Does fiscal decentralization improve healthcare outcomes empirical evidence from china

DOES FISCAL DECENTRALIZATION IMPROVE HEALTHCARE OUTCOMES? EMPIRICAL EVIDENCE FROM CHINA Yinghua Jin School of Economic Development Georgia Southern University Statesboro, Georgia Rui Sun Department of Public Administration University of Central Florida Orlando, Florida ABSTRACT Since the late 1970s, China has adopted a variety of economic reforms that have led to its overall economic success The Tax Sharing System (TSS) reform, as part of the gradual fiscal decentralization policy, was initiated in 1994 The conventional theory claims that fiscal decentralization could result in various potential benefits including increased responsiveness of local governments to deliver public goods However, very little empirical work has examined the impact of fiscal decentralization on health outcomes in China In this study, we use the infant mortality rate (IMR) as an indicator of healthcare outcomes and provide a quantitative measurement of the impact of fiscal decentralization on infant mortality at the provincial government level We measure fiscal decentralization as both a dummy and a ratio and estimate the infant mortality rate production function using both Ordinary Least Squares (OLS) and Panel Feasible Generalized Least Squares (FGLS) approaches We find that, contradictory to the predictions made by the conventional theories, fiscal decentralization has generated an overall adverse impact on the IMR in China INTRODUCTION Living a longer and healthier life has become the foremost choice and purpose of human development (United Nations Development Programme, 19902008) Among different measures of human health, the lifespan of babies is considered ―the most delicate test of health conditions‖ (Liu, Hsiao, & Eggleston, 1999) As the starting stage of life, an infant is the most vulnerable Thus, improved health conditions may have far-reaching positive effects in reducing infant mortality Blaxter (1981) and Sen (1998) argue that the quality of life depends heavily on healthcare, medical knowledge, and medical insurance They also find that the statistics on infant mortality reflect all of these policy issues According to the United Nations Development Programme (UNDP), infant mortality rate (IMR) is defined as the number of infant deaths Public Finance and Management Volume 11, Number 3, pp 234-261 2011 PFM 11/3 235 per 1,000 live births under a year of age in the same year (UNDP, 1990-2008) This indicator has been widely used for cross-country comparisons and trend analysis of healthcare outcomes A number of studies have tried to associate healthcare with fiscal decentralization (Asfaw, Frohberg, James, & Jutting, 2007; Cantarero & Pascual, 2008; Duret, 1999; Uchimura & Jutting, 2007) Within the healthcare sector, fiscal decentralization specifically refers to the decentralization of financial resources and expenditure responsibilities for healthcare from central government to sub-national governments (Mills, Vaughan, Smith, & Tabibzadeh, 1990) This area of decentralization becomes an important component of policy reforms in many countries including China, Ghana, Indonesia, the Philippines, Uganda, and Zambia Using different measures of decentralization, scholars generally find that higher fiscal decentralization leads to a lower IMR (Asfaw, Frohberg, James, & Jutting, 2007; Cantarero & Pascual, 2008; Duret, 1999; Uchimura & Jutting, 2007) However, there are few studies that have explored the impact of fiscal decentralization on the IMR in China The purpose of this study is to provide a quantitative measurement of the impact of fiscal decentralization on the IMR in China using provincial government data Since 1978 China has moved away from a centralized fiscal system to a decentralized one The systematic change to a decentralized fiscal system occurred upon the passage of the 1994 Tax Sharing System (TSS) reform To capture the impact of the TSS reform, we developed a generalized model using a panel provincial dataset for the period of 1980 to 2003 that encompasses both the pre-TSS and post-TSS eras Using a framework of IMR production function, we analyze both the direct and indirect channels such as income and medical facilities For the purpose of comparison, we employ two measures of fiscal decentralization: first, we treat the 1994 TSS reform as a natural experiment and use an interactive term of a fiscal decentralization dummy and a geographical location dummy to gauge the effect of fiscal decentralization on the IMR in different regions; second, we measure the degree of fiscal decentralization using the ratio of per capita provincial budgetary expenditures to the sum of per capita central budgetary expenditures and per capita provincial budgetary expenditures, as developed by Qiao, MartinezVazquez & Xu (2008) Both measures are analyzed through Ordinary Least Squares (OLS) and Panel Feasible Generalized Least Squares (FGLS) regressions There are two main reasons to focus on the relationship between infant mortality and fiscal decentralization in China First, China has achieved remarkable progress in reducing the IMR from 1949 to1978, which was the planned economy period with a low level of personal income With the reforms of 1978, China’s economy started to boom in the 1980s and maintained a high growth rate — an average of about 9% growth in real Gross Domestic 236 Jin & Sun Product (GDP) throughout the 1990s and into the 21st century.1 According to conventional views, higher economic development should be associated with the reduction of infant mortality (World Bank, 1993) In China, however, infant mortality stayed around 29 infant deaths per 1,000 live births from the late 1980s until present, and did not see further large-scale reductions despite high economic growth during that period of time (United Nations, 2005) Second, the 1994 TSS reform in China recentralized government revenues while keeping major healthcare expenditure responsibilities on the shoulders of sub-national governments without providing adequate funding support from the central government Conventional theories of fiscal decentralization predict that sub-national governments would be more responsive to local needs including healthcare delivery (Oates, 1993) Unlike other health indicators such as life expectancy and maternal mortality, infant mortality could be more sensitive to public health investments in the form of health expenditures by governments According to Barker (1997), Wagstaff (2001), and Case, le Roux, and Menendez (2004), prenatal