Bj?rn Gustafsson (corresponding author)
    Department of Social Work
    University of G?teborg
    P.O. Box 720
    SE 405 30 G?teborg
    Sweden
    Fax (+46) 31 773 18 88
    &
    Institute for the Study of Labor (IZA), Bonn, Germany
    E-mail: Bjorn.Gustafsson@socwork.gu.se
    and
    Li Shi
    Institute of Economics
    Chinese Academy of Social Sciences
    2 Yuetan Beixiaojie
    Beijing
    P.R. China
    Fax (+86) 10 68032473
    &
    Institute for the Study of Labor (IZA), Bonn, Germany
    E-mail: lishi89@263.net
    June 2003
     
     
     
     
     
     
     
     
    Abstract
    Household health and education expenditures have increased rapidly in rural China. Based on data from households in eighteen provinces in 1988 and in 1995 we investigate how such expenditures affect poverty assessments. After accounting for these expenditures we cannot state that poverty decreased during this period of rapid economic growth. Further, poverty in China in 1995 appears to be even more concentrated in the west and in officially designated poor areas when education and health expenditures are considered.
    JEL code: I32, O53, P24
    Key words: China, Education, Health Care, Poverty
    1.      Introduction
    The introduction of the household responsibility system in China at the end of the 70s, as well as rural industrialisation has generated  rapid growth in household income. As a consequence, rural poverty (as commonly assessed) decreased rapidly in the earliest phase of reform, although later phases showed little poverty reduction.[1] Still, there is another side to how the transition has affected poverty among rural inhabitants that has not attracted attention in earlier writings on poverty in rural China; this situation has never been assessed, nor can it be inferred from official statistics. Rural inhabitants today spend a larger proportion of their budget on education and health care than was necessary before.
    Before economic reform was introduced, rural inhabitants typically had access to collectively financed health care. In addition, the central government channelled considerable resources for education to lower levels of government. The abolition of the collective organisation as well as decreased public funding from the central level have put the provision of health care and education under financial pressure.[2]Some rural inhabitants might have fallen into poverty after being required to pay for health care and education. This paper investigates how household expenditures for health and education affect the assessment of poverty in rural China. Our household data covers eighteen provinces in 1988 and 1995. We analyse changes in the extent of poverty as well as the profile of poverty. 
    2.      Sample and definitions
    Our data come from two surveys conducted by the Institute of Economics, Chinese Academy of Social Sciences, with the assistance of the State Statistical Bureau in Beijing. The first survey of household income in 1988 was implemented in the spring of 1989; the second surveys income in 1995 and was collected between January and March 1996. Both samples were derived from larger samples of the State Statistical Bureau. Members of the research team designed the questionnaires which contain fairly comprehensive questions about household income and its components. We can use these answers to derive the subsistence income of rural residents. The questionnaires asked for the present value of private houses, which enables us to impute rents by adopting a discount rate. Household expenditures were recorded and grouped into categories.[3]
    From the two surveys we derive our working sample of households from the same provinces for both years, a process which yields households from eighteen provinces. These are then grouped into three regions: Liaoning, Jiangsu, Zhejiang, Shandong and Guangdong are grouped into the eastern region; the central region includes Hebei, Shanxi, Jilin, Anhui, Jiangxi, Henan, Hubei and Hunan; Sichuan, Guizhou, Yunnan, Shaanxi and Gansu are in the western region. There are 4,178 households and 20,502 individuals in the 1988 sample and 7,888 households and 34,318 individuals in the 1995 sample.
    Our analysis on the importance of education and health expenditures when assessing poverty in rural China is based on four measures of disposable income defined at the household level. First there is the sum of money income (wages, salaries, net revenue from farming and business), subsistence income, income in-kind from owner-occupied housing as well as transfer income net of income taxes and fees. This measure (Disposable income I) closely resembles that which is used when producing official statistics for rural China.
    Assume that education is compulsory for children of a certain age and households are requested to pay schooling fees. This means that such educational expenditures can be regarded as taxes. Further assume that health care utilisation is a response to an unforeseen and unsolicited "shock" and can be so costly as to represent a threat to the household's ability to purchase other goods and services. Under these conditions, it is important to include health and education expenditures with a negative sign into the definition of disposable income.[4]  By subtracting the two expenditure groups individually and jointly we arrive at three additional definitions of disposable income.[5]
    In the analysis we classify households and their members as poor or not poor according to each of the four definitions of income. We apply the World Bank World Poverty line which is defined as 1 USD Purchasing Poverty Parity per person per day (1985 prices), see World Bank (2000). [6]  This poverty line has the advantage of allowing for international comparisons. This poverty line is higher than the poverty line used when producing official statistics for rural China, which we also employ as asensitivity test.. We measure the extent of poverty using the family of indices suggested by Foster et al (1984).
