Price effect and income distribution effect of replacing business tax with VAT

NI Hongfu1 GONG Liutang2 WANG Ximeng2

(1.National Academy of Economic Strategy, CASS, Beijing , China 100045)
(2.Guanghua School of Management, Peking University, Beijing , China 100871)

【Abstract】The reform of replacing business tax with VAT is an important change of China’s tax system, which has great impact on tax burden, welfare and income distribution of Chinese households. This paper embedded the VAT deduction mechanism into the input-output price model, and used the CFPS data and China input-output table of 2012 to analyze the price effect and income distribution effect of the reform of replacing business tax with VAT. The results show that in the context of maintaining the existing tax collection and administration ability of China, the reform of replacing business tax with VAT may generally make the price decline and reduce the tax burden on enterprises. If the tax collection and administration ability has increased to 100%, there are some labor-intensive sectors whose price and cost could rise highly, such as transportation, postal service and education. After the reform of replacing business tax with VAT, the tax burden and its rate of households in cities and towns both decline. In the baseline scenario that the current tax collection and administration ability remains, the reform of replacing business tax with VAT has slightly improved the status of income distribution from the perspective of MT index. However, once the tax collection and administration ability has been greatly improved, due to the strong regressive nature of VAT, it is possible to increase the tax burden of residents and worsen the situation of income distribution.

【Keywords】 tax burden; input-output model; value-added tax ; income distribution;


【Funds】 Major Project of the National Social Science Foundation (14ZDA084) Youth Project of the National Natural Science Foundation (71401009) Youth Project of the Natural Science Foundation of Beijing (9154028)

Download this article

(Translated by ZHANG Xiaoyan)


    [1]. ① VAT is mainly levied on manufacturing industry, while business tax is mainly levied on service industry. Generally speaking, wealthy households have a relatively higher proportion of service products consumption. For example, the lodging and catering industry belongs to the service industry which is imposed with business tax. The wealthy households are more likely to go out eating and travelling, so they bear the business tax of lodging and catering industry indirectly. Thus, intuitively, VAT is regressive and business tax is progressive. [^Back]

    [2]. ② CFPS aims to provide a data base for academic research and public policy analysis by tracking and collecting the data at three levels of individual, household and community, and reflecting changes in China’s society, economy, population, education and health. CFPS is conducted by the Institute of Social Science Survey (ISSS) of Peking University. For more information, please refer to the website: [^Back]

    [3]. ③ VAT is a type of tax which is levied on the added value realized by the units and individuals that sell goods, provide processing and repair service and import goods. Actually, it is difficult to accurately calculate the newly added value or additional value of a commodity in the manufacturing and circulation process. Therefore, China also adopts the tax deduction method (VAT deduction mechanism) which is generally utilized in international society, i.e. according to the sales of commodity or labor service, calculate the sales tax at the specified tax rate; then deduct the paid VAT when obtaining that commodity or service, which is input tax; and the balance is the tax that the added value part should pay. This calculation method reflects the principle of calculating tax based on the added value element. [^Back]

    [4]. ① To measure the household welfare effect caused by tax, the theory basis generally relies on the duality theory. The household expenditure function is y= e (p, u), which indicates the minimum expenditure to achieve the effect u at the provided price p. The price will change when imposing tax on economic system. To maintain the previous effect level, the currency that shall be compensated for households is. Obviously, if the expenditure function can be estimated accurately and the impact of tax on the price can be acknowledged, CV can be measured and the welfare change (actual change of purchasing power, i.e. the welfare change measured by currency) caused by tax can also be measured. But as limited by data and method, it is very difficult to accurately estimate the expenditure function (the effect function), i.e. the data requirement of measuring the welfare effect of indirect tax is high. Therefore, according to the Shepard lemma, CV can be obtained by the first-order approximation, i.e. conducting Taylor expansion of the above equation and we can get,In the equation, the first item, which is the first-order approximation of CV, represents the change of resident household’s expenditure caused by the price change which is resulted from tax when resident household does not change its consumption quantity. Furthermore, assuming that the price change margin generated by tax is exactly equal to the government’s tax rate, i.e. the tax completely shifts to resident’s final consumption products (the papers studying the tax burden generally assume that tax completely shifts), the resident’s tax burden is the first-order approximation of resident’s welfare change. [^Back]

    [5]. ② The CGE model can also be used to analyze the impact of income distribution. Generally, the impact of tax system change on the welfare of each group is analyzed after dividing households into some even hundreds of representative ones according the income level. The CGE model determines the impact of the income distribution mainly according to the welfare change of representative households at different income levels, which cannot further analyze the income distribution effect of income groups at a much more micro-level. [^Back]