healthcare, baby delivery facilities and personnel, infant nutrition, and public sanitation are all possible channels through which infant health could be affected These factors are also direct outcomes of government healthcare expenditures The increased responsibility coupled with inadequate funding at the sub-national level could contribute to the stagnation of infant mortality abatement since the late 1980s in China Thus, this study attempts to quantify whether the high-speed economic development during the 1990s and early 21st century, as well as the fiscal decentralization represented by the TSS reform of 1994 have affected infant mortality in China This study intends to improve existing studies in several ways First, we use a panel province-wide dataset that allows for the effects of time-varying unobservables Second, we measure health expenditures in total amount of expenditures, as a percentage of total government expenditures, and as a ratio to nominal Gross Regional Product (GRP) Third, we include several control variables such as a regional dummy, healthcare human capital, healthcare physical capital, urbanization, and fertility Finally, in addition to a traditional dummy measure, we also measure the degree of fiscal decentralization using the ratio of per capita provincial budgetary expenditures to the sum of per capita central budgetary expenditures and provincial budgetary expenditures The remainder of the paper is organized as follows Section briefly describes China’s healthcare delivery system Section surveys IMR production functions and possible channels through which infant mortality could be determined Section develops empirical models and introduces data sources See http://www.chinability.com/GDP.htm PFM 11/3 237 Section reports the results and section concludes with policy implications and suggestions for future research HEALTHCARE SYSTEM IN CHINA China has a unitary form of government with five levels hierarchically arranged in a pyramid-like fashion with the central government at the apex, sitting atop sub-national levels that consist of provincial, prefectural (including prefectural-level cities), county (including county-level cities), and township governments Provincial-level governments include 22 provinces, five ethnic minority autonomous regions, and four municipalities directly administered by the State Council Figure Infant mortality rate in China, 1950-2006 Data Source: Population Division of the Department of Economic and Social Affairs of the United Nations Secretariat, World Population Prospects As a part of public welfare during the planned economy period from 1949 through 1978, healthcare delivery was innovatively designed by the central government and successfully carried out by the sub-national governments The lower levels provided a public medical system in the urban areas and relied primarily on part-time peasant doctors (or ―barefoot doctors‖) in the rural areas The training and services of barefoot doctors were subsidized by the sub-national governments Sidel and Sidel (1975) summarize this type of medical system as a combination of traditional Chinese medicine and modern Western medicine: preventive, labor-intensive, cooperative-oriented, mass- 238 Jin & Sun based collectivism, and egalitarianism This system was proven effective in that it quickly reduced infant mortality from over 200 per 1,000 live births in 1950 to around 50 in 1978, about a three-quarter reduction in magnitude (see Figure 1) The average life expectancy in China has increased from about 35 in 1949 to about 70 in the early 1980s The overall health conditions in China improved significantly and many contagious diseases were eradicated in less than 30 years Due to its remarkable achievements, this system was recognized as a grassroots healthcare model by the World Health Organization (WHO) at the Alma Ata Conference in 1978 (WHO, 2008) However, this relatively successful centralized medical system did not survive the economic reforms of 1978, which promoted profit-seeking, privatization, commercialization, and marketization in the healthcare sector All medical institutions such as the Centers for Disease Control and Prevention (CDC) now have to be responsible for their own profits and losses in accordance with economic reforms without public financial support or any other sort of government subsidies Healthcare services including infant healthcare and immunization are charged at market prices As a result, the previous preventive and cooperative-oriented low cost medical system has been dissolved and replaced with a market-oriented medical system at soaring prices Less than one tenth of Chinese population, the majority of which are public servants or employees in state-owned enterprises (SOEs), has medical insurance (Bertelsmann Stiftung, 2010) Along with the marketization of healthcare goods, services and institutions, government healthcare expenditures have shrunk by nearly half As shown in Figure 2, total national expenditure on healthcare is composed of government budget expenditure, government extra-budget expenditure, and personal expenditure Among them, the share of government budgetary expenditure has decreased from about 39% in 1982 to about 18% in 2006, and government extra-budget expenditure has also reduced from more than 47% at the end of 1970s to 32% in 2006.2 In contrast, personal expenditure on healthcare has more than doubled over the past three decades from about 20% in 1978 to nearly 50% in 2006 After the 1994 TSS reform, healthcare expenditures were shifted from the central government to sub-national governments The officials of sub-national The account for public finance in China has dual tracks: budget account and extra-budget account Both include revenue and expenditure accounts Budgetary expenditure refers to the distribution and use of the funds that the government has raised based on Budget Law, so as to meet the needs of economic construction and various causes Extra-budgetary expenditure refers to the expenditure that is arranged in line with the extra-budget plans and appropriated from the special accounts at the same administrative level See World Bank (2000, 2001) for the role of extra-budget funds in China PFM 11/3 239 governments have been pursuing GDP-centered economic growth for the purpose of being promoted, at the cost of public healthcare investment Based on the relevant government regulations, the healthcare expenditure responsibility is jointly assumed by central, provincial, prefectural, and county levels of governments.3 In fact, the expenditures of