WPS5788 Policy Research Working Paper 5788 The Value-Added Tax Reform Puzzle Jing Cai Ann E. Harrison The World Bank Development Economics Vice Presidency September 2011 Policy Research Working Paper 5788 Abstract This explores the impact of a tax reform in some The reform reduced the total number of employees provinces of China which eliminated the value-added for all types of firms. For domestic firms, it reduced tax on some investment goods. While the goal of the employment by almost 8 percent. The results are robust experiment was to encourage upgrading of technology, to a variety of approaches, and suggest that the primary the results suggest that there was no evident increase impact of the policy has been to induce labor-saving overall in fixed investment, and employment fell growth. This experiment has since been extended to the significantly in the treated provinces and sectors. rest of China. This paper is a product of the Development Economics Vice Presidency. It is part of a larger effort by the World Bank to provide open access to its research and make a contribution to development policy discussions around the world. Policy Research Working Papers are also posted on the Web at http://econ.worldbank.org. The author may be contacted at ann. harrison@berkeley.edu. The Policy Research Working Paper Series disseminates the findings of work in progress to encourage the exchange of ideas about development issues. An objective of the series is to get the findings out quickly, even if the presentations are less than fully polished. The papers carry the names of the authors and should be cited accordingly. The findings, interpretations, and conclusions expressed in this paper are entirely those of the authors. They do not necessarily represent the views of the International Bank for Reconstruction and Development/World Bank and its affiliated organizations, or those of the Executive Directors of the World Bank or the governments they represent. Produced by the Research Support Team The Value-Added Tax Reform Puzzle Jing Cai University of California, Berkeley Ann Harrison University of California, Berkeley and NBER August, 2011 Abstract In this paper, we explore the impact of a tax reform in some provinces of China which eliminated the value-added tax on some investment goods. While the goal of the experiment was to encourage upgrading of technology, our results suggest that there was no evident increase overall in fixed investment, and employment fell significantly in the treated provinces and sectors. The reform reduced the total number of employees for all types of firms. For domestic firms, it reduced employment by almost 8%. Our results are robust to a variety of approaches, and suggest that the primary impact of the policy has been to induce labor-saving growth. This experiment has since been extended to the rest of China. 1     1. Introduction Much of the literature in public finance focuses on the role of tax reforms in affecting firm behavior. Policy makers in many countries use tax incentives to encourage firm investment, and China is no different. In China, the value-added tax is the major source of fiscal revenue for the Chinese government, generating much more revenues than any other types of tax. In 2002, the revenue from value-added tax was 814.4 billion RMB, accounting for around 48% of the state total tax revenue in that year. In 2009, the Chinese Ministry of Finance estimated that VAT revenue accounted for approximately 31 percent of China’s overall revenue.1 Beginning in 2004, the Chinese government implemented a value-added tax reform in three northeast provinces which removes fixed asset investment from the value-added tax base. The reform has since been extended to the whole country, beginning in 2009. The objective of the 2004 reform was to encourage firms to raise investment on fixed assets for production (excluding structures) and to upgrade their machinery and equipment. The goals of the 2009 reform were similar, but in addition the government expressed the need to provide additional assistance to domestic enterprises to help them weather the adverse effects of the crisis, as well as to encourage fixed asset investments to promote an industrial policy now focused on more technologically advanced sectors. 2 In this paper, we use a firm-level panel dataset ranging from 1998 to 2007 to identify the effect of the 2004 value-added tax reduction in selected provinces on firm behavior, 1 See www.dorsey.com/china_vat_reform 2 According to the People’s Daily Online, December 9, 2008, quoting Zheng Jianxin, deputy director general of the taxation department of China’s Ministry of Finance, “The VAT reform would encourage investment and technological upgrading at Chinese companies, boost domestic demand, improve companies’ competitive strength and play a positive role in helping companies tackle the financial crisisâ€?. The article also states that “The reform was aimed at a shift from the existing production-based to a consumption-based VAT regime, which would enable companies to get tax deductions on spending on fixed assets, Zheng said, adding that this would reduce the tax burden on companies by more than 123 billion Yuan.â€? 2     including employment, investment, profit, productivity, and exports. Although tax reduction is an important fiscal policy, identifying the effect of it is challenging because of the endogeneity of taxes. The government’s choice of tax reduction in areas or sectors is non-random, and usually depends on sector or firm attributes such as size, productivity, capital intensity, ownership, etc. This creates a potential selection bias in policy treatment and makes it hard to identify the causal effect of tax policies. Consider the value-added tax reform of China in 2004: the three northeast provinces were chosen as the first pilot group because while many coastal cities had undergone rapid changes and upgrades in both capital assets and technology after the opening-up of the Chinese economy to the world, the traditional industrial base in northeast regions are left behind in the race of technological advancement and prosperity. Encouraging firms in these provinces to invest more on fixed productive assets to upgrade their technology, and to revitalize these old industrial bases was the main reason to implement the value- added tax reform in these provinces first.3 As a result, we cannot identify the causal effect of the value-added tax reform without first addressing potential endogeneity issues. We use a nonparametric technique, propensity score matching combined with difference-in-difference estimation, to identify the causal effect of value-added tax reduction. This method has two advantages. First, it emphasizes the comparability of the treated and control firms by excluding firms that are not comparable. Second, it relaxes the parametric assumptions associated with regression-based techniques such as the linear regression framework. We assess the credibility of the matching procedure using absolute standardized bias measure and formal paired t-tests. Moreover, we combine the matching 3 According to the Xinhua News Agency on December 22, 2005, “The experiment, which moves the tax from production to tax on consumer spending has encouraged northeast China to increase investment in machinery and equipment and phase out outdated equipmentâ€?. 