about Development,” Journal of Economic Literature 48(2): 424-455. Department of Social, Science and Technology Statistics of the National Bureau of Statistics (NBS) (2008), Zhongguo shehui tongji nianjian 2008 (China Social Statistical Yearbook 2008), Beijing: Zhongguo tongji chubanshe. Gao, Qin, Irwin Garfinkel and Fuhua Zhai, 2009, “Anti-poverty Effectiveness of the Minimum Living Standard Assistance Program in Urban China,” Review of Income and Wealth 55 (special issue 1): 630-655. Hvistendahl, Mara, 2013, “The Numbers Game,” Science 340(6136): 1037-1039, May. Li, Shi, Hiroshi Sato and Terry Sicular, eds., forthcoming 2013, Inequality in China: Challenges to a Harmonious Society, New York: Cambridge University Press. Lin, Wanlong and Christine Wong, 2012, “Are Beijing’s Equalization Policies Reaching the Poor? An Analysis of Direct Subsidies under the “Three Rurals” (sannong),” The China Journal 67: 23- 45. Luo, Chuliang and Terry Sicular, 2013, “Inequality and Poverty in Rural China,” chapter 5 in Li, Shi, Hiroshi Sato and Terry Sicular, eds., forthcoming 2013, Inequality in China: Challenges to a Harmonious Society, New York: Cambridge University Press. Ministry of Civil Affairs, 2012, 2011 Social Services Development Statistical Report (2011 年社会 服务发展统计报告), http://cws.mca.gov.cn/article/tjbg/. Accessed April 8, 2013. Ministry of Civil Affairs, 2011, 2010 Social Services Development Statistical Report (2010 年社会 服务发展统计报告), http://cws.mca.gov.cn/article/tjbg/ . Accessed April 8, 2013. Ministry of Civil Affairs, 2010, 2009 Social Services Development Statistical Report (2009 年社会 服务发展统计报告), http://cws.mca.gov.cn/article/tjbg/ . Accessed April 8, 2013. 36 Ministry of Civil Affairs, 2009, 2008 Social Services Development Statistical Report (2008 年社会 服务发展统计报告), http://cws.mca.gov.cn/article/tjbg/ . Accessed April 8, 2013. Ministry of Civil Affairs, 2008, 2007 Social Services Development Statistical Report (2007 年社会 服务发展统计报告), http://cws.mca.gov.cn/article/tjbg/ . Accessed April 8, 2013. Ministry of Civil Affairs, 2007, 2006 Social Services Development Statistical Report (2006 年社会 服务发展统计报告), http://cws.mca.gov.cn/article/tjbg/ . Accessed April 8, 2013. National Bureau of Statistics Rural Socioeconomic Survey Department (various years), Zhongguo nongcun zhuhu diaocha nianjian (China Yearbook of Rural Household Survey), Beijing: Zhongguo tongji chubanshe. National Bureau of Statistics (NBS) (various years), Zhongguo tongji nianjian (China Statistical Yearbook), Beijing: Zhongguo tongji chubanshe. O’Keefe, Philip, 2004, “Social Assistance in China: An Evolving System,” mimeo, World Bank, Washington DC. Poverty Alleviation Office of the State Council, 2010, “China Rural Poverty Alleviation and Development Program (2000-2010) Implementation Evaluation Report (《中国农村扶贫开发 纲要(2001-2010 年)》实施效果的评估报告),” Beijing. Ravallion, Martin, 2008, “Miss-targeted or miss-measured?” Economics Letters 100: 9-12. Ravallion, Martin, 2009, “Decentralized Eligibility for a Federal Antipoverty Program: A Case Study for China,” The World Bank Economic Review 23: 1-30 Wang, Meiyan, 2007, “Emerging Urban Poverty and Effects of the Dibao Program on Alleviating Poverty in China,” China and the World Economy 15(2): 74-88. World Bank. 2009. Main report. Vol. 1 of China - From Poor Areas to Poor People: China's Evolving Poverty Reduction Agenda - an Assessment of Poverty and inequality in China. Washington D.C.: The World Bank. http://documents.worldbank.org/curated/en/2009/03/10444409/china-poor-areas-poor- people-chinas-evolving-poverty-reduction-agenda-assessment-poverty-inequality-china-vol-1- 2-main-report World Bank. 2011. Reducing Inequality for Shared Growth in China: Strategy and Policy Options for Guangdong Province. Washington, D.C.: The World Bank. https://openknowledge.worldbank.org/handle/10986/2251. Accessed April 8, 2013. 37 World Bank Social Protection Group, Human Development Unit, EASHD, 2010, “Social Assistance in Rural China: Tackling Poverty through Rural Dibao,” World Bank, Washington, D.C. Xinhuanet, 2013, “Ministry of Civil Affairs: The Effectiveness of Dibao Work is Notable, National Error Rate is 4% ( 民政部：低保工作成效显著 全国低保错保率约为 4 ％ ), February 25. http://www.chinanews.com/gn/2013/03-13/4638980.shtml . Accessed April 8, 2013. Xinhua, 2007a, “77% of Rural Poor Covered by Allowance System,” May 8. http://www.china.org.cn/english/government/212112.htm. Accessed August 13, 2013. Xinhua, 2007b, “Chinese Government Decides to Subsidize All Rural Poor,” People’s Daily Online, May 24. http://english.peopledaily.com.cn/200705/24/eng20070524_377380.html. Accessed August 13, 2013. Xu, Yuebin and Xiulan Zhang, 2010, “Rural Social Protection in China: Reform, Performance and Problems,” chapter 6 in James Midgley and Kwong-leung Tang, ed.s, Social Policy and Poverty in East Asia: The Role of Social Security, New York: Routledge, pp. 116-128. Zhu, Wurong, 2012, “Strictly Oppose Unhealthy Practices in Urban and Rural Dibao ( 严防城乡 低 保 中 的 不 正 之 风 )”, Renminwang, May 29. http://news.xinmin.cn/rollnews/2012/05/29/14933396.html. Accessed April 8, 2013. 38 XIII. Figures Figure 1a: Growth incidence curve, 2007 to 2009 10 5 0 -5 0 20 40 60 80 100 Percentiles Notes: Weighted. This is a plot of average annual growth in household income per capita from 2007 through 2009, in constant prices, of each percentile group in the income distribution of the CHIP rural sample. Incomes are deflated using the rural consumer price index (NBS, various years). 39 Figure 2: The Distribution of County-level Dibao Thresholds, by Year (yuan per person per year) .0008 .0006 Density .0004 .0002 834 10681200 0 0 1000 2000 3000 4000 Yuan Dibao threshold, 2007 Dibao threshold, 2008 Dibao threshold, 2009 Note: This figure shows the distribution of dibao thresholds for counties covered in the CHIP rural sample. For the year 2007, the January 2008 dibao threshold values were used. For 2008 and 2009, the December 2008 and 2009 threshold values were used. Vertical lines represent the yearly median threshold values, which were 834, 1,068 and 1,200 yuan for 2007, 2008 and 2009, respectively. Source: MOCA (various years). 