财政分权与卫生绩效

财政分权与卫生绩效 //.paper.edu.cn -1- Fiscal Decentralization and Health Outcomes Yan Qi?? School of Economics & Management, Southeast University, Nanjing, Jiangsu, PRC (211189) Abstract China went a tough way to undertake the health reform, which was featured as fiscal decentralization. To capture the health effects of fiscal decentralization, which are analyzed in two dimensions: health status and health equality, this paper first sets a theoretical model. And theoretical proposition highlights that the effect of fiscal decentralization on health outcomes meets a trade-off between health status and health equality. Then provincial panel data sets in 31 provinces from 2002 to 2006 are used, only to find that fiscal decentralization is detrimental to health status but good for health equality. However, health decentralization, a direct indicator for fiscal decentralization in health sector, is good for health status but detrimental to health equality. Keywords: fiscal decentralization; health level; health equality 1. Introduction Fiscal decentralization is a worldwide trend in developing countries (World Bank, 2000). And reforming the government is a crucial component of both the transition from a planned to a market economy and economic development (Jin, Qian and Weingast, 2005). As a transition country, China is an outstanding case. China, with its large size and population, is one of the most decentralized countries in the world measured in terms of spending authority assigned to the local governments. Transferring authority to lower levels of governments, is expected to improve the provision of local public goods, such as education and health (Bardhan, 2002). This is almost based on two ideas. First, as Hayek(1945) put it, local governments provide public goods and services that better match local preferences than the central government owing to information advantage. Second, citizens can “vote with their feet”. They will select a local jurisdiction that best balances their aspiration for local public goods and the tax level that goes with it. In the long-term, competition for citizens between the different local jurisdictions will drive down costs for the production of local public goods and services1 (Tiebout, 1956). Health, as well as education, is considered to be a vital component of human capital (Mushkin,1962), which serves the leading role in the theory of endogenous growth. Various empirical studies also showed that health has an obvious and robust positive effect on the aggregate output (Bloom et al, 2001). On the contrary, the low level of health and poor performance of health sectors may contribute to a slowdown (Zon et al, 2001). Then what effects would fiscal decentralization bring to health? Does the classic theory of decentralization work? Theoretically, Tiebout’s paradigm for fiscal decentralization might fail in China. First, for local governments, they may lack financial and technical skills to manage health sector (Crook, 1999). As to the central government, the reduction of health expense by superior government will weaken the superior government’s control power in this sector (Ye, 2000). This will harm the function of basic health care service system. Second, even if the local officials are indeed better informed, they may have not enough incentive to respond to the information because of political reasons(Lin, 2000). Local officials in China are not elected publicly by citizens, but appointed by superior government. And the current performance rating system adds too much weight on economic performance, while the efforts on health are not able to generate GDP in short-term(Wang et al, 200<>7). Under health decentralization, this will result in under-provision and deteriorate health status. Further more, given that governments ????????????????????????????????????????????????????????????? ???This essay is the working paper of National College Student Innovative Project (Project NO. G200<>735). 1 Some areas fit the theory well, like Bolivia( Faguet, 2004) //.paper.edu.cn -2- are “capable and loyal”, there still remains possibility that some specific health services with interjurisdictional externalities are under-provided by local governments. Empirical studies on this topic are really scant2. Somke(2001) pointed out that the effect of fiscal decentralization on health in China is not clear yet. However, A report by The World Bank(199<>7) showed that fiscal decentralization did harm to rural health care, especially in poor regions. But Emily Yee’s (2001) analysis concluded that fiscal decentralization had not been detrimental to health care. And Hiroko Uchimura& Johannes Jütting(200<>7) used county-level panel data analysis and finds that fiscal decentralized provinces lowered mortality rate in corresponding counties. The above researches mainly focus on the impact of fiscal decentralization on the health status, but leave out that on health equity. What’s more, some important socioeconomic and political factors are not included, like marketization in health sector-another feature of health reform. In this paper, the effects of fiscal decentralization on health are analyzed in two dimensions: the effect on health status and the effect on health equity. First, I make a mathematical model to identify respective effect of fiscal decentralization on health status and health equity. Second, I construct two indicators for health equity in empirical study. The results show that fiscal decentralization is detrimental to health status but good for health equity. However, health decentralization is detrimental to health equity but good for health status. The remainder of this paper is organized as follows: Section 2 presents an outlook of health status in China. Section 3 sets a mathematical model to analyze the impact of fiscal decentralization. Section 4 discusses sets out the econometric model and the data. Section 5 reports the results of the estimation. And the last section contains a brief conclusion and some further work to do in future study. 