3     technique with difference-in-difference estimation to deal with concerns about possible unobservable firm characteristics that share the same time dynamics for both treatment and control firms. Our estimation results suggest that while the reform was effective in reducing the value-added tax paid by firms, its impact on firm behavior in other ways is puzzling. The policy significantly reduced firms’ total number of employees for both domestic and foreign firms. Second, the impact of the tax reduction on firm productive investment was limited: while there is some evidence that SOEs increased their investment, the impact was not significant for most other types of enterprises. The impact of the policy on firm profits was similar: while most of the benefits accrued to SOEs, other firms were generally unaffected. Finally, the tax reform did not have any significant effect on firm productivity for all types of firms, and it decreased export intensity for most types of firms. Our results differ in significant ways from Nie, Fang, and Lie (2010), who also explore the impact of the value-added tax reform in China. They find smaller, but still negative effects on employment and positive, significant effects on fixed asset investment. Our results differ in large part from Nie, Fang, and Lie (2010) for two reasons. First, they only had one year of data following the 2004 reform, while we have three years, which allows us to identify the longer term effects. In addition, they do not address the potential endogeneity of the reform targets, while we explicitly address this through nonparametric propensity score matching techniques. Using their same approach, we show that the beneficial effects of the reform on investment was limited to SOEs and was 4     very short-term. Even for SOEs, the positive impact on fixed investment disappears if we include 2006 and 2007. Our evidence suggests that the primary effects of the tax reform were to reduce value-added tax payments and cut employment, as firms shifted to more capital-intensive or labor-saving technologies. While there were some small increases in productive investments, they appear to have been concentrated in SOEs and were not large. Instead, SOEs appear to have benefited primarily through increased profits. One puzzle is why, in light of these limited gains, the policy was extended to the rest of China. One possible explanation is that the VAT reform was part of a package of measures for fiscal stimulus during the 2008-2009 financial crisis. The remainder of the paper is organized as follows. In Section 2, we discuss the value-added tax system and the tax reform in China. Section 3 discusses the identification strategy. Section 4 presents estimation results. Section 5 shows robustness checks and results. Section 6 concludes. 2. Background Mainland China introduced the value-added tax as part of a major general tax reform initiative in 1994. There are three types of value-added taxes with different tax bases: the first type is a GDP-type value-added tax with GDP as the tax base; under that system, no deductions are allowed for capital investment and depreciation when calculating the tax base. The tax is equivalent to a sales tax applicable to both consumer and capital goods. The second one is an income-type value-added tax with income or saving plus 5     consumption as the tax base. The third type is a consumption-based value-added tax, for which the tax base is consumption with investment excluded. Prior to 2004, China adopted the first type of value-added tax. The value-added tax rate is 17% for most products and 13% for some products such as agricultural products. Export enterprises receive value-added tax refunds as an export incentive, with refund rates ranging from 9% to 17%. The GDP-type value-added tax is effective in ensuring fiscal revenue for the country but does not encourage capital investment. When firms purchase equipment, they pay value-added taxes on the input which is included in the purchase price of the equipment. But this part is not deductible against output value- added taxes associated with the finished goods. When they sell the product, consumers indirectly pay value-added taxes again for the same equipment in the form of higher sales prices. As a result, there is a double taxation on equipment purchases. In order to promote an equitable market environment and to stimulate investment, in July 2004, the Chinese government selected three northeastern provinces as a pilot area to implement the consumption-type of value-added tax. In these provinces, value-added tax payers in six selected industries, including agricultural product processing, equipment manufacturing, petrochemical, metallurgy, ship building and automobile manufacturing, were allowed to deduct expenditure on fixed assets from the value-added tax base. This reform was expected to eliminate double taxation and alleviate firms’ tax burden, which as a result could lower prices for consumers and encourage more investment on fixed assets. At the end of that year, the government further included military products and high-tech products as pilot sectors and extended the scope of the tax deduction from incremental quantity to the full amount of fixed assets. In 2007, the reform was extended 6     to six provinces in the central area, including 26 cities. In 2008, eastern Inner Mongolia was further included, and finally in January 2009, the consumption-based value-added tax policy was implemented in all sectors and provinces of China. 3. Data, Variable Definitions and Summary Statistics 3.1 Dataset The data for this analysis comes from a large dataset developed and maintained by the National Bureau of Statistics of China (NBS). The NBS dataset contains annual firm- level unbalanced survey data of all “above scaleâ€? industrial firms with annual sales of more than 5 million RMB. On average, around 220,000 firms per year from 1998 to 2007 are included in the dataset, spanning 37 two-digit manufacturing industries and 31 provinces or province-equivalent municipal cities. They account for most of China’s industrial value added and have 22% of China’s urban employment in 2005. The combined dataset contains detailed information about each firm’s identity, address, industry classification, incorporation year, ownership types, new products and total value of output, total fixed assets, fixed assets for production, sales revenue, profit, total workforce, export sales, total industrial sales, employee educations, income and value added tax payable, etc. These are the key variables based on which we estimate firm level total factor productivity and value added tax reform impacts. The original dataset includes 2,226,104 firm-year observations. Since the paper focuses on manufacturing firms, we eliminate non-manufacturing observations. To further clean the sample, we delete observations whose information on variables such as firm identifiers, county code, sector id, year established are missing, or observations with 7     negative or zero values for key variables such as output, total workforce, capital, and total wages. In addition, observations are dropped if total assets are less than liquid assets or total fixed assets. After implementing these data cleaning procedures, we obtain a sample of 1,894,660 observations for analysis. 3.2 Variable Definitions and Summary Statistics In Table 1, we provide summary statistics for key variables of the analysis. The main outcome variables we consider include employment, investment, profit, total factor productivity (TFP), and export intensity. Employment is defined as total number of employees. Investment is measured in two ways: the first measure is fixed assets for production, and the second measure is the growth of fixed assets plus depreciation. Profit is rescaled by industrial sales. We did not delete firms with negative profits because otherwise it is not a random sample. Value of total fixed assets and fixed assets for production is deflated by the fixed assets investment index. TFP is the firm level total factor productivity estimated using OLS with firm fixed effects. Export intensity is calculated by the ratio of export procurement to total industrial sales. Our key controls include firm size, age, HKTM share, foreign share and state shares. Firm size is measured by total values of output, which are deflated by the sector-specific ex-factory price index of industrial products. HKTM share, foreign share, and state share are defined as the share of the firm’s total equity owned by Hong Kong-Taiwan-Macau investors, investors from other countries, and the state, respectively. These three firm level controls are continuous variables ranging from 0 to 1. In Figure 1, we illustrate the evolution of the value added tax in treated and control groups (rescaled by industrial sales) from 1998 to 2007. First, over the sample period, the 8     value added tax of firms in treated sectors in control provinces does not change much from year to year. Second, we compare the evolution of value added tax between firms in treated sectors in treated provinces and firms in treated sectors in control provinces. We can see that before 2004, the value added tax of treated firms is always higher than that of control firms in treated sectors. However, one year after the value added tax reform was implemented in 2004, there is a significant decrease in the value added tax paid by treated firms, it reaches a similar value to control firms in treated sectors in 2005, and falls in 2006 and 2007. In Table 2, we compare the evolution of the value added tax from 1998 to 2007 between firms with different ownership. For foreign invested firms, the value added tax did not decrease after the reform was implemented; it even increased a little bit after 2004, and the trend is very similar between treatment and control firms. This is because before the reform, there were already some tax exemption policies for foreign invested firms. For domestic firms, there is no significant change in value added tax before and after 2004 for control firms; for treated firms, the value added tax was stable before 2004 but fell annually beginning in 2004. This trend holds for both state owned and non-state owned domestic firms, and is more significant for non state owned domestic firms. 