|
| HOME 主页 VITA 简历 RESEARCH 研究 TEACHING 教学 RESOURCES 资源 WUHAN 武汉 历史地图 |
|
Guo, Gang. 2022. "Race to the Top or Bottom? Globalization and Education Spending in China." Journal of Education Finance, Volume 48, Issue 1, Summer, Pages 94-111.
The question of whether economic globalization promotes or depresses governments’ education spending has attracted social scientists’ attention for decades. Existing literature presents an interesting contrast between two theoretical scenarios, namely race to the top and race to the bottom. This paper attempts to adapt the debate to the subnational context of China and argues that, under this decentralized authoritarian setting, economic globalization could boost the absolute levels of education funding by incentivizing human capital formation and by contributing to local government coffers but shrink its relative share in overall spending by shifting government priorities away from education to other budget items such as infrastructure that relate more closely to foreign investment. A dynamic panel data analysis of provincial-level statistics from China over an 11-year period confirms that inflow of foreign direct investment increases the absolute level but decreases the relative weight of education spending in the overall provincial budget, essentially a race to the top and to the bottom at the same time.
Keywords
China, economic globalization, foreign direct investment, education spending
The crucial benefits of public education for individuals and societies have been generally accepted among academic researchers, practitioners, and other stakeholders, yet in practice government commitment to education varies widely around the world. In 2018, nearly a third of government expenditure in the West African country of Sierra Leone went to education while its neighbor Liberia spent only 8 per cent of government expenditure on education (World Bank 2022). Likewise in Asia, the proportions of government budget spent on education for Malaysia (20 per cent) and Vietnam (14 per cent) are much higher than those for Myanmar (10 per cent) and Cambodia (9 per cent). Even for the same country, government spending on education could also fluctuate significantly over time. China’s education system, the largest in the world with a quarter billion students, has accounted for between 13.7 per cent (2009) and 16.9 per cent (2012) of total government expenditure from 2007 through 2019 (National Bureau of Statistics 2022).
Previous studies on education funding have examined various underlying factors behind those variations. Of particular interest here is the intersection of that literature with the research on the social and welfare implications of globalization. In the past half century, globalization has probably been the most significant social economic force that shapes the world we live in. Especially economic globalization, generally defined as the increasing integration of almost every country on earth into the world markets for goods, services, capital, and labor, has had tremendous impact on governments and peoples. Existing studies on the effect of economic globalization on education spending have often produced ambiguous and/or divergent theoretical predictions and empirical findings, essentially on whether such spending expands or contracts under globalization (e.g. Busemeyer and Garritzmann 2017; Baskaran and Hessami 2012; Busemeyer and Trampusch 2011; Nooruddin and Simmons 2009; Ansell 2008). This paper attempts to contribute to the current debate by extending the theoretical argument and empirical test to the subnational cases of China. In particular, we argue that economic globalization could boost the absolute levels of education funding but shrink its relative proportions in the overall budget, in a sense a race to the top and to the bottom at the same time. The study also attempts to move beyond examining just the national aggregates and to leverage the subnational variations in both education funding and economic globalization over time where many of the confounding institutional factors are held constant within the same country. Finally, the authoritarian but decentralized setting of China provides a unique research opportunity that has thus far been underutilized in existing studies on globalization and education funding.
The rest of the paper is organized as follows. The next section reexamines the theoretical predictions of the “race to the top” and “race to the bottom” scenarios, especially with respect to absolute levels versus relative proportions of education spending under economic globalization and also with respect to subnational variation, competition, and government incentives. Then the theoretical arguments are extended to the subnational cases of China, where economic globalization, especially foreign direct investment, could lead the provinces to “race to the top” in absolute levels of education spending but “race to the bottom” in relative budget shares. These two hypotheses are then tested on a panel data set of Chinese provinces over an 11-year period and the findings interpreted and discussed. The last section concludes and also explores some of the implications of this study.