2. An outlook on health outcomes China made a dramatic success in improving health conditions in before reform era and in early reform era–as reflected in life expectancy's rise from less than forty years in 1950 to sixty-nine years in Bank, 199<>7). But after the economic reform, China 1982(World carried out a market-oriented health sector reform which, in general, is claimed to be a failure(DRC,2005). According to some major reports, four aspects should be referred. 2.1 Financing of health care Under the reform of fiscal decentralization, heath expenditure is ascribed to local government budget and central transfers account for little. But local government focuses too much on economic growth and the share of government in NHE (national health expenditure) declined3 (DRC, 2005). According to WHO (2004), public expenditure’s proportion in NHE of China is lower than that of other similar developed countries. Besides, public hospitals are encouraged to compete in market with private medical institutions to ease the burden of government. These market-oriented financing reforms improved the productivity of public health institutions, but several unintended consequences became evident. User fees reduced the take-up of preventive services with positive externalities. The lack of government funds resulted in under-provision of services with public goods’ characteristics (Liu&Mills, 2002). 2.2 Health level In the long run, the production of new knowledge and new treatment will accelerate improvement of health (Cutler et al, 2006). However, things like these do not happen often. In Deaton’s (2003) study, income plays an important role in health status. During the post-reform time, average individual ????????????????????????????????????????????????????????????? 2 A general topic-the effects of health decentralization-was studied, like Arredondo’s work(2006) with space decision approach created by Bossert(1998) 3 See Figure 1A in Appendix 1 //.paper.edu.cn -3- income has increased, but indicators of health sometimes deteriorate (Hossain, 199<>7). There are some explanations, e.g., Poel, et al(2009) finds that urbanization of China has a robust and negative causal effect on health. But in the language of World Bank (199<>7), this erosion in health gains stems from changes in government financing of the health and the shift to a more market-oriented economy. 2.3 Health inequality Hossain (199<>7) outlined the inequity problem in China by three dimensions: regressive public health expenditures4, unequal access, disparity in the health status. Because of regressive public health expenditure, World Health Report 2000 ranked China 188th among world’s 191 countries in accordance with health performance represented by health equity. Gao et al(2001) found that the rich and poor made different medical decision owing to their financial budgets5, i.e., there emerged unequal access to the rich and the poor. China’s current policies favor urban human capital investment over rural human capital investment( Heckman, 2004). Thus, evidence of a widening gap in health status between urban and rural residents in the transitional period was found6. (Smith, 1998 &Liu et al, 1999). 2.4 Rising price of medical care Growing rate of medical expenditure doubles that of income of urban citizens and triples that of rural residents(Shi&Gong, 2005). And medical care demand is price inelastic, but price elasticity is larger in absolute value for poorer households(Mocan et al,2000). This generated unequal access to health for the rich and the poor. Though, induced demands by suppliers are to blamed for the high price(Shi&Gong, 2005). Chow(2006) argued that rapid increase in health care demand resulted in rapid increase in price when supply failed to increase because of financing system of health. Taking synthesis of the above problems, we can see that the reform of government financing is where the problem rooted. In the following section, I constructed a model to analyze the micro mechanism through which fiscal decentralization affects health outcomes. 3. The model This model is based on Gerhard Glomm and B. Ravikumar(1992) who constructed a model to compare the outcomes of growth and income inequality under two extremely different financing systems for human capitals. Unlike Gerhard Glomm and B. Ravikumar(1992), who studied the two ways of investment-public offered and private offered-in human capital separately, I focused on the health outcomes under the system of a combined economy, i.e., health is both supported by government and individuals. Thus we can analyze the effects of fiscal decentralization have bring to health status. 3.1 The basic frame The world comprises many independent jurisdictions. In each jurisdiction, there are individuals who live for two periods and die at the end of the second period. In the second period of life, each individual gives birth to another so that the population of the specific jurisdiction remains constant over time At time t, there is an initial generation of old agents endowed with health status ht, and the ith member endowed with hit. Health status of the members of the initial generation is distributed according to the distribution function ft(ht). Here we take the form of lognormal distribution function with parameters ??t and ζt2. All individuals born at t=0、1、2„ have the same preferences over leisure in youth and consumption when old, and the bequest left to their offspring. Formally, the preferences of an individual i born at ????????????????????????????????????????????????????????????? 