    3.      Results on how poverty has changed
    /Table 1 about here/
    First we provide a general overview of the distribution of education and health expenditures and Disposable income I in rural China . Table 1 provides information on mean values for various deciles defined by Disposable income I for 1988 and 1995. There was rapid expenditure growth between those years as average health expenditures increased by as much as 192 percent and average education expenditures by 165 percent. Expenditure shares also increased rapidly to 2.2 percent for educational expenditures and 3.6 for health expenditures.
    /Figure 1 about here/
    Although mean expenditures on education and health increase with disposable income, expenditure shares decrease rapidly by decile . This is true for both years under study. For example, while thelowest decile spent 4.6 percent on education and 7.3 percent on health expenditures in 1995, the corresponding percentages for the highest decile was 1.0 and 2.2 percent respectively. Figure 1 illustrates the strong negative relation between income and expenditure shares. It also illustrates that for the same level of income, expenditure shares were considerably higher in 1995 than in 1988. [7]   
    /Table 2 about here/
    Table 2 provides our poverty estimates for rural China as a whole in 1988 and 1995. We highlight several observations.[8]  First, education expenditures have a stronger impact than health expenditures on the poverty indices . Together the two expenditures pushed 1.9 percent of the sample into poverty in 1988 while the corresponding proportion reached 4.3 percent in 1995.[9] Second,  alternative measures of disposable income show no change in poverty rates although we observe a moderate decrease in the poverty using Disposable income I.[10]
    We also conclude that solely examining the proportion of inhabitants pushed into poverty provides only part of the picture of how health and education expenditures affect poverty. Such expenditures also push some people further into poverty. This effect is illustrated by an increase in poverty gaps using the alternative disposable income measures.
    A fourth conclusion is that most of the results are also obtainable when setting the poverty line at the level used when producing official statistics (see Table 2b). However, in this case the head count ratio in the last column of the table indicates a small decrease from 1988 to 1995 while the poverty gap squared shows a rapid increase.
    [BF1]
    4.      Results on the poverty profile
    /Table 3 about here/
    Table 3 reports estimates of poverty indices for the three regions of China in 1995 using the four definitions of disposable income. The estimates show health and education expenditures force more people into poverty in the less developed western region compared to other regions. The proportion is 6.0 percent for the west compared with 4.6 percent in the middle region and not more than 2.7 percent in the eastern region. These findings are robust with regard to choice of poverty index, except when we account for education expendituresand measure the poverty gap squared, in which case the difference is largest in the eastern region.
    /Table 4 about here/
    Much of China's poverty reduction strategy targets resources to officially designated poor counties. Poor areas are designated at both the central and the provincial level. In 1995 one out of four rural inhabitants in China lived in an officially designated poor area. Table 4 shows that poverty was greater in the poor counties than in other areas.  More important however, is that the proportion of the population pushed into poverty by health and education expenditures was larger in the poor counties than in other locations. The other poverty indices indicate the same results,  except for the poverty gap squared index, where the consequences of expenditures on education were not larger in the poor areas than elsewhere.
    5.      Conclusions
    In this paper we have shown that education and health expenditures in rural China increased rapidly between 1988 and 1995 and constitute a much higher share of expenditures among people in low-income households than among high income households. Taking these expenditures into consideration means that it is no longer possible to state that poverty decreased during this period of rapid economic growth. Further, taking health and education expenditures into consideration shows rural poverty in China in 1995 was more concentrated in the western region and in officially designated poor areas. We feel that the results of this exercise have shown that poverty measures in rural China should account for  household expenditures on education and health care.  
    References
    Blom, G. and Gu, X. (1997) "Health Sector Reform: Lessons from China", Social Science and Medicine, 45, 351 - 360.
    Bloom, G. and Tang, S. (1999) "Rural Health Prepayment Schemes in China: Towards a More Active Role for Government", Social Science and Medicine, 951 - 960.
    Bray, M. and Borevskaya, N. (2001) "Financing Education in Transitional Societies: lessons from Russia and China", Comparative Education, 37, 345 - 365.
    Beach, M. (2001) "China's Rural Health Care Gradually Worsens", Lancet, 358, 567.
    Citro, C. and Michael, R. (1995) Measuring Poverty. A New Approach, Washington D. C., National Academy Press.
    Deng, Quheng (2003) Determinants of Educational Attainment in Rural People in China,  (Thesis in Chinese), Beijing: Graduate School of Chinese Academy of Social Sciences. 
    Eichen, M. and Zhang, M. (1993), "The 1988 Household Sample Survey - Data Description and Availability.” In Keith Griffin and Zhao Renwei, eds., The Distribution of Income in China, pp. 331-46. London: Macmillan Press.
    Gustafsson, B. and Li, S (1998) "The Structure of Chinese Poverty 1988", The Developing Economies, 35 (4) 387 - 406.
    Gustafsson, B. and Wei Z. (2000) How and Why has Poverty in China Changed? A Study based on Microdata for 1988 and 1995", The China Quarterly, 164, 983 - 1006.
     
