A New Perspective on Regional Tax Burden Differences in China

1 Introduction

Tax burden has always been one of the most acute and sensitive issues in people’s livelihood. Research in taxation from almost all schools of economics will inevitably involve tax burden issue. The moderation of tax burden in the economy, and the fairness of tax distribution in regions not only determine the efficiency of resource allocation, but also affect the sustainability of economic growth, as well as the robustness and coordination of social functioning.

Since the tax reform in 1994, the total tax revenue of China has shown a trend of rapid growth on the basis of rapid economic development. However, from perspective of regional equilibrium, the regional distributions of China’s tax burden have showed great differences. In real terms, the maximum difference in regional GDP (ratio of the highest and lowest areas) is more than 87 times every year, and the highest one among all years is 101 times. However, the maximum difference in regional tax revenues is 169 times on average, with a maximum difference of 281 times in 2007.

Although the regional imbalance is a common phenomenon in the process of economic development in both developed and developing countries, the regional difference of tax burden in China is extremely prominent compared with other countries. Besides, the regional difference of tax burden is not only reflected in the difference determined by the economic performance, but also in the divergences between economic performance and tax revenue. Specifically, China’s regional tax burden and economic level showed a negative correlation. It presents a unique phenomenon that the higher a region’s economic level, the lower is its tax burden.

Table 1 shows the sorted result of regional economic share and tax burden by size order and geographical distribution since 1995. From the order of economic share, eastern region ranks first, followed by middle region, and western region ranks at the bottom. From the order of tax burden, the sort of the three parts is more uniform. For some regions, there even shows a reverse order. For example, the economic shares in Shandong, Hebei, Henan and Hubei enter into the national top ten, but their tax burden rank at the bottom ten in the country. The economic shares of Yunnan, Guizhou, Xinjiang, Tianjin and Hainan rank at the bottom ten of the country, but their tax burden rank into the national top ten. For the other regions, they also showed this deviation phenomenon in some degree. Although there are reasons like special taxes (e.g, resource tax) and the centralized collection of tax in headquarters economy, it is only limited in few areas, and can not been fully explained in most regions.

Revenue comes from economy, that means tax revenue and economic growth are positively correlated. Regardless of the growth speed of regional economy, the tax burden should approximate a stable value in the same tax system. In other words, in the absence of regional tax legislative authority, there is no difference in regional tax burden, and regional tax share should be consistent with their GDP share. Obviously, the actual situation of China’s regional tax burden does not comply with this law, which means there is a deviation mechanism between regional economic growth and tax revenue. So, what factors influence the abnormal distribution of regional tax burden in China is a subject worthy of further explore.

2 Literature Review

On the formation mechanism of regional difference in tax burden, different scholars have different interpretations. Starting from relations between tax and economic, researchers find that economic growth is in an inverted U-shaped relationship with tax revenue, so different levels of economic level corresponds to different tax burden^[1-4] Tanzi^[5] used social and economic factors as explanatory variables for regression analysis, and holds that the urbanization and government expenditure increase lead to the expansion in government debt, resulting in high tax burden; Ansari^[6] considered that high population density will increase the tax exemption and thereby reduce tax burden, because overpopulation is a symbol of underdevelopment; Stotsky and Woldemariam^[7] found that the economic structure has a significant impact on the whole tax burden, raising the proportion of agriculture will reduce tax burden, while raising the proportion of manufacturing sector will increase tax burden. Teera and Hudson^[8] found that per capita income, net exports, population density, scale of foreign debt and tax administration have an important influence on tax burden through empirical analysis. Garcia and Von Haldenwang^[9] used panel data of 131 countries in 1990–2008 and find that in addition to per capita income, demographic factors and economic structure, the type of political system also plays a role in tax burden. However, the findings of these studies usually support the conclusion that the more developed the economy is, the heavier its tax burden is. Besides, these studies mainly use the comparisons between countries, tax disparities within a country under the same tax system are few involved.