4 Greater health expenditures are made in richer provinces. 5 Compare Table 1A and 2A (cited from Ping(2003) ) in Appendix 1 6 See Figure 2A in Appendix 1 //.paper.edu.cn -4- time t are represented by<>7 1 1 1( ) ln ln ln ( )t t t tU l c E+ +?? = + + ?? Where l1t is leisure at time t, ct+1 is consumption at time t + 1, and Et+1 , the only bequest to their offspring, is the quality of health at time t + 1. Individuals are endowed with one divisible unit of time in their youth. Young individuals at time t allocate l1t units of their endowment toward leisure at time t and devote l2t units to taking care of their own health care, including personal hygiene and exercise (Agénor, 2009). All these help to accumulate health capital (Grossman, 19<>72) 8 according to 1 2 0 , , (0,1) (1) t t t th l E h here and β γ δθ θ β γ δ+ = > ? Here ht represented the health capital of the corresponding parents. This indicates that the stock of parents’ health capital will affect their children’s health status9. Then they spend the remaining 1- l1t -l2t units of time on work, while the production function satisfies 1 1 1 2 1 1 2( ( , (1 ) ) ,1 ) ( , (1 ) )(1 ) (2)t t t t t t ty F p h w l l p h w l l+ + += ?? ?? ?? = ?? ?? ?? Where p(ht+1,(1-w) ) stands for productivity10 and later we will explain why in this model it depends on ht+1 and (1-w). In the classic paper of Zodrow and Mieszkowski (1986), public expenditures are classified into two stereotypical forms: one, denoted by G, enters into the utilities of agents. This corresponds broadly to such consumption items as recreational facilities or social services etc. Second, denoted by P enters into the output function, and corresponds broadly to such items as roads, sewers and the like. Under current health financing system, I assume that the local government devotes w units of its spending (defined as one unit) into G, of which ?? percent are invested in health sector and the rest (1-w) units into P. Thus ?? w units of public expenditure enter in health sectors like aiding medical research, buying advanced equipments and training doctors. All of these are standards to grade the hospitals. We can easily see that identical personal health inputs will generate different health outcomes in medical institutions of different levels. In fact, as classified in the World Development Report (1993), public health programs work in three ways: they deliver specific health services to populations (for example, immunizations), they promote healthy behavior, and they promote healthy environments. So it’s reasonable to assume that public spending and private spending work complementarily to generate health quality (Bhattacharya&Xue, 200<>7)11. Based on that, we employ the power function form for the quality of health at time t+1. 1 1 ( ) (0 1) (3)t tE w e where α?? α+ += ? ， Here the coefficient of et+1 can be interpreted as productivity in health sector. On the other hand, infrastructure services have a strong growth promoting effect through their impact on production costs, the productivity of private inputs, and the rate of return on capital-particularly when, to begin with, stocks of infrastructure assets are relatively low (Agénor, 2006). Thus, public capital in P will bring ????????????????????????????????????????????????????????????? <>7? There are at least three formulations of intergenerational altruism in the distribution and growth literature. One of these is that members of current generation may value the wealth they pass on to their descendants(Glomn& Ravikumar,1992)? 8? Gross investments in health capital are produced by household production functions whose direct inputs include the own time of the consumer and market goods such as medical care, diet, exercise, recreation, and housing. 9 The well-known positive association between health and income in adulthood has antecedents in childhood. Part of the intergenerational transmission of socioeconomic status may work through the impact of parents' income on children's health(Case et al,2002)? 10 Jack &Lewis(2009) find robust evidence on the micro linkage from health investment to productivity. Actually, investment in young children pay off in healthier and more productive adults. 11? We do not wish to interpret the public component as a competitor to the private input in health markets; viewed in this way, the two inputs would have to be seen as being substitutes in health service provision. Instead, we 6><#00aa00'>view public health investments as being complementary to private health investments. //.paper.edu.cn -5- productivity effect to firms. What’s more, the stock of health capital also tends to increase productivity both in the short-term and in the long-term (Arora, 2001). Therefore, here we take synthesis of the two different effects on productivity and then assume the following functional form for p 1 1( , (1 )) (1 ) (4) t tp h w h w πρ+ +?? = ?? 3.2 Equilibrium At time t+1, every agent will be taxed at rate η which is determined by local government. Thus, the agent faces the budget constraint 1 1 1(1 ) (5)t t tc e yη+ + ++ = ?? The young agent at t time has to choose l1t , l2t ,et+1,ct+1 to optimize his utility. Put equations (*), (1), (2), (3), (4), (5) together, then we get target mathematical planning 1 2 1 1 1 1 1{ , , , } 1 1 1 1 1 1 1 1 2 1 2 ln ln ln (1 ) . . (1 ) (1 ) , , , max t t t t t t tl l c e t t t t t t t t t t t t t t t l c E c e y E w e s t y h w l l h l E h w e h given α π β γ δ γ δ η ?? ρ θ ?? η + + + + + + + + + + + + + + + = ?? = = ?? ?? ?? = ) ( , Work with the mathematic trick12, and we will soon get the optimal solutions to the planning: 1 2 2 1 * * * *1 2 1, , 1 3 2 3 2 3 2t t t t l l l lββ β β= = ?? ?? =+ + + The agent’s stock of health capital at time t+1 is 1 1( ) ( ) ( ) ( ) [( ) (1 ) ] (6) 3 2 3 2 3 2 2t t t t h e h w w hβ γ δ β γ γ α π γ δ γβ β ρ ηθ θ ??