40 Figure 3a: County and Village Average Dibao Transfers, 2008 (yuan per recipient) .002 .0015 .001 .0005 0 0 1000 2000 3000 4000 Yuan County Village Figure 3b: County and Village Average Dibao transfers, 2009 (yuan per recipient) .002 .0015 .001 .0005 0 0 1000 2000 3000 4000 Yuan County Village Note: County transfers shown in Figures 3a and 3b are restricted to counties covered in the CHIP survey. Village transfers are for villages covered in the CHIP survey. Outliers (higher than 4000 yuan) have been removed. The dashed vertical lines represent the average village transfer for CHIP villages; the dotted vertical lines represent the average county transfer for CHIP counties. Source: Authors’ calculation based on data from CHIP and MOCA (various years). 41 Figure 4: Dibao Participation Rates by Ex Ante Income Decile (%) 8% 6% Percent 4% 2% 0% 1 2 3 4 5 6 7 8 9 10 Deciles ex ante income (minus county transfer), 2007 ex ante income (minus village transfers), 2008 ex ante income (minus county transfer), 2008 ex ante income (minus village transfers), 2009 ex ante income (minus county transfer), 2009 Notes: Weighted. This shows dibao participation rates for individuals by decile groups based on ex ante income per capita. Estimates of ex ante income are calculated using the CHIP income data and average transfers at the village level (from CHIP) or county level (MOCA). 42 XIV. Tables Table 1: Official Statistics for China’s Rural Dibao Program 2006 2007 2008 2009 2010 2011 2012 2013 rural dibao recipients (millions) 15.93 35.66 43.06 47.60 52.14 53.06 53.45 53.88 rural dibao transfers (million yuan) na 10910 22873 36300 44500 66770 71820 86690 national average rural dibao na 840 988 1210 1404 1718 2003 2434 threshold (yuan per person per year) national average rural dibao transfer na 466 605 816 888 1273 1344 1609 (yuan per person per year) Note: The Ministry of Civil Affairs only started publishing data on transfers and thresholds for the rural dibao program after 2007, so data for transfers and thresholds for earlier years are missing or incomplete. Dibao transfers are the sum of dibao transfers from all levels of government. The average transfer is calculated as total rural dibao transfers divided by the number of recipients. Sources: NBS (2012); Ministry of Civil Affairs (various years). 43 Table 2: The CHIP Rural Survey: Sample Sizes, Dibao Participation and Mean Income 2007 2008 2009 Sample size: number of individuals Present in the current year 31791 31506 31317 Present in the current year and prior year -- 30877 30208 Present all three years -- -- 29720 Sample size: number of households Present in the current year 8000 7994 7955 Present in the current year and prior year -- 7946 7882 Present all three years -- -- 7858 Dibao participation Number of individuals 531 662 910 Number of households 145 176 240 Mean income per capita (yuan, current prices) CHIP sample 4429 5096 5629 NBS 4140 4761 5153 Annual growth in mean income per capita (%, constant prices) CHIP sample na 8.0 10.8 NBS 9.5 7.9 8.6 Notes: Here and elsewhere, income is household net income per capita as measured by the NBS income definition. Constant-price growth rates are calculated using the NBS rural consumer price index (1.054 in 2007, 1.065 in 2008, and 0.997 in 2009). Sample sizes and the numbers of dibao participants are not weighted. CHIP sample mean incomes are weighted using two-level (province x region) weights. Sources: NBS income statistics are from NBS (various years). CHIP sample sizes and income per capita are calculated by the authors using the CHIP dataset. 44 Table 3: Inequality in Rural Household Income per Capita, 2007-2009 2007 2008 2009 Inequality in CHIP samples Gini 0.345 0.352 0.365 MLD 0.201 0.208 0.227 Theil 0.213 0.217 0.235 Coefficient of variation 0.798 0.803 0.866 Top 20% to bottom 20% 5.89 6.15 6.69 Top 10% to bottom 10% 10.06 10.76 12.00 Inequality from official publications (national) Gini 0.374 0.378 0.385 Top 20% to bottom 20% 7.27 7.53 7.95 Notes: Inequality in the CHIP samples is estimated over individuals in the nine provinces that are covered in all years of the CHIP survey (see text). Here and elsewhere, unless noted otherwise, all calculations using the CHIP data are weighted using two-level (province x region) population weights. Inequality from official publications is based on nationwide data covering all provinces. Sources: Official Gini coefficients are from the National Bureau of Statistics Rural Socioeconomic Survey Department (2010). Official income ratios of the top 20% to bottom 20% are calculated using average rural incomes of the top and bottom 20% reported by the NBS (various years). Estimates of inequality in the CHIP samples are calculated using the CHIP dataset. 45 Table 4: Poverty Lines (yuan per person per year) 2007 2008 2009 Absolute poverty lines Official poverty line 1976 2105 2098 $1.25 poverty line 1995 2125 2118 $2.00 poverty line 3191 3398 3388 Relative poverty lines 0.5 of median income 1808 2072 2245 0.6 of median income 2170 2487 2694 Notes: All poverty lines are in current prices. The official poverty line is the new official poverty line of 2300 yuan announced in 2011. We adjust this back to 2007, 2008 and 2009 using the rural consumer price index published by the NBS (various years). The $1.25 and $2 international poverty lines are converted to yuan using the 2005 PPP exchange rate of 4.09 (LCU per international dollar, World Development Indicators 2013, http://data.worldbank.org/data-catalog/world-development-indicators), and then adjusted forward to 2007, 2008 and 2009 using the rural consumer price index. Table 5: Poverty Incidence in Rural China, 2007-2009 (%) 2007 2008 2009 Absolute poverty lines Official poverty line 14.77 12.52 11.23 $1.25 poverty line 15.01 12.83 11.40 $2.00 poverty line 40.91 36.64 32.57 Relative poverty lines 0.5 of median income 11.84 12.12 13.29 0.6 of median income 18.46 19.34 20.35 Note: Calculated using reported incomes (including dibao transfers), with weights. 