Does value-added tax reform increase the sophistication of enterprise exports? Evidence from China

doi:10.21203/rs.3.rs-2242598/v1

Download PDF

Research Article

Does value-added tax reform increase the sophistication of enterprise exports? Evidence from China

https://doi.org/10.21203/rs.3.rs-2242598/v1

This work is licensed under a CC BY 4.0 License

Version 1

posted

You are reading this latest preprint version

Value-added tax (VAT) is a neutral tax widely adopted around the world that is conducive to easing the tax burden of central and local governments and promoting enterprise innovation. Using the micro-data of China's industrial enterprises from 2000 to 2007,this paper takes the VAT reform in Northeast China as a quasi-natural experiment to explore the impact of reform on the export sophistication of enterprises. This study reveals that the VAT reform increases the sophistication of enterprises' export, and the influence of reform on the export sophistication is found to be insignificant not only for state-owned enterprises but also for medium and low-tech sectors’ enterprises. Moreover, the mechanism analysis shows that the VAT reform stimulates the vitality of enterprise innovation and increases investment in R&D, thereby increasing the sophistication of exported products. The research in this paper is helpful not only for deepening our understanding of the significance of VAT reform but also for further improving policy measures to promote the high-quality development of export trade.

JEL:H25,P25,D22

VAT reform

Export sophistication

Revitalizing of Northeast China

Value-added tax (VAT) is a tax levied on the value added at each of the multiple links in the commodity production chain, during product circulation, and from labour services, i.e., on the value added to commodities. Value-added tax originated in France in 1954. Since its inception, value-added tax has become a popular form of taxation for governments all over the world due to its inherent neutrality and ability to generate a stable income. Existing literature focuses on the impact of tax reform on labour market and economic growth, involving dimensions such as employment scale and structure, enterprise transformation and upgrading, and productivity (Adhikari, 2020; Yang and Zhang, 2021; Peng et al., 2022). In particular, this paper focuses on the impact of tax reform on export complexity. Export complexity reflects an economy's export technology content, export structure, production efficiency and other important economic characteristics (Hidalgo and Hausmann, 2009). Increasing export complexity will help industries move to the upper end of the global value chain and enhance their competitiveness in the international market. Therefore, this study from the perspective of export complexity will help us understand how to better promote tax reform and achieve high-quality economic development.

The effect of the tax reform has long been an important topic of discussion among scholars. Research on enterprises has mainly focused on its impact on investment and production behaviour (including labour employment). On the one hand, according to neoclassical investment theory, proposed by Jorgenson (1963), tax reforms generally change the "capital user cost" and further affects the investment decisions of enterprises. Wang (2013) used a difference-in-differences (DID) model to study the impact of the 2009 VAT reform and found that the policy had a significantly positive impact on corporate investment. Auerbach et al. (1995) confirmed the positive effect of Sweden's tax system reform on corporate fixed asset investment. Da Rin et al. (2006) reported that lowering the capital gains tax rate in European countries can induce venture capitalists to raise new funds and scale up their investments. In addition, Liu (2019) found that the VAT reform led to a significant 8.8% increase in investment among firms affected by the VAT reform relative to the investment of enterprises that were not covered by the policy implementation and that this positive effect was stronger for enterprises with liquidity constraints. On the other hand, some scholars have found that tax reform policies affect employment, but the conclusions drawn from these studies are inconsistent. The initial stock of capital, the speed of economic development, and industry heterogeneity all play a role (Hutton and Ruocco, 1999). Nie et al. (2010) found that tax cuts significantly reduce labour employment and that there is a capital substitution effect, which is consistent with the findings in Chirinko's (2002) research. As the price of labour relative to capital increases, enterprises choose to use assets to replace labour in production and thus reduce their demand for labour. Moreover, when the corporate tax burden (such as that from a VAT) increases, the difficulty of transferring the tax burden increases production costs for the enterprises and encourages enterprises to increase their employment of informal workers in order to reduce their production costs (Lora, Gonz, 2016). Garrett et al. (2020) examined the impact of the accelerated depreciation policy for fixed assets on the labour market in the United States and found that this policy improved regional employment to an extent.