In view of the circumstances of China, domestic scholars Pan and Huang^[10] studied the issue of regional differences in tax burden in earlier time. They calculated China’s regional tax burden by Theil Index, and attributed the differences to the ownership structure, extra budgetary revenue, income distribution, price indices and tax incentives. Subsequently, Wu^[11] broke down the Gini coefficient of total tax burden, and he believes that the ownership structure, industrial structure, income distribution, price indices, fiscal system and tax incentives are the main factors that cause the regional differences in tax burden. Zhang, Liu^[12] broke down the tax burden differences by regions, and consider the case of Beijing, Tianjin and Shanghai alone. They hold that tax burden has positively correlated with economic development, proportion of secondary industry, price indices and tax incentives, and has negatively correlated with extra-budgetary revenue. Zhang, Mo^[13] found that the tax burden has negatively correlated with proportion of primary industry, and that economic development and tax system are the determinants of tax burden. Lü and Guo^[14] believed that in addition to the factors as economic growth, rising prices, industrial structure change, progressive taxation and taxation structure, the incentive function of tax decentralization and the focus releasing of revenue growth dividend are the main reasons that lead to the rapid growth of practical tax burden. Huang^[15] broke down Gini coefficient by tax categories, and found that indirect tax make a great contribution to the regional difference of tax burden, while the contribution of direct tax is increasingly important.

In the empirical analysis, Lin^[16] took empirical analysis with panel data in 1994–2006 and found that tax burden has a significant positive correlation with industrial structure, ownership structure and tax collection. Wang and Liu^[17] believed that the proportion of non-taxable GDP affects regional tax burden based on the framework of national accounting, they using input-output analysis found that industry structure has an important influence on tax burden, finance and insurance sector has the highest tax burden, while agriculture sector has the lowest tax burden. Huang, Hu and Liu^[18] found that per capita GDP, population density and total trade have significantly positive impacts on regional tax burden, and agriculture sector has the lowest tax capacity, while tax burden in industrial sector and service sector are relatively heavier. From the literature above, we find that most domestic studies focus on the comparison of tax burden difference (such as using statistical index to break total difference into within-group difference and between-group difference), as well as the qualitative analysis on its formation mechanism.

Furthermore, even if there is no conclusive support from quantitative test, the current studies also agree that the high tax burden comes from high economic growth, non-agricultural industries and other social economic factors. However, this does not fully explain the regional differences in tax burden showing in Table 1, especially the serious deviation between economic development and tax burden like Shandong, etc. In current studies, there is only one paper refers to the reverse distribution in regional tax burden^[15], but it does not give any qualitative or quantitative interpretation. This requires us to find a new outlet. One of the ideas is to retrospect the accuracy of statistical data, especially regional data of GDP. In general, tax data comes from micro-economic entities, and it will be deducted from the economic benefits of enterprises or residents. The possibility of exaggerating tax data is not very high when taxpayer’s own economic benefits are considered, so the accuracy of tax data is relatively reliable. In contrast, under the current promotion model of local officials in China, which is termed as “promotion tournament model”, economic indicators are often manipulated at each administration level^[19], thus the authenticity of GDP has always been questioned^[20-25]

In the long-time query of the authenticity of China’s GDP data, the most intuitive and prominent one is the issue that regional data of GDP significantly differ from the national data. Since China established GDP accounting system formally, there has always been a gap between national data and regional data on GDP. We sum up regional GDP and compare it with national GDP in National Statistical Yearbook over years, find that the gap between them has been more than 1 trillion yuan since 2001, then the gap continued expansion. The gap then reached 6.12 trillion yuan by 2013, accounting for 10.75% of GDP. This gap weighted highest in 2004, which accounts for 19.26% of GDP, far exceeding the normal statistical error. Although, part of this gap is caused by survey system, in general it can be considered that there is an overestimate tendency in regional GDP figures. In this paper we argue that it is the unbalanced overestimating of GDP in regions resulting in the abnormal distribution of tax burden.

Following the literature review above, we will have a theoretical analysis on the major factors and its mechanism affecting regional differences in tax burden. Then we will do the empirical analysis and robustness test on various hypotheses using spatial panel model. Lastly, we will give appropriate suggestions based on the results. Our study discussed the divergence of tax burden and economic development in view of the accuracy of GDP for the first time, which may enrich the empirical study on influence factors of regional tax burden.