46 Table 6: Poverty Gaps (trillion yuan) % % change, change, Poverty line 2007 2008 2009 2007-08 2008-09 Official 60.506 58. 504 55.633 -9.21% -4.66% $1.25 62.448 60.245 57.159 -9.42% -4.84% $2.00 294.150 274.918 242.316 -12.24% -11.59% Note: Poverty gaps are in current prices; change over time is in constant prices. Calculated using reported incomes (including dibao transfers), with weights. The rural consumer price index is from NBS (various years). Table 7: Rural Dibao Thresholds and Transfers (yuan per capita per year) 2007 2008 2009 Dibao thresholds Average, all provinces 1064 1166 1428 Average, 9 provinces 1051 1151 1395 Dibao transfers Average county transfer, all provinces 580 707 979 Average county transfer, 9 provinces 569 697 974 Average village transfer, CHIP sample -- 732 845 Notes: Not weighted. Dibao thresholds and county-level transfers are calculated using official monthly county-level data and cover all counties, not just the CHIP counties. MOCA county- level data are monthly data. In this table, for the year 2007, we report the averages across counties for January 2008, multiplied by 12. For 2008 and 2009 we report the averages for December 2008 and December 2009, multiplied by 12. Sources: Thresholds and county transfers are from MOCA (various years); village transfers are calculated using the CHIP village-level data. 47 Table 8: Rural Dibao Participation Rates, 2007-2009 (%) 2007 2008 2009 Dibao participation rates, CHIP data Full sample 1.91 2.03 3.01 Hebei 0.493 0.337 0.869 Jiangsu 0.969 0.653 1.114 Zhejiang 0.934 0.894 0.667 Anhui 2.036 2.815 3.593 Henan 2.568 3.965 3.738 Hubei 1.366 1.298 1.509 Guangdong 1.259 3.476 4.847 Chongqing 2.020 3.473 6.483 Sichuan 2.859 1.223 3.184 Dibao participation rates, official data National 4.99 6.12 6.90 9 provinces in CHIP sample -- 4.60 5.57 Hebei -- 4.22 4.39 Jiangsu -- 3.59 4.03 Zhejiang -- 2.57 2.63 Anhui -- 5.10 5.96 Henan -- 4.48 6.16 Hubei -- 4.77 5.78 Guangdong -- 4.59 4.85 Chongqing -- 5.49 8.43 Sichuan -- 6.59 7.92 Notes: Calculated over individuals. The national CHIP participation rates are calculated using weights; the provincial CHIP rates are unweighted. CHIP dibao participation rates are self- reported by households; members of households that report participation in either the dibao or wubao program are counted as dibao participants. The NBS publishes statistics on the national number of rural dibao participants; we divide these by the rural population to obtain the official national participation rates. MOCA issues the provincial numbers of rural dibao participants by month. We use the December numbers divided by NBS provincial rural population statistics to calculate the official provincial and 9-province participation rates. As a check, we calculated the national participation rates using the MOCA December numbers, which gives participation rates of 5.94 in 2008 and 6.68 in 2009; these are consistent with the participation rates based on the NBS annual participation numbers shown in the table. Sources: NBS (various years); MOCA (various years), available only since 2007 48 Table 9: Proportion of Individuals with Income below the Local Dibao Threshold (%) % of all % of dibao Year individuals recipients 2007 2.42 4.52 Ex post income < dibao threshold (includes dibao transfer) 2008 2.64 2.42 2009 3.77 5.71 2007 -- -- Ex ante income < dibao threshold (net of village average dibao transfer) 2008 2.79 9.82 2009 3.97 12.53 2007 2.49 9.23 Ex ante income < dibao threshold (net of county average dibao expenditure) 2008 2.79 9.82 2009 4.05 15.27 Notes: Not weighted. For dibao lines we use the county-level December dibao thresholds from MOCA, which are available for 2008 and 2009; for 2007 we use the county-level dibao thresholds for January, 2008. Ex ante incomes net of village-level dibao transfers cannot be calculated for 2007 as village dibao transfer data are not available for that year. 49 Table 10: Poverty Incidence Calculated Using Ex Post and Ex Ante Incomes (%) 2007 2008 2009 Official poverty line Ex post income per capita 14.77 12.52 11.23 Ex ante income per capita (net of village avg. dibao transfer) -- 12.75 11.44 Ex ante income per capita (net of county avg. dibao expenditure) 14.92 12.68 11.62 $1.25 poverty line Ex post income per capita (including dibao transfer) 15.01 12.83 11.40 Ex ante income per capita (net of village avg. dibao transfer) -- 13.05 11.64 Ex ante income per capita (net of county avg. dibao expenditure) 15.16 13.01 11.79 $2.00 poverty line Ex post income per capita (including dibao transfer) 40.91 36.64 32.57 Ex ante income per capita (net of village avg. dibao transfer) -- 36.94 32.78 Ex ante income per capita (net of county avg. dibao expenditure) 41.07 36.90 33.04 Notes: Weighted. Sources: Authors’ calculations using the CHIP dataset and MOCA (various years) data on county average dibao transfers. 50 Table 11: The Poverty Gap and Dibao Expenditures 2007 2008 2009 Poverty gap (million yuan) Ex post income per capita 60506 58504 55633 Ex ante income per capita (net of county avg. dibao transfer) 61923 60222 59273 difference (%) 2.34% 2.94% 6.54% Total dibao expenditures MOCA total dibao expenditures (million yuan) 10910 22873 36300 as a % of ex ante poverty gap 17.6% 38.0% 61.2% CHIP total dibao expenditures (million yuan) 4950 6299 15261 as a % of ex ante poverty gap 8.0% 10.5% 25.7% Average reduction in the poverty gap per yuan dibao expenditure (yuan) Calculated using MOCA total expenditures 0.13 0.04 0.06 Calculated using CHIP total expenditures 0.29 0.27 0.24 Notes: Weighted. The poverty gap is calculated using the official poverty line (Table 4). Ex ante incomes are calculated by subtracting county average dibao expenditures from incomes reported in the CHIP data. MOCA total dibao expenditures are the official national totals (Table 1). CHIP total dibao expenditures are calculated as the (weighted) sum over all individuals receiving dibao in the CHIP sample of the county average transfer in the location of residence. Note that for dibao recipients who live in counties for which MOCA county average transfer data are not available, we use the village average transfers from CHIP (which are available only in 2008 and 2009). 