Economic globalization has gained an unscrupulous reputation in recent decades, especially in rich industrialized countries (Rudra et al. 2021). An interesting research question that would have direct bearing on the normative debate over the human cost of globalization is whether exposure to foreign trade and investment stimulates or constrains government spending on education. Theoretically both scenarios have plausible foundations, and each has received some empirical support in cross-national and within-nation studies. For the sake of convenience, I shall call these two scenarios "race to the top" and "race to the bottom" respectively and discuss them in turn next.
According to the scenario of "race to the top," globalization intensifies the competition between governments in public spending on human capital formation. Theoretically it is based on simple assumptions about the incentives of investors and governments. The decision by multinational corporations to choose the location of their investment is based on a wide variety of considerations (Plouffe 2019; Blonigen 2005), and the level of human capital in a prospective country is an important factor, although the empirical evidence has been rather mixed (Garriga 2021). Domestic firms exposed to international market also benefit from and thus would welcome government spending to increase human capital. Human capital formation especially in the areas of education and health care generally contributes to a high-quality and productive workforce that help both domestic and foreign-invested companies remain competitive in the world market. Given the value of human capital to investors abroad and at home, governments will spend more on such budget items as education and health care, and the race to attract foreign capital and to remain globally competitive is in effect turned into a "race to the top" of domestic human capital formation.
It should be pointed out that a "compensation" hypothesis in the literature makes similar predictions to that in the scenario of "race to the top" but is based on quite different theoretical foundations and clearly should not apply to this paper. According to the "compensation" hypothesis, globalization exacerbates economic inequality and insecurity, which in turn prompts governments to increase social spending in order to compensate the losers from globalization (Garrett 2001: 3) and to prevent political instability (Kaufman and Segura-Ubiergo 2001: 557), especially in welfare expenditures on such programs as social security, unemployment benefits, retraining, etc. that provide income supplements (Cameron 1978: 1258). Even in its relative infancy in 1960, economic globalization already became “the best single predictor” of the extent of expansion of the public economy in industrialized Western countries (Cameron 1978: 1254). In contrast, the "race to the top" scenario in this paper mostly concerns government investment on human capital formation such as education and health care. In other words, government spending on education and health care does not necessarily "compensate" those at risk due to increasing integration into the world market (Hecock 2006: 952).
Moreover, the "compensation" hypothesis may be inapplicable and untestable in developing countries. Inspired initially by the experiences of rich industrialized democracies, the "compensation" hypothesis does not travel well to non-democracies, where public demands for compensation do not readily translate into government spending programs. Findings by Kaufman and Segura-Ubiergo, for instance, "suggest that Latin American democracies do generally support demands for more progressive forms of social spending" and that “regimes matter” (2001: 584). Likewise, Bhagwati argued that “a set of strong institutions, including labor unions and social democratic parties” can neutralize the globalization pressures for countries to race to the bottom (2004: 101). Such institutions would obviously be absent in an authoritarian regime, which calls into question the applicability of the “compensation” hypothesis (Dreher et al. 2008, Nooruddin and Simmons 2009). Besides, state capacity in developing countries to intervene in their economy may be more constrained by various obstacles. For example, agricultural subsidies are probably one of the most important mechanisms used by rich nations to compensate farmers exposed to the world market, but building an adequate infrastructure for even information collection and implementation of such compensations has proven to be a daunting task for many poor developing countries. In all, compared with advanced industrial democracies, developing countries may lack the incentive, capacity, or popular demand to compensate their domestic losers from globalization (Potrafke 2018).
The scenario of "race to the bottom" predicts the opposite effect of globalization on social spending (Rudra 2008). According to this scenario, international and domestic firms are more interested in lowering their cost, especially their tax burden, than in having access to a high-quality productive (but more expensive) labor force (Plouffe 2019; Harish & Plouffe 2020). Faced with the pressure to remain competitive in attracting foreign investment and in helping domestic businesses, governments will choose to constrain their fiscal capacity by cutting down expenditures on social programs. In the end, states converge to the lowest common denominator in public spending. This theoretical scenario is consistent with the argument that competition between governments results in more efficient public spending and therefore can be called the "efficiency" hypothesis (Garrett 2001: 6, Hansson and Olofsdotter 2008: 1004). Besides explaining government spending, the scenario of "race to the bottom" has also been applied to other areas of government behaviors, such as reduced regulation, lower environmental standards, and violation of labor rights (Mosley and Uno 2007).