3 Research Design

In order to investigate the extent of deviation and influence factors of tax burden, the previous expression of tax burden that use regional tax revenue as the numerator and regional GDP as the denominator, is unable to measure whether the economy is overvalued or not. Here we use the deviation between tax share and economic share as the dependent variable, and its specific calculation equation is as follows:

Tax_it represents the tax revenue of region i in year t, Tax_t represents total tax revenue of all the regions in year t, so the first term in right side of Equation 1 represents the tax share of region i in year t. Similarly, GDP_it represents the size of GDP of region i in year t, GDP_t represents total GDP of all the regions in year t, the second term in right side of Equation 1 represents GDP share of region i in year t. Using regional tax burden deviation as dependent variable, by share calculation, we can present regions whose tax burden are significantly different from the average level. If the regional tax deviation is zero, it indicates that the tax share of this region is in line with its GDP share, and its development is relatively compatible in both taxation and economy. If the deviation is positive, it indicates that the regional tax burden is higher than the average level of the country, and GDP figure in this region may be underestimated relatively. If the deviation is negative, it indicates the regional tax burden is below the average level, and its GDP figure may be overestimated relatively.

4 Empirical Analysis

5 Conclusions and Implications

Tax burden is related to the sustainable development of economic as well as the Equilibrium and stability of social interests, therefore the issues of regional differences in tax burden have been always concerned by researchers. By computing the tax share deviation of all regions in China, the phenomenon of regional GDP overestimate is found an important cause of the regional differences in tax burden, applying the space panel data model to analyze the determinant of area tax share deviation from 1995–2013, this study found that:

First, long-term loans, electricity and rail freight, these three variables that characterize real economic development shows that the aggregate figures of GDP in China is overestimated. In the GDP-centered governance structure of China’s local governments, which is termed as “promotion tournament model”, to get praise from the performance evaluation, local government officials often deliberately cater to the GDP measurement that deviate from the original purpose of the evaluation. We can now consider taking the real economic development indicators into the official assessment system to avoid the inaccuracy of regional GDP figures. However, in humans’ economic and social life, the interaction between measure behavior and measure objects is inevitable. So in the long run, we also need to introduce a variety of assessment methods and index, checks and balances the measure behavior, then to form a healthy competition ultimately.

Secondly, per capita income and R&D expenditure are positively related with the original tax burden. In the areas of higher per capita income and more adequate R&D expenditure, enterprises have strong profitability, and the whole area has a good economic performance, therefore the motivation of GDP figures fraud is relatively weak, per capita income and R&D expenditure are real factors for the differences in regional tax burden. Moreover, this difference in tax burden is an important manifestation of taxation function in income distribution adjustment. The crucial policy orientation is not to reduce the tax burden, but rather to build a scientific and rational system of local fiscal expenditure, so that the fiscal expenditure will be more inclined to the people’s livelihood, especially improving the living standards of low-income groups who live in these areas, improving social security, regulate the underlying people’s feeling to pay taxes.

Besides, the regional spatial effect presents negative coefficients, which can be explained from two aspects. On the one hand, the negative deviation is brought by the share shift of region economy. From the formula of Yit, there is a tradeoff between regions. On the other hand, due to the difference in production elements and institutional factors, different regions have become independent economic space units, which means there are negative dynamic competitive relationship with each other. For example, although there is no independent taxation legislative authority for the local governments, but they may attract production elements from the surrounding area to promote their own economic performance through tax credits, government subsidies and other economic tools, while the surrounding areas may be impaired due to the loss of production elements. Thus, by building a scientific and rational revenue distribution system, encouraging interregional cooperation, and establishing a unified market, we can promote the balanced development among regions.

Finally, investment in fixed assets, proportion of secondary industry and proportion of state-owned economic have a positive impact on regional GDP overestimated. Previous studies have generally considered that these three factors play an important role in pushing up the local tax burden, but it is under the assumption that local economic has not been overestimated. Negative regression coefficients in empirical analysis confirmed the existence of GDP overestimated phenomenon, and their negative impact on regional tax burden have surpass their positive effects under the theoretical analysis. The extensive economic growth pattern over depend on investment is deficient not only in revenue contribution, but also in enterprises’ sustainable development. The rapid development of local economy needs layout at a higher level on industry structure and industrial patterns.