4. Identification Strategy In order to identify the causal effect of value added tax reform on a firm’s behavior, the best way is to compare the behavior of a firm that is exposed to the reform with that of the same firm if it had not experienced the reform. However, this sort of counterfactual is rarely observable. In our case, although the policy was only implemented on certain 9     sectors in certain provinces, the assignment was not random. It is possible that these sectors or provinces were chosen because they fell behind other sectors or provinces in investment, or because they are more capital intensive and reducing the value added tax is more important for these types of enterprises. As a result, the key difficulty with identifying the causal effect of value added tax reform is endogenous selection. To solve the endogeneity problem, we use a two-stage identification approach. First, we use nonlinear propensity score matching techniques to construct a control group of firms that match most closely firms that have been treated based on observable characteristics; Second, we estimate the program impact using the difference-in-difference approach to remove all unobservable effects that have the same time dynamics in the treatment and matched control group. 4.1 Propensity Score Matching While a number of methods are available for estimating treatment effects using nonlinear matching techniques, we adopt a procedure using Gaussian kernel matching introduced by Becker and Ichino (2002). To identify the most appropriate control group, we need to specify a list of covariates as key determinants of policy assignment. Here we use sector, foreign share, state share, export share, firm size, age, capital, and productivity as matching covariates, or in other words, firms in the control group are matched to the treatment group on the basis of the pre-treatment (1998-2003) mean of these observables. There are two steps to test whether the propensity score matching method works well. First, we need to estimate whether the covariates we chose are actually important determinants of policy treatment. For this, we estimate a probit model for the likelihood of the value added tax reform treatment: 10     !!"#$%"&$!" = !! + !! !"#$%&'(â„Ž!"!!" + !! !"#$%â„Ž!"!!" + !! !"#"!"â„Ž!"!!" + !! !"#$%&'â„Ž!"!!" + !! !"#!"#$"#!" + !! !"#$%"&!" + !! !"!!" + !! !"#$%!!" + !! !"#$%!!" + !!"  (1)   Where Treatmentij is a dummy variable which equals 1 if firm i in sector j was exposed to the tax reform in 2004 and 0 otherwise, and Sectorij includes a set of two-digit sector dummies. Second, we calculate the standardized differences for covariates in the probit regression to assess the performance of our propensity score matching. Specifically, for each covariate, we take the average difference between treated and matched control firms, and then normalize it by the pooled standard deviation of the covariate in the treatment and control groups, which is referred as the absolute standardized bias (ASB). While there is no clear criterion for the value of ASB, Rosenbaum and Rubin (1985) suggests that a value of 20 is large. Moreover, we perform a formal pairwise t-test comparison between treated and matched control firms to see whether there are any significant differences. Throughout we impose the common support condition and confine our attention to the matched firms falling within the support of the propensity score distribution of the treated group. 4.2 Difference-in-Difference Estimation Using difference-in-differences is likely to improve the quality of non-experimental evaluation studies because it removes the time invariant unobserved heterogeneity across firms, such as sector specific effects, managerial behavior, etc. Here we define the first difference of outcome variables, including total number of employees, investment, profit, TFP, and export in two ways. The first method is to take the difference between post- treatment (2005-2007) and pre-treatment (1998-2003) means of outcome variables, which means we only keep a balanced sample with firms exist in the sample both before and 11     after the policy treatment. The second method is to take the growth of outcome variables from 2003, which is one year before the policy enacted, to 2007, which is the end of the sample period. In this case, we only keep firms that exist in the data for both 2003 and 2007. We now explicitly show the formula we use to combine propensity score matching with difference-in-difference estimation. In the standard difference-in-difference estimation, we treat each of the firms linearly and with the same weight, while the difference-in-difference estimator paired with propensity score matching allows us to include only treated firms within the common support and picks control firms according to the metric function specific to the matching method. The estimator is as follows: ! !!!" = ! !!" − !!"!! − !(!!" , !!" )(! − ! ) !" !"!! ! !∈!! ∩!! !∈!! ∩!! Where is the set of treated firms that falls within the common support !! , !! is the set of control firms, and !! if the number of treated firms in the common support set. ! is outcome variables and ! measures the probability of receiving treatment based on the vector of firm characteristics !!"!! : !!" = !(!!" = 1|!!"!!   ) !(. ) is a Gaussian kernel weighting function that depends on the propensity score distance between the treated and control firms: !!" − !!" !( !! ) ! !!" , !!" = !!" − !!" !∈!! ∩!! !( !! ) 12     !!   Where !(. ) is the Gaussian normal function where ! ! = ! ! ! and !! is a bandwidth parameter. !!!" is the estimator of the causal effect of the value added tax reform, and we obtain standard error using bootstrap procedure. 5. Estimation Results 5.1 Propensity Score Matching Estimation Results In this section, we analyze the estimation result of the probit model for the policy treatment and the matching balance test. First, in Table 3, we show the results of the probit regression. The dependent variable is a dummy variable which takes a value of 1 if a firm was included in the value added tax reform treatment group and 0 otherwise. The objective is to check whether the covariates we chose are important determinants of policy treatment. All covariates are measured by the mean before the policy treatment. Columns (1) to (3) reports estimation results when we define first difference of outcome variables as the growth from 2003 to 2007, while columns (4) to (6) displays results if we define it as the difference between pre and post treatment means. We find that for both domestic and foreign firms, most covariates are significant determinants of policy treatment. Specifically, firms are more likely to receive policy treatment if they have lower foreign shares or HKTM shares, or higher state shares. Export-oriented firms are less likely to be treated, and firms with less output or more capital are more likely to be included. Younger firms or firms with higher productivity have a higher probability of being selected. The results confirm that the focus of the reform was on regions with less foreign investment, a larger state enterprise presence, 13     less competition through export activity, and smaller firms. These results are consistent with the anecdotal and press reports that the goal of the initial 2004 reform was to encourage upgrading in the more backward northeastern provinces. Table 4 reports the balancing test results based on the Gaussian kernel matching. The ASB measures reported in column (3) are all below 5% in absolute value in the matched sample. Adopting the matching method reduced bias substantially as shown in column (4). Moreover, there’s no significant difference in covariates we chosen between treated and matched samples. The only exception is that the ASB measure of logTFP increases greatly after matching. However, after matching the measure is still well below 20, and the t-test of difference between treatment and matched groups is not significant. Overall, the quality of the matching procedure is good and provides a solid foundation for the difference-in-difference estimation in the next stage. 