51 Table 12: Targeting Errors (%) Measure of income per capita Error 2007 2008 2009 Inclusion 97.3 97.8 94.4 Ex post Exclusion 97.3 98.0 95.4 Ex ante, net of village avg. dibao Inclusion -- 92.1 89.4 transfer Exclusion -- 93.2 91.6 Ex ante, net of county avg. dibao Inclusion 93.6 92.3 85.7 expenditure Exclusion 93.7 93.3 89.1 Note: Weighted. Inclusion error equals the percent of dibao recipients who are not eligible (whose incomes are above the dibao thresholds); exclusion error equals the percent of eligible individuals (with incomes below the dibao thresholds) who do not receive dibao transfers. Table 13: Shares of Poor and Nonpoor Individuals Who Receive Dibao (%) Ex ante income estimated using village average dibao transfers 2007 2008 2009 Nonpoor Poor Nonpoor Poor Nonpoor Poor Official poverty line -- -- 1.58 5.16 2.55 6.57 $1.25 Poverty line -- -- 1.59 5.05 2.53 6.67 $2 Poverty line -- -- 1.24 3.40 2.35 4.36 Ex ante income estimated using county average dibao expenditures 2007 2008 2009 Nonpoor Poor Nonpoor Poor Nonpoor Poor Official poverty line 1.43 4.65 1.66 4.64 2.36 7.98 $1.25 Poverty line 1.43 4.58 1.63 4.76 2.36 7.86 $2 Poverty line 1.03 3.17 1.31 3.29 1.97 5.12 Notes: Weighted. Poverty classifications are based on ex ante incomes. See Table 4 for the poverty lines. 52 Table 14: Targeting Errors Relative to the Official Poverty Line (%) Measure of income per capita Error 2007 2008 2009 Ex ante, net of village avg. dibao Inclusion -- 67.7 75.0 transfer Exclusion -- 94.8 93.4 Ex ante, net of county avg. dibao Inclusion 63.6 71.1 69.2 expenditure Exclusion 95.3 95.4 92.0 Notes: Weighted. In this table inclusion and exclusion errors measure whether or not poor households receive dibao transfers. In other words, the inclusion error is the % of dibao recipients who had income above the poverty line, and the exclusion error is the % of individuals with income below the poverty line who were not dibao recipients. Poverty classifications are carried out using ex ante incomes. 53 Table 15: Characteristics of Dibao and Non-dibao Households, 2007 Dibao mean Non- as a dibao Dibao % of mean SD mean SD non-dibao Household characteristics Per capita income 5263 4347 3789 2859 72% Ex ante per capita income (village correction) . . . . Ex ante per capita income (county correction) 5263 4347 3369 2821 64% Household size 3.980 1.359 3.662 1.464 92% Average age of adult household members 41.71 9.568 45.79 11.661 110% Years of schooling of household head 7.487 2.337 6.752 2.503 90% Share of male household members 0.523 0.146 0.504 0.181 96% Share of household members age > 60 0.102 0.222 0.195 0.300 191% Share of household members age < 16 0.150 0.172 0.161 0.183 107% Existence of bad health household member (dummy) 0.137 0.344 0.407 0.493 297% Existence of disabled household member (dummy) 0.116 0.321 0.352 0.479 303% Existence of household member with migrant job (dummy) 0.408 0.491 0.352 0.479 86% Share net income from wages 0.426 0.414 0.315 0.293 74% Share net income from non-agricultural business 0.094 0.399 0.025 0.099 27% Household has no major appliance (refrigerators, etc) (dummy) 0.370 0.483 0.641 0.481 173% Household has motorized transport means (dummy) 0.475 0.499 0.193 0.396 41% Marriage in household (dummy) 0.046 0.21 0.062 0.242 135% Death in household (dummy) 0.036 0.185 0.034 0.183 94% Log housing area 4.798 0.518 4.476 0.532 93% Share of housing area that is multi-story 0.492 0.47 0.303 0.447 62% Household cultivated land area . . . . Water flush toilet (dummy) 0.271 0.444 0.131 0.339 48% Existence of piped water (dummy) 0.416 0.493 0.234 0.425 56% Village characteristics Natural disaster occurrence (dummy) 0.551 0.497 0.683 0.467 124% Revolutionary area (dummy) 0.028 0.164 0.048 0.215 171% Mountainous area (dummy) 0.015 0.123 0.014 0.117 93% Road covered by asphalt/cement (dummy) 0.437 0.496 0.297 0.458 68% Distance to township gov't > 10 km 0.012 0.108 0.007 0.083 58% Distance to county seat > 20 km 0.052 0.222 0.083 0.276 160% 54 Table 16: Characteristics of Dibao and Non-dibao Households, 2008 Dibao mean Non- as a dibao Dibao % of mean SD mean SD non-dibao Household characteristics Per capita income 6030 4893 4253 2778 71% Ex ante per capita income (village correction) 6030 4893 3608 2737 60% Ex ante per capita income (county correction) 6030 4893 3694 2745 61% Household size 3.945 1.39 3.761 1.481 95% Average age of adult household members 42.46 9.836 46.18 11.988 109% Years of schooling of household head 7.501 2.312 6.519 2.409 87% Share of male household members 0.522 0.148 0.527 0.199 101% Share of household members age > 60 0.116 0.239 0.213 0.312 184% Share of household members age < 16 0.139 0.166 0.133 0.174 96% Existence of bad health household member (dummy) 0.153 0.360 0.455 0.499 297% Existence of disabled household member (dummy) 0.120 0.325 0.358 0.481 298% Existence of household member with migrant job (dummy) 0.374 0.484 0.330 0.471 88% Share net income from wages 0.472 2.000 0.330 0.285 70% Share net income from non-agricultural business 0.065 1.899 0.030 0.125 46% Household has no major appliance (refrigerators, etc) (dummy) 0.331 0.471 0.585 0.494 177% Household has motorized transport means (dummy) 0.490 0.500 0.358 0.481 73% Marriage in household (dummy) 0.043 0.204 0.040 0.196 93% Death in household (dummy) 0.022 0.146 0.023 0.149 105% Log housing area 4.812 0.534 4.597 0.590 96% Share of housing area that is multi-story 0.511 0.465 0.335 0.456 66% Household cultivated land area 4.452 5.302 4.357 3.805 98% Water flush toilet (dummy) 0.293 0.455 0.119 0.325 41% Existence of piped water (dummy) 0.428 0.495 0.273 0.447 64% Village characteristics Natural disaster occurrence (dummy) 0.377 0.485 0.369 0.484 98% Revolutionary area (dummy) 0.028 0.165 0.051 0.221 182% Mountainous area (dummy) 0.015 0.121 0.028 0.167 187% Road covered by asphalt/cement (dummy) 0.468 0.499 0.415 0.494 89% Distance to township gov't > 10 km 0.012 0.107 0.023 0.149 192% Distance to county seat > 20 km 0.052 0.223 0.074 0.262 142% 55 Table 17: Characteristics of Dibao and Non-dibao Households, 2009 Dibao mean Non- as a dibao Dibao % of mean SD mean SD non-dibao Household characteristics Per capita income 6652 6033 4725 3282 71% Ex ante per capita income (village correction) 6652 6033 4130 3241 62% Ex ante per capita income (county correction) 6652 6033 3856 3146 58% Household size 3.94 1.42 3.79 1.555 96% Average age of adult household members 43.05 9.976 47.15 12.652 110% Years of schooling of household head 7.467 2.336 6.725 2.526 90% Share of male household members 0.522 0.149 0.511 0.179 98% Share of household members age > 60 0.128 0.252 0.251 0.333 196% Share of household members age < 16 0.129 0.162 0.123 0.161 95% Existence of bad health household member (dummy) 0.139 0.346 0.346 0.477 249% Existence of disabled household member (dummy) 0.089 0.285 0.267 0.443 300% Existence of household member with migrant job (dummy) 0.169 0.375 0.242 0.429 143% Share net income from wages 0.462 0.399 0.362 0.311 78% Share net income from non-agricultural business 0.067 0.269 0.015 0.072 22% Household has no major appliance (refrigerators, etc) (dummy) 0.259 0.438 0.486 0.501 188% Household has motorized transport means (dummy) 0.517 0.500 0.329 0.471 64% Marriage in household (dummy) 0.050 0.218 0.021 0.143 42% Death in household (dummy) 0.018 0.132 0.046 0.210 256% Log housing area 4.852 0.526 4.596 0.547 95% Share of housing area that is multi-story 0.511 0.465 0.361 0.456 71% Household cultivated land area 4.551 4.290 3.708 3.012 81% Water flush toilet (dummy) 0.364 0.481 0.231 0.422 63% Existence of piped water (dummy) 0.542 0.498 0.430 0.496 79% Village characteristics Natural disaster occurrence (dummy) 0.326 0.469 0.412 0.493 126% Revolutionary area (dummy) 0.036 0.187 0.045 0.207 125% Mountainous area (dummy) 0.021 0.144 0.039 0.193 186% Road covered by asphalt/cement (dummy) 0.506 0.500 0.408 0.493 81% Distance to township gov't > 10 km 0.014 0.119 0.033 0.18 236% Distance to county seat > 20 km 0.066 0.249 0.104 0.306 158% Note to Tables 15, 16 and 17: Unweighted. We also calculated the descriptive statistics using household size weights, with little difference in results, so we do not report them. 2007 values are calculated over 7855 non-dibao and 145 dibao households; 2008 and 2009 values are 56 calculated over 7818 and 176, and 7715 and 240, non-dibao and dibao households, respectively. For some variables the number of observations is lower due to some missing values. Table 18: Results of probit regressions (dependent variable =1 if the household receives dibao, =0 otherwise) 2007 2008 2009 (2) (1) (2) (1) (2) Log ex ante per capita income -0.0096*** -0.0086*** (village correction) (0.001) (0.002) Log ex ante per capita income -0.0068*** -0.0093*** -0.0108*** (county correction) (0.001) (0.002) (0.002) Household size -0.0027*** -0.0030*** -0.0031*** -0.0021** -0.0025** (0.001) (0.001) (0.001) (0.001) (0.001) Average age of adult household -0.0001 -0.0002 -0.0002 -0.0002 -0.0002 Members (0.000) (0.000) (0.000) (0.000) (0.000) Share of male household members -0.0130** -0.0007 -0.0005 -0.0080 -0.0080 (0.006) (0.005) (0.005) (0.007) (0.007) Share of household members age > 60 -0.0001 0.0017 0.0019 0.0129** 0.0110* (0.005) (0.005) (0.005) (0.006) (0.006) Share of household members age < 16 0.0078 -0.0007 -0.0002 0.0058 0.0046 (0.005) (0.006) (0.006) (0.008) (0.007) Existence of bad health household 0.0113*** 0.0162*** 0.0171*** 0.0101** 0.0108** member (0.004) (0.004) (0.004) (0.004) (0.004) Existence of disabled household 0.0159*** 0.0144*** 0.0146*** 0.0333*** 0.0319*** member (0.005) (0.005) (0.005) (0.009) (0.009) Household member with migrant job -0.0010 -0.0011 -0.0009 0.0105** 0.0101** (0.002) (0.002) (0.002) (0.005) (0.005) Share net income from wages -0.0075** -0.0084*** -0.0090*** -0.0169*** -0.0157*** (0.003) (0.003) (0.003) (0.004) (0.004) Share net income from non-agricultural -0.0124*** -0.0079*** -0.0085*** -0.0180*** -0.0158*** business (0.005) (0.003) (0.003) (0.005) (0.005) Household has no major appliance 0.0039* 0.0041* 0.0041* 0.0036 0.0033 (0.002) (0.002) (0.002) (0.003) (0.003) Household has motorized transport -0.0067*** -0.0014 -0.0015 -0.0044 -0.0038 (0.002) (0.002) (0.002) (0.003) (0.003) Natural disaster occurrence 0.0027 -0.0012 -0.0011 0.0009 0.0014 (0.002) (0.002) (0.002) (0.002) (0.002) Marriage in household 0.0005 0.0046 0.0048 -0.0090*** -0.0076** (0.004) (0.006) (0.006) (0.003) (0.003) Death in household 0.0029 0.0003 -0.0008 0.0397** 0.0381** (0.006) (0.006) (0.005) (0.018) (0.018) Log housing area -0.0040* 0.0012 0.0011 -0.0030 -0.0027 (0.002) (0.002) (0.002) (0.003) (0.003) Share multi-story area -0.0013 -0.0032 -0.0030 0.0014 0.0013 (0.002) (0.002) (0.002) (0.003) (0.003) Household cultivated land area (mu) -0.0000 -0.0000 -0.0012*** -0.0011*** (0.000) (0.000) (0.000) (0.000) Water flush toilet 0.0027 -0.0070*** -0.0071*** -0.0023 -0.0018 (0.003) (0.002) (0.002) (0.003) (0.003) 57 Piped water -0.0018 -0.0012 -0.0013 -0.0026 -0.0024 (0.002) (0.002) (0.002) (0.003) (0.002) Revolutionary area 0.0101 0.0148 0.0152 -0.0033 -0.0034 (0.010) (0.012) (0.012) (0.006) (0.005) Mountainous area -0.0063** -0.0011 -0.0012 0.0004 -0.0008 (0.003) (0.006) (0.006) (0.009) (0.008) Road covered by asphalt/cement 0.0002 0.0032* 0.0034* 0.0003 0.0007 (0.002) (0.002) (0.002) (0.003) (0.002) Distance to township gov't > 10 km -0.0054 0.0029 0.0029 0.0108 0.0108 (0.004) (0.009) (0.009) (0.013) (0.013) Distance to county seat > 20 km 0.0066 -0.0002 -0.0003 0.0093 0.0099 (0.006) (0.004) (0.004) (0.007) (0.007) Log likelihood -601.07 -673.97 -679.95 -778.72 -775.27 2 Likelihood ratio test χ (33) 239.19 333.91 321.97 321.91 335.96 2 Pseudo R .166 .199 .191 .171 .178 Observations 7,971 7,952 7,952 7,358 7,359 Notes: Estimated over households, without weights. Estimation was also done using the household-size as weights; the coefficients were very similar and equality could not be rejected using a Hausman test. The table reports marginal effects, evaluated at the mean of the data. Standard errors are in parentheses. The regressions also included controls for province fixed effects (not reported). Anhui, Henan, Chongqing and Guangdong had significant, positive coefficients with Hebei as reference province. The regressions were estimated including some additional explanatory variables such as years of schooling, but since the coefficients were uniformly not significant, these variables were dropped. Statistically significant coefficients are shown in red. *** p<0.01, ** p<0.05, * p<0.1. 58 Table 19: Targeting Performance Based on Propensity Scores (% of individuals) All Receiving dibao Not receiving dibao Year Ineligible Eligible Total Ineligible Eligible Total Ineligible Eligible Total 2007 98.09 1.91 100.00 1.59 0.32 1.91 96.50 1.59 98.09 2008 97.97 2.03 100.00 1.63 0.40 2.03 96.34 1.63 97.97 2009 96.99 3.01 100.00 2.51 0.50 3.01 94.48 2.51 96.99 Notes: Percentages of CHIP sample individuals in each year, calculated using weights. Eligibility is determined using the propensity score method. Propensity scores are calculated from the probit estimates in Table 18 (specification 2). The propensity score threshold for each year is created by counting off individuals ranked from highest to lowest propensity score, starting from the highest propensity score, until reaching the number of dibao individuals in the survey that year. By construction, in the propensity score approach the number of eligible individuals is exactly equal to the number of recipient individuals. Consequently, column 2 (all ineligible) is identical to the last column (total not receiving dibao), and column 3 (all eligible) is identical to column 6 (eligible receiving dibao). In reality, the number of eligible individuals exceeds the number of recipients. Table 20: Targeting Errors Using Propensity Scores (% of households) Error 2007 2008 2009 Inclusion 83.2 80.3 83.4 Exclusion 83.2 80.3 83.4 Note: Weighted. Calculated from the numbers in Table 19. Inclusion error is the percent of dibao recipients who are not eligible according to the propensity score method; exclusion error equals the percent of eligible individuals (according to the propensity score method) who did not receive dibao transfers. By construction, in the propensity score approach inclusion and exclusion errors are the same because the number of eligible individuals is exactly equal to the number of recipient individuals. In reality, the number of eligible individuals exceeds the number of recipients, so in that in the conventional targeting analysis the exclusion errors are larger than the inclusion errors (Table 12). 59 Table 21: Simulations: Expanding Coverage versus Increasing Transfers, or a Uniform Transfer (1) (2) (3) (4) (5) (6) (7) Change in poverty relative to baseline (%) Number Povert Squared Budget (mill. of Transfer Poverty y gap Poverty Headcou Square Yuan) recipients amounts headcount index gap nt Gap d gap Baseline (“observed”) 13580 20398820 Local 11.23 3.91 7.67 a) Expand coverage (to all eligible) 2.54 x base 44211569 Local 10.64 2.97 6.33 -5.25 -24.04 -17.47 b) Expand coverage (lottery) 2.54 x base 52322970 Local 10.88 3.79 7.58 -3.12 -3.07 -1.17 2.54 x c) Increase transfer 2.54 x base 20398820 local 10.89 3.78 7.59 -3.03 -3.32 -1.04 d) Uniform transfer base 20398820 666 Yuan 11.17 3.93 7.67 -0.53 0.51 0.00 Notes: The baseline case is calculated using rural population weights and observed dibao participation in the 2009 CHIP data. Dibao transfer amounts are assumed to equal the local average in the village of residence (where village data are missing, we use the county average from MOCA). Poverty is calculated using the official poverty line. Simulations (a) – (d) assume that dibao transfers continue to go to all recipients in the baseline case. The expanded budget used in these simulations is 2.54 times the base budget, which is the amount of funding required by simulation (a) in which transfers continue to go to original recipients as well as to any other individuals who were not original recipients but who are eligible, i.e., whose incomes are below their local dibao thresholds. Simulation (b) assumes that the program is expanded by adding additional recipients who are selected randomly from among all non-recipients until the budget is exhausted. Simulations (c) and (d) do not add any new dibao recipients, but explore changing the amount of the dibao transfers. Simulation (c) increases the transfer received by each dibao recipient in the baseline by 2.54 times. Simulation (d) replaces the transfer received by each dibao recipient in the baseline case with a uniform transfer equal to the average baseline transfer of 666 Yuan. 60 Table 22: Simulations: Uniform Transfer versus a Uniform Eligibility Threshold (1) (2) (3) (4) (5) (6) (7) Change in poverty relative to baseline (%) Budget Transfer Poverty Squared (mill. Number of amount Poverty gap Poverty Headco Square yuan) recipients in Yuan headcount index gap unt Gap d gap Baseline (perfect dibao targeting) 23710 26717666 Local 10.85 3.04 6.35 e) Uniform transfer base 26717666 