central government on healthcare have been minimized It is provincial and sub-provincial level governments that shoulder the task Specifically, sub-national governments have assumed 97% of healthcare expenditures in recent years while the central government shares only 3% However, the major part of sub-national government revenues has been used to initiate large-scale infrastructure construction and investment projects in addition to funding administrative expenses The remaining public funds available for healthcare are minimal In contrast with the average 9% growth rate of nominal GDP annually, total healthcare expenditures as a percentage of nominal GDP decreases— from about more than 1% in 1981 to less than 1% in 2006.4 Furthermore, the share of healthcare expenditures in total government expenditures also decreased from more than 5% in 1981 to less than 5% in 2006 (see Figure 3) Figure Compositions of total healthcare expenditures, 1978-2006 Due to the reduction in government expenditures for healthcare and the increasing marketization of medical services, the overall performance of health3 For example, The State Council Document No 3, ―The Decision on Public Health Reform and Development‖, issued in January 1997, requires that public spending on health care at both the central government and sub-national governments increase at a higher rate of growth than general budgetary expenditures See http://www.china.com.cn/chinese/2006/Jan/1087140.htm 240 Jin & Sun care is unsurprisingly deteriorating Reduced government healthcare expenditures directly restrain the healthcare capital accumulation, which could result in deteriorating healthcare outcomes such as stagnating IMR reduction in the 1990s and 2000s In a health fairness assessment conducted by the WHO in 2002, China was ranked 144th among the 191 countries in the world Besides overall poor performance, the disparities in healthcare expenditures are also widening Government healthcare expenditures have shifted from rural areas to urban areas in order to train professional medical staff, purchase capitalintensive medical facilities, and finance advanced medical research The gap between urban and rural areas on healthcare expenditures is ever increasing in terms of healthcare expenditures per capita (see Figure 4) Hillier and Shen (1996) estimate that the gap of healthcare expenditures per capita between urban and rural areas increased four folds in 1981 and six folds in the 1990s Figure Shares of total health expenditures in total government expenditures and nominal GDP, 1981-2006 Through many years’ trials, China has now endeavored to realize a comprehensive healthcare system In urban areas, it is combined from socially accumulated funds and personal accounts with minimum compulsory medical insurance, employer’s compensatory medical insurance, and individual commercial medical insurance In rural areas, a new rural cooperative medical system has been carried out with joint funding from rural residents, sub-national governments, and the central government Although the 1978 economic reforms have brought remarkable economic growth in China, the healthcare delivery has not seen much improvement The IMR reduction has stagnated after 1980, as shown in Figure Meanwhile, the PFM 11/3 241 life expectancy has remained roughly the same from 68 in 1982 to 69 in 1993 (Hsiao & Liu, 1996) Furthermore, as Bloom and Gu (1997) and Liu et al (1999) reported, almost every single healthcare indicator is better for urban residents than for rural residents after the economic reforms For example, infant mortality in urban areas has been consistently dropping albeit at a slower rate compared to the speed before 1978 while the IMR in rural areas has been constantly increasing since the 1990s Figure Total health expenditures per capita in urban and rural areas Notes: (1) The total health expenditures include government budgetary expenditures on health, extra-budgetary expenditures on health and personal health expenditures; (2) the measurement unit is yuan per person LITERATURE REVIEW Infant mortality and life expectancy are frequently chosen as measurements of the level of overall health in current literatures In contrast with life expectancy’s being influenced more by personal health investment, accumulation of positive or negative factors during one’s life and individual living habits, infant mortality is impacted more by income level, public spending, and local medical facilities Accordingly, to quantify the impact of income and public healthcare expenditures, this study focuses on infant mortality only Infant mortality can be a result of both direct and indirect causes The former is mainly medical including immediate causes (such as immaturity, birth injury, genetic disease, and congenital anomaly) and chronic causes (such as malnutrition, prenatal care, availability of all vaccines, and infection) The indirect causes of infant mortality consist of social, economic and environmental factors that cause infants to be more exposable and sensitive to direct causes These factors include, but are not limited to, income level, income 242 Jin & Sun distribution, public healthcare expenditures, healthcare human capital, healthcare physical capital, women’s labor force participation, urbanization, ethnolinguistic fractionalization, quality of governance, public sanitation and other issues dealing with infrastructure such as access to safe water and electricity, and so forth Among them, public healthcare expenditures are direct input, whereas healthcare human capital (such as the number of doctors or nurses per one thousand persons) and physical capital (such as the number of hospital beds per one thousand persons) are healthcare direct output The overall effects of fiscal decentralization on the healthcare outcome include direct effects such as cost saving in production and delivery of healthcare services as well as indirect effects such as increased healthcare expenditures or improved healthcare capital Although historically IMRs fluctuate with wars, famines, epidemics and social turmoil, as the general welfare of a society improves its IMR declines Therefore, rich countries tend to have a lower IMR than poor ones Flegg (1982) conducts a cross-underdeveloped-country study over the period of 1968-1972 and uses OLS estimations controlling for income inequality, female fertility rate, female illiteracy rate, and healthcare human capital (measured by the number of doctors per 1,000 persons and the number of nurses per 1,000 persons) The result shows that the impact of per capita real