    Gustafsson, B. and Wei, Z (2002) "Determinants of Rural Poverty in China", China & World Economy, 10, (2) 41 - 49.
    Khan, A. R., and Riskin, C. (1998) Income Inequality in China: Composition, Distribution and Growth of Household Income, 1988 to 1995." The China Quarterly. 154, 221 - 253.
    Khan, A.R. and Riskin, C. (2001) Inequality and Poverty in China in the Age of Globalization, Oxford: Oxford University Press.
    Mok, K. H. (2000) "Marketizing Higher Education in post-Mao China", International Journal of Educational Development, 20, 109 - 126. 
    Peters, D., Yazbeck, A., Sharma, R., Pritchett, L. and Wagstaff, A. (2001) Better Health Systems for India's Poor. Findings, Analysis, and Options, Washington.D. C.
    Riskin, C. (1994) "Chinese Rural Poverty: Marginalized or Dispersed?", American Economic Review, 84, (2) 281 - 84.
    Riskin, C. and Li Shi. (2001) Chinese Rural Poverty Inside and Outside the Poor Regions", in Riskin, C., Zhao, R. and Li Shi (Eds) China's Retreat from Equality. Income Distribution and Economic Transition, Armonk, M.E. Sharp.
    Yao, S. (2000) "Economic Development and Poverty Reduction in China over 20 Years of Reform", Economic Development and Cultural Change, 48, 447 - 474.
    Zhao, Renwei, Li, Shi, and Riskin, Carl, eds., 1999, Zhongguo jumin shouru fenpei zai yangju (Re-study on Income Distribution of Chinese Households). Beijing: Publishing House of Chinese Finance and Economy.
    Wagstaff, A. and van Doorslaer, E. (2001) "Paying for Health Care: Quantifying Fairness, Catastrophe, and Impoverishment, with Applications to Vietnam, 1993 - 98", Working Paper, World Bank: Policy Research Working Paper No 2715.
    World Bank (1997) China 2020 - Sharing Rising Incomes, Washington D.C. 
    World Bank (2000) Attaching Poverty. World Development Report 2000/2001, Washington D.C.
    World Bank (2001) China. Overcoming Rural Poverty, Washington D.C
    
    

    
    

Table 1a. Education and health care expenditures in rural China: by decile,  1988

Decile	(1)          Mean income	(2)          Mean value of education expenditure	(3)          Mean value of health care expenditure	(4)          =(2)/(1)          (%)	(5)          =(3)/(1)          (%)
1	228	4	9	1.8	3.9
2	367	5	8	1.4	2.2
3	480	7	9	1.5	1.9
4	575	6	11	1.0	1.9
5	653	8	10	1.2	1.5
6	741	8	13	1.1	1.8
7	843	9	13	1.1	1.5
8	1002	14	14	1.4	1.4
9	1220	10	21	0.8	1.7
10	2082	9	26	0.4	1.2
Average	819	8	13	1.0	1.6

    

Table 1b Education and health care expenditures in rural China: by decile,  1995

Decile	(1)          Mean income	(2)          Mean value of education expenditure	(3)          Mean value of health care expenditure	(4)          =(2)/(1)          (%)	(5)          =(3)/(1)          (%)
1	259	12	19	4.6	7.3
2	398	11	23	2.8	5.8
3	495	17	25	3.4	5.1
4	590	18	25	3.1	4.2
5	693	18	29	2.6	4.2
6	813	17	34	2.1	4.2
7	963	27	45	2.8	4.7
8	1174	26	44	2.2	3.7
9	1571	28	61	1.8	3.9
10	3559	37	79	1.0	2.2
Average	1052	21	38	2.2	3.6