887 10.75 2.68 6.19 -0.92 -11.84 -2.52 f) Uniform threshold (distance to poverty line) base 32405357 Local 10.90 2.72 6.19 0.46 -10.53 -2.52 g) Uniform threshold (lottery among the poor) base 34302984 Local 9.94 2.81 6.53 -8.39 -7.57 2.83 h) Uniform threshold and uniform transfer base 26705957 887 10.96 2.49 6.07 1.01 -18.09 -4.41 (distance to poverty line) i) Uniform threshold and uniform transfer base 26699889 887 10.00 2.61 6.43 -7.83 -14.14 1.26 (lottery among the poor) Notes: The baseline in this table is a simulation in which there is perfect targeting based on the dibao thresholds: all individuals with income below their local dibao thresholds receive the local dibao transfers, and no individuals with income above their local dibao thresholds receive dibao transfers. Simulation (e) is the same as the baseline except local transfer amounts are replaced with a uniform transfer equal to the average transfer in the baseline (887 Yuan). Simulation (f) assumes a uniform threshold equal to the official poverty line, with perfect targeting based on depth of poverty. Recipients are selected starting with the poorest (those furthest below the official poverty line) and given the local transfer until the baseline budget is used up. Simulation (g) also assumes a uniform threshold equal to the official poverty line, but here dibao recipients are randomly selected from among the poor and given the local transfer until the baseline budget is used up. Simulation (h) is the same as simulation (f) but transfers are now uniform and equal to the average transfer in the baseline. Simulation (i) is the same as simulation (g) but transfers are now uniform and equal to the average transfer in the baseline. In all cases poverty levels are calculated using the official poverty line. 61 Social Protection & Labor Discussion Paper Series Titles 2012-2014 No. Title 1423 Any Guarantees? An Analysis of China’s Rural Minimum Living Standard Guarantee Program by Jennifer Golan, Terry Sicular and Nithin Umapathi, August 2014 1422 World Bank Support for Social Safety Nets 2007-2013: A Review of Financing, Knowledge Services and Results by Colin Andrews, Adea Kryeziu and Dahye Seo, June 2014 1421 STEP Skills Measurement Surveys: Innovative Tools for Assessing Skills by Gaëlle Pierre, Maria Laura Sanchez Puerta, Alexandria Valerio and Tania Rajadel, July 2014 1420 Our Daily Bread: What is the Evidence on Comparing Cash versus Food Transfers? by Ugo Gentilini, July 2014 1417 Benin: Les Filets Sociaux au Bénin Outil de Réduction de la Pauvreté par Andrea Borgarello et Damien Mededji, Mai 2011 1416 Madagascar Three Years into the Crisis: An Assessment of Vulnerability and Social Policies and Prospects for the Future by Philippe Auffret, May 2012 1415 Sudan Social Safety Net Assessment by Annika Kjellgren, Christina Jones-Pauly, Hadyiat El-Tayeb Alyn, Endashaw Tadesse and Andrea Vermehren, May 2014 1414 Tanzania Poverty, Growth, and Public Transfers: Options for a National Productive Safety Net Program by W. James Smith, September 2011 1413 Zambia: Using Social Safety Nets to Accelerate Poverty Reduction and Share Prosperity by Cornelia Tesliuc, W. James Smith and Musonda Rosemary Sunkutu, March 2013 1412 Mali Social Safety Nets by Cécile Cherrier, Carlo del Ninno and Setareh Razmara, January 2011 1411 Swaziland: Using Public Transfers to Reduce Extreme Poverty by Lorraine Blank, Emma Mistiaen and Jeanine Braithwaite, November 2012 1410 Togo: Towards a National Social Protection Policy and Strategy by Julie van Domelen, June 2012 1409 Lesotho: A Safety Net to End Extreme Poverty by W. James Smith, Emma Mistiaen, Melis Guven and Morabo Morojele, June 2013 1408 Mozambique Social Protection Assessment: Review of Social Assistance Programs and Social Protection Expenditures by Jose Silveiro Marques, October 2012 1407 Liberia: A Diagnostic of Social Protection by Andrea Borgarello, Laura Figazzolo and Emily Weedon, December 2011 1406 Sierra Leone Social Protection Assessment by José Silvério Marques, John Van Dyck, Suleiman Namara, Rita Costa and Sybil Bailor, June 2013 1405 Botswana Social Protection by Cornelia Tesliuc, José Silvério Marques, Lillian Mookodi, Jeanine Braithwaite, Siddarth Sharma and Dolly Ntseane, December 2013 1404 Cameroon Social Safety Nets by Carlo del Ninno and Kaleb Tamiru, June 2012 1403 Burkina Faso Social Safety Nets by Cécile Cherrier, Carlo del Ninno and Setareh Razmara, January 2011 1402 Social Insurance Reform in Jordan: Awareness and Perceptions of Employment Opportunities for Women by Stefanie Brodmann, Irene Jillson and Nahla Hassan, June 2014 1401 Social Assistance and Labor Market Programs in Latin America: Methodology and Key Findings from the Social Protection Database by Paula Cerutti, Anna Fruttero, Margaret Grosh, Silvana Kostenbaum, Maria Laura Oliveri, Claudia Rodriguez-Alas, Victoria Strokova, June 2014 1308 Youth Employment: A Human Development Agenda for the Next Decade by David Robalino, David Margolis, Friederike Rother, David Newhouse and Mattias Lundberg, June 2013 1307 Eligibility Thresholds for Minimum Living Guarantee Programs: International Practices and Implications for China by Nithin Umapathi, Dewen Wang and Philip O’Keefe, November 2013 1306 Tailoring Social Protection to Small Island Developing States: Lessons Learned from the Caribbean by Asha Williams, Timothy Cheston, Aline Coudouela and Ludovic Subran, August 2013 1305 Improving Payment Mechanisms in Cash-Based Safety Net Programs by Carlo del Ninno, Kalanidhi Subbarao, Annika Kjellgren and Rodrigo Quintana, August 2013 1304 The Nuts and Bolts of Designing and Implementing Training Programs in Developing Countries by Maddalena Honorati and Thomas P. McArdle, June 2013 1303 Designing and Implementing Unemployment Benefit Systems in Middle and Low Income Countries: Key Choices between Insurance and Savings Accounts by David A. Robalino and Michael Weber, May 