1. The effects of Social Assistance on Poverty Reduction in Urban China Du Yang Institute of Population and Labor Economics, CASS

2. Main contents The pattern of social assistance in the past decade Data Poverty targeting and coverage The effects of social assistance programs Conclusion

3. Social Assistance in Urban Areas During the period of planning economy, comprehensive employment had been implemented in urban areas through dual economic system Urban workers, mostly workers in SOEs, have attached their welfare with job, namely, work unit (danwei) In other words, the assistance system was not socialized

4. Social Assistance in Urban Areas A radical economic restructuring happened in urban China in the middle of 1990s, especially in SOEs One of the direct outcomes is labor market dislocation in urban areas, which generated massive unemployment and decreasing labor force participation

5. Unemployment rates and LFP in urban labor market

6. Social Assistance in Urban Areas some laid-off or unemployed workers fell into poverty, because unemployment disconnected any welfare they had before Some temporary measures were implemented to eradicate the poverty due to labor market shocks A widely used approach is to establish reemployment center in firms that pays living allowance and Or lump-sum payment to who lost jobs: buying out working ages, early retirement,….

7. Social Assistance in Urban Areas The temporary measures are not compatible with principles on labor market Shanghai firstly piloted the Minimum Living Standard Guarantee (dibao) program in 1993, which support the poor whose income below local dibao line In the next year, Ministry of Civil Affair proposed to extend Shanghais practice to the other urban areas of China All cities and the towns where county government locate were required to set up the program since 1999 In 2003, the Ministry of Civil Affair claimed that in 2002 Dibao program has covered all the poor whose income are below local Dibao line

8. the coverage and subsidy of Dibao: 1999-2006

9. Data Data used in this paper collected from five big cities, Shanghai, Wuhan, Shenyang, Fuzhou, Xian, in 2001 and 2005 respectively In both waves of survey, proportional population sampling approach was used to sample an average of 15 households in each neighborhood clusters and 10 of the households were interviewed We may take advantage of the timing of data collection to observe: (1) the way of social assistance changes (2) the effects of social assistance changes

10. Poverty Lines in Urban Areas No universal poverty lines for urban areas Dibao line has been frequently employed to evaluate poverty but is criticized for laid down by fiscal capacity of local government instead of poverty status Meng et al (2005) proposed line based on CBN methods The World Bank lines

11. Various Poverty Lines Used in this Study ShanghaiWuhanShenyangFuzhouXian CULS1: 2001 Dibao33602340 24001872 CBN_up40682472231626402376 $/day1124 2$/day2247 3$/day3371 Subjective3225 CULS2: 2005 Dibao34802316246025252400 CBN_up41882616238827242424 $/day1124 2$/day2247 3$/day3371 Subjective3225

12. Dibao line was published by city governments at current price in 2001 and 2005 respectively The Bank lines are adjusted as a comparable price When applying fixed lines, we convert the CULS incomes and transfers to 2003 prices by dividing them by provincial urban CPI indices for 2005 to create values in 2004 urban prices, then dividing those values in 2004 yuan by the variable of spatial price indice data provided by Brant and Holz (2004).

13. Targeting and Coverage Using NBS data, Chen et al. (2005) find that 43% households who are covered by Dibao actually are not qualified and meanwhile 72% household that should be covered are ignored In our sample, the leakage rate is about 48.5%, which is lower that Chen et al. 2005 and we found that 41.7% targeted households are not qualified, which is similar to their study. An international comparison study (Coady et al, 2004.) displays that some other programs similar to Dibao have large rates of leakage The coverage of dibao is not perfect, but pretty good.

14. The Effects on Poverty Reduction: Poverty Incidence For poverty incidence changes, in 2001 Xiagang subsidy had most significant impacts on poverty reduction while in 2005 Dibao is the major tool Poverty incidence rates are sensitive to the lines, which means research on reasonable poverty line is essential for urban poverty

15. Concentration Curve of Income Transfer

16. The Effects on Poverty Reduction: Income Transfer In 2001, the three curves have similar shape, which means diversity of means to poverty reduction. the poorest 20 percent population gets about 35% unemployment transfers, about 40% xiagang subsidy, and 55% Dibao transfers In 2005, the poorest 20 percent population gets about 25% unemployment transfers, about 25% xiagang subsidy, and more than 80% Dibao transfers. This also indicates that Dibao has been the dominant social assistance program in urban China

17. The Effects on Poverty Reduction: Behaviors We care about the behavior changes after receiving income transfer, which also indicates the effects of social assistance Such as, labor supply, education, health, and food consumption PSM is applied to match the treated and non-treated poor

18. Average Treatment Effects of Dibao Receipt, Matching Estimates E[Y 1 |D=1] (1) E[Y 0 |D=0] (2) E[Y 0 |D=1] (3) ATT (1)-(3) Nearest neighboring Working0.4690.5860.512-0.042 (1.13) Labor market participation0.6090.6260.627-0.017 (0.56) Health expenditure788.0724.9546.1241.9 (1.00) Food expenditure/total expenditure 0.5310.4600.4890.041(2.27) Education expenditure/total expenditure 0.1710.0750.0970.074 (5.02) Kernel matching Working0.4690.5860.546-0.077 (2.15) Labor market participation0.6090.6260.629-0.02 (0.49) Health expenditure per capita788.0724.9628.2159.8 (0.61) Food expenditure/total expenditure 0.5310.4600.4990.032 (1.93) Education expenditure/total expenditure 0.1710.0750.0970.074 (5.11)

19. Conclusions Although the social assistance system is still a combination of several policy instruments, Dibao has played as a dominant role other than other tools targeting is not as perfect as claimed, but it still has a relative good coverage compared to some other programs implemented in other economies distribution of transfer also shows that the poorest 20% population gets more than 80% of Dibao transfer, which means that the implementation of the program functions pretty well we found that Dibao increased the poors consumption on food, health, and education. Meanwhile, it also decreased labor supply of the treated families on average