β β β + + ??= = ??+ + + From the above equation, we can see that, if ln(ht) is normally distributed with mean ??t and variance ζt2, then ln(ht+1) is also normally distributed with mean ??t+1 and variance ζt+12, where 1 1ln ( ) ( ) ( ) [( ) (1 ) ] ( ) (<>7) 3 2 3 2 2t t w wβ γ γ α π γβ ρ η?? θ ?? δ γ ??β β+ ?? ????= ?? + +?? ??+ +?? ??>7 2 2 2 1 ( ) (8)t tζ δ γ ζ+ = + 3.3 Results from the model Here, 1 1 1 1( )t t t tEh h df h+ + + += ? and 21 1 1 1 1( ) ( ) ( )t t t t tVar h h Eh df h+ + + + += ??? are used to measure average health level and health inequality. For lognormal distribution, Eht+1=exp (u t+1+ζt+12/2) and Var(ht+1)=exp(2u t+1+ζt+12)(exp(ζt+12)-1). With comparative static analysis, some propositions were put forward. The following propositions characterized the government’s effect in the evolution of health outcomes. Proposition 1:(1) If / ( )w α α π< + ，average health level increases with the expansion of government spending on health and vice versa. (2) Average health status declines as the tax rate rises. Proof: See in Appendix 2. Proposition 2: (1) If / ( )w α α π< + , health inequality increases as the government expenditure on ????????????????????????????????????????????????????????????? 12 See the Appendix. //.paper.edu.cn -6- health rises and vice versa. (2) Health inequality declines with the rise of tax rate. Proof: This proof is like that in proposition 1. From proposition 1 and 2, we can deduce that government’s effect on the health outcome depends on the parameters α and π , because government expenditure affects both individual income and health sector productivity. Besides, the local government falls into a trap of trade-off between efficiency and equity. One possible explanation is that professionals and middle classes made better use of the Health Service than those in lower social groups (LeGrand, 19<>78) Fiscal decentralization means that every jurisdiction has the power to decide w and η . As Keen and Merchand (199<>7) put it, non-cooperative equilibrium in government fiscal competition will be attained at the point of under-provided G, i.e., a smaller w. As a support, Fu and Zhang (200<>7) proved that in China local governments bias their expenditure towards infrastructure. However, there is no direct answer to the effect of a smaller w on health status and health equity. Hence, it is an empirical question whether fiscal decentralization leads to an improvement in health outcomes. 4. Empirical methodology and Data 4.1 Basic model To empirically assess the effects of fiscal decentralization on health, my analysis are based on the following basic model for panel dataset: ( )it it i it it itY a X C Bβ γ ε= + + + where i indexes the province and t is time. Xit, denotes decentralization indicators and Cit, a vector, denotes the controlled variables. Here Yit are indicators for health outcomes and εit is a disturbance term. Dependent variables: As has been analyzed in part 3, health outcomes are assessed in two dimensions: health status and health equality. 1) Health status As for health status, three variables are used like Yee(2001). The first variable is the number of doctors per thousand people (PD). And here assumes an increase of this variable indicates an improvement in health status of the whole province. The second variable I used is the number of beds per thousand people (PB). It is assumed to have position correlation with provincial health status. The third variable is the provincial mortality rate (MR) which assesses the direct outcome of health status. Obviously, a smaller MR means better health status. 2) Health equality Like PD and PB, the number of doctors per thousand agricultural people (PAD) and the number of beds per thousand agricultural people (PAB) reflect the health status of rural residents. To measure health inequality, I employed two proxies because of the limited accessibility to data. The first is denoted by RD, where /RD PD PAD= .The second is denoted by RB, where /RB PB PAB= . They both represent unequal access to health resources. I assume that, a larger of either indicates a bigger gap of health status between rural residents and urban residents13. Explanatory variables: 1) decentralization indices To quantitatively measure the health effect of fiscal decentralization, two indicators are employed. The first is the measurement of the standard fiscal decentralization: the ratio of province expenditure ????????????????????????????????????????????????????????????? 13 Zhang(2003) also used PD(PB) and PAD(PAB) to discuss medical care disparity in China. Here I define health equality to be equality of health access. //.paper.edu.cn -<>7- per capita to central government expenditure per capita14(FDEC). And similarly, I invent the ratio of provincial health expenditure per capita to central government health expenditure per capita (HDEC)15 to capture the characteristics of fiscal decentralization in health sector. 