Fiscal decentralization, equalization, and intra-provincial inequality in China

The recent work argues that the negative effect of fiscal decentralization on regional inequality is only significant for high-income countries, while in poor countries, fiscal decentralization may well increase regional inequality.

Recent studies by Uchimura and Jütting (2009), Brehm (2013), and Wu and Wang (2013) are notable exceptions in this regard.

These studies, based on data for OECD countries or European Union countries, tend to find a negative relationship between fiscal decentralization and regional inequality. In advanced democracies, the detrimental effect of fiscal decentralization on regional equity may be reduced by various institutional factors, including, of course, equalization transfer programs.

There may be circumstances under which fiscal decentralization has the opposite effect on regional inequality. For instance, as pointed out by Rodríguez-Pose and Ezcurra (2010), decentralization is likely to enhance fiscal transparency and place central and local governments under pressure to equalize public goods and services around the country, which eventually may contribute to lower levels of regional inequality. Nevertheless, this argument appears to be weak in the Chinese context where the main focus of the central government has been to foster strong fiscal and political incentives for local governments to promote local economic development as opposed to any emphasis on equalizing public goods and services delivery.

The 25-member Politburo is one of the most powerful decision-making bodies in China.

Drawing on a political economy model, Kotsogiannis and Schwager (2008) also point out that equalization programs may improve local accountability through enabling voters to compare public good supplies between regions thus keeping politicians more accountable. As fiscal equalization efforts usually aim to ensure the same standards for public services across regions, politicians will have to be more self-restrained and be less rent-seeking; otherwise, poor public goods provision may make rational voters punish the incumbents.

See the next subsection for detailed definitions of the intra-provincial inequality, fiscal decentralization indicator, and equalization effort index.

The Kuznets curve hypothesis suggests that inequality first increases with economic development and then decreases with its further development.

Data on human capital are derived from Chen et al. (2004) for the year before 2001, and they are augmented by the authors for the years after 2001.

Similar to other relevant studies that have used Chinese county data (e.g., Zhang 2006; Uchimura and Jütting 2009), our basic unit for calculating intra-provincial inequality is county j, which does not include the city district (“qu”), as the latter differs significantly from the former in many aspects including administrative, economic, and fiscal aspects. In this regard, it has been highlighted in the literature that when calculating regional inequality, it is necessary to use a territorial classification that creates relatively homogenous regions (Lessmann 2009).

In contrast, local governments in China have significant levels of autonomy and discretion on the expenditure side of the budget.

Uchimura and Jütting (2009) and Wu and Wang (2013) are notable exceptions in that they also measure fiscal decentralization at the sub-provincial level in China.

Separate transfers from prefectural governments should be negligible, if any. Under the current institutional setup in China, revenue assignments at the sub-provincial level are exclusively at the direction of the provincial governments, and the prefectural-county fiscal relations are in general explicitly defined by the provincial governments. Although, in some cases, prefectural governments do have some discretion on the assignment of revenues between them and county governments, prefectural governments offer little help to county governments financially (World Bank 2002). First, sub-provincial level governments have been assigned with very heavy expenditure responsibilities. Prefectural governments usually cannot help county governments out; instead, both levels of government rely heavily on transfers from the central government. Second, no institutional rules obligate prefectural governments to transfer their own revenues to lower-level governments.

Total revenue=own source revenue+equalization transfers.

The tax rebate was a compromised outcome to smooth the implementation of the Tax-Sharing reform in 1994. Specifically, the tax rebate was introduced to guarantee the vested interests of richer provincial governments prior to the 1994 reform. Its essence was to return to the provinces the amounts of VAT, consumption taxes, and income taxes that otherwise would have gone to these provinces under the old system. By design the tax rebates did not perform any equalization role. The ad hoc transfers typically involve the central government response to high-priority emergencies or are generally associated with particular programmatic objectives; this type of transfer is usually endowed with strong bargaining features. See Zhang and Martinez-Vazquez (2003) for a detailed discussion on the fiscal transfer system in China and Huang and Chen (2012) and Liu et al. (2014) for analyses of its equalization effects.

For example, Shanghai has only one county and Beijing has only two by the year of 2009.

The tax-sharing system reform was implemented in 1994.

Under the current Chinese system, tax legislation is highly centralized at the central level and there is virtually no autonomy for local governments to determinate either tax bases or tax rates, leading to little formal, institutionalized variations of revenue decentralization across provinces. On the contrary, expenditure assignment has never been clearly defined by the central government and it was also largely ignored by the central government in the last three decades of intergovernmental fiscal relations reform in China, resulting in large direction for provincial governments to set up their own arrangements within the provinces. Thus, it appears that measuring fiscal decentralization on the expenditure side is more meaningful in the Chinese context. Additionally, while the current tax-sharing system (which was implemented in 1994) re-centralized revenue assignments, there has been considerable devolution of expenditure responsibilities to local governments. More importantly, the vertical imbalances caused by the opposite trends in revenue centralization and expenditure decentralization have not been adequately offset by intergovernmental transfers (see for example, Zhang and Martinez-Vazquez 2003; Liu et al. 2015). Our calculation of the simple correlation coefficient (based on our sample data) also indicates a modest correlation between revenue decentralization and expenditure decentralization (i.e., 0.65).

In specifications with interaction terms, the statistical significance of an individual variable does not necessarily imply that the marginal effect of it is statistically significant (see Brambor et al. 2006).

Fisman and Gatti (2002) and Iimi (2005) are the two exceptions in the literature.

That is, the weight is a product of the inverse of the distance between provinces and the population size of the provinces. By this construction, we assume that geographically closer provinces interact more strongly with each other, and provinces with larger population exert more influence on the settings of the neighboring provinces’ policies.

Data on minority population of the provinces were obtained from various issues of the “Major Figures on Population Census of China.”

As mentioned previously, under the current hierarchical system of China, although the central government directly determines the amount of transfers to local governments, practically, all central government transfers have to go through the provincial governments before reaching the various layers of sub-provincial governments (prefectures, counties and townships); thus, provincial governments enjoy a substantial degree of discretion in designing their own equalization policies to meet their diversified policy objectives.

For space reason, we only report the first-stage estimation results for specifications controlling for the full list of explanatory variables in the Appendix Table 9.

Shea’s partial R-squared can pinpoint the possible presence of weak instruments if the partial R-squared is sufficiently small.

A different pattern is observed in developed economies (Akai and Sakata 2009; Lessmann 2012)