Given the directly opposite predictions of the two theoretical scenarios, it may seem a simple and straightforward empirical question as to whether governments are racing to the top or bottom in the face of globalization. However, there are three important complications that have to be taken into consideration before we can set up empirical tests of the above scenarios using real-world data. First, the theories mainly imply changes in government spending on specific categories such as welfare, health care, and education rather than total spending. The empirical test would require detailed statistical data on disaggregated government budget allocation which however may not be readily available especially in developing countries (UNESCO Institute for Statistics 2022).
More importantly, it is often ambiguous in the existing literature whether the expected changes in government spending on such budget items as education should be measured as absolute levels or relative proportions in total budget (Dreher et al. 2008). The race to the top theory of human capital formation explicitly predicts a rise in education spending levels but leaves largely unanswered the question whether the proportion of education in total budget would rise as well, since it depends on changes in the rest of the budget. Likewise, although the race to the bottom theory predicts shrinking government spending levels on education, it is unclear to what extent the proportion of education spending in total budget would suffer, since globalization should also pressure governments to cut other spending items in pursuit of fiscal efficiency and discipline. As will be discussed in the next section, this ambiguity in the empirical implications of the two theoretical scenarios leaves open the possibility that governments could race to the top and bottom at the same time, in levels of education spending and in its proportions in total budget respectively.
Finally, the worldwide trend of fiscal decentralization in recent decades has further complicated the comparative study of the impact of globalization on government social spending. To varying degrees many national governments especially in the developing world have delegated spending responsibilities on such budget items as education and health care to sub-national levels (OECD/UCLG 2019: 44). That adds another layer of analysis to the already intricate nexus between globalization and social spending. Sub-national governments within a same country may be affected by globalization in rather different ways and thus may develop quite different incentives concerning social expenditures. In a highly decentralized system, for instance, the fiscal impact of globalization may be prominent in only a few locations that are well exposed to international trade and investment while the country as a whole does not experience any significant variation in government social expenditures as a result of globalization. So far cross-national studies do not seem to have paid adequate attention to the domestic allocation of social spending responsibilities in the analysis of government incentives in the face of globalization (Ansell 2008).
During the past four decades, perhaps no other country has been more profoundly transformed by economic globalization than China does. When the Chinese Communist Party launched its ambitious program of “reform and opening up” in 1978, China’s total foreign trade was a meager 20 billion dollars, representing less than 10 per cent of China’s equally meager GDP at the time. Fast forward to 2014, and China exchanged 4.3 trillion dollars’ worth of merchandise with the rest of the world, surpassing the United States' 4 trillion dollars to become the largest trading nation in the world according to the World Trade Organization. Since then China's foreign trade figures have decreased somewhat as world trade contracted, but still remains the top merchandise exporter in the world, accounting for nearly 15 per cent of the global total in 2020 (World Trade Organization 2021:80). China's dependence on foreign trade, measured as percentage of GDP, peaked at 64 per cent in 2006. Even at the much lower level of 35 per cent in 2020, China remains more dependent on trade than such major economies as Brazil (32 per cent), Japan (31 per cent), or the United States (23 per cent) (World Bank 2022).
What is even more dramatic is the inflow of foreign direct investment into China from virtually zero in 1978 to over 100 billion US dollars annually since 2010 (National Bureau of Statistics 2021). China even surpassed the United States to become the top host economy of FDI inflows in 2014 (United Nations Conference on Trade and Development 2015) and again in 2020 as the COVID-19 pandemic shrank global FDI by 42 per cent from 2019 (United Nations Conference on Trade and Development 2021). Such huge sums of inward foreign direct investment, most of which are in the non-financial sector (National Bureau of Statistics 2022), have a tremendous impact on China's economy as well as on government coffers. One study suggests that foreign invested enterprises may have contributed over 40 per cent of China’s economic growth in 2003 and 2004 (Whalley and Xin 2010). According to a report to China's national legislature by the then Minister of Finance, in 2006 foreign invested companies accounted for 21 per cent of all tax receipts in China (Jin 2007). In August 2021, China's Minister of Commerce stated at a press conference that foreign-invested enterprises account for one tenth of China's urban employment, one sixth of China's tax revenues, and 40 per cent of China's foreign trade (Wang 2021).