2) socioeconomic and political characteristics Social characteristics are measured by education level and the degree of marketization in health sector16. The provincial illiteracy rate (IR) is the percentage ratio of the number of illiterates to the total population aged 15 and over, which is used in our model as a proxy of the education level. And the ratio (MAR) of the number of for-profit health institutions to not-for-profit institutions is taken as a measurement of the degree of health marketization1<>7. Economic characteristics are measured by the economic level of the province. Economic level is measured by the provincial per capita GDP (PGDP). Urbanization captures both social and economic characteristics of the province, and is measured by the ratio (URB) of urban people to total people in the province. Along with fiscal decentralization, the central government has been in a strong position both to reward or to punish local administrations based on the local economic performance (Blanchard &Shleifer, 2000; Tsui&Wang, 2004). Thus, with financial and political incentives, yardstick competition between local governments takes place. So the degree to which the province involved in yardstick competition captures the some political characteristics of the specific province, and is measured by actural foreign capital usage per capita (COM) like in Zhang’s(200<>7) study18. 4.2 Data and descriptive analysis Panel data that cover 31 provinces for five years (2002~2006) are used for this quantitative analysis. The data here all come from Statistical Year Book of China, Health Statistical Year Book of China and ACMR (supported system for China statistics application). Table 1 descriptive statistics of major variables obs Mean Maximum Minimum Std. Dev PD 146 4.65 10.52 2.31 1.65 PB 146 2.<>79 5.29 1.53 0.88 MR 146 5.9<>7 <>7.30 2.54 0.<>70 RD 146 3.82 5.8<>7 1.<>76 0.98 RB 146 3.56 6.88 1.11 1.08 FDEC 146 3.26 13.80 1.23 2.35 HDEC 146 2.68 14.99 0.83 2.26 PGDP 146 14461.53 5<>7114.99 3240.63 10380.52 COM 146 454.08 3155.9<>7 3.65 665.64 URB 146 41.46 89.09 20.39 15.82 MAR 146 0.99 2.<>7<>7 0.01 0.69 IR 146 10.68 24.<>7<>7 3.85 5.12 Table 1 reported some descriptive statistics of the panel data sets. The mortality rate varies across ????????????????????????????????????????????????????????????? 14 There are disputes on the measurements of fiscal decentralization. The indicator I use here eliminates the effects population-which may have positive correlation with public expenditure(Fu &Zhang,200<>7) 15 Though we may come across a problem of endogeneity(Chen, 2004), it is attractive to use HD to capture the direct effect of health decentralization to health outcomes. 16 Duggan’s(2000) study showed that increased for-profit penetration makes not-for-profit hospitals more profit-oriented. It is reasonable to include the factor of marketization. 1<>7 Health institutions are divided into for-profit institutions and not-for-profit institution in 2000, when a market-oriented reform of large scale really started(Liu,2005). In yardstick competition, local governments focus on attracting 18 business capitals, especially FDI. So foreign investors serve the role of “voters” in Tiebout’ model. //.paper.edu.cn -8- provinces and over the years. The lowest MR is for Heilongjiang province in 2002. And the highest MR is for Yunnan province in 2002. The lowest (2.54) is better than those of other Asian countries-for example, Japan’s MR was 6.0019 in 2002, while the highest (<>7.3) overruns them. Socioeconomic and political characteristics also differ between provinces and years. It is well known that significant economic inequality exists between regions in China. The extent to which a province is involved in the yardstick competition varies with the regional natural endowments, some disadvantaged provinces may each take the strategy of “make badness worse” in the competition (Wang et al, 200<>7). Table 1 gives some information of the variation of FDEC and HDEC between provinces. Since they are major variables, more details are represented in the following figure. Figure 1 FDEC and HDEC in 2006 Figure 1 provides the FDEC and HDEC of 31 provinces in 2006. The FDEC and HDEC are both more than one which mean that local governments are main public providers in local public goods including health care. Though the two indicators varies across provinces, it is proved that they are strongly positive correlated 20 . This suggests that provinces with high fiscal capacity invest correspondingly more in health sector. 5. Empirical analysis and Results Based on basic model(B), we examine the following set of models that focus on FDEC and HDEC respectively: 1 1 2 3 4 5 1 1 2 3 4 5 ln ( ) ln ( ) it it it it it it it it it it it it it it it it Y a FDEC PGDP COM URB MAR IR a Y a HDEC PGDP COM URB MAR IR b β γ γ γ γ γ ε β γ γ γ γ γ ε = + + + + + + + = + + + + + + + Here Yit is a vector of variables measuring health outcomes. And it includes PD, PB, MR, RD and RB. The first three represents health status and the last two health inequality. Since panel data model is used, the selection between fixed-effects and random-effects should be taken into account. If there are significant provincial specific effects that, if not accounted for, could bias the estimates of the model. Otherwise a random-effects model is preferred. I estimated my model both with a random-effects model and fixed-effects model. Then Hausman test and Redundant fixed-effects test were conducted to select the most fitted model. The regression results showed that all the estimated models have fixed-effects and the fixed-effects estimations have well-performed statistics with higher R-squares than random-effects model. Only one of the estimations has significant random-effects, but the R-squares is no more than 0.05521. Two-way ????????????????????????????????????????????????????????????? 19 Calculated with the data from health statistical yearbook in china(2008) 20 The correlation coefficient reaches as high as 0.91. Hence, the problem of endogeneity may occur. 