    [6]. ① In China’s tax system, there are several kinds of indirect tax, such as consumption tax, resource tax, VAT, business tax and city maintenance and construction tax etc. As this article focuses on the price impact of replacing business tax with VAT, only VAT and business tax are taken into consideration in the theoretical model and the empirical research application model. The consumption tax, resource tax and city maintenance and construction tax etc. are classified as other indirect tax. [^Back]

    [7]. ② If assuming that there is only one kind of input element (labor), viis the element input of producing one unit Industry i (such as the number of laborers). In the model system of this article, the added value element price (wage rate) of 1 is taken as the valuation standard, thus, vai is the value produced by the (labor) element. [^Back]

    [8]. ① There are two obvious disadvantages: first, the impact of tax collection and administration ability and tax preference during tax collection and administration are not taken into consideration, which results in the actual tax rate borne by resident is lower than the statutory tax rate. Second, the impact of the intermediate input-output structure on the shift of indirect tax is also not taken into consideration. Assuming that business tax is levied on the communication and transportation industry but not on the food industry, according to the simple statutory tax rate method, only the household’s consumer spending part of communication and transportation industry has business tax burden. But actually, the manufacturing of food needs the intermediate input of communication and transportation industry, thus the business tax of communication and transportation industry is hidden in the food products. While according to the statutory tax rate method, household’s food spending does not have business tax burden, which is obviously unreasonable. So if not considering the intermediate input structure, the statutory tax rate method is very likely to underestimate the indirect tax burden. [^Back]

    [9]. ② When studying the indirect tax burden, it is generally assumed that the indirect tax completely shifts forwards to consumers. Although the tax shifts may be different because of different market structure, product manufacturing process and government regulation environment, it is still the common method to assume that the tax completely shifts forwards. [^Back]

    [10]. ① If the VAT rate of the intermediate input purchased by an enterprise (input VAT rate) is same with the VAT rate of selling products (output VAT rate) and the input tax can be totally deducted. Thus, in theory, the imposed form of the deduction mechanism of corporate product is equivalent with the collection method of corporate added value. [^Back]

    [11]. ② This method is widely used in the calibration of model parameter in the CGE model. Generally, the price is assumed as 1. Essentially, this method changes the measuring unit. For example, assuming it costs two yuan to buy one piece of bread, if the physical measuring unit is one piece of bread, the unit price of bread is two yuan; if the physical measuring unit is 1/2 piece of bread, the unit price of bread is one yuan. It should be noted that when calibrating the model parameter, assuming the price is standardized as 1, the economic system has various kinds of indirect tax. If in turn, these calibration parameter is utilized to calculate the price, the price should be 1. But, after replacing business tax with VAT, i.e. replacing the actual business tax rate calibrated before with VAT rate, and the model structure has changed, the price calculated through Equation (6) may not be 1. [^Back]

    [12]. ③ As limited by the length of this article, the corresponding relation of specific industries can be obtained from the author. [^Back]

    [13]. China Tax Statistical Yearbook 2013 records the taxation in 2012. [^Back]

    [14]. ① This method is commonly used in designing the social accounting matrix. [^Back]

    [15]. ② As these minor tax accounts for a very small ratio, which does not surpass 5%, this method will not affect the research conclusions of this article. [^Back]

    [16]. ③ Thanks to the reviewers for their valuable comments. Actually, there is another method to solve the actual VAT (output) rate (ГVA): through standardizing all the price as 1, in the context of determining A, V, ГB and ГO, the actual VAT (output) rate (ГVA) is determined by inversely solving Equation (6). Actually, these model parameters A, V, ГB and ГO are all solved step by step according to the construction idea of existed original data and model. While in this article, ГVA is obtained by directly using the original data, according to that output VAT minus input VAT is tax, to solve the equation. Therefore, these two methods are consistent essentially, and the solved ГVA is the same. [^Back]

    [17]. ① Generally, the eight categories are food, clothing, living, household facility and service, health care, transportation and communication, education, culture and entertainment service and other commodities and services. [^Back]

    [18]. China Yearbook of Household Survey (2013) provides urban and rural households’ per capital income and consumer spending data of various income groups. According to income, urban households can be divided into lowest income households (10%, including financially difficult households of 5%), lower income households (10%), lower middle income households (20%), middle income households (20%), higher middle income households (20%), higher income households (10%) and highest income households (10%). According to income, rural households can be divided into low income households (20%), lower middle income households (20%), middle income households (20%), higher middle income households (20%) and high income households (20%). [^Back]