    Table 2a Poverty indices computed using various definitions of disposable income for rural China 1988 and 1995. Poverty line at 1 USD PPP per day and per capita
    

	Disposable income          (DI)	DI minusExpenditures on Education	DI minusExpenditures on Health care	DI minus Expenditures on Education and Health care
Head count ratio (%)
1988	19.8	19.9	20.8	21.4
1995	17.6	19.3	20.1	21.9
Difference 95 - 88	- 2.2	-0.6	-0.7	+0.5
1995 as % of 1988	89	97	97	102
Poverty gap
1988	0.0598	0.0620	0.0642	0.0666
1995	0.0459	0.0535	0.0576	0.0664
Difference 95 - 88	-0.0139	-0.0085	- 0.0066	-0.002
1995 as % of 1988	77	86	90	100
Poverty gap squared
1988	0.0260	0.0276	0.0292	0.0309
1995	0.0184	0.0290	0.0257	0.0374
Difference 95 - 88	-0.0076	0.0014	- 0.035	0.0065
1995 as % of 1988	71	105	88	121

     
     
     
     
     
     
     
     
     
     
     
    Table 2b Poverty indices computed using various definitions of disposable income for rural China 1988 and 1995. Poverty line is that which is used to produce official statistics. 
    

	Disposable income          (DI)	DI minusExpenditures on Education	DI minusExpenditures on Health care	DI minusExpenditures on Education and Health care
Head count ratio (%)
1988	14.0	14.3	15.0	15.7
1995	11.1	12.5	13.2	14.8
Difference 95 - 88	- 2.9	-1.8	-1.8	-0.9
1995 as % of 1988	79	87	88	94
Poverty gap (%)
1988	0.0596	0.0625	0.0659	0.0691
1995	0.0396	0.0514	0.0606	0.0741
Difference 95 - 88	-0.0200	-0.0085	- 0.0066	-0.002
1995 as % of 1988	77	86	90	100
Poverty gap squared (%)
1988	0.0421	0.0462	0.0479	0.0523
1995	0.0211	0.0523	0.0564	0.0915
Difference 95 - 88	-0.0210	0.0061	0.0085	0.0392
1995 as % of 1988	50	113	118	175

     
     
     
     
     
     
     
     
     
     
     
     
    Table 3 Poverty indices computed using various definitions of disposable income for various regions of rural China 1995. Poverty line at 1 USD PPP per day and per capita
    

	Disposable income          (DI)	DI minusExpenditures on Education          (Difference from column I)	DI minusExpenditures on Health care          (Difference from column I)	DI minusExpenditures on Education and Health care          (Difference from column I)
Head count ratio          (%)
East	6.2	7.2         (1.0)	7.6      (1.4)	8.9      (2.7)
Central	16.4	18.1       (1.7)	19.1    (2.7)	21.0    (4.6)
West	35.5	38.2       (2.7)	39.3    (3.8)	41.5    (6.0)
Poverty gap (%)
East	0.0147	0.0203  (0.0056)	0.0211  (0.0064)	0.0279  (0.0132)
Central	0.0422	0.0497  (0.0075)	0.0566  (0.0144)	0.0654  (0.0232)
West	0.0954	0.1059  (0.0105)	0.1105  (0.0151)	0.1222  (0.0268)
Poverty gap squared (%)
East	0.0057	0.0248 (0.0191)	0.0097 (0.0040)	0.0297 (0.0240)
Central	0.0169	0.0215 (0.0046)	0.0266 (0.0097)	0.0326 (0.0157)
West	0.0386	0.0461 (0.0075)	0.0470 (0.0084)	0.0554 (0.0168)

     
     
     
     
     
     
     
     
     
     
     
     
    Table 4 Poverty indices computed using various definitions of disposable income for people living in poor counties and in non-poor counties of rural China 1995. Poverty line at 1 USD PPP per day and per capita
    