5.2 Difference-in-Difference Estimation Results Having demonstrated the quality of the matching procedure, we then present the difference-in-difference matching estimation results. As discussed earlier, we use two specifications to define first difference of outcome variables. Results are listed in Tables 5.1 and 5.2 for these two specifications, respectively. Since the estimation results do not vary much between different specifications, we will focus on Table 5.1. We present results for the overall sample, domestic (state-owned and non-state-owned), and foreign firms separately. We begin by looking at the effect of the tax reform on value-added taxes paid by firms. The estimates show that overall, the reported value-added tax paid (rescaled by industrial sales) by treated firms becomes 0.008 less than that paid by control firms. This 14     effect is statistically and economically significant given that the mean of value-added tax before the policy was implemented was around 0.09. This suggests that the value-added tax reform effectively decreased taxes paid by treated firms. The magnitude of effect varies by firm ownership. The effect is largest for domestic state-owned firms: the reform reduced the value-added tax ratio by 7.6% for treated firms. For domestic non-state- owned firms, the reform also reduced tax paid by around 0.7%. These numbers indicate that the tax reform reduced the tax burden for SOEs ten times more than for non-SOEs. Taxes paid by treated foreign firms also decreased by 0.5% but the effect is not significant. This is consistent with the fact that the reform itself focused on domestic and particularly state owned enterprises, while foreign firms already faced favorable value- added tax policies before the reform. In summary, value-added tax reform significantly reduced the tax paid by domestic firms, but did not have a large impact on foreign firms. We then turn to the impact of the tax reduction on firm behavior. We consider employment first. Results suggest that the reform reduced the total number of employees for all types of firms. For domestic firms, it reduced employment by almost 8%, but the effect is only significant for domestic non-state-owned firms. For state-owned firms, the magnitude is around 6% but it is not statistically significant. This might because we do not have sufficient power for state-owned firms. The tax reform has a smaller effect on employment of foreign firms: they reduced employment by around 6%. Second, we estimate the impact of tax reform on firm investment. In this paper, we use two measures for firm investment. The first one is fixed assets for production, which includes equipment, machinery, etc.; the second measurement is the growth of fixed assets plus depreciation. Based on the first measure of investment, we see that the tax 15     reform did not influence the investment behavior of domestic non-state-owned firms. It raised state-owned firm investment by 12.26%, but the effect is only significant at the 10 percent level. For foreign invested firms, after the reform, their investment was reduced by 8%. However, if we define investment as growth of fixed assets plus depreciation, the impact of value-added tax reform on firm investment is insignificant for all types of firms. Taken together, the results suggest a shift in the composition of investment, which is most evident for SOEs but is not significant at conventional levels. Third, the value-added tax reform affected the profit of domestic state-owned firms and foreign firms. It raised the ratio of profit to industrial sales of domestic state-owned firms by 0.48 (the sample mean of profit ratio is around 6.66), but did not affect the profit of domestic non-state-owned firms. However, foreign firms’ profit was reduced by a small amount of around 0.02 after the policy change. Overall, domestically owned firm profits increased significantly, but the gains were concentrated in SOEs. Fourth, we also check whether the tax reform affected firm productivity. However, the result suggests that the reform did not have any effect on firm productivity for any types of firms. The effect is small and insignificant for firms with any types of ownership. Finally, we consider export activity. According to the estimation results, firms’ export intensity, which is measured by the share of export procurement in industrial sales, significantly fell after the tax reform policy. The effect holds for all firms except for state-owned firms. The export intensity of domestic non-state-owned firms decreased by 0.008 (sample mean of export intensity equals 0.17) after 2004. The effect on foreign firms is very large. The reform reduced export intensity of foreign firms by 0.06. 16     In summary, the value-added tax reform reduced firms’ tax burden significantly. It had a significantly negative effect on employment, and the effect is very robust. While the main stated objective of implementing value-added tax reform was to encourage firms to invest more on machinery and equipment, we do not find a large and significant effect of the reform on firm investment for most firms. It only raised investment for domestic state-owned firms. Similarly, the reform only raised profits for state-owned firms, while profits of foreign firms fell. We did not find any significantly effect of tax reform on firm productivity. Finally, the policy reduced export intensity for most firms. 6. Alternative Specifications 6.1 Instrumental Variable Estimation The results indicate that the reform did not generate the expected large positive effect on firm investment. Moreover, even for foreign firms where the value-added tax did not fall significantly after the reform, we see a similar effect of the policy treatment on employment reduction, investment and even export reduction. This suggests maybe there are some other targeted changes that occurred together with the value-added tax reform in these treated sectors and provinces, which could influence firm behavior in similar ways. In order to check whether the above results are actually driven by value- added tax reduction, we use a more direct method to study the tax impact in this section. While in the last section, we defined policy treatment based on the sector and province in which a firm is located, it is not necessarily true that for all these firms, the value-added tax actually fell after 2004. A more direct way is to see whether a firm actually paid a smaller amount of value-added tax, and estimate the effect of tax 17     reduction on firm behavior to see whether we can get similar results as above. The OLS estimating equation is as follows: !_!"#!" = !! + !! !"#$%&'()*+!!" + !! !!" + !! !! + !!"        (2) Where !_!"!!" is the difference in outcome variables before and after year 2004. We consider the same set of outcome variables as before, including employment, investment, profit, TFP, and export. As before, we use two specifications to define the difference: one is to take the growth from 2003 to 2007, and the other is to take the difference between post-treatment mean and pre-treatment mean. TaxReduction!" is a dummy variable which equals one if the value-added tax fell after 2004 and zero otherwise. !!" includes firm level controls including foreign share, state share, export share, firm size, age, capital, and productivity. !! is a set of sector dummies. In order to solve the endogeneity problem, we use the policy change, which is a binary variable that takes the value of one if a firm falls in treated sectors in treated provinces and zero otherwise, as the instrument for the variable !!"