However, the existing literature has focused more on the investment decisions of the enterprises themselves and has not further explored whether tax reforms improve the innovation behaviour of the observed enterprises or the technological sophistication of the product. The VAT reform, on the one hand, reduced the cost of upgrading technical equipment for enterprises and enhanced the speed with which equipment can be technologically upgraded for enterprises in the pilot area; we call this the innovation effect. On the other hand, the VAT reform had an effect on the labour market, which may have improved the quality of company products by expanding employment. The new policy guarantees an operating cash flow for enterprises, thus prompting the enterprise to invest more in labour and fixed assets and to expand the scale of its operations. The above behaviours enhance enterprise innovation and research and development and expand the number of employed labourers, which could result in a production ramp-up and the improvement of techniques. In this way, the quality of the products exported by enterprises could be improved. In the age of globalization, because enterprises are major players in terms of economic activity and technological innovation, the technical complexity of their products is an important factor determinant of the level of their exports. Improving enterprise sophistication equalizes the competition between domestic products and imported foreign commodities. Enterprises benefit when the industrial structure is adjusted, technology is upgraded and product competitiveness is improved, and these benefits can play an important role in promoting regional economic development. However, much existing research has ignored the technical sophistication of enterprise exports. This paper explores how this outcome was affected by the pilot VAT reform that occurred from 2000 to 2007.

This paper makes the following contributions to the existing literature. First, this paper enriches the research on the sophistication of enterprise exports and the impact of the VAT reform on that sophistication. Second, we consider the confounding effect of the corporate income tax policy during the same period, as the implementation of the northeast revitalization strategy should have had a positive effect on most enterprises in northeast China to at least a certain extent. Therefore, we expand our research sample to include six provinces in central China and use a difference-in-differences-in-differences (DDD) model to control for the deviation caused by the above factors. Finally, this study is helpful for deepening the relationship between tax reform and enterprise development, improving the ability of tax theory to guide tax reform in practice, and contributing to the further implementation of the revitalization strategy in Northeast China.

The remainder of this paper is organized as follows: the second section describes the theoretical mechanism and hypothesis. The third section presents the research design, a description of the data, and the variables. The fourth section analyses the empirical results. The fifth section contains the discussion and mechanism tests. The sixth section concludes.

2.1 International VAT model comparison: EU model and New Zealand model

At present, the value-added tax is one of the most frequently implemented taxes globally and has been levied in more than 170 countries and regions around the world. From an international perspective, the VAT system has derived two models during its development: the EU model and the New Zealand model.

The EU (European Union) model has undergone a certain evolutionary process. In 1977, the European Community issued a directive to clarify the basic connotation of consumption-based value-added tax. In 1992, member states reached a directive on making tax rates as uniform as possible, stipulating a maximum standard rate of 25% and a minimum of 15%. The EU stipulated that the value-added tax in the import and export links should be abolished in 1997; in 1999, the council decided to implement a low value-added tax rate for some labour-intensive service industries, but agreed that the tax rate should not be lower than 5%. In 2007, EU member states reached an agreement on the basic elements of the VAT system: each member state stipulates taxation on the sales of goods and the provision of labour services; in terms of tax rates, one standard tax rate is set, supplemented by one or two low tax rates, and only individual member states (such as Italy and Luxembourg) have three tax rates. At present, the VAT model in China that is credited with invoices is the EU model.

The most notable feature of the New Zealand model is that it has a wide range of taxation, a single tax rate, and maintains tax neutrality as much as possible. In October 1986, New Zealand began to levy a value-added tax (goods and labour tax) nationwide, and all sales of goods and provision of labour services and imported goods that occurred in the country were included in the scope of taxation. The VAT rate in New Zealand is uniformly 15%. Except for small businesses with annual sales below NZ$60,000, which do not need to pay VAT, all other businesses are required to pay VAT. To a certain extent, the New Zealand model can solve the problems of repeated taxation and price distortion inherent in the EU model with multiple tax rates, and can ensure the integrity of the value-added tax deduction chain as much as possible. From the perspective of economic efficiency, the New Zealand model has been hailed as the development direction and the best model for the VAT system.

Table 1
VAT Model	EU model	New Zealand model
Formation Time	1977	1986
Feature	Multiple tax rates	Single tax rate
Reform Contents	Set one standard tax rate, supplemented by one or two low tax rates that mainly applies to basic daily necessities, as well as medicines, catering services, etc.	The VAT rate in New Zealand is uniformly 15%. Except for small businesses with annual sales below NZ$60,000, which do not need to pay VAT, all other businesses are required to pay VAT.