GDP on IMR is not statistically significant, which suggests that income level (measured by per capita real GDP) is not a direct determinant of infant mortality and may affect infant mortality only indirectly through such factors as healthcare human capital In fact, using the WHO’s cross-country data in 2004, Anand and Barnighausen (2004) have confirmed the significantly positive relationship between healthcare human capital and infant mortality reduction Previous studies also find that public healthcare expenditures have a positive impact on infant mortality For example, Corman, Grossman, and Joyce (1987) use the 1977 cross-county neonatal mortality rates in the U.S and find that poverty-related public health expenditure programs play an important role in reducing neonatal mortality The World Bank (1995) also documents the significant effect of public health expenditure on infant mortality reduction in backward areas of the Philippines Using demographic and health survey data from over 60 low-income countries between 1990 and 1999, Wang (2003) finds that infant mortality in rural areas is substantially higher than in urban areas A recent study by Bokhari, Gai, and Gottret (2007) estimates the elasticity of under-five-year-old child mortality with respect to both income and government health expenditures using instrumental variable techniques and finds that mortality is affected by government health expenditure but not by economic growth PFM 11/3 243 On the other hand, some studies have produced opposing results For example, Filmer and Pritchett (1999) use the United Nations Children's Fund (UNICEF) and World Bank cross-country data with IV estimation and find that the effects of public health expenditures on infant mortality are both statistically and economically insignificant Musgrove (1996) summarizes that among the determinants of infant mortality, the income variable is always significant while healthcare expenditure share in GDP, healthcare expenditures in total government expenditures, and the government expenditure share in GDP are all insignificant Using a cross-sectional sample containing 117 countries in 1993 and a model correcting for heteroscedasticity, Zakir and Wunnava (1999) find that government healthcare expenditure and its share of GNP not play a role in determining infant mortality Berger and Messer (2002) also argue that the reverse relationship between healthcare expenditures and IMR reduction by previous studies does not hold based on their analysis of 19601992 data across 20 Organization for Economic Co-operation and Development (OECD) countries using OLS estimation On the contrary, they find that an increase in public healthcare expenditures is associated with an increase in IMRs In addition, their research suggests that increases in income inequality are related to lower mortality rates In regard to the impact of fiscal decentralization on IMR reduction, several studies argue that fiscal decentralization could lead to increased local governments’ responsiveness and accountability by providing preference-matching local public goods such as particular vaccination initiatives (Alesina & Spolaore, 1997; Faguet, 2004; Lockwood, 2002; Oates, 1972; Silverman, 1992) This effect is known as ―allocative efficiency.‖ Seabright (1996), Persson and Tabellini (2000), and Hindriks and Lockwood (2005) contend that fiscal decentralization could also reduce the incumbents’ rent diversion out of tax revenues Hayek (1945) argues that the provision by local governments regarding residents’ preferences saves information transmission costs from sub-national governments to the central government This effect is related to the ―productive efficiency.‖ For example, targeting a low-income population and nutritionally at-risk infants, local special welfare programs such as supplemental foods, health care referrals, and nutrition education for low-income pregnant women can be initiated immediately rather than waiting for approval from the central government Besides allocative efficiency and productive efficiency, the third possible gain of fiscal decentralization in a large country such as China is that various sub-national governments can experiment with alternative ways of reducing the IMR This type of effect could be called lab or ―experimental efficiency‖ (Garzarelli, 2006; Oates, 1999) All of the above are direct effects of fiscal decentralization on IMR reduction There exist other mechanisms through which fiscal decentralization could have indirect impact on the IMR For example, fiscal devolution could alter the local healthcare expenditure structure, thus affecting local healthcare human capital and subsequently the IMR 248 Jin & Sun FERit is the natural population growth rate (i.e., population growth rate minus death rate) used as a proxy for fertility rate It is expected that the higher the fertility rate, the higher the IMR Figure Average IMRs for four groups of provinces or province-level regions Notes: The geographical location dummy ―1‖represents four municipalities that are directly administered by the State Council – Beijing, Tianjin, Shanghai and Chongqing; ―2‖represents eight coastal provinces including Liaoning, Hebei, Shandong, Jiangsu, Zhejiang, Fujian, Guangdong and Hainan; ―3‖represents thirteen inland provinces including Jilin, Heilongjiang, Anhui, Jiangxi, Henan, Hubei, Hunan, Shanxi, Sichuan, Guizhou, Yunnan, Shaanxi and Gansu; and ―4‖represents five ethnically minority populated autonomous regions – Inner Mongolia, Guangxi, Tibet, Ningxia and Xinjiang Finally, the error terms υi is time-invariable region specific shock and ℮it is white noise Table provides summary statistics of the variables The IMRs range from 3.66 (in 2000 of Beijing) to around 122 (in 1981 of Xinjiang) with the mean value of 32 deaths per 1,000 live births The numerical degree of fiscal decentralization ranges from 0.33 to about 0.93 with an average of 0.63.The healthcare expenditures per capita range from about 10 yuan to 341 yuan and averages at 49 yuan The natural population growth rate ranges from -1.35 to 23.57 with a mean of 11 The urbanization rate ranges from 10 to about 82% and its average is around 27% PFM 11/3 249 RESULTS Since the modified Wald test suggests that our data are subject to heteroskedasticity and the Wooldridge test for autocorrelation in panel data indicates the existence of serial correlation in our data, we estimate standard errors with robust options in the OLS models and use panel FGLS technique correcting for heteroskedasticity and panel-specific AR(1) autocorrelation URBANit