#$!"#$%&'!" . The advantage of this method is that we can directly go from policy change to value-added tax changes to specific outcomes, while the disadvantage is that there is still potential endogeneity if the program was not completely exogenous. The OLS estimation results are listed in table 6.1 and 6.2. In Table 6.1, we show the result if we specify the first difference of outcome variables as the growth from 2003 to 2007, while Table 6.2 presents results when we define it as the difference between before and after 2004. We will focus on Table 6.1 here since estimation results do not vary significantly across the two different specifications. The results in the first column indicate that for firms which experienced a fall in value-added taxes, employment fell. 18     The average fall in employment for these firms ranged from 2 to nearly 6 percentage points. The biggest reduction in employment in conjunction with the falling value-added taxes where experienced by SOEs. However, there is no evidence of increasing physical capital or overall investment increase associated with the value-added tax reduction. Consistent with the reduction in employment, total factor productivity increased for those firms, but profits were not significantly affected. The IV estimation results, where value-added taxes paid by the enterprise are instrumented with the 2004 policy change, are reported in Tables 7.1 and 7.2. In Table 7.1 we define differences in outcome variables as the growth from 2003 to 2007. The first column reports the first stage results of regressing the dummy variable for VAT tax reduction on the treatment. As expected, the coefficient is positive and statistically significant for most firms, indicating that the share of value-added taxes in sales fell for firms in the treatment group. The results for employment are similar to our difference-in-difference matching estimation as well as the OLS results in Tables 6.1 and 6.2. Overall, the tax reduction had a negative effect on the number of employees. Second, consider the impact of tax reduction on firm investment, when using IV estimation, we cannot find a significant effect. Similarly, the IV results suggest after the value-added tax was reduced, firms’ profit (standardized by industrial sales) did not change significantly. 6.2 Other Specifications and Robustness Tests In July 2007, the value-added tax reform was extended to 26 middle cities. We did not exclude 2007 in our results. However, in unreported extensions we tested the 19     robustness of our results in two ways. First we restricted the sample to 1998 through 2006 and redid our estimates. Second, we used 1998-2007 but excluded the 26 middle cities. The results are robust to both these extensions. We also reproduced the specifications employed by Nie, Fang, and Li (2010) in Appendix Tables 1 through 3. In order to see why we get different results, we use their variable definitions, with investment defined as annual growth in fixed assets, and the same control variables. The dependent variable is the mean post 2004 – the mean before 2004 and the independent variables are the means before 2004. According to Appendix Table 1 columns (1) - (3), using the same years 1999-2003 and 2005, we can get the same sign and similar magnitudes of effects as Nie, Fang, and Li (2010). However, if we extend the sample to 1998-2007 and do the same estimation, the results reported in columns (4) - (6) show that the effect on fixed investment is much smaller and becomes insignificant. There are other problems with their approach in addition to lack of robustness over the longer time horizon. For example, they do not address the potential endogeneity of treatment, and their definition of investment is problematic too (the value of fixed assets is measured in nominal terms and depreciation is not taken into account). Appendix Tables A.2 and A.3 separate the results into SOEs and non-SOEs. The results in Appendix Table A.2 show that the results in Nie et al are driven by the changes in fixed investment across SOEs. Comparing the first three and last three columns, which include 2006 and 2007, we see that the significant effects on investment were only present in 2005, and disappear if we add 2006 and 2007. Appendix Table A.3 reports the results for non-SOEs, for which the impact on fixed investment was negative. Overall, these additional tables confirm that the positive impact on investment was short-lived and 20     limited to SOEs, while the negative and significant impact on employment is evident across all ownership types and increases in magnitude over time. 7. Conclusions This paper analyzes the impact of the value-added tax reform in China on firm employment, investment, profit, TFP, and export intensity. We use a difference-in- difference propensity score matching approach to identify the causal effect. We find that the reform significantly reduced firms’ tax burden. The tax reduction also reduced firm employment for both domestic and foreign firms, while its effect on firm investment was limited, and only positively significant for state-owned enterprises (SOEs). The profits of domestic firms improved after the tax burden was partially removed, while the profits and investments of foreign firms fell. The impact on firm productivity is only significant for domestic firms and the effect is negative. Finally, for most firms, their exports fell in conjunction with the value-added tax reform. Our results suggest that the reform was targeted at provinces with a large number of state owned enterprises, less foreign ownership, more capital intensity and less outward orientation. Using our matching estimation as well as other approaches such as instrumental variables using the targeted sectors and provinces as instruments for the change in value-added taxes, we find that the 2004 tax reform led firms to use fewer workers, increased profits for SOEs, and led to some increase in productive investment among SOEs but no significant change for other enterprises. The insignificant effects that we find on productive investment for non-SOEs, combined with the fall in 21     employment across the board, are puzzling. We intend to do further research to identify whether more detailed data on investment categories could show a shift among non-SOEs towards labor-saving technology as a result of the reform. Our most robust finding is the significant reduction in employment among treated firms. Treated firms reduced employment between 6 and 8 percentage points. One policy problem that should be considered for future research is whether encouraging such labor-saving changes are optimal. Policy changes in both developed and developing countries appear to be encouraging manufacturing growth which leads to small increases in employment. For the US, for example, Ebenstein, Harrison, McMillan and Phillips (2011) show that falling prices of investment goods led to a reduction in domestic manufacturing employment. Since the benefits from the reform in terms of increasing aggregate investment and even productive investment seem quite limited, one question is why the reform was extended to the rest of China. One likely explanation is that extending the reform to the rest of China was part of a comprehensive stimulus package in response to the 2008-2009 financial crisis. 