2.2 China VAT system evolution

China introduced a value-added tax in 1979. After 40 years of reform and development, VAT has become the largest type of tax, and China has established a modern VAT system that is in line with international standards and unique to the Chinese context. VAT revenue increased from 14.77 billion yuan in 1984 to 6,234.736 billion yuan in 2019, accounting for 7.24% of the country's total tax revenue in 2019, up from 7.24% in 1984.

The establishment and reform of China's VAT system can be divided into three stages: (1) Groping stage (1979–1993): In 1984, China officially proposed a value-added tax. China established a production-based VAT system and adopted the "tax included in price" during this period, which posed some challenges for the collection and management of VAT. For example, The important factors that affect the profit indicators of enterprises not only cause great pressure on enterprises but also bring unstable factors to the government's tax revenue. (2) Establishment stage (1994–2003): In 1994, China implemented a comprehensive reform of the VAT system, which laid the foundation of China's modern VAT system. First, China expanded the scope of taxation. The scope has been expanded to include sales, processing, repair and replacement services, etc., covering the basic links of commodity circulation. Second, China simplified the tax structure. In accordance with the principle that the actual tax burden was basically the same, China implemented a new tax rate structure. Finally, China reformed the tax calculation method. Specifically, there was a shift from "tax included in price" to "tax excluded in price", and taxpayers were divided into general taxpayers and small-scale taxpayers, and a relatively complete value-added tax system was gradually formed. (3) Transition stage(2004–2011): Under the production VAT, the taxes paid on the value of fixed assets (e.g., machines and equipment) were not deductible, resulting in the double taxation of fixed assets, which affected the efficiency of enterprise production and increased their tax burden. Therefore, starting in 2004, China introduced a pilot VAT reform, changing production VAT to consumption VAT¹. The pilot VAT reform was first conducted in eight major industries² in three provinces of Northeast China (i.e., Heilongjiang, Jilin, Liaoning).. After the pilot reform, the Chinese government implemented the newly revised "Provisional Regulations on Value-Added Tax" in 2008, which announced the decision to implement the VAT reform nationwide beginning on January 1, 2009. At that point, the VAT reform was fully rolled out across China. Figure 1 shows the three stages of China's taxation reform.

¹According to the different methods for deducting purchased fixed assets from the tax liability, VAT can be divided into three categories: production-type VAT, income-type VAT, and consumption-type VAT
²The eight major industries include the equipment manufacturing industry, the petrochemical industry, the metallurgy industry, the shipbuilding industry, the automobile industry, the agricultural products processing industry, the military production industry, and the high-tech product industry. Due to the complicated policy provisions for military production and high-tech products, it is difficult to define these industries accurately, so they are not considered in this paper.

3.1. Theory on the impact of a tax reform on technological sophistication

Purchased fixed assets and their depreciation are not deducted from the calculation of the production-type VAT, which not only results in redundant taxation and inhibits investment but also inhibits enterprise investment in high-tech processes and improvements in the technological sophistication of products. The reformed tax system introduced the consumption-type VAT, which allows companies to deduct the depreciation or amortization period of the fixed and intangible assets that they hold or that were newly purchased in the current depreciation period. That is, companies can increase the depreciation rate on their fixed assets, and the fixed assets purchased by the enterprises can be deducted from the tax base.

Innovation is a key determinant of product complexity (Hobday,1998). As a result of the VAT reform, net cash flows for companies increased, which facilitates improvements in enterprise technology, enabling firms to quickly introduce new technologies and equipment into their production and operation processes to increase the technical content of their products and improve their innovation capabilities and market competitiveness (Ding et al.,2021).

Consumption-type VAT can increase the intrinsic usefulness of goods so that the price of goods can accurately reflect the status of production within a country. Consumption-type VAT also enriches the types of products that are exported and increases their technical complexity, thus improving international competitiveness. Additionally, consumption VATs are conducive to adjusting the industrial structure, expanding the scale of assets, and accelerating the replacement of equipment (Aneja et al.,2021). Equipment upgrading improves firm technology, increases the share of high-tech products produced regionally, and ultimately promotes the upgrading of the industrial structure. Hypothesis H1 is proposed:

H1. The VAT reform increased the sophistication of enterprise exports.