is defined as the ratio of urban population to total population It represents the level of urbanization and captures the disparity of the IMR between urban and rural areas with a possibly negative β11 We first run the regressions with fiscal decentralization measured by a dummy variable Using OLS with robust option, we estimate the impact of income and fiscal decentralization in Model (i); and then add three health expenditure variables: lnHEPCit, HESEit, and HESGit in Model (ii), among which high multicollinearity could exist; in Model (iii), we add two health expenditure output variables: healthcare physical variable represented by BEDPit and healthcare human capital variable represented by DOCPit; in Model (iv), we include all other control variables: geographical dummy, interactive term between fiscal decentralization and geographical dummy, urbanization, and fertility rate Lastly, we apply panel FGLS technique to the model after correcting for heteroskedasticity and panel-specific AR(1) autocorrelation The FGLS method has the advantage over a fixed-effects model in that the unobserved, time-invariant heterogeneity assumed by the fixed-effects model does not necessarily apply to different provinces because the objectives of sub-national governments in different regions are changing (WHO, 2008) Contrary to our expectations, as shown in Table 2, fiscal decentralization represented by the 1994 TSS reform, FDit, has increased infant mortality nation wide with significantly positive signs in each of the OLS models and the FGLS model of panel estimation at or 5% levels The adding of health expenditure-related variables, health expenditure output variables and other control variables step by step (stepwise regression) does not influence the magnitude and sign of FDit ’s effects on IMRit at all (Efroymson, 1960) With inclusion of all control variables and other things being equal, the TSS reform increases infant mortality by about 24 per 1,000 live births under a year of age in the same year, which is not only statistically significant at the 1% level but also both economically and sociologically significant Fortunately, after controlling for heteroskedasticity and panel-specific AR(1) autocorrelation, this adverse effect of FDit on IMRit is reduced to about 13, which is still statistically significant at the 1% level 250 Jin & Sun Table Summary statistics Obs Mean Std Dev Min Max 178 32.00 21.18 3.66 121.92 FD dummy ("0" before the 1994 TSS and "1" after the 1994 TSS) 186 0.33 0.47 FD ratio (defined as the ratio of per capita provincial budgetary expenditures to the sum of per capita central budgetary expenditures and provincial budgetary expenditures) 182 0.63 0.14 0.33 0.93 182 7.50 1.32 5.31 10.76 Public healthcare expenditures per capita in log form (lnHEPC) 178 3.60 0.74 2.27 5.83 The share of public health expenditures in total public expenditures (HESE) 178 0.20 0.18 0.02 1.61 The share of public health expenditures in Nominal Gross Regional Product (HESG) 178 0.03 0.02 0.002 0.12 The number of medical beds per ten thousand persons (BEDP) 182 25.02 11.27 0.17 62.10 The number of doctors per ten thousand persons (DOCP) 182 17.63 9.44 0.13 46.30 Geographical location for each province (GEO) 186 2.65 0.90 Natural population growth rate (FER) 180 11 5.35 -1.35 23.57 0.27 0.17 0.09 0.82 Variables Dependent variable Infant Mortality Rate (IMR) (‰) Independent variables Fiscal decentralization (FD) with two measures: Real Gross Regional Product per capita in log form (lnGRPPC) Urbanization rate (URBAN, the percentage of urban population in total popula182 tion) Data Source: National Bureau of Statistics in China Notes: IMR is census data PFM 11/3 251 Table Regression Result (FD as a Dummy) Dependent variable: yearly provincial infant mortality rate (‰) (ii) 40.79*** (5.95) -8.64*** (3.19) 0.34 (2.93) OLS (iii) 41.61*** (6.12) -9.60*** (3.29) 0.14 (2.97) (iv) 24.13*** (7.91) -5.13 (3.55) -1.41 (3.11) Panel Estimation FGLS 13.09*** (3.69) -6.09*** (1.61) 0.32 (1.29) HESE 44.27*** (14.51) 40.61*** (13.54) 14.22** (5.78) -25.35* (13.21) HESG 872.45*** (98.25) 876.30*** (95.59) 798.20*** (120.76) 523.22*** (96.80) 0.34* (0.19) -0.27* (0.15) 0.37** (0.18) -0.09 (0.18) 2.88 (2.47) 0.24*** (0.09) -0.09 (0.11) 3.56*** (1.25) 0.79 (2.25) -0.01 (1.27) 0.18 (0.19) -8.64* (5.25) 47.46*** (11.07) 172 FD dummy lnGRPPC (i) 13.69** (4.04) -12.69*** (1.73) lnHEPC BEDP DOCP GEO FD*GEO 122.88** (12.52) 49.60*** (16.88) 53.74*** (17.25) 0.15 (0.30) -4.83** (7.21) 29.13 (19.07) Observation number 178 174 174 172 R-squared 0.30 0.60 0.61 0.63 FER URBAN Constant Wald Chi-squared 340.81 (1) ***statistically significant at 1%, **significant at 5%, *significant at 10%; (2) In parentheses are standard errors of coefficients; (3) OLS are estimated with robust option; (4) FGLS is estimated correcting for heteroskedasticity and autocorrelation AR(1) This implies that since 1994 TSS reform, the sub-national governments in China have focused on GDP-centered economic growth while ignoring the basic livelihood of local residents The potential benefits of fiscal decentralization have not been realized as predicted by the conventional theories Table also indicates that income level represented by real GRP per capita, lnGRPPCit, has expected negative signs in all of the five models and is statis- 252 Jin & Sun tically significant in the first three OLS models and the FGLS model After controlling other variables step by step, the negative effect of income level on infant mortality has decreased, but is still statistically significant at the 1% level in the FGLS and three OLS models This illustrates that the income level still contributes to infant mortality rate reduction although this effect could be offset by relevant health expenditure input variables and health expenditure output variables upon inclusion The healthcare expenditure per capita, lnHEPCit, has mixed signs but none are statistically significant The share of health expenditure in total expenditure: HESEit, and the share of health expenditure in nominal GRP, HESGit, have positive signs and statistically significant at the one or 5% levels in both OLS and FGLS models with only one exception: the HESEit estimate in the FGLS model is negatively signed and are statistically significant at the 10% level This shows that an increase in public healthcare expenditure share in total public expenditure or in total gross regional product is associated with an increase in infant mortality rates This echoes part of the findings by Berger and Messer (2002) in a cross-country study To