The important case of China has nevertheless been largely overlooked in the existing literature on economic globalization (Tan 2020). A single country study would seem at first blush less powerful than a cross-national study, yet the former actually offers important advantages that could contribute to our understanding of the fiscal consequences of economic globalization. Some of the potentially complicating factors mentioned above are conveniently controlled for in the context of a single country (Hecock 2006). Democratic institutions, political party competition, and some powerful interest groups such as independent labor unions are conspicuously absent throughout mainland China. That suppresses political demands for government compensation in the form of income supplements and removes one of the key components of the causal story in the “compensation” hypothesis. In the absence of compensation pressures and mechanisms, the mutually neutralizing effects of compensation and disciplining that have complicated the existing empirical literature on the topic (Dreher et al. 2008) become less of a concern for this study. The theoretically relevant scenarios in China thus form an interesting competition between a “race to the top” of human capital formation and a “race to the bottom” of government efficiency and fiscal discipline.
Secondly, the abovementioned difficulty of accounting for significant cross-national variation in fiscal decentralization would manifest less prominently within the Chinese context. China’s policy implementation has been highly decentralized in many areas, including both FDI (World Bank 2010; Malesky 2008: 97) and education (Guo 2007a: 216). On the former, in most industries, provincial governments approve almost all foreign investment projects except those worth more than 300 million US dollars (Ministry of Commerce 2010). On the latter, the national government has accounted for only a small fraction of total education spending in China, and even that proportion has been decreasing in recent years, from 6.1 per cent in 2011 to 4.6 per cent in 2020 (National Bureau of Statistics 2022). Both trends of decentralization point to a study at the subnational level as a potentially fruitful avenue of research. The enormous variation both across regions within China and over time as regards both FDI inflows and education spending also seems a helpful feature from the perspective of empirical research.
China’s 31 provinces, including five autonomous regions for ethnic minorities and four directly administered municipalities, are country-sized entities. In terms of population, if Chinese provinces were independent nations, ten of them would each rank among the 25 most populous countries of the world, and all except Tibet would rank above most countries. In terms of land area, only eight provinces would rank below 100 among all the independent countries of the world (Central Intelligence Agency 2022). Each province has distinct natural endowment such as geographic location, mineral resources, etc. which determines to a large extent their comparative advantages in attracting foreign direct investment. Urban population centers on the Pacific coast of China with proximity or convenient transportation links to advanced East Asian economies have been magnets of FDI inflows. Two such provinces, Guangdong and Jiangsu, have consistently accounted for about a third of all FDI inflows to mainland China annually since the 1990s (National Bureau of Statistics 2022). While the exogenous factors of natural endowment are certainly crucial, the regional competition to attract FDI also hinges on the supply of high-quality physical infrastructure such as development zones (Zhang 2011; Zhang 2001), transportation (Fung et al. 2005; Zhang 2001), energy, utilities, and telecommunications, all of which are expensive and would require significant government budget outlays. That adds an important twist to the theoretical scenario of “race to the top”. In the original formulation mentioned above, governments increase public spending on human capital formation in the face of competitive pressure brought about by economic globalization. However, it may well be the case that governments respond to intense competition for FDI by boosting infrastructure spending at a similar or even higher rate than they do education spending, which could effectively “crowd out” human capital expenditures in relative terms.