	Disposable income          (DI)	DI minusExpenditures on Education          (Difference fromfrom column I)	DI minusExpenditures on Health care          (Difference from column I)	DI minusExpenditures on Education and Health care          (Difference fromfrom column I)
Head count ratio          (%)
Poor counties	35.8	38.5 (2.7)	39.5 (3.7)	42.3 (6.5)
Non-poor counties	11.9	13.3 (1.4)	14.0 (2.1)	15.5 (3.6)
Poverty gap (%)
Poor counties	0.0997	0.1116 (0.0119)	0.1177 (0.0180)	0.1310 (0.0313)
Non-poor counties	0.0290	0.0352 (0.0062)	0.0387 (0.0097)	0.0461 (0.0171)
Poverty gap squared (%)
Poor counties	0.0413	0.0504 (0.0091)	0.0518 (0.0105)	0.0622 (0.0209)
Non-poor counties	0.0112	0.0223 (0.0111)	0.0175 (0.0063)	0.0296 (0.0184)

     
     
     
     
     
     
     
     
    

---

    [1] See for example Yao (2000) and World Bank (1997, 2001).
    [2] See Bloom and Tang (1999), Bloom and Gu (1997), Beach (2001), Bray and Borevskaya (2001), Mok (2000).  
    [3] For more information on the first survey see Eichen and Zhang (1993) and on the second see Khan and Riskin (1998). The questionnaires are reproduced in Zhao et al (1999). While several authors (Riskin,1994, Gustafsson and Li, 1998, Gustafsson and Wei, 2000, Khan and Riskin, 2001, Riskin and Li, 2001 and  Gustafsson and Wei, 2002) have used one or both of the surveys to study rural poverty in China, none have taken household expenditures on education and health care into account when defining poverty.
    [4] For example when evaluating the method of producing official estimates of poverty in the United States, the National Academy Panel recommended out-of-pocket medical care expenditures to be subtracted from income, see Cirto and Michael (1995).
    [5] The motivation for considering health and education expenditures when assessing poverty is that they are unavoidable. In our opinion this is a good first approximation of the reality most rural households in China met during the years studied. However, we cannot rule out that some households choose high educational expenditures whichpush them under the poverty line. While many of these households may have left poverty during a later period, they have very little to live on during the period we assess. 
    [6] The World Development Report 2000/2001 (World Bank 2000, p 280) presents PPP units to USD for China, which is  $ 1 = RMB 1.2 in 1990. The inflation rate in the United States was 21% for the period 1985-1990 respectively 19.5% for the period 1990-1995 while the inflation rate in rural China was 76% during the period 1990-1995. Based on this information we computed PPP units to 1 USD 1985 to 2.56 RMB 1995. Based on this we arrive at the 1995 poverty line of 934 yuan per year (which is  to 1 USD per day and person in 1985). The 1988 poverty line was computed as 425 yuan, under120% inflation in China between 1988-1995.
    [7] From this it follows that inequality indices computed for the different definitions of income will show different values. Income inequality will be lowest for Disposable income I, and the difference between the alternatives is largest in 1995. While we advocate replacing poverty assessments based on Disposable income I, the arguments are not equally strong regarding assessments of income inequality (see footnote 5). We do not suggest that Gini coefficients based on the definitions of disposable income used in this paper are superior to one using Disposable income I.
    [8] The poverty rates based on Disposable income I are somewhat lower than those reported by World Bank 2001, which may result from including the imputed rents of owner occupied housing in our income definition.
    [9] These numbers are similar to those reported for health expenditures in other developing Asian countries. Peters et al. (2002 p 290) report that direct out-of-pocket medical costs push 2.2 percent of Indians into poverty per year. Wagstaff and van Doorslaer (2001) report that in Vietnam health care expenditures pushed 4.4 (3.4) percent of the households under the food poverty line in 1993 (1998). 
    [10] Although our poverty rates based on disposable income less expenditures on education and health care should be an improvement when assessing change over time they are not without limitations. Increased expenditures might indicate increased availability and/or quality in provision. In our judgement such changes are not the main reasons why rural households in China have larger expenditures on health and education in 1995 than in 1988. For example based on the same data as for this study Deng (2003) reports that the household expenditures per pupil in primary school increased from 14 to 141 yuan by 4.6 times in constant prices, and in middle school from 51 yuan to 408 yuan, by 3.6 times in constant prices. It is highly unlikely that such an enormous increase in school quality would have passed unnoticed. 
    

---

    

请文明参与讨论，禁止漫骂攻击。 昵称：（注册  登录）