explore the indirect impact of FDit on IMRit through health spending, we use lnHEPCit as the dependent variable and regress it on FDit and lnGRPPCit, and then obtain the predicted value, The result is reported in the following equation: (2) (0.23) (0.12) (0.03) n=178, R2=0.71 In parentheses are robust standard errors We then run the main model specified in equation (1), with the inclusion of the predicted value, and obtain the coefficient of The result is reported in the following equation: (3) (10.18) (2.62) n=172, R2=0.62 The total effect of FDit on IMRit is the sum of the direct effect, the coefficient of FDit in the main model (iv), and the indirect effect, the product of the coefficient of FDit in equation (2) and the coefficient of in equation (3) That is, PFM 11/3 253 All of these three estimates are statistically significant at the 1% level This result strengthens the positive effect of fiscal decentralization on infant mortality, in terms of either direct effects or indirect effects through health expenditures In addition, the healthcare physical capital BEDPit has an unexpected positive sign and is statistically significant at the 5% level in model (iv) and the 1% level in the FGLS model This surprising result could be explained by the dependency hypothesis stating that as healthcare facility supplies increase, the population becomes increasingly dependent upon healthcare facilities to maintain their health and neglects the more important lifestyle factors and nutrition (Sidel & Sidel, 1975) Another possible explanation is that the utilization rate of hospital beds by infants is very low even with increasing hospital beds because many infants are cared for in the residents’ homes in China The healthcare human capital variable DOCPit has the expected negative sign although it is statistically significant at 10% level only in the OLS model (iii) The geographical location variable GEOit has the expected positive signs The coefficient of this geographical dummy is statistically insignificant in the OLS model but is statistically significant at the 1% level in the FGLS model This indicates that the infant mortality rate is higher in western and inland regions than that in eastern developed regions of China The interactive term of fiscal decentralization and geographical dummy FDt*GEOit has mixed signs but neither of them is statistically significant The fertility rate, FERit, approximated by the population natural growth rate, has the expected positive sign but is not statistically significant It could be because the population natural growth rate is a poor proxy of fertility Urbanization rate, URBANit, has the negative sign as predicted by Weng and Wang (1993) and the effect is statistically significant at the and 10% levels in the OLS and FGLS models, respectively This suggests that the gap in healthcare outcomes between urban and rural areas is still big Table displays the results of OLS and FGLS regressions using fiscal decentralization measured as the ratio of per capita provincial budgetary expenditures to the sum of per capita central budgetary expenditures and per capita provincial budgetary expenditures Again, fiscal decentralization shows positive signs in all models and is statistically significant at the 1, and 10% levels in all OLS models while insignificant in the FGLS model 254 Jin & Sun Table Regression result (FD as a ratio) Dependent variable: yearly provincial infant mortality rate (‰) Panel Estimation FGLS 27.74 (21.26) 0.85 (1.94) (ii) 74.35** (35.65) 9.03*** (2.45) OLS (iii) 85.84** (34.48) 7.72*** (2.42) -26.35*** (6.75) -25.48*** (6.32) 23.93** (10.83) 986.83*** (111.06) 22.31** (10.57) 938.61*** (110.44) (iv) 86.05* (45.64) 2.01 (2.63) 16.31*** (5.61) 1.49 (8.73) 671.00*** (119.11) BEDP 0.20 (0.17) 0.28* (0.16) 0.22** (0.10) DOCP -0.50*** (0.17) -0.02 (0.18) -0.10 (0.13) GEO 8.47 (6.50) 2.09 (4.08) FD*GEO -4.02 (8.90) 5.43 (5.69) FER -0.24 (0.30) -0.20 (0.19) -36.16*** (9.91) -12.80* (6.78) -7.50 (21.40) 172 0.62 19.43 (15.98) 172 FD ratio lnGRPPC (i) 55.28*** (16.02) -12.33*** (1.60) lnHEPC HESE HESG URBAN Constant 89.90*** (7.17) 178 0.35 -19.31 (13.91) 174 0.52 -14.41 (13.76) 174 0.54 Observation number R-squared Wald Chi-squared (1) ***statistically significant at 1%, **significant at 5%, *significant at 10%; (2) In parentheses are standard errors of coefficients; (3) OLS are estimated with robust option; (4) FGLS is estimated correcting for heteroskedasticity and autocorrelation AR(1) -10.63*** (2.60) -18.81 (12.12) 586.91*** (88.40) 339.22 To explore the impact of fiscal decentralization measured as a ratio on IMRs through health expenditures, we follow the same procedure described above and regress lnHEPCit, on FDit and lnGRPPCit with robust option, obtain the predicted values for lnHEPCit,, plug it into the original OLS model (iv) PFM 11/3 255 with the full set of other control variables, and calculate the total effect of of FDit on IMRit as follows: Again, the total effect of FDit on IMRit is still positive and this overall positive size is even bigger than that obtained with the dummy measure of fiscal decentralization The estimate of real GRP per capita has a negative sign and is statistically significant for model (i) while becoming positive and statistically significant for model (ii) and (iii) This seemingly contradicting result could be associated with the estimates for health expenditures per capita With the numerical measurement of fiscal decentralization, health expenditures per capita have expected negative signs and are statistically significant at the 1% level in all models This effect induced by healthcare expenditures could be the indirect effect of real GRP per capita on infant mortality reduction through healthcare expenditures This is also in line with our conclusion about income factor mentioned above: the diminishing effect of income variable on infant mortality could be due to the inclusion of correlated health expenditure input variables and health expenditure output variables The other two health expenditure-related ratios, HESEit, and HESGit, are very similar with the estimates in Table The higher shares of healthcare expenditures in total public expenditures and total income are related to higher infant mortality It could be a vicious cycle of outcomes for poor countries BEDPit and DOCPit have similar results as in Table GEOit has positive signs but is statistically insignificant even in the FGLS model FDt*GEOit, FERit, and URBANit have similar estimates and signs as in Table as well Furthermore, in view of potential endogeneity of the income