An important empirical implication of the above discussion is that it is possible for Chinese provinces to “race to the top” and “race to the bottom” simultaneously. In the face of rising FDI inflows and increased competition pressure, provincial governments may decide to spend more money on education in response as investment in human capital. Foreign investment also contributes to local government coffers and indeed to one sixth of China's tax revenues (Wang 2021) and thus makes governments better able to spend on education. At the same time, however, provincial governments may also have to boost other budget items such as infrastructure to remain competitive in attracting foreign investment. Therefore, while the absolute level of education spending may rise with the inflow of FDI, that "race to the top" may not keep pace with rising spending on other budget items. Education may thus be "crowded out" in the overall budget as provincial governments shift budget priorities away from it, which in effect constitutes a "race to the bottom" in relative terms.
Moreover, there is a political logic behind the relative budget priorities of education versus other expenditures. From the perspective of provincial leaders, education spending does not represent a very “efficient” way to use budget funds. Political leaders are motivated by career advancement, and in an authoritarian regime the leaders’ career advancement does not require popular endorsement in regular, free, fair, and competitive elections. In the case of China, the appointment and removal of government leaders are determined by superior communist party committees in the nomenklatura system, which was borrowed from the former Soviet Union. During the post-Mao reform era, one of the most important criteria used in the evaluation of government leaders is the record of economic performance of the jurisdiction that they are in charge of (Li and Zhou 2005, Guo 2007b, Choi 2012). While education and other human capital investment certainly can help economic performance in the long term, provincial leaders are rarely concerned about the long term. The term of office for provincial governors and party secretaries is five years according to Article 106 of China's constitution and Article 26 of the constitution of the Chinese Communist Party respectively, but in practice most of them do not serve that long in their position. Among the 79 provincial party committee secretaries who left office in one way or another during 2000-2011, the average time in office was only 3.8 years and the median was four years. Given the extremely short time horizon of provincial leaders, it should come as no surprise that short-term economic stimulus such as physical infrastructure projects become more attractive areas for politicians to devote budgetary resources to than long-term investment in human capital, whose economic benefits would likely only surface many years later and thus be of little help to the political careers of the incumbents. Public spending on education also offers less lucrative opportunities for politicians' rent-seeking than investment in other areas such as infrastructure does, which further reduces government incentive to prioritize education.
We derive the following two hypotheses based on the above discussion:
Hypothesis 1: All else being equal, education spending rises as a result of increased inflow of foreign direct investment.
Hypothesis 2: All else being equal, education spending decreases relative to other budget items as a result of increased inflow of foreign direct investment.
To test the above two hypotheses on the impact of FDI on education spending, we compiled a panel data set derived from provincial-level statistics from China. In accordance with the two hypotheses above, the dependent variable, education spending, is measured in two different ways, that is, per capita amounts and proportion of total budget, respectively. It should be pointed out that there are some disagreements over how to measure education spending in relative terms. Some of the previous studies used education spending as a proportion of GDP (Ansell 2008, Baskaran and Hessami 2012), while some others measured it as a proportion of total government spending since this reflects governments' budget priorities more directly (Rudra and Haggard 2005, Nooruddin and Simmons 2009). For our purpose in this study, we concur with the latter approach, as budget share rather than GDP share aligns better with the theoretical arguments outlined in the previous section.
The panel data statistics on education spending used in this study are derived from the publicly available annual reports on education funding published every December by China's Ministry of Education and the National Bureau of Statistics, which contain official statistics on budgetary spending on education for each of China's 31 provinces. Unfortunately, the Ministry of Education and the National Bureau of Statistics changed the way they measure education spending in their 2012 report and thereafter, which means later statistics for 2012 and onwards are no longer comparable with earlier time series, and so the annual longitudinal data used in this analysis ends in 2011. The values of the in-budget spending on education are then transformed into per capita constant 1990 values using the annual provincial population and consumer price index (CPI) numbers published by China's National Bureau of Statistics.