    [19]. ① Concentration curves are extremely similar with the Lorenz curve, and the Suit index is similar with the Gini coefficient. After sorting all the households according to the rich (income) level from poor (small) to rich (big), the abscissa represents the ratio of the accumulated income in total income starting from the poorest (lowest income) household, the corresponding ordinate represents the ratio of the accumulated tax burden in total taxation starting from the poorest (lowest income) household. If the tax is in proportion to the income, the concentration curves of tax will coincide with the diagonal, which means that the tax is proportional tax. If the tax is regressive, the ratio of the tax burden borne by the low income households is bigger than the ratio of their income in the total income. At this time, the tax concentration curve is above the diagonal. If the tax is progressive, the tax concentration curve will be below the diagonal. Similar with the definition method of the Gini coefficient, the Suit index is defined as 1 minus the quotient of the area below the tax concentration curve divided by the area below the diagonal, i.e. S = 1 − L/K, among which S indicates the Suit index, L indicates the area below the tax concentration curve, and K indicates the area below the diagonal. S is between −1 and 1, and zero indicates proportional tax. The positive number means progressivity, and the negative number means regression. [^Back]

    [20]. ② The difference between the before-tax income and after-tax income caused by tax will further change the income distribution and Gini coefficient. If the tax is progressive, the inequality of tax will decrease and the Gini coefficient will be smaller. In turn, if the tax is regressive, the after-tax income of low income people will be smaller, the income gap will be wider and the Gini coefficient will be bigger. [^Back]

    [21]. ③ As limited by the length of this article, the table of actual industrial tax rate which are changed to be imposed with VAT can be obtained by asking from the author. [^Back]

    [22]. ① The income distribution effect of tax is generally measured by the progressive (regressive) nature of tax. The tax whose tax burden rate increases along with the increased income is progressive, otherwise it is regressive. The tax whose burden rate is irrelevant with income is the proportional tax. The progressive tax can improve income distribution, the regressive tax deteriorates income distribution, and the proportional tax has no influence on the income distribution. [^Back]

    [23]. ① For other situations, when the VAT collection and administration ability becomes stronger, the regressive nature of VAT will deteriorate the income distribution. This part is omitted as limited by the article length. Anyone who need it can ask from the author. [^Back]


    [1] Nie, H. & Liu, Y. Economic Research Journal (经济研究), (7): 31–42 (2010).

    [2] Nie, H. & Liu, Y. Taxation Research (税务研究), (8): 44–47 (2009).

    [3] Liu, Y. & Nie, H. Economic Research Journal (经济研究), (5): 22–30 (2004).

    [4] Rajemison, H., S. Haggblade, and S. D. Younger. Indirect Tax Incidence in Madagascar: Updated Estimates Using the Input-Output Table. CFNPP Working Paper, 2003.

    [5] Zhang, B. doctoral thesis, Renmin University of China, (2012).

    [6] Ping, X., Liang, S., Hao, Z. et al. Economic Research Journal (经济研究), (9): 66–80 (2009).

    [7] Auerbach, A.J., and L. J. Kotlikoff. Dynamic Fiscal Policy. Cambridge: Cambridge University Press, 1987.

    [8] Chen, Y., Zhang, X., Kou, E. et al. Economic Research Journal (经济研究), (9): 29–42 (2010).

    [9] Younger, S.D., E. S. David, S. Haggblade, and A. D. Paul. Tax Incidence in Madagascar: Analysis Using Household Data.World Bank Economic Review, 1999, 13 (2): 303–31.

    [10] Scutella, R. The Final Incidence of Australian Indirect Taxes. Australian Economic Review, 1999, 32 (4): 349–368.

    [11] Nie. H. China Economic Quarterly (经济学(季刊)), (1): 287–312 (2012).

    [12] Warren, N. Australian Tax Incidence in 1975-76: Some Preliminary Results. Australian Economic Review, 1979, 12 (3): 19–30.

    [13] Suits, D. Measurement of Tax Progressivity. American Economic Review, 1997, 76 (4): 747–752.

    [14] Musgrave, R.A., and T. Thin. Income Tax Progression, 1929–48. Journal of Political Economy, 1949, (56): 498–514.

    [15] Caspersen, R., and G. Metcalf. Is a Value Added Tax Progressive? Annual Versus Lifetime Incidence Measures. NBER Working Paper, 1993.

This Article


CN: 11-3536/F

Vol , No. 12, Pages 23-39

December 2016


Article Outline



  • 1 Introduction
  • 2 Model, method and statistics
  • 3 Empirical analyses
  • 4 Conclusion
  • Footnote