22     References Sascha Becker and Andrea Ichino. (2002). “Estimation of Average Treatment Effects Based on Propensity Scoresâ€?. The Stata Journal, SataCorp LP, vol. 2(4), pages 358-377, November. Alvaro Bustos, Eduardo M.R.A. Engel, and Alexander Galetovic. (2004). “Can Higher Taxes Increase the Long-run Demand for Capital? Theory and Evidence for Chileâ€?. Journal of Development Economics, 73(2004): 675-697. Robert Carroll, Douglas Holtz-Eakin, Mark Rider, and Harvey S. Rosen. (2010). “Income Taxes and Entrepreneurs’ Use of Laborâ€?. Journal of Labor Economics, 18(2): 324-351. Rodrigo A. Cerda and Felipe Larrain. (2010). “Corporate Taxes and the Demand for Labor and Capital in Developing Countriesâ€?. Small Business Economics, 34: 187-201. Raj Chetty and Emmanuel Saez. (2005). “Dividend Taxes and Corporate Behavior: Evidence From the 2003 Dividend Tax Cutâ€?. Quarterly Journal of Economics, CXX(3): 791-833. Ebenstein, Avraham & Harrison, Ann & McMillan, Margaret & Phillips, Shannon, 2011. "Estimating the impact of trade and offshoring on American workers using the current population surveys," Policy Research Working Paper Series 5750, The World Bank. Nada Eissa. (1995). “Taxation and Labor Supply of Married Women: The Tax Reform Act of 1986 as a Natural Experimentâ€?. NBER working paper, No. 5023. Sourafel Girma and Holger Gorg. (2007). “Evaluating the Foreign Ownership Wage Premium Using a Difference-inDifference Matching Approachâ€?. Journal of International Economics, 72(2007): 97-112. Ann Harrison and Jason Scorse. (2010). “Multinationals and Anti-Sweatshop Activismâ€?. American Economic Review, 100: 247-274. Kevin A. Hassett and R, Glenn Hubbard. (2002). “Tax Policy and Business Investmentâ€?. Handbook of Public Economics, Chapter 20: 1293-1343. Shuanglin Lin. (2008). “China’s Value-added Tax Reform, Capital Accumulation, and Welfare Implicationsâ€?. China Economic Review, 19(2008): 197-214. Karel Mertins and Morten Ravn. (2010). “Empirical Evidence on the Aggregate Effects of Anticipated and Unanticipated U.S. Tax Policy Shocksâ€?. NBER working paper, No. 16289. 23     Huihua Nie, Mingyue Fang, and Tao Li. (2010). “China’s Value-Added Tax Reform, Firm Behavior and Performanceâ€?. Frontiers of Economics in China, 5(3): 445-463. Paul Rosenbaum and Donald Rubin. (1985). “Constructing a Control Group Using Multivariate Matched Sampling Methods that Incorporate the Propensity Scoreâ€?. The American Statistician, volume 39, No. 1 (Feb 1985), pp. 33-38. 24     Figures and Tables Figure 1. Evolution of Value Added Tax: 1998-2007 !"!$$% Tax rates !"!'$% !"!&$% !"!#$% ())*% ()))% #!!!% #!!(% #!!#% #!!&% #!!'% #!!$% #!!+% #!!,% Year Treated sectors in treated prov Control sectors in treated prov Treated sectors in control prov Control sectors in control prov Note: The variable value added tax is defined as the ratio of reported value added tax to industrial sales. 25     !"#$%&'(&)*++",-&)."./0./10&23&4%-&5",/"#$%06&'778&9&:;;< # of Obs. Mean St. Dev. Minimum Maximum log(Labor) 1894660 4.75 1.15 0.00 12.15 log(Fixedasset) 1894660 8.28 1.71 -0.16 18.03 log(Fixed assets for production) 1825161 8.44 1.70 -1.21 18.68 log(Investment) 823890 6.72 2.12 -6.55 17.26 log(Output) 1894660 9.96 1.36 0.62 19.49 Profit/Industrial sales 1893780 0.00 6.66 -7710.80 2515.00 log(TFP) 1825161 2.00 0.37 -0.19 14.06 Export Intensity 1893780 0.17 0.34 0 1 Foreign Share 1894645 0.07 0.24 0 1 HKTM Share 1590548 0.08 0.25 0 1 State Share 1894627 0.10 0.29 0 1 Age 1892729 14.65 12.57 1 819 Notes: Labor is measured by the total number of employees. Total fixed asset, fixed asset for production, and value of output are deflated values. Investment is calculated as the growth of fixed assets for production plus depreciation. TFP is estimated using OLS fixed effect method. Export intensity is defined as the export procurement divided by industrial sales. Foreign share contributed by HK-Taiwan-Macau equals the share of firms' total equity owned by investors from HK-Taiwan-Macau. foreign share contributed by other countries is defined as the share of firms’ total equity owned by investors outside HK- Taiwan-Macau, principally from OECD countries. State share equals the proportion of firms' state assets to its total equity. Table 2. Evolution of Value Added Tax, 1998 - 2007 All Sample Foreign invested Domestic Domestic Non-SOE Domestic SOE Treated Control Treated Control Treated Control Treated Control Treated Control 1998 0.0888 0.0915 0.0734 0.0782 0.0919 0.0947 0.0955 0.0960 0.0869 0.0922 1999 0.0888 0.0976 0.0773 0.0798 0.0912 0.1021 0.0940 0.0965 0.0867 0.1150 2000 0.0925 0.1289 0.0776 0.2268 0.0964 0.1025 0.1012 0.1066 0.0855 0.0895 2001 0.0921 0.0985 0.0831 0.0902 0.0943 0.1008 0.0965 0.1040 0.0883 0.0866 2002 0.0890 0.0972 0.0799 0.0923 0.0914 0.0986 0.0918 0.1002 0.0898 0.0891 2003 0.0920 0.1010 0.0795 0.0902 0.0953 0.1040 0.0954 0.1056 0.0947 0.0904 2004 0.1032 0.1118 0.0896 0.1002 0.1067 0.1150 0.1055 0.1161 0.1157 0.1003 2005 0.0952 0.1033 0.0955 0.0930 0.0952 0.1063 0.0936 0.1070 0.1181 0.0920 2006 0.0872 0.1083 0.0875 0.0951 0.0872 0.1119 0.0867 0.1126 0.0976 0.0933 2007 0.0849 0.1040 0.0854 0.0980 0.0849 0.1047 0.0844 0.1043 0.0999 0.1222 Observations 88,200 1,805,580 15,466 353,294 72,733 1,452,275 62,479 1,284,952 10,254 167,323 Total 0.0911 0.1048 0.0848 0.1022 0.0924 0.1054 0.0923 0.1065 0.0936 0.0964 Note: This table compares the evolution of value added tax between treated and control firms. The variable value added tax is constrcted from the ratio of reported value added tax to reported industrial sales. 26     Table 3. Determinants of Value Added Tax Reform Policy Treatment Variables Value Added Tax Reform (1 = Yes, 0 = No) Outcome = DIF (2003, 2007) Outcome = DIF (After 2004, Before 2004) All Domestic Foreign All Domestic Foreign (1) (2) !"# !$# (5) !%# Foreign Share -0.178 -3.453 &'()*+ -0.159 -3.766 &'()*+ (0.30)*** (3.023) !'('$,# (0.027)*** (2.884) !'('$,# HKTM Share -0.662 -0.862 &)('$- -0.671 -0.884 &)('$- (0.044)*** (1.180) !'(',*#... (0.040)*** (1.120) !'(',*#... State Share 0.231 0.173 '(,*$ 0.160 0.113 '(,*$ (0.033)*** (0.035)*** !'('-,#... (0.026)*** (0.028)*** !'('-,#... Export Share -0.027 -0.282 '()", -0.048 -0.290 '()", (0.027) (0.041)*** !'('"%#... (0.024)** (0.0366)*** !'('"%#... log(Output) -0.181 -0.180 &'()-) -0.149 -0.142 &'()-) (0.011)*** (0.012)*** !'(')-#... (0.009)*** (0.010)*** !'(')-#... log(Fixedasset) 0.122 0.130 '('+, 0.113 0.118 '('+, (0.007) (0.008)*** !'(')/#... (0.006)*** (0.007)*** !'(')/#... Age -0.004 -0.003 &'(''+ -0.004 -0.003 &'(''+ (0.001)*** (0.001)*** !'(''/#... (0.001)*** (0.001)*** !'(''/#... log(TFP) 0.198 0.189 '(/,$ 0.161 0.143 '(/,$ (0.032)*** (0.036)*** !'(',$#... (0.027)*** (0.031)*** !'(',$#... Obervations 97916 70851 "",-- 126580 92992 "",-- Sector dummies Yes Yes Yes Yes Yes Yes Pseudo R2 0.0743 0.0671 '()//- 0.0719 0.0645 '()//- Note: This table tests whether variables we used for matching are important determinants of policy treatment. Four different variable specifications were used: Specification 1 from column (1) to (3) defines outcome variables as the growth from 2003 to 2007, and control variables as the mean between 1998 and 2003; specification 2 in column (4) to (6) defines outcome variables as the difference between 2003 and 2007, and control variables as the average between 1998 and 2003. * significant at 10% level, ** significant at 5% level, *** significant at 1% level. 27     Table 4. Balancing tests for propensity score matching Mean % Bias % Bias t-test Treated Matched Reduction t-stat p-value Outcome = DIF(2003, 2007) All Sample Foreign Share 0.125 0.125 -0.300 97.6 -0.130 0.893 HKTM Share 0.035 0.033 0.700 97.9 0.460 0.643 State Share 0.114 0.112 0.700 96.4 0.280 0.780 Export Share 0.165 0.162 0.900 93.6 0.470 0.642 log(Output) 10.018 10.016 0.200 97.5 0.070 0.940 log(Fixedasset) 8.707 8.692 0.900 94.0 0.420 0.673 Age 15.199 15.251 -0.500 52.8 -0.210 0.835 log(TFP) 1.935 1.929 2.300 -374.9 1.080 0.278 Domestic Foreign Share 0.000 0.000 -0.100 98.0 -0.060 0.953 HKTM Share 0.000 0.000 -0.100 97.1 -0.040 0.966 State Share 0.130 0.133 -1.000 94.1 -0.370 0.710 Export Share 0.063 0.064 -0.600 97.8 -0.270 0.789 log(Output) 9.872 9.863 0.800 82.6 0.300 0.764 log(Fixedasset) 8.536 8.533 0.200 98.9 0.080 0.934 Age 16.228 16.331 -0.800 64.3 -0.320 0.751 log(TFP) 1.910 1.902 3.100 -213.2 1.260 0.209 Foreign Foreign Share 0.4745 0.4733 0.3000 99.1 0.08 0.939 HKTM Share 0.1275 0.1307 -1.0000 98.7 -0.31 0.755 State Share 0.0658 0.0723 -4.3000 82.6 -0.92 0.357 Export Share 0.4430 0.4273 3.7000 -15.7 0.90 0.370 log(Output) 10.3850 10.4120 -2.1000 81.0 -0.51 0.611 log(Fixedasset) 9.1595 9.1984 -2.3000 71.6 -0.55 0.582 Age 12.3980 12.4370 -0.6000 91.5 -0.14 0.886 log(TFP) 2.0042 2.0049 -0.3000 -43.7 -0.06 0.954 Outcome = DIF(Before 2004, After 2004) All Sample Foreign Share 0.113 0.115 -0.600 94.3 -0.310 0.755 HKTM Share 0.033 0.031 0.800 97.4 0.680 0.497 State Share 0.137 0.135 0.700 96.2 0.330 0.742 Export Share 0.151 0.149 0.700 96.0 0.380 0.707 log(Output) 9.922 9.930 -0.600 88.4 -0.320 0.749 log(Fixedasset) 8.638 8.624 0.800 94.8 0.450 0.654 Age 15.973 15.980 -0.100 97.7 -0.030 0.975 log(TFP) 1.914 1.911 1.100 -201.3 0.610 0.542 Domestic Foreign Share 0.000 0.000 -0.100 97.9 -0.080 0.933 HKTM Share 0.000 0.000 0.000 98.5 -0.030 0.980 State Share 0.158 0.163 -1.500 90.4 -0.650 0.514 Export Share 0.059 0.061 -0.800 96.8 -0.430 0.665 log(Output) 9.779 9.775 0.400 85.9 0.180 0.854 log(Fixedasset) 8.482 8.484 -0.100 99.4 -0.050 0.958 Age 17.075 17.165 -0.700 79.4 -0.300 0.761 log(TFP) 1.891 1.884 2.400 -450.2 1.080 0.280 Foreign Foreign Share !"