The two measures of education spending do not show any immediately obvious distribution patterns. The overall average spending level for a province was 183 yuan per capita in constant 1990 value, and the measure varies widely both across provinces and over time. It has a standard deviation of 212 yuan, ranging from 52 yuan per capita in the central province of Hubei in 2001 to 2,079 yuan per capita in the western Tibet Autonomous Region in 2011. The proportion of education spending in provincial budget saw much less variation across provinces and over time, with a mean of 19 per cent and a standard deviation of 2.6 per cent. It ranges from 9 per cent in the western Tibet Autonomous Region in 2001 to 26 per cent in the southeast coastal province of Fujian in 2002.
The key explanatory variable in this study is the inflow of foreign direct investment to the provinces. The official statistics on the total registered capital of foreign invested enterprises in each province measured in U.S. dollars is listed in the annual editions of the China Statistical Yearbook. This was then converted into proportions of provincial GDP using annual data on year-end exchange rates between US dollar and Chinese RMB yuan and annual data on regional GDP. The average value of this FDI/GDP ratio is 55 per cent and the standard deviation is 77 per cent. This variable ranges from 5 per cent in the western Xinjiang Autonomous Region in 2011 to 594 per cent in the southern island province of Hainan in 2007. Since annual FDI inflow is unlikely to affect public spending in the same year and its effect should take some time to manifest, this variable is lagged by one year. Using time-lagged values of this variable could also help alleviate a potential and legitimate concern that reverse causality in the model could lead to biased estimation results in the data analysis. While foreign direct investment could change local governments' spending behavior and priorities, government spending on education could also affect the incentive and decisions of foreign investors (Hecock and Jepsen 2013). That theoretical scenario can not be ruled out, but at least by using time-lagged values of the independent variable we have more confidence that the current year's education spending (the dependent variable) is unlikely to cause changes in the previous year's foreign direct investment (the independent variable).
We also controlled for other important factors that previous studies on the effects of economic globalization on education spending have customarily included, such as economic development level, illiteracy rate, and youth population (e.g. Ansell 2008, Baskaran and Hessami 2012, Nooruddin and Simmons 2009). The first is measured by provincial GDP obtained from the annual editions of the China Statistical Yearbook, again divided by both provincial population and discount factor calculated from annual consumer price index figures for each province. The resulting variable has a mean of 5,890 yuan and a standard deviation of 5,190 yuan. Illiteracy rates for each province are also listed in the annual editions of the China Statistical Yearbook. Finally, proportion of the population under the age of 15 for each province was calculated from the results of the annual population sample survey reported in China Statistical Yearbooks. All three control variables are expected to be positively correlated with government spending on education. Higher levels of GDP per capita enables provincial governments to spend more on education, and higher levels of illiteracy and youth population indicate more needs for education funding. The descriptive statistics for the dependent and explanatory variables are listed in Table 1 below.
| unit | mean | standard deviation | |
| Dependent variables | |||
| In-budget spending per capita on education in constant 1990 value | yuan | 183 | 212 |
| % of total government budget spent on education | % | 19 | 2.6 |
| Explanatory variables | |||
| FDI stock/GDP | % | 55 | 77 |
| GDP per capita in constant 1990 value | yuan | 5890 | 5190 |
| Illiteracy rate | % | 12 | 9 |
| % of population under 15 | % | 23 | 8 |
Since both dependent variables, namely education spending per capita and education's share in total expenditures, are derived from government budgetary figures, the best predictors for them are the values of the respective variables for the previous year. This is based on the assumption that governments are unlikely to make new allocation decisions on education funding every year from the ground up. Instead, their best reference point is the level and budget share of education spending in the previous year. Previous studies on the effect of economic globalization on education spending have commonly included the lagged dependent variable to model the "sticky" and dynamic nature of education spending (e.g. Ansell 2008, Baskaran and Hessami 2012, Nooruddin and Simmons 2009). From the perspective of Chinese provincial leaders when they make annual budget decisions, the most readily available reference is the budget of the previous year. The concept of “zero-based budgeting” has often been mentioned in Chinese official documents but rarely used in practice. Therefore, we model the impact of FDI on education spending using a dynamic panel-data framework that regresses education spending for year t on education spending for year t-1 and the main independent of interest, FDI to GDP ratio for year t-1. We also added a series of control variables in the model, including GDP per capita, illiteracy rate, and youth population, which were discussed in the previous section of this paper. To capture the possible impact of national funding cycles in China and of exogenous shocks on the international stage such as the 2007-2009 Global Financial Crisis, a full set of dummy variables for the 10 years from 2002 through 2011 are included.