3.2. Mechanism of influence and hypothesis

3.2.1. VAT reform and R&D

The VAT reform may have increased the sophistication of enterprise exports through its effects on innovation. Innovation from R&D is foundational for technological progress, which is usually achieved through complex industrial production processes. Financing constraints are an important factor affecting enterprise innovation and directly determine its sustainability. However, compared with the outcomes from fixed asset investment, the outcomes of innovative R&D investment are uncertain. Therefore, when companies lack funding, they are likely to experience financing constraints (Holmstrom, 1989). The VAT reform guaranteed an operating cash flow for enterprises, making it easier for enterprises to raise funds. Enterprises that are highly dependent on external financing have stronger incentives to innovate (Song et al., 2015). Additionally, the implementation of various policies eased the opportunity cost of capital among companies in the pilot industries (Howell,2016), reducing the cost of modifying equipment used in high-tech innovation. This also shows that for enterprises, the costs of production and innovation declined by a large amount, enabling firms to increase their investments in innovation and R&D, which in turn promotes enterprise innovation. Moreover, this creates a positive feedback loop; that is, companies apply R&D and innovative technological inventions to their business operations. The profits and cost savings created by these inventions encourage companies to engage in additional innovation activities.

In addition, innovation promotes research into and the development of products with high value-added, increases the heterogeneity among products, and increases the technical content of exported products (Fortunato and Razo,2014), thereby enhancing the international competitiveness of the exporting enterprises. In addition, the innovation behaviour of neighbouring enterprises in Northeast China generates knowledge spillovers. As a result, not only can enterprises that invest in innovation improve the sophistication of their export products by accumulating production experience and technology, but spatially adjacent enterprises can also absorb the knowledge and technology of the innovative enterprises at a lower cost due to the spillover effects. This absorption improves the product technology level (Arrow,1971). This can invigorate market competition and exacerbate spillover effects and can also increase the returns to industrial activities, improve the quality of export products and result in the coordinated development of various enterprises. Moreover, the internal innovation and coordinated development of various enterprises can also help improve the overall level of regional economic development, thereby increasing the sophistication of exported products (Wang and Yin,2020). Therefore, hypothesis H2a is developed as follows:

H2a. The VAT reform stimulated the innovation vitality of enterprises, thus improving the sophistication of enterprise exports by improving enterprise productivity and increasing the knowledge spillover effects.

3.2.2. The VAT reform and its labour market effects

This paper argues that the VAT reform had two effects on the employment behaviour of enterprises: a scale effect and a substitution effect. (1) The scale effect: A consumption VAT reduces the redundant taxation of enterprises, thereby reducing the capital user cost of fixed assets, guaranteeing an operating cash flow for the enterprise, and effectively ensuring the stability of the enterprise’s capital chain. A consumption VAT can stimulate large-scale production among enterprises and lead to industrial restructuring. In addition, smaller companies and companies with larger cash flows respond more strongly to tax credits (Zhang,2018), which should promote employment growth. (2) The substitution effect: Labour and capital are mutual substitutes, and the VAT reform created very powerful tax incentives for enterprises to invest in fixed assets, greatly increasing such investment among enterprises eligible for credit. An increase in the relative price of capital allows firms to reduce their input of labour. However, industrial enterprises require machinery and equipment to complete their production processes, meaning that there is little substitution between labour and capital. Compared with the scale effect, the substitution effect is small.

Therefore, due to the labour market effect, enterprises can expand the scale of their production, which can generate economies of scale and help increase the sophistication of enterprise exports. The production of highly technically complex export products requires immense investments in advanced equipment, new product development, marketing, etc., all of which generate very large costs (Johnson,2012), and exports are more costly than domestic sales in terms of logistics, business management, etc. In addition, there are risks arising from the uncertainty surrounding export policy. Economies of scale can create very large cost advantages for enterprises, enhance the competitiveness of enterprises and their ability to take risks, and create favourable conditions for the technical content of products to be improved. Moreover, highly productive enterprises are more motivated to improve the technical quality of their products and further increase their technological competitiveness (Atkinson, 2013). This in turn encourages specialized production and R&D and the upgrading of enterprise product structures so that enterprise investments can be made in capital- and technology-intensive industries (Ketels, 2017). Such changes to enterprise investment improve the technological level of the products produced by the enterprises and eventually promote an increase in export sophistication. Therefore, hypothesis H2b is stated as follows:

H2b. The VAT reform increased the scale of labour and improved the sophistication of enterprise exports through its effects on the labour market.