variable, the instrument variable estimation is also applied with Generalized Methods of Moments (GMM) using default heteroskedasticity-robust weight matrix (not reported here) Fiscal decentralization measured by the dummy variable has the same expected signs and is also statistically and economically significant In testing the potential endogeneity of real GRP per capita, we failed to reject the null that the income variable can be assumed exogenous CONCLUSIONS The preceding study examined whether sub-national governments in China are becoming more responsive to local healthcare needs after fiscal decentralization represented by the 1994 TSS reform The findings are opposite to the predictions by conventional theories of fiscal decentralization and the empirical evidence presented in many previous research Our result shows that fiscal 256 Jin & Sun decentralization has played an overall adverse impact in reducing the IMRs in China, either with a dummy measure or a ratio measure We also find that income level plays a role in reducing infant mortality and can be assumed exogenous in the IMR production function However, the income effect can be dwindling after control for other income-related variables such as healthcare expenditures per capita The shares of healthcare expenditures in total public expenditures and total gross regional product have an adverse effect on infant mortality reduction Urbanization has the expected effects as predicted by previous studies The increase in healthcare physical capital is positively associated with the IMRs while that in healthcare human capital is negatively associated with the IMRs This study has important policy implications The result suggests that fiscal decentralization needs to be more deliberately designed in China to balance local residents’ healthcare needs and economic development Specifically, we make the following policy recommendations: first, a comprehensive performance appraisal system instead of the GDP-centered economic growth appraisal system for sub-national government officials should be established to achieve a sustainable livelihoods as well as local economic development Second, a more equalized intergovernmental transfer system diverting certain public resources from relatively developed regions to inland and minority concentrated regions could be a helpful tool to narrow the gap in health outcomes among different geographical locations Third, urbanization seems to be a feasible channel to lower the IMRs more effectively over the course of economic development than the share raise of healthcare expenditures in total public expenditures and total income Despite of the above important findings, this study is limited in several ways First, the IMR census data are available only for six years since the 1980s The low number of observations will not allow much variation in this key variable Second, female literacy data are not available, which constrains the inclusion of education effects for pregnant women on the IMRs Finally, fiscal decentralization is measured from the expenditure side with a policy variable and in aggregate level only and does not take into account the possible effects arising from revenue structure and expenditure structure Considering these limitations, we suggest that further research and analysis include more indicators of healthcare outcomes such as under-five-year-old child mortality, maternal mortality, and life expectancy, and further explore the impact of different fiscal revenue structure and fiscal expenditure structure on these healthcare outcomes REFERENCES PFM 11/3 257 Akai, Nobuo and Sakata, Masayo (2002) ―Fiscal Decentralization Contributes to Economic Growth: Evidence form State-Level Cross-Section Data for the United States‖, Journal of Urban Economics, Vol 52, No 1, pp 93-108 Alesina, Alberto and Spolaore, Enrico (1997) ―On the Number and Size of Nations‖, Quarterly Journal of Economics, Vol 112, No 4, pp 1027-1056 Anand, Sudhir and Bärnighausen, Till (2004) ―Human Resources and Health Outcomes: Cross-Country Econometric Study‖, Lancet, Vol 364, No 9445, pp 1603-1609 Asfaw, Abay, Frohberg, Klaus, James, K S and Jutting, Johannes (2007) ―Fiscal Decentralization and Infant Mortality: Empirical Evidence from Rural India‖, Journal of Developing Areas, Vol 41, No 1, pp 17-36 Barker, David J P (1997) ―Maternal Nutrition, Fetal Nutrition and Diseases in Later Life‖, Nutition, Vol 13, No 9, pp 807-813 Berger, Mark C and Messer, Jodi (2002) ―Public Financing of Health Expenditures, Insurance, and Health Outcomes‖, Applied Economics, Vol 34, No 17, pp 2105 – 2113 Bertelsmann Stiftung (2010) BTI 2010 — China Country Report, Gütersloh: Bertelsmann Stiftung Blaxter, Mildred (1981) The Health of Children: A Review of Research on The Place of Health in Cycles of Disadvantage, London: Heinemann Educational Books Bloom, Gerald and Gu, Xingyuan (1997) ―Health Sector Reform: Lessons from China‖, Social Science & Medicine, Vol 45, No 3, pp 351-360 Bokhari, Farasat A S., Gai, Yunwei and Gottret, Pablo (2007) ―Government Health Expenditures and Health Outcomes‖, Health Economics, Vol 16, No 3, pp 257-273 Cantarero, David and Pascual, Marta (2008) ―Analysing the Impact of Fiscal Decentralization on Health Outcomes: Empirical Evidence from Spain‖, Applied Economics Letters, Vol 15, No 2, pp 109-111 Case, Anne, le Roux, Ingrid and Menendez, Alicia (2004) ―Medical Compliance and Income-Health Gradients‖, American Economic Association Papers and Proceedings, pp 331-335 258 Jin & Sun Corman, Hope, Grossman, Michael and Joyce, Theodore (1987) ―Birth Outcome Production Functions in the U.S.‖ The Journal of Human Resources, Vol 22, No 3, pp 339-360 Duret, Elsa (1999) ―Depenses Publiques et Mortalite Infantile: Les Effets de la Decentralisation‖ (Public Expenditure and Infant Mortality Rates (IMR): The Effects of Fiscal Decentralization With English Summary), Revue d'Economie du Developpement, pp 39-68 Ebel, Robert D and Yilmaz, Serdar (2002) ―On the Measurement and Impact of Fiscal Decentralization‖, Policy Research Working Paper No 2809, Washington, DC: World Bank Efroymson, M A (1960) ―Multiple Regression Analysis‖, in A Ralston and H.S Will (eds.), Mathematical Methods for Digital Computers, New York: Wiley Faguet, Jean-Paul (2004) ―Does Decentralization Increase Government Responsiveness to Local Needs? Evidence from Bolivia‖, Journal of Public Economics, Vol 88, No 3-4, pp 867-893 Filmer, Doen and Pritchett, Lant (1999) ―The Impact of Public Spending on Health: Does Money Matter?