2013 1302 Entrepreneurship Programs in Developing Countries: A Meta Regression Analysis by Yoonyoung Cho and Maddalena Honorati, April 2013 1301 Skilled Labor Flows: Lessons from the European Union by Martin Kahanec, February 2013 1220 Evaluating the Efficacy of Mass Media and Social Marketing Campaigns in Changing Consumer Financial Behavior by Florentina Mulaj and William Jack, November 2012 1219 Do Social Benefits Respond to Crises? Evidence from Europe & Central Asia During the Global Crisis by Aylin Isik-Dikmelik, November 2012 1218 Building Results Frameworks for Safety Nets Projects by Gloria M. Rubio, October 2012 1217 Pension Coverage in Latin America: Trends and Determinants by Rafael Rofman and Maria Laura Oliveri, June 2012 1216 Cash for Work in Sierra Leone: A Case Study on the Design and Implementation of a Safety Net in Response to a Crisis by Colin Andrews, Mirey Ovadiya, Christophe Ribes Ros and Quentin Wodon, November 2012 1215 Public Employment Services, and Activation Policies by Arvo Kuddo, May 2012 1214 Private Pension Systems: Cross-Country Investment Performance by Alberto R. Musalem and Ricardo Pasquini, May 2012 1213 Global Pension Systems and Their Reform: Worldwide Drivers, Trends, and Challenges by Robert Holzmann, May 2012 1212 Towards Smarter Worker Protection Systems: Improving Labor Regulations and Social Insurance Systems while Creating (Good) Jobs by David A. Robalino, Michael Weber, Arvo Kuddo, Friederike Rother, Aleksandra Posarac and Kwabena Otoo 1211 International Patterns of Pension Provision II: A Worldwide Overview of Facts and Figures by Montserrat Pallares-Miralles, Carolina Romero and Edward Whitehouse, June 2012 1210 Climate-Responsive Social Protection by Anne T. Kuriakose, Rasmus Heltberg, William Wiseman, Cecilia Costella, Rachel Cipryk and Sabine Cornelius, March 2012 1209 Social Protection in Low Income Countries and Fragile Situations: Challenges and Future Directions by Colin Andrews, Maitreyi Das, John Elder, Mirey Ovadiya and Giuseppe Zampaglione, March 2012 1208 World Bank Support for Pensions and Social Security by Mark Dorfman and Robert Palacios, March 2012 1207 Labor Markets in Middle and Low Income Countries: Trends and Implications for Social Protection and Labor Policies by Yoonyoung Cho, David Margolis, David Newhouse and David Robalino, March 2012 1206 Rules, Roles and Controls: Governance in Social Protection with an Application to Social Assistance by Lucy Bassett, Sara Giannozzi, Lucian Pop and Dena Ringold, March 2012 1205 Crisis Response in Social Protection by Federica Marzo and Hideki Mori, March 2012 1204 Improving Access to Jobs and Earnings Opportunities: The Role of Activation and Graduation Policies in Developing Countries by Rita Almeida, Juliana Arbelaez, Maddalena Honorati, Arvo Kuddo, Tanja Lohmann, Mirey Ovadiya, Lucian Pop, Maria Laura Sanchez Puerta and Michael Weber, March 2012 1203 Productive Role of Safety Nets by Harold Alderman and Ruslan Yemtsov, March 2012 1202 Building Social Protection and Labor Systems: Concepts and Operational Implications by David A. Robalino, Laura Rawlings and Ian Walker, March 2012 1201 MicroDeterminants of Informal Employment in the Middle East and North Africa Region by Diego F. Angel-Urdinola and Kimie Tanabe, January 2012 To view Social Protection & Labor Discussion papers published prior to 2012, please visit www.worldbank.org/spl Abstract This paper examines China’s rural minimum living standard guarantee (dibao) program, one of the largest targeted transfer schemes in the world. Using household survey data matched with published administrative data, we provide background on the patterns of inequality and poverty in rural China, describe the dibao program, estimate the program’s impact on poverty, and carry out targeting analysis. We find that the program provides sufficient income to poor beneficiaries but does not substantially reduce the overall level of poverty, in part because the number of beneficiaries is small relative to the number of poor. Conventional targeting analysis reveals rather large inclusionary and exclusionary targeting errors; propensity score targeting analysis yields smaller but still large targeting errors. Simulations of possible reforms to the dibao program indicate that expanding coverage can potentially yield greater poverty reduction than increasing transfer amounts. In addition, replacing locally diverse dibao lines with a nationally uniform dibao threshold could in theory reduce poverty. The potential gains in poverty reduction, however, depend on the effectiveness of targeting. About this series... Social Protection & Labor Discussion Papers are published to communicate the results of The World Bank’s work to the development community with the least possible delay. This paper therefore has not been prepared in accordance with the procedures appropriate for formally edited texts. The findings, interpretations, and conclusions expressed herein are those of the author(s), and do not necessarily reflect the views of the International Bank for Reconstruction and Development/The World Bank and its affiliated organizations, or those of the Executive Directors of The World Bank or the governments they represent. The World Bank does not guarantee the accuracy of the data included in this work. The boundaries, colors, denominations, and other information shown on any map in this work do not imply any judgement on the part of The World Bank concerning the legal status of any territory or the endorsement or acceptance of such boundaries. For more information, please contact the Social Protection Advisory Service, The World Bank, 1818 H Street, N.W., Room G7-803, Washington, DC 20433 USA. Telephone: (202) 458-5267, Fax: (202) 614-0471, E-mail: socialprotection@worldbank.org or visit us on-line at www.worldbank.org/spl. © 2013 International Bank for Reconstruction and Development / The World Bank