#$%&'( !"#$%&'( &'") **"! &'"'# '"*+! HKTM Share '"($'(, '"($(#+ &'"! **"! &'"(+ '",-, State Share '"'-!)( '"'-,(! &)"! *'", &'"+- '"+## Export Share '"!)--$ '"!()#- $"+ $,"* '"*+ '"$!) log(Output) ('"$) ('"$!$ &("- ,(") &'"!# '"#!! log(Fixedasset) *"'*-- *"()!$ &("+ ,+ &'"!( '"#,( Age ()"#*) ()"#,) '"( *#"- '"'! '"*- log(TFP) ("**$- ("**)) '"# -'"* '"(! '",,# Note: This table tests whether there’s significant difference between treated and matched groups on potential determinants of policy treatment. Four different variable specifications were used: Specification 1 defines outcome variables as the growth from 2003 to 2007, and control variables as the mean between 1998 and 2003; specification 2 defines outcome variables as the difference between means post and before treatment, and control variables as the average between 1998 and 2003. 28     Table 5.1. The Impact of Value-Added Tax Reform on Firm Behavior -- Outcome = DIF(2003, 2007) Common Support Off Support Matching Estimate Std. Err. T-stat Untreated Treated Untreated Treated All Sample VAT/Industrial sales -0.008 (0.002)*** -3.31 93525 4304 0 87 log(Labor) -0.077 (0.011)*** -7.30 93536 4304 0 87 log(Fixed assets for production) -0.047 (0.019)** -2.43 88045 3944 0 80 log(Investment) 0.014 (0.069) 0.2 31876 1200 0 24 Profit/Industrial sales 0.021 (0.012)* 1.81 93525 4304 0 87 log(TFP) 0.010 (0.006) 1.6 88045 3944 0 80 Export Intensity -0.021 (0.003)*** -6.25 93525 4304 0 87 All Domestic VAT/Industrial sales -0.013 (0.003)*** -4.39 67642 3145 0 64 log(Labor) -0.0763 (0.012)*** -6.18 67649 3145 0 64 log(Fixed assets for production) -0.0229 (0.023) -0.98 63776 2884 0 58 log(Investment) -0.0440 (0.084) -0.52 21772 860 0 17 Profit/Industrial sales 0.0376 (0.016)** 2.37 67642 3145 0 64 log(TFP) 0.0055 (0.007) 0.74 63776 2884 0 58 Export Intensity -0.0075 (0.003)*** -2.57 67642 3145 0 64 Domestic Non-SOE VAT/Industrial sales -0.007 (0.003)*** -2.91 63327 2867 0 58 log(Labor) -0.078 (0.013)*** -6.09 63327 2867 0 58 log(Fixed assets for production) -0.032 (0.025) -1.32 59825 2622 0 53 log(Investment) -0.031 (0.089) -0.34 20465 790 0 16 Profit/Industrial sales -0.004 (0.010) -0.38 63327 2867 0 58 log(TFP) 0.005 (0.008) 0.65 59825 2622 0 53 Export Intensity -0.008 (0.003)** -2.39 63327 2867 0 58 Domestic SOE VAT/Industrial sales -0.076 (0.030)** -2.53 3975 279 0 5 log(Labor) -0.0597 (0.045) -1.33 3975 279 0 5 log(Fixed assets for production) 0.1226 (0.073)* 1.66 3644 262 0 5 log(Investment) 0.1680 (0.264) 0.64 1165 70 0 1 Profit/Industrial sales 0.4754 (0.234)** 2.03 3975 279 0 5 log(TFP) 0.0026 (0.031) 0.08 3644 262 0 5 Export Intensity -0.0034 (0.006) -0.59 3975 279 0 5 All Foreign VAT/Industrial sales 0.005 (0.004) 1.12 25883 1159 0 23 log(Labor) -0.0616 (0.030)*** -3.03 25887 1159 0 23 log(Fixed assets for production) -0.0801 (0.032)** -2.50 24269 1061 0 21 log(Investment) 0.1849 (0.121) 1.53 9171 341 0 6 Profit/Industrial sales -0.0163 (0.005)*** -3.02 25883 1159 0 23 log(TFP) 0.0172 (0.012) 1.39 24269 1061 0 21 Export Intensity -0.0601 (0.010)*** -5.98 25883 1159 0 23 Note: This table presents matching estimation results. * significant at 10% level, ** significant at 5% level, *** significant at 1% level. 29     Table 5.2. The Impact of Value-Added Tax Reform on Firm Behavior -- Outcome = DIF(Before 2004, After 2004) Common Support Off Support Matching Estimate Std. Err. T-stat Untreated Treated Untreated Treated All Sample VAT/Industrial sales -0.003 (0.009) -0.35 120787 5678 0 115 log(Labor) -0.118 (0.009)*** -13.22 120793 5678 0 115 log(Fixed assets for production) -0.023 (0.014) -1.61 118833 5598 0 114 log(Investment) -0.029 (0.038) -0.77 72096 2940 0 59 Profit/Industrial sales 0.139 (0.081)* 1.72 120787 5678 0 115 log(TFP) -0.009 (0.005)* -1.93 118833 5598 0 114 Export Intensity -0.019 (0.002)*** -7.67 120787 5678 0 115 All Domestic VAT/Industrial sales -0.020 (0.06)*** -3.39 88651 4255 0 86 log(Labor) -0.116 (0.010)*** -11.47 88657 4255 0 86 log(Fixed assets for production) 0.008 (0.017) 0.46 87098 4192 0 85 log(Investment) -0.049 (0.046) -1.08 50625 2121 0 43 Profit/Industrial sales 0.196 (0.107)* 1.82 88651 4255 0 86 log(TFP) -0.012 (0.005)** -2.21 87098 4192 0 85 Export Intensity -0.011 (0.002)*** -5.06 88651 4255 0 86 Domestic Non-SOE VAT/Industrial sales -0.021 (0.007)*** -3.27 80526 3744 0 76 log(Labor) -0.124 (0.010)*** -12 80529 3744 0 76 log(Fixed assets for production) 0.002 (0.018) 0.12 79273 3689 0 75 log(Investment) -0.026 (0.049) -0.53 45953 1863 0 38 Profit/Industrial sales -0.017 (0.005)*** -3.11 80526 3744 0 76 log(TFP) -0.019 (0.005)*** -3.51 79273 3689 0 75 Export Intensity -0.012 (0.002)*** -5.07 80526 3744 0 76 Domestic SOE VAT/Industrial sales -0.0229 (0.011)** -2.02 7342 511 0 10 log(Labor) -0.0793 (0.032)** -2.41 7342 511 0 10 log(Fixed assets for production) 0.0474 (0.043) 1.09 7072 503 0 10 log(Investment) -0.1833 (0.128) -1.43 4159 258 0 5 Profit/Industrial sales 1.7379 (1.221) 1.42 7342 511 0 10 log(TFP) 0.0284 (0.020) 1.42 7072 503 0 10 Export Intensity 0.0005 (0.004) 0.11 7341 511 0 10 All Foreign VAT/Industrial sales 0.054 (0.034) 1.58 32136 1423 0 29 log(Labor) -0.092 (0.019)*** -4.96 32136 1423 0 29 log(Fixed assets for production) -0.081 (0.026)*** -3.13 31735 1407 0 28 log(Investment) 0.055 (0.067) 0.82 20369 819 0 16 Profit/Industrial sales 0.023 (0.030) 0.76 32136 1423 0 29 log(TFP) 0.003 (0.009) 0.28 31735 1407 0 28 Export Intensity -0.043 (0.007)*** -5.72 32136 1423 0 29 Note: This table presents matching estimation results. * significant at 10% level, ** significant at 5% level, *** significant at 1% level. 30     Table 6.1. OLS Estimation: The Impact of Value-Added Tax Reform on Firm Behavior -- VAT/Sales Change Dummy, Outcome = DIF(2003, 2007) VARIABLES log(Labor) log(K for production) log(Investment) Profit/Sales log(TFP) Export Intensity (1) (2) (3) (4) (5) (6) All Sample VAT/Sales Reduction !"#"$"$%%% !"#"&'(%%% !"#"$&) !"#"*+$ "#"$,$%%% "#""--$%%% (1 = Yes, 0 = No) ."#""$,+/ ."#"",$)/ ."#"**'/ ."#"*('/ ."#""*&$/ ."#""&(,/ Observations ')0'&, '*0",$ $$0"'- ')0'&, '*0",$ ')0'&, R-squared "#*&+ "#&+$ "#"(* "#"", "#*"& "#"", All Domestic VAT/Sales Reduction !"#"*+'%%% !"#"**,%%% !"#"&'" !"#"-*( "#"$))%%% !"#""$''%% (1 = Yes, 0 = No) ."#""-*)/ ."#"")(+/ ."#"*)*/ ."#"$(*/ ."#""*-+/ ."#""&(+/ Observations )"0+(& ,,0)&$ *$0"), )"0+(& ,,0)&$ )"0+(& R-squared "#*"( "#&+' "#"(' "#"") "#*&$ "#"", Domestic Non-SOE VAT/Sales Reduction !"#"*)$%%% !"#"*&+%%% !"#"&+& !"#"&', "#"$('%%% !"#""$'&%% (1 = Yes, 0 = No) ."#""-$'/ ."#"")+&/ ."#"*+&/ ."#"*("/ ."#""*("/ ."#""&,+/ Observations ,,0*(* ,*0-', *&0,', ,,0*(* ,*0-', ,,0*(* R-squared "#*"& "#&'& "#",$ "#""- "#*** "#"", Domestic SOE VAT/Sales Reduction !"#"(-"%%% !"#"-*" !"#")&( !"#$+& "#",$(%%% !"#""('$% (1 = Yes, 0 = No) ."#"&),/ ."#"$&&/ ."#&&"/ ."#$'+/ ."#"&*+/ ."#""$-$/ Observations -0('' -0*&) &0$+" -0('' -0*&) -0('' R-squared "#&(- "#"+' "#"$, "#"(+ "#&*+ "#"*" All Foreign VAT/Sales Reduction !"#"$("%%% !"#""'++ !"#"-"& "#"&*'%%% "#"$$)%%% "#"*)$%%% (1 = Yes, 0 = No) ."#"")&'/ ."#"&&)/ ."#"-*-/ ."#""*-$/ ."#""-&$/ ."#""$+,/ Observations *)0",( *(0$(" &"0"&+ *)0",( *(0$(" *)0",( R-squared "#*(+ "#&(- "#"$, "#"-& "#&)- "#"&& Notes: This table presents OLS estimation results. * significant at 10% level, ** significant at 5% level, *** significant at 1% level. 31     Table 6.2. OLS Estimation: The Impact of Value-Added Tax Reform on Firm Behavior -- VAT/Sales Change Dummy, DIF(Before 2004, After 2004) VARIABLES log(Labor) log(K for production) log(Investment) Profit/Sales log(TFP) Export Intensity (1) (2) (3) (4) (5) (6) All Sample VAT/Sales Reduction !"