The two linear regression models to be estimated are specified as dynamic panel data models with the lagged dependent variable on the right-hand side of the regression equations. The two regression equations are identical except for the dependent variables:
(Education spending per capita)i,t = α(Education spending per capita)i,t-1 + β1(FDI stock/GDP)i,t-1 + β2(GDP per capita)i,t + β3(illiteracy rate)i,t + β4(% population below age 15)i,t + β5(year dummy variables) + β0 + εi,t
(Education's share in total spending)i,t = α(Education's share in total spending)i,t-1 + β1(FDI stock/GDP)i,t-1 + β2(GDP per capita)i,t + β3(illiteracy rate)i,t + β4(% population below age 15)i,t + β5(year dummy variables) + β0 + εi,t
In both regression equations, the main coefficient of interest is β1 on foreign direct investment. A negative and statistically significant β1 estimate would lend support to the theoretical scenario of race to the bottom, while a positive and statistically significant β1 estimate would conform to the scenario of race to the top. The statistical estimation, however, is complicated by the fact that both regression equations include the lagged dependent variables as explanatory variables, which by construction are correlated with the unobserved panel-level effects. Therefore, ordinary regression would produce inconsistent estimators. Instead, we will obtain the Arellano-Bond estimators that use a generalized method of moments (GMM) framework (StataCorp 2021). The results are shown in Table 2 below.
| Dependent variable | In-budget spending per capita on education in constant 1990 value | % of total government budget spent on education | ||
| Explanatory variables | Coefficient estimate | Robust standard error | Coefficient estimate | Robust standard error |
| Lagged dependent variable | 1.029*** | 0.013 | 0.430*** | 0.061 |
| Lagged FDI stock/GDP | 0.050** | 0.020 | -0.003** | 0.001 |
| GDP per capita in constant 1990 value | 0.001 | 0.002 | 0.000 | 0.000 |
| Illiteracy rate | -0.972 | 0.620 | -0.004 | 0.025 |
| % of population under age 15 | 4.765 | 3.097 | 0.013 | 0.056 |
| Year 2001 | Base | |||
| Year 2002 | 5.686* | 3.350 | 0.588** | 0.251 |
| Year 2003 | -8.582** | 4.359 | 0.227 | 0.249 |
| Year 2004 | -8.172 | 7.935 | 0.200 | 0.351 |
| Year 2005 | 2.897 | 6.550 | -0.180 | 0.271 |
| Year 2006 | -11.782 | 19.445 | -0.119 | 0.357 |
| Year 2007 | 21.261** | 10.556 | 1.193** | 0.470 |
| Year 2008 | -2.561 | 19.158 | 0.353 | 0.402 |
| Year 2009 | 2.523 | 19.661 | -0.502 | 0.496 |
| Year 2010 | -18.077 | 26.690 | -0.442 | 0.600 |
| Year 2011 | 28.805 | 22.690 | 0.001 | 0.720 |
| Constant | 119.070* | 70.568 | 10.439*** | 1.877 |
| Number of observations | 341 | 341 | ||
| Number of provinces | 31 | 31 | ||
| Arellano-Bond Test for Autocorrelation in First-Differenced Errors H0: no autocorrelation | ||||
| Order | z | Prob > z | z | Prob > z |
| 1 | -1.2184 | 0.2231 | -3.9518*** | 0.0001 |
| 2 | 1.0988 | 0.2719 | -0.38569 | 0.6997 |
Unsurprisingly the effect of the lagged dependent variables did turn out to be statistically significant, suggesting a certain degree of persistence in Chinese provinces' education funding from year to year. However, the most important and interesting finding is that FDI inflows increases the absolute level but decreases the relative proportion of education spending even when other economic and demographic factors are controlled for, thus providing support for both theoretical scenarios of race to the top and to the bottom. On one hand, as FDI/GDP ratio rises by one standard deviation, government spending on education would rise by 3.85 yuan per capita (in constant 1990 value) in the following year, all else being equal. The magnitude of that effect may appear small at first glance but would actually represent a boost of more than 2 per cent in education spending for an average province during the time period under study. The effect is statistically significant at the 5 per cent level. This result confirms the theoretical scenario of the “race to the top” in which Chinese provincial governments respond to rising FDI inflow with increased investment in human capital formation.