‖ Social Science & Medicine, Vol 49, pp 13091323 Flegg, A.T (1982) ―Inequality of Income, Illiteracy and Medical Care as Determinants of Infant Mortality in Underdeveloped Countries‖, Population Studies, Vol 36, No 3, pp 441-458 Garzarelli, Giampaolo (2006) ―Cognition, Incentives, and Public Governance: Laboratory Federalism from the Organizational Viewpoint‖, Public Finance Review, Vol 34, No 3, pp 235-257 Green, A and Collins, C (1994) ―Decentralization and Primary Health Care: Some Negative Implications in Developing Countries‖, International Journal of Health Services, Vol 24, No 3, pp 459-475 Guldner, Mattias (1995) ―Health Care in Transition in Vietnam: Equity and Sustainability‖, Health Policy and Planning, Vol 10 (Supplement), pp 49-62 Halder, Pragna (2007) ―Measures of Fiscal Decentralization‖, Retrieved from http://aysps.gsu.edu/intern/papers/2007_HalderPragna_Measures.pdf Hayek, Friedrich A (1945) ―The Use of Knowledge in Society‖, The American Economic Review, Vol 35, No 4, pp 519-530 PFM 11/3 259 Hillier, S and Shen, J (1996) ―Health Care Systems in Transition: People's Republic of China‖, Journal of Public Health Medicine, Vol 18, No 3, pp 258-265 Hindriks, Jean and Ben Lockwood (2005) ―Decentralization and Electoral Accountability: Incentives, Separation and Voter Welfare‖, CEPR Discussion Paper 5125, England Hsiao, William and Liu, Yuanli (1996) ―Economic Reform and Health — Lessons from China‖, New England Journal of Medicine, Vol 335, No 6, pp 430-432 Lin, Justin Yifu and Liu, Zhiqiang (2000) ―Fiscal Decentralization and Economic Growth in China‖, Economic Development and Cultural Change, Vol 49, No 1, pp 1-21 Liu, Yuanli, Hsiao, William C and Eggleston, Karen (1999) ―Equity in Health and Health Care: The Chinese Experience‖, Social Science & Medicine, Vol 49, No 10, pp 1349-1356 Lockwood, Ben (2002) ―Distributive Politics and the Costs of Centralization‖, Review of Economic Studies, Vol 69, No 2, pp 313-337 Mills, A., Vaughan, J P., Smith, D., and Tabibzadeh, I (1990) Health System Decentralization: Concepts, Issues and Country Experiences, Geneva: World Health Organization (WHO) Musgrove, Philip (1996) ―Public and Private Roles in Health: Theory and Financing Patterns‖, World Bank Discussion Paper No 339, Washington, DC: World Bank Oates, Wallace E (1972) Fiscal Federalism New York: Harcourt Brace Jovanovich Oates, Wallace E (1993) ―Fiscal Decentralization and Economic Development‖, National Tax Journal, Vol 46, No 2, pp 237-243 Oates, Wallace E (1999) ―An Essay on Fiscal Federalism‖, Journal of Economic Literature, Vol 37, No 3, pp 1120-1149 Persson, Torsten and Tabellini, Guido (2000) Political Economics: Explaining Economic Policy, Boston, MA: MIT Press 260 Jin & Sun Prud'homme, Remy (1995) ―The Dangers of Decentralization‖, World Bank Research Observer, Vol 10, No 2, pp 201-220 Qiao, Baoyun, Martinez-Vazquez, Jorge and Xu, Yongsheng (2008) ―The Tradeoff between Growth and Equity in Decentralization Policy: China's Experience‖, Journal of Development Economics, Vol 86, No 1, pp 112-128 Robalino, David A., Picazo, Oscar F., and Voetberg, Albertus (2001) ―Does Fiscal Decentralization Improve Health Outcomes? Evidence from a CrossCountry Analysis‖, Policy Research Working Paper Series No 2565, Washington, DC: World Bank Seabright, Paul (1996) ―Accountability and Decentralisation in Government: An Incomplete Contracts Model‖, European Economic Review, Vol 40, No 1, pp 61-89 Sen, Amartya K (1998) ―Mortality as an Indicator of Economic Success and Failure‖, Economic Journal, Vol 108, No 446, pp 1-25 Sidel, Victor W and Sidel, Ruth (1975) ―The Development of Health Care Services in the People's Republic of China‖, World Development, Vol 3, No 7-8, pp 539-549 Silverman, Jerry M (1992) Public Sector Decentralization: Economic Policy Reform and Sector Investment Programs, Washington DC: World Bank Tang, Shenglan and Bloom, Gerald (2000) ―Decentralizing Rural Health Services: A Case Study in China‖, The International Journal of Health Planning and Management, Vol 15, No 3, pp 189-200 Tanzi, Vito (1996) ―Fiscal Federalism and Decentralization: A Review of Some Efficiency and Macroeconomic Aspects‖, Annual World Bank Conference on Development Economics, 1995 (pp 295-316), Washington, DC: World Bank Uchimura, Hiroko and Jutting, Johannes (2007) ―Fiscal Decentralization, Chinese Style: Good for Health Outcomes?‖ IDE Discussion Paper No.111, Institute of Developing Economies, Japan United Nations (2005) World Population Prospects United Nations Development Programme (UNDP) (1990-2008) Human Development Reports, United Nations PFM 11/3 261 Wagstaff, Adam (2001) ―Poverty and Health‖, Working Group No.1, Working Paper No Boston, MA: WHO Commission on Macroeconomics and Health Wang, Limin (2003) ―Determinants of Child Mortality in LDCs: Empirical Findings from Demographic and Health Surveys‖, Health Policy, Vol 65, No 3, pp 277-299 Weng, S and Wang, S (1993) ―An Analysis of Infant Mortality in China‖, Chinese Journal of Population Science, Vol 5, No 1, pp 75-81 World Bank (1993) World Development Report: Investing in Health, Washington, DC: World Bank World Bank (1995) ―Colombia Local Government Capacity: Beyond Technical Assistance‖, World Bank Report 14085-C, Washington DC: World Bank World Bank (2000) China: Managing Public Expenditures for Better Results, Washington, DC: World Bank World Bank (2001) China: Provincial Expenditure Review, Washington, DC: World Bank World Health Organization (WHO) (2008) ―Consensus During the Cold War: Back to Alma-Ata‖, Bulletin of the World Health Organization, Vol 86, pp 737-816 Zakir, Mohammed and Wunnava, Phanindra V (1999) ―Factors Affecting Infant Mortality Rates: Evidence from Cross-sectional Data‖, Applied Economics Letters, Vol 6, No 5, pp 271-273 Zhang, Tao and Zou, Heng-Fu (1998) ―Fiscal Decentralization, Public Spending, and Economic Growth in China‖, Journal of Public Economics, Vol 67, No 2, pp 221-240 Reproduced with permission of the copyright owner Further reproduction prohibited without permission ... 2007) Within the healthcare sector, fiscal decentralization specifically refers to the decentralization of financial resources and expenditure responsibilities for healthcare from central government... the impact of fiscal decentralization on the IMR in China The purpose of this study is to provide a quantitative measurement of the impact of fiscal decentralization on the IMR in China using provincial... China and find little evidence that radical fiscal decentralization leads to increased healthcare outcomes Their research cautions against attempts to implement radical decentralization without