21 When RB serves as independent variable and model (b) is applied, the random-effects model is significant. But its R-squared is only 0.055. //.paper.edu.cn -9- fixed-effects are only managed in regressions with PD as the independent variable. The others reject period fixed effects and accept cross-section fixed effects. Table 2 summarizes the main results. And it only shows the sign of coefficients and significance, more details like t-statistics and R-squared are exhibited in Table 3A an Table 4A in appendix. Table 2 impact of fiscal decentralization on health outcomes22 (a) (b) FDEC lnPGDP COM MAR URB IR HDEC lnPGDP COM MAR URB IR PD?? -?????? +?????? + + +???? +???? +?????? +?????? - +???? +?????? +?????? PB?? -?????? +?????? - -?????? +?????? + + +?????? -?? -?????? +?????? + Health status MR?? +?????? + - +?????? -?????? +?????? -?????? +?????? - + -?????? +?????? RD?? -?????? +?????? -?????? -?????? +?????? + +?????? +?????? -?????? -?????? +?????? + Health inequality RB?? -?????? +?????? + -?????? - + + +?????? - - - + Note: 1. *, **, *** indicate significant at 10%, 5%, 1% confidence level respectively. 2. ?? indicates that the regression is estimated with two-way fixed-effects model, while ?? indicates the model is estimated with cross section fixed-effects model. Heteroscedasticity is corrected with cross-section weights and White cross-section coefficient variance method is used to get a robust estimator. First, we examine the effect of FDEC on health outcomes. FDEC is the only variable that remains significant in all my models. When health status is measured by PD and PB, the coefficients of FDEC are negative and statistically significant. And when MR is used to measure health status, the coefficient of FDEC is positive and significant. Both factors suggest that fiscal deteriorated health status. As to health inequality, whether it is measured by RD or by RB, the coefficient of FDEC is negative and significant, i.e., fiscal decentralization tends to reduce health inequality. Second, we focus on the effect of HDEC on health outcomes. Actually, health decentralization has the very opposite effects on health status and health inequality compared to fiscal decentralization, i.e., health decentralization improves health status, but enlarges health inequality. But when health status and health inequality are measured by “beds”, the coefficient of HDEC is no longer significant23. Regarding the controlled variables, some of the effects are expected. Economic development leads to a better health outcome (better health status and smaller health inequality). Though COM is not significant in most of the models, but it reports to be a helping hand to reduce health inequality with high significance. The impact of MAR on health status is not clear, but it helps to improve health equality. Besides, the more urbanized, the better for health and the worse for health equality. The results show that a lower indicator of illiteracy (IR) means better health status. 6. Conclusions and Further work Fiscal decentralization is an important component for China’s transition. This paper investigates the effects of fiscal decentralization on health outcomes, which are analyzed in two dimensions: health status and health equality. To identify the effects through micro mechanism, I construct a theoretical model and find that fiscal decentralization affected health outcomes through two channels: raising individual income and lower health sector productivity, and the net effect depends on which weights more, i.e., the net effect is ambiguous. But this model comes to a proposition that fiscal decentralization will bring opposite impacts to health status and health equality at the same time. Then I construct RD and RB to get respective health effects of fiscal decentralization on health status and health inequality, which distinguished current studies. ????????????????????????????????????????????????????????????? 22 A log transformation for PGDP is conducted here, not only to manage heteroskedasticity, but also to reduce collinearity between PGDP and other variables like FDEC (from 0.<>7 to 0.6). 23 It may be an implication for the duplication problem of beds in statistical materials (Yee, 2001). //.paper.edu.cn -10- Empirical study shows that fiscal decentralization is detrimental to health status but helps to reduce health inequality, which coincides with the theoretical proposition. But health decentralization has the very opposite effects on health outcomes compared to fiscal decentralization, which means health decentralization does good to improving health status but does harm to health equality. However, the mechanism is not revealed here. We also find that economic development, urbanization and a higher education level could enhance health status. Regarding health inequality, effects of economic development, local government competition and health marketization is positive but urbanization negative. Despite the results attained above, some problems remains to be further studied. First, the indicators of health outcomes this paper use are not direct indicators to measure health outcomes except mortality rate. Some accurate and direct indictors for health outcomes would be helpful to empirical analysis. Second, health decentralization can be affected by fiscal decentralization. Then HDEC may better be the independent variable in the regression. Third, more details for decision making of local governments like finance, service organization, human resources, governance rules should be analyzed to comprehend the effect of fiscal decentralization. Fourth, the effect of medical insurance is omitted in this paper which needs further study. References [1] Agénor, P-R, Public Capital, Health Persistence and Poverty Traps[R]. Economics Discussion Paper of Manchester University No.115, February 2009. [2] Arora, S., health, human productivity and long-term economic growth [J]. The