On the other hand, however, the results also suggest that FDI inflow actually decreases the relative weight of education funding in the overall budget. This lends support for the “race to the bottom” scenario in which education spending is seen as an inefficient way of using public funds and, in response to the competitive pressure brought about by rising inflows of foreign direct investment, governments divert budgetary priorities from education to other more productive areas of expenditures. Again, the regression coefficient estimate on foreign direct investment shows an effect that may seem small in magnitude but statistically significant at the 5 per cent level. As the FDI/GDP ratio rises by one standard deviation, the proportion of total government budget spent on education is expected to decrease by a quarter of a percentage point. To put that in perspective, the standard deviation of the latter variable is only 2.6 percentage points, and so the estimated effect on education's share in total budget may not be as minuscule as it appears at first glance.
None of the coefficient estimates for the three control variables attained statistical significance even at the 10 per cent level in either of the regression models. It may seem somewhat counterintuitive that budgetary spending on education in Chinese provinces correlate with neither the capacity nor the needs for education funding. However, since provincial leaders in China are effectively shielded from upward political pressure and only accountable to superior party committees, their budgetary decision making is probably unresponsive to either funding capacity or funding needs.
Finally, the set of dummy variables for each year from 2002 through 2011 reveals an intriguing temporal pattern of five-year cycles of funding surges in 2002 and 2007. Not coincidentally both years were the time of turnover of party and government leaders in China. In 2002 twelve provincial party secretaries were replaced, and in 2007 more than half of the provincial party secretaries were replaced. The number of turnovers for other years was no more than 8. The exact reason for rising education funding during times of leadership turnover still needs further research. It should also be pointed out that although we have controlled for variables such as economic development level, illiteracy rate, and youth population that previous empirical studies on globalization's effect on education funding usually include, our regression models may still have omitted some important determinants of education spending, which causes bias in the estimates of regression coefficients, and so the findings need to be interpreted with some caution.
The Arellano-Bond test for first- and second-order autocorrelation in the first-differenced errors was conducted on the dynamic panel data regression models to see whether the moment conditions used by the method is valid. As shown in Table 2 the test results present no significant evidence of serial correlation in the first-differenced errors at order 2 and of model misspecification.
The question of whether economic globalization promotes or depresses governments’ social spending has attracted social scientists’ attention for decades. This study attempts to adapt the debate to the subnational context of China, and argues that, while the “compensation” hypothesis is no longer a viable theoretical contender due to the authoritarian nature of the Chinese regime, the “race to the top” and “race to the bottom” scenarios still present testable hypotheses on the relationship between the inflow of foreign direct investment and government funding for education. Foreign investment could certainly contribute to government coffers and also incentivize governments to "race to the top" of human capital formation by boosting education funding. However, the decentralized authoritarian setting constitutes a fertile ground for competitive pressure on subnational governments and for a "race to the bottom" of fiscal efficiency with diminished budgetary priority for public education. Therefore, as the statistical analysis shows, Chinese provinces are engaged in a race to the top and the bottom at the same time, with higher levels but lower shares of education spending in the face of rising inflows of foreign direct investment. The theoretical arguments and empirical analysis in this study could also suggest a cautionary tale beyond the Chinese context. That is, in the absence of effective popular participation and democratic institutions but with intense competition between subnational units, government spending priority on education would suffer losses under economic globalization that the small boost in absolute level of spending is not sufficient to compensate.