#"$%&''' !"#""(%) "#""$(" !"#$*& "#"%(&''' "#""&*"''' (1 = Yes, 0 = No) +"#""%"*, +"#"")-*, +"#"(%%, +"#$$$, +"#""(*), +"#""((*, Observations ($&.*/" ($).*)" -*."0% ($&.*/" ($).*)" ($&.*/" R-squared "#$0* "#$$$ "#"*0 "#""$ "#$** "#""- All Domestic VAT/Sales Reduction !"#"$(*''' !"#""%-$ "#"(/% !"#%$) "#"%)(''' !"#""()% (1 = Yes, 0 = No) +"#""%*), +"#""*&%, +"#"(*0, +"#$/-, +"#""(/$, +"#""((-, Observations 0$.00$ 0(.%-" *$.-/- 0$.00$ 0(.%-" 0$.00$ R-squared "#$-* "#$$* "#"&& "#""$ "#$&( "#""& Domestic Non-SOE VAT/Sales Reduction !"#"(/-''' !"#""&%* "#"$%- !"#""*$$ "#"%$)''' !"#""(// (1 = Yes, 0 = No) +"#""%&/, +"#""*/&, +"#"(&&, +"#""&&/, +"#""(-/, +"#""($-, Observations /).%)& /%."%) )-./*% /).%)& /%."%) /).%)& R-squared "#$*) "#$$/ "#"-$ "#""( "#$/" "#""& Domestic SOE VAT/Sales Reduction !"#"*&"''' !"#"(/& !"#"**% !$#*** "#")//''' "#""$&" (1 = Yes, 0 = No) +"#"($*, +"#"(0), +"#"*)&, +$#$0$, +"#""0$*, +"#""$(%, Observations /.&)& /.%%& ).0%) /.&)& /.%%& /.&)& R-squared "#$)) "#((* "#"%& "#"(% "#(/% "#"(- All Foreign VAT/Sales Reduction !"#"%$%''' !"#"""&$" !"#"$(" !"#"$/) "#"$-)''' "#"$/"''' (1 = Yes, 0 = No) +"#""*0&, +"#""/-&, +"#"$)", +"#")$*, +"#""$0$, +"#""$0%, Observations %%.*// %%.(-" $$.%"& %%.*// %%.(-" %%.*// R-squared "#%)& "#$(* "#")% "#""( "#$)* "#"() Notes: This table presents OLS estimation results. * significant at 10% level, ** significant at 5% level, *** significant at 1% level. 32     Table 7.1. IV Estimation: The Impact of Value-Added Tax Reform on Firm Behavior -- VAT/Sales Change Dummy, Outcome = DIF(2003, 2007) First stage Second stage VARIABLES log(Labor) log(K for production) log(Investment) Profit/Sales log(TFP) Export Intensity (1) (2) (3) (4) (5) (6) (7) All Sample VAT/Sales Reduction !"!#$%&& '(")$$&& '%"$$* #"++* !"(++ #"#*+ '#"(!!&& (1 = Yes, 0 = No) ,!"!!))%- ,%".!!- ,%"#$#- ,%"++#- ,#".*!- ,!"*)#- ,!")!)- Observations *)/*#$ *)/*#$ *%/!$. ../!*( *)/*#$ *%/!$. *)/*#$ All Domestic VAT/Sales Reduction !"!%)+&&& '."!!(&&& '!"$() '!"..# !"()% !"(+# '!".!$&& (1 = Yes, 0 = No) ,!"!!*!+- ,#"!%0- ,!"0*#- ,#"*!*- ,#"!0(- ,!"..$- ,!"#(*- Observations )!/0+# )!/0+# $$/)#. %./!)$ )!/0+# $$/)#. )!/0+# Domestic Non-SOE VAT/Sales Reduction !"!%0)&&& '%"0%%&&& '!"0*$ '!"+%) '!"*)# !"+.+ '!"%*$&& (1 = Yes, 0 = No) ,!"!!*(0- ,!"*)+- ,!"*#0- ,#")*0- ,!"*0#- ,!".+%- ,!"#+!- Observations $$/%+% $$/%+% $%/(*$ %#/$*$ $$/%+% $%/(*$ $$/%+% Domestic SOE VAT/Sales Reduction !"!#($ '$"!*( 0"#(+ '.+"+0 *"%%! '#"!.) '!"%0. (1 = Yes, 0 = No) ,!"!.!*- ,#."!(- ,%$"()- ,()!"*- ,%%"$#- ,("+0+- ,!")#$- Observations (/+** (/+** (/%#) #/.0! (/+** (/%#) (/+** All Foreign VAT/Sales Reduction '!"!#%* ("%%* ."*0$ '()"!. !"+%+ '#"#)# ."%%$ (1 = Yes, 0 = No) ,!"!#(0- ,+"!))- ,("!!.- ,%*)"$- ,!"0+$- ,#"%#(- ,%"+))- Observations %)/!$+ %)/!$+ %+/.+! #!/!#0 %)/!$+ %+/.+! %)/!$+ Notes: This table presents IV estimation results. * significant at 10% level, ** significant at 5% level, *** significant at 1% level. Table 7.2. IV Estimation: The Impact of Value-Added Tax Reform on Firm Behavior -- VAT/Sales Change Dummy, DIF(Before 2004, After 2004) First stage Second stage VARIABLES log(Labor) log(K for production) log(Investment) Profit/Sales log(TFP) Export Intensity (1) (2) (3) (4) (5) (6) (7) All Sample VAT/Sales Reduction !"!#$%&&& '("))*&&& '!"!%+% !"$+* +"*#, !"!##! '!")*(&&& (1 = Yes, 0 = No) -!"!!.+,/ -$",$+/ -!",%*/ -$"$#+/ -+"*!)/ -!"#!!/ -!"#+)/ Observations $#.0,)! $#.0,)! $#(0,(! +,0!%* $#.0,)! $#(0,(! $#.0,)! All Domestic VAT/Sales Reduction !"!*(*&&& '*"((.&&& !"(.$ '!"((% ,".,+ '!"$*+ '!"**%&&& (1 = Yes, 0 = No) -!"!!+)$/ -!")!%/ -!"(,#/ -!"+%(/ -,"+!%/ -!"$((/ -!"$!$/ Observations %#0%%# %#0%%# %$0*+! ,#0+)+ %#0%%# %$0*+! %#0%%# Domestic Non-SOE VAT/Sales Reduction !"!*.)&&& '*"!#,&&& !")*#& !"$+$ '!",!, '!"$., '!"**,&&& (1 = Yes, 0 = No) -!"!!)**/ -!"+$#/ -!"()%/ -!")()/ -!"*$*/ -!"$(#/ -!"$!$/ Observations )(0*(. )(0*(. )*0!*( (+0),* )(0*(. )*0!*( )(0*(. Domestic SOE VAT/Sales Reduction !"!*!. '("*)! '!"#$! '#"+$$ '#",,* '!"!##% '!"!)*) (1 = Yes, 0 = No) -!"!##./ -*"*$(/ -$"$*$/ -#"$*#/ -$("+(/ -!"(%./ -!"$.$/ Observations )0.(. )0.(. )0**. (0%*( )0.(. )0**. )0.(. All Foreign VAT/Sales Reduction '!"!!))! .",#+ *"+%) '("*)# '*"!%# '$"#)) #"+(! (1 = Yes, 0 = No) -!"!$**/ -$!"#$/ -,"(%+/ -,"(!#/ -."*,,/ -$",),/ -#".$*/ Observations **0,)) **0,)) **0$+! ##0*!. **0,)) **0$+! **0,)) Notes: This table presents IV estimation results. * significant at 10% level, ** significant at 5% level, *** significant at 1% level. 33     Appendix Tables Table A1. Replicate Nie et al (2010) results: All Sample 1999-2003 and 2005 1998-2007 Fixed assets Fixed assets Fixed assets Fixed assets VARIABLES growth (nominal) growth (reall) LogL growth (nominal) growth (reall) LogL (1) (2) (3) (4) (5) (6) VAT policy 8,738* 8,857* -0.0925*** 2,331 2,935 -0.111*** (4,863) (4,780) (0.00843) (2,471) (2,782) (0.00855) Firm size (logSales) 397.9 490.6 0.00892*** 3,843*** 2,934** 0.00745*** (3,561) (3,187) (0.00153) (1,480) (1,447) (0.00162) Profit 0.870* 0.692 0.284 0.115 (0.509) (0.451) (0.174) (0.167) Foreign Share -8,722*** -7,914*** 0.146*** -6,455*** -5,241*** 0.182*** (2,423) (2,280) (0.00751) (1,622) (1,545) (0.00794) HKTM Share -1,778* -1,599* 0.0672*** -2,438*** -1,987** 0.0956*** (976.4) (934.4) (0.00701) (905.2) (849.4) (0.00732) State Share 10,257*** 9,330*** -0.352*** 9,385*** 7,307*** -0.428*** (3,173) (3,005) (0.00698) (2,422) (2,198) (0.00717) Export Share 2,051 1,702 0.0209*** -324.2 -431.9 0.00127 (1,670) (1,472) (0.00587) (644.7) (607.1) (0.00607) Profit-sales ratio 0.000218 -0.00102 (0.00125) (0.00161) Observations 91,578 90,975 121,597 96,222 95,459 128,777 Sector dummies Yes Yes Yes Yes Yes Yes R-squared 0.087 0.062 0.036 0.018 0.007 0.047 Table A2. Replicate Nie et al (2010) results: Stateowned Firms 1999-2003 and 2005 1998-2007 Fixed assets Fixed assets Fixed assets Fixed assets VARIABLES growth (nominal) growth (reall) LogL growth (nominal) growth (reall) LogL (1) (2) (3) (4) (5) (6) VAT policy 91,828** 94,221** -0.0785** 28,810 37,324 -0.0767** (40,327) (39,737) (0.0313) (20,795) (24,395) (0.0315) Firm size (logSales) 14,816*** 14,032*** 0.0140*** 10,016** 8,651** 0.00940** (4,148) (3,977) (0.00385) (3,982) (3,793) (0.00406) Profit 0.765 0.692 1.559*** 1.291*** (0.691) (0.637) (0.590) (0.500) Foreign Share -153,072 -141,586 -0.622 -289,068** -273,298** -0.321 (133,346) (129,895) (0.465) (129,113) (132,694) (0.376) HKTM Share -78,407 -112,568 -0.119 -319,417* -312,192* -0.0675 (128,457) (124,240) (0.302) (188,349) (172,661) (0.332) State Share 8,366 237.5 -0.0364 -5,317 -11,736 -0.0458* (12,497) (16,769) (0.0222) (10,809) (15,119) (0.0244) Export Share -36,952** -39,383*** -0.0673 -26,949** -31,486** -0.117** (14,490) (13,981) (0.0532) (13,437) (13,086) (0.0563) Profit-sales ratio 0.0112* 0.0142 (0.00639) (0.00899) Observations 7,423 7,344 8,217 7,996 7,882 8,995 Sector dummies Yes Yes Yes Yes Yes Yes R-squared 0.055 0.054 0.008 0.151 0.119 0.010 34     Table A3. Replicate Nie et al (2010) results: Non-stateowned Firms 1999-2003 and 2005 1998-2007 Fixed assets Fixed assets Fixed assets Fixed assets VARIABLES growth (nominal) growth (reall) LogL growth (nominal) growth (reall) LogL (1) (2) (3) (4) (5) (6) VAT policy -2,174 -2,214 -0.0920*** -993.1 -1,262 -0.112*** (1,941) (1,741) (0.00871) (1,087) (977.6) (0.00885) Firm size (logSales) -3,995 -3,843 0.0101*** 1,938 1,011 0.00926*** (4,482) (4,005) (0.00170) (1,267) (1,139) (0.00180) Profit 0.898 0.694 -0.0306 -0.173** (0.610) (0.543) (0.0746) (0.0753) Foreign Share -5,085** -4,027** 0.141*** -2,078* -997.5 0.174*** (2,154) (1,979) (0.00754) (1,253) (1,137) (0.00796) HKTM Share -335.2 -128.0 0.0626*** -1,181 -800.8 0.0895*** (1,142) (1,043) (0.00703) (881.8) (793.3) (0.00734) State Share 12,089** 9,538** -0.348*** 1,954 -156.0 -0.410*** (5,287) (4,777) (0.0123) (2,680) (2,276) (0.0124) Export Share 2,350 1,919 0.0202*** -787.5 -834.3* 0.000763 (1,831) (1,628) (0.00591) (478.9) (438.1) (0.00611) Profit-sales ratio -0.000215 -0.00151 (0.000849) (0.00117) Observations 84,155 83,631 113,380 88,226 87,577 119,782 Sector dummies Yes Yes Yes Yes Yes Yes R-squared 0.136 0.102 0.019 0.002 0.011 0.024 35   Â