Journal of Economic History, Vol. 61, No. 3 (Sep., 2001), pp. 699-<>749 [3] Agénor, P-R, a theory of infrastructure-led development[R]. Economics Discussion Paper of Manchester University. December, 2006. [4] Arredondo, A. and E. Orozco, Effects of health decentralization, financing and governance in Mexico [J]. Rev Saude Publica,2006;40(1);pp:151-160 [5] Bardhan, P., Decentralization of governance and development [J]. Journal of Economic Perspective ,2002,16 (4), pp:185–205 [6] Bhattacharya J. and Xue Qiao, Public and private expenditures on health in a growth model [J]. Journal of Economic Dynamics & Control ,199<>7,31 ,pp:2519–2535 [<>7] Blanchard, O. and A. Shleifer, Federalism with and without Political Centralization: China versus Russia[R], IMF Staff Papers, 2000, pp: 1<>71-1<>79. [8] Bloom, D.E., Canning D. and J. Sevilla, The effect of health on economic growth: theory and evidence[R]. NBER working paper NO.858<>7, November 2001 [9] Bossert, T., Analyzing the decentralization of health systems in developing countries: decision space, innovation and performance [J]. Soc. Sci. Med, 1998,Vol. 4<>7, No. 10, pp:1513?152<>7. [10] Case, A., Lubotsky, D. and C. Paxso, Economic Status and Health in Childhood: The Origins of the Gradient[J].The American Economic Review, 2002,Vol. 92, No. 5, pp. 1308-1334 [11] Chen Qiulin, The Effect of Decentralization on Health: A Case of China[C].The fourth annual meeting of Chinese economics society, 2004. [12] Chow, G.C., An Economic Analysis of Health Care in China[R].Princeton Working Paper series NO.132,2006 [13] Crook, R.C. and A.S. Sverrisson, To what extent can decentralized forms of government enhance the development of pro-poor policies and improve poverty-alleviated outcomes?[R]. World Bank Research Papers, 1999. [14] Cutler, D., Deaton, A. and Adriana L-M, The Determinants of Mortality [J]. The Journal of Economic Perspectives, 2006,Vol. 20, No. 3, pp. 9<>7-120 [15] Deaton, A., Health, Inequality, and Economic Development [J]. Journal of Economic Literature, 2003,Vol. 41, No. 1, pp. 113-158 [16] Duggan, M., Hospital market structure and the behavior of not-for-profit hospitals: evidence from responses to California’s disproportionate share program[R].NBER Working Paper NO.<>7966, October, 2000. [1<>7] Faguet, J-P, Does decentralization increase government responsiveness to local needs? Evidence from Bolivia[J], Journal of Public Economics ,2004,(88),(pp:86<>7– 893 [18] Fu Yong and Zhang Yan, fiscal decentralization in Chinese style and biased public spending pattern: the cost of competition for growth [J]. Management World,NO.3 ,200<>7.pp:4-11 [19] Gao Jun, Shenglan Tang, Tolhurst, R. and K. Rao, Changing access to health services in urban china: implications for equity [J]. Health Policy Planning,2001;16(3):302-312 [20] Glomm, G. and B. Ravikumar, Public vs private investment in human capital: endogenous growth and //.paper.edu.cn -11- income inequality [J]. Journal of Political Economy, 1992,Vol. 100, No. 4, pp. 818-834 [21] Grossman, M., On the concept of health capital and the demand for health [J]. The Journal of Political Economy, 19<>72,Vol. 80, No. 2, pp. 223-255 [22] Hayek, F.A., The Use of Knowledge in Society[J]. The American Economic Review, 1945,Vol. 35, No. 4, pp. 519-530 [23] Heckman, J.J., China’s human capital investment [J]. China Economic Review ,2005, (16) pp:50–<>70 [24] Hossain, S.I., Tackling Health Transition In China[R].the world bank policy research working paper 1813,199<>7 [25] Jack, W. and L. Maureen, Health Investments and Economic Growth: Macroeconomic Evidence and Microeconomic Foundations[R]. World Bank Policy Research Working Paper 48<>7<>7, 2009 [26] Jin Hehui, Qian Yingyi, T. Barry and R. Weingast, Regional decentralization and fiscal incentives: Federalism, Chinese style[J]. Journal of Public Economics,2005, (89),pp:1<>719–1<>742 [2<>7] Keen, M. and M. Marchand, Fiscal competition and the pattern of public spending [J]. Journal of Public Economics,199<>7,( 66) ,pp:33-53 [28] LeGrand Julian, The Distribution of Public Expenditure: The Case of Health Care [J]. Economica, New Series, 19<>78,Vol. 45, No. 1<>78, pp. 125-142 [29] Liu Junming, Excess marketization and decentralization - the double mistakes in health reform in China [J]. Health Economic Research, 2005, Vol12, pp:3-10. [30] Liu Xingzhu, Anne Mills, Financing reforms of public health services in China: lessons for other nations [J]. Social Science & Medicine ,2002,54 , pp:1691–1698 [31] Liu Yuanli, Hsiao, W.C., and K. Eggleston, Equity in health and health care: the Chinese experience [J]. Social Science & Medicine 49 (1999) 1349?1356 [32] Mocan, H.N., Tekin, E. and J.S. Zax, The demand for medical care in urban China[R].NBER Working Paper NO.<>76<>73,April 2000. [33] Mushkin, S.J., Health as an Investment [J]. The Journal of Political Economy, Vol. <>70, No. 5, Part 2: Investment in Human Beings (Oct., 1962), pp. 129-15<>7 [34] Ping Xinqiao, the selection of financing system of medical and health service in the rural areas-a <#00aa00'>view based on the expending behavior in medical and health work for China’s peasants[J]. Management World, NO.11,2003.pp:52-63 [35] Shi, Guang and Sen Gong, China’s health input and its outcomes in post-reform era[R]. DRC Reports,2005 [36] Smith, C.J., Modernization and health care in contemporary China [J]. Health& Place, 1998,Vol4. NO.2, pp125-139. [3<>7] Smoke, P., fiscal decentralization in developing countries[R].Democracy, governance and human rights program paper. NO.2. February 2001. [38] Tiebout, Charles M., A Pure Theory of Local Expenditures [J]. The Journal of Political Economy,1956, Vol. 64, No. 5, pp. 416-424 [39] Tsui, Kai-yuen and Y. Wang, Between Separate Stoves and a Single Menu: Fiscal Decentralization in China [J]. China Quarterly, 2004, pp: <>71-90. [40] Uchimura, Hiroko and Jütting, Johannes, Fiscal decentralization, Chinese Style: good for health outcomes? [R].OECD Development Center Working Paper, November 200<>7. [41] Van de Poel, E., O’Donnell, O. and, E. Van Doorslaera, The health penalty of China’s rapid urbanization[R]. Tinbergen Institute Discussion Paper 2009-016/3 [42] Van Zon, A. and J. Muysken, Health and endogenous growth [J]. Journal of Health Economics,2001,(20) pp:169–185 [43] Wang Yongqin, Zhang Yan, Zhang Yuan, Chen Zhao, Lu Ming, on the China’s development model: the costs and benefits of China’s decentralization approach to transition [J]. Economic Research Journal, Vol1,200<>7,pp:4-16 [44] World Bank. Financing Health Care: Issues and Options for China. Washington DC: World Bank, 199<>7. [45] Ye Lei. Workshop on Health Reform and Development Strategy, MOH of China, 2000. [46] Yee, E., The Effects of Fiscal Decentralization on Health Care in China，Undergraduate Journal of Economics, University Avenue, 2001. [4<>7] Zhang Jun, Gao Yuan, Fu Yong and Hong Zhang, Why Does China Enjoy So Much Better Physical Infrastructure [J]. Economic Research Journal, Vol 3, 200<>7, pp:4-19 [48] Zhang Xiaobo, Education and health inequality in China [J]. China Economic Quarterly, 2003, Vol.2 [49] Zodrow, G. and P. Mimeszkowski, Pigou, Tiebout, Property Taxation, and the Underprovision of Local Public Goods[J].Journal of Urban Economics, 1986,(19), pp:356-3<>70. //.paper.edu.cn -12- Appendix: 1. Figures and Tables Figure 1A Percentage of national health expenditure Figure 2A Maternal mortality in different developed provinces Table 1A Financing of rural residents’ health care 199<>7 1998 1999 2000 (a)=local government health expenditure per capita (yuan) 41.20 44.00 4<>7.<>70 45.60 (b)= government health expenditure per capita (yuan) 43.00 49.36 54.81 61.6<>7 (c)=out-of-pocket expenditure per capita (yuan) 66.00 <>72.00 <>75.00 9<>7.00 (a)/(c) 0.62 0.61 0.63 0.4<>7 (b)/(c) 0.65 0.68 0.<>73 0.63 Table 2A Financing of urban residents’ health care 199<>7 1998 1999 2000 (a)=public fund 159.8 1<>76.<>7 191.3 211.00 (b)=social funds 93<>7.<>70 1006.00 1064.60 116<>7.<>70 (c)=out-of-pocket expenditure 1369.54 1584.52 185<>7.82 20<>7<>7.35 (a) +(b)/ (c) 0.80 0.<>75 0.6<>7 0.6<>7 //.paper.edu.cn -13- Table 3A The effect of decentralization on health status PD ?? PB?? MR?? (a) (b) (a) (b) (a) (b) FDEC -4.1<>7?????? -3.10?????? 3.50?????? HDEC 2.88?????? 1.56 -3.90?????? lnPGDP 4.44?????? 4.86?????? 11.48?????? 15.36?????? 0.612 3.<>76?????? COM 1.20 -0.46 -1.64 -1.<>79?? -1.46 -1.34 MAR 1.14 2.15???? -4.<>70?????? -3.81?????? 2.68?????? 0.52 URB 2.44???? 2.84?????? 4.21?????? 3.19?????? -4.64?????? -4.62?????? IR 3.43?????? 2.98<>7?????? 0.42 0.68 5.46?????? 8.29?????? -squared 0.996 0.995 0.99<>7 0.996 0.9<>76 0.9<>78 R D-W 1.33 1.28 1.95 1.8<>7 2.11 2.12 Obs 146 146 146 146 146 146 Note: 1. *, **, *** indicate significant at 10%, 5%, 1% confidence level respectively. 2. ?? indicates that the regression is estimated with two-way fixed-effects model, while ?? indicates the model is estimated with cross section fixed-effects model. 3. The numbers in parentheses are t-statistics, corrected for panel heteroscedasticity. Table 4A The effect of decentralization on health equity RD?? RB?? (a) (b) (a) (b) FDEC -5.46?????? -3.50?????? HDEC 3.0<>7?????? 0.64 lnPGDP 15.<>7<>7?????? 9.85?????? 5.12?????? 3.15?????? COM -5.82?????? -4.9<>7?????? 0.63 -0.05 MAR -3.<>7<>7?????? -4.18?????? -3.81?????? -0.0<>7 URB 3.11?????? 3.35?????? -0.99 -0.01 IR 0.36 0.<>71 0.30 0.54 R-squared 0.988 0.991 0.989 0.992 D-W 1.<>70 1.68 1.<>78 1.<>7<>7 Obs 146 146 146 146 Note: 1. *, **, *** indicate significant at 10%, 5%, 1% confidence level respectively. 2. ?? indicates that the regression is estimated with two-way fixed-effects model, while ?? indicates the model is estimated with cross section fixed-effects model. 3. The numbers in parentheses are t-statistics, corrected for panel heteroscedasticity. 2. Solutions to mathematical planning Define the Lagrange function : 1 1 1 1 1 1 2 2ln ln ln { (1 ) (1 )(1 ) } t t t t t t t t t tL l c E c e l l w l E h π β γ δλ η ρ θ+ + + += + + + + ?? ?? ?? ?? ?? According to first order conditions, we have: 2 1 1 1 2 2 2 1 1 1 1 1 (1 ) (1 ) 0 (1 ) 0 1 0 1 0 t t t t t t t t t t t t t t t t t L w l E h l l L l l E h l E h l L c c L e e π β γ δ γ δ β γ δ λ η ρ θ β λ λ + + + + ?? = + ?? ?? =?? ?? = ?? ?? ?? =?? ?? = + =?? ?? = + =?? //.paper.edu.cn -14- Thus, we get optimal solutions: 1 2 2 1 * * * * 1 1 1 1 2 1, , 1 3 2 3 2 3 2 (1 ) / 2 t t t t t t t l l l l c e y β β β β η+ + + = = ?? ?? =+ + + = = ?? Substituting for lt+1 , et+1 in equation (1), we have 1 1( ) ( ) ( ) ( ) [( ) (1 ) ] 3 2 3 2 3 2 2t t t t h e h w w hβ γ δ β γ γ α π γ δ γβ β ρ ηθ θ ??β β β + + ??= = ??+ + + Q.E.D. 3. Proof of proposition 1 Substituting for ??t+1 and ζ t+1 in Eht+1 andVar(ht+1), we get 2( ) /2 1 1( ) ( ) ( ) [( ) (1 ) ] ( ) e 3 2 3 2 2 t t tEh w w γ γ α π γβ ρ ηθ ?? γβ β γβ δ+ ζ + ?? ????= ?? + δ+ ???? ??+ +?? ??>7 2 2( ) ( ) 1 1( ) 2 ( ) ( ) ( ) [( ) (1 ) ] ( ) e (e 1) 3 2 3 2 2 t t t tVar h w w γ γβ γ γ α π γβ ρ ηθ ?? γβ β δ+ ζ δ+ ζ + ?? ????= ?? + δ+ ?? ???? ??+ +?? ??>7 (1) Differentiate Eht+1 with respect to w, we get 1 11 [ (1 ) ( ) (1 ) ( ) ]tEh A w w w w w πγ αγ πγ αγαγ?? ?? πγ ???? ??+?? = ?? ?? ???? Where A represents 2( ) /21( ) ( ) ( ) e 3 2 3 2 2 tγβ γ γβ ρ ηθ β β δ+ ζ?? + + , and A>0 Thus 1 0tEh w +?? >?? if and only if 1 1(1 ) ( ) (1 ) ( ) 0w w w wπγ αγ πγ αγαγ?? ?? πγ ???? ???? ?? ?? > It can be reduced to / ( )w α α π< + (2) Differentiate Eht+1 with respect to η , we get -11 1-( ) 0 2 tEh B for allγηγ ηη +?? = >?? Q.E.D.