4.1. Research model

First, the probit model is used to test whether the VAT reform promotes the export behaviour of enterprises. Then, a DID model is adopted to study the effect of the VAT reform on the sophistication of enterprise exports. The probit model is given as follows:

$$\Pr (\exp or{t_{it}}=1)=\Phi ({\alpha _0}+{\alpha _1}VAT * year+{\alpha _2}{Z_{it}}+{e_{it}})$$ 1

where $\exp or{t_{it}}$ is a binary variable that indicates whether the enterprise is an exporter. If the company exports at least some products in year t, its value is 1; if the company does not export any products in year t, its value is 0. $VAT * year$ is a dummy variable. Specifically, the treatment group includes firms in the six pilot industries: equipment manufacturing, the petrochemical industry, metallurgy, shipbuilding, automobile manufacturing, and the agricultural processing industry. If an enterprise belongs to one of the six pilot industries, then $VAT$= 1; otherwise, $VAT$= 0. The "Regulations on Several Issues Concerning the Expansion of the Scope of the Value-Added Tax Deduction in the Northeast Region" was promulgated and implemented in 2004, so 2004 is defined as the "event year" in this paper. Relatedly, year is a dummy variable that is used to capture the policy effect. It equals one for 2004 and after and 0 otherwise. Moreover,${Z_{it}}$ indicates a vector of control variables such as enterprise size (the logarithm of the enterprise’s total assets), the assets-to-liabilities ratio (the ratio of the enterprise’s total liabilities to total assets), total factor productivity LP (calculated by using the LP method and taking the enterprise’s intermediate input as a proxy variable for productivity), the number of employees, the foreign capital share (the ratio of foreign capital to paid-in capital), and the total subsidies received by the firm (in logarithmic form).

In this paper, the VAT reform is regarded as a quasi-natural experiment and studied through the use of a DID model. The sample was divided into two groups, with enterprises in the pilot industries in northeast China defined as the treatment group and enterprises in the nonpilot industries in northeast China defined as the control group. After dividing the sample, we defined the following benchmark model:

$$fc{i_{it}}={\beta _0}+{\beta _1}(VAT * year)+{\beta _2}{X_{it}}+{\alpha _{it}}+{\varepsilon _{it}}$$ 2

where $VAT * year$ is the policy dummy variable and is the same as in model (1). ${X_{it}}$ represents the individual characteristics of enterprises, ${\alpha _{it}}$ represents regional, industry, and time trends used to test the robustness of the model, and${\varepsilon _{it}}$ is the error term. In addition to the VAT reform variables, the individual characteristics of each company, such as company size, the assets-to-liabilities ratio, the input of intermediate products, and employment costs, may be important factors affecting the sophistication of exports. Export sophistication depends largely on the production, operations, and export situation of each individual company, and these features are related to the characteristics of the company.

4.2. Export sophistication

Export sophistication can be used to measure the composition of products exported by a country or region, thus reflecting the technological level of the exported products. Scientifically measuring export sophistication is a prerequisite for studying the influence of any policy on that sophistication. Products can reflect the knowledge and capabilities of a country, and only when that country has all the necessary capabilities can any product be produced (Erkan,2015). On the basis of these ideas, Hidalgo et al. (2009) proposed the reflection method to generate a graph of the country–product network, which links the diversity of a country's exports with the universality of its products.

The essence of production is the combination of various capabilities, and technologically complex products are usually produced in countries whose production capacity is above a certain level (Hausmann and Hidalgo, 2010). The graph of a country–product network can be constructed based on the revealed comparative advantage index (RCA), which measures whether a country is an effective exporter. If a country effectively exports a product, it means that that country has the necessary capacity to produce that product, and so RCA > 1; otherwise, RCA < 1. National diversity ${K_{c,0}}$is defined as the number of types of products that country c exports with a relative comparative advantage, and product universality${K_{p,0}}$ is defined as the number of countries exporting the product p for RCA > 1. The two indicators are negatively related. It is generally believed that the higher the export diversity of a country is, the higher is the technical content of the exported products and the lower the universality of the product (Cristelli, 2013). RCA, ${K_{c,0}}$and ${K_{p,0}}$ are calculated as follows:

$$RC{A_{cp}}=\frac{{{{{X_{cp}}} \mathord{\left/ {\vphantom {{{X_{cp}}} {\sum {{X_{cp}}} }}} \right. \kern-0pt} {\sum {{X_{cp}}} }}}}{{{{\sum {{X_{cp}}} } \mathord{\left/ {\vphantom {{\sum {{X_{cp}}} } {\sum {\sum {{X_{cp}}} } }}} \right. \kern-0pt} {\sum {\sum {{X_{cp}}} } }}}}$$ 3

$${K_{c,0}}=\sum\nolimits_{p} {{M_{cp}}}$$ 4

$${K_{p,0}}=\sum\nolimits_{c} {{M_{cp}}}$$ 5

where c indexes countries, p indexes products, X represents the export volume, and ${M_{cp}}$ is a dummy variable that links the country and product trade information. It equals one if country c can produce product p and 0 if country c does not have the capacity to produce product p. National complexity and product complexity can be obtained by iterating the calculations for national diversity and product universality. Specifically, because ${K_{p,n}}$ involves information about diversification and ${K_{c,n - 1}}$ reflects the degree to which a country specializes in the production of diversified products, ${K_{p,n}}$ is more accurate than ${K_{c,n - 1}}$. When ${K_{c,n+2}}={K_{c,n}}$ and ${K_{p,n+2}}={K_{p,n}}$, the iteration stops, as all information has been obtained.

$${K_{c,n}}=\frac{1}{{{K_{c,0}}}}\sum\nolimits_{p} {{M_{cp}}} \times {K_{p,n - 1}}$$ 6

$${K_{p,n}}=\frac{1}{{{K_{p,0}}}}\sum\nolimits_{c} {{M_{cp}}} \times {K_{c,n - 1}}$$ 7

This article uses data from the Observatory of Economic Complexity database provided by the Growth Lab at Harvard University. International trade data for 5109 products with an HS 6-digit code ranging from 2000 to 2007 are used to construct the above indicators. The focus of this article is on standardized product complexity, indicated by pci. The greater the pci values are, the greater the requirements of the product in terms of technical production. pci values are calculated as follows:

$$pci=\frac{{{K_{p,n}} - K_{{p,n}}^{{mean}}}}{{K_{{p,n}}^{{s\tan darddeviation}}}}$$ 8

The reflection method introduces the concept of complexity on the country and product levels only, but the corporate capability endowments are still unknown. At the micro level, the enterprises within a country can be redefined as a large number of small economies, which allows us to infer the complexity of each enterprise from the information on the complexity of its export products. This is possible because all information about the universality of the products and the diversity of the exporting countries is obtained through reflection. On the one hand, product complexity can reflect technical characteristics such as specific inputs. On the other hand, product complexity also provides information about market conditions such as market competition. Enterprise-level information on exported products is used to calculate enterprise-level complexity. Therefore, given the previously proposed measure of product complexity, Guan's (2020) approach is used to calculate the complexity of each company’s export products in terms of a weighted average of the value of the company exports. In the equation below, $valu{e_{i,j}}$represents the value of enterprise i's export product j, and N represents the product categories for enterprise i's exports.

$$fc{i_i}=\sum {\frac{{valu{e_{i,j}}}}{{\sum\nolimits_{N} {valu{e_{i,j}}} }}} \times pc{i_j}$$ 9

4.3. Data and independent variables

4.3.1. Samples and data

In this article, six major industries are studied: the equipment manufacturing industry, the petrochemical industry, the metallurgy industry, the shipbuilding industry, the automobile industry, and the agricultural products processing industry. Table 2 shows the number of companies in each industry in the Northeast.

Table 2

Number of enterprises in each industry in Northeast China.
Total	Nonpilot industry	Pilot industry	Equipment manufacturing	Petrochemical industry	Metallurgy	Shipbuilding	Automobile manufacturing	Agricultural product processing industry
4809	680	4129	1099	619	208	34	174	2175

The study reported in this article is conducted at the enterprise level. Most of the data come from the 2000–2007 China Industrial Enterprise Database and the China Customs Database. The trade data needed to calculate the product complexity and the enterprise export complexity come from the United Nations Commodity Trade Statistics Database (UN COMTRADE). The China Industrial Enterprise Database covers all state-owned and privately owned industrial enterprises above a designated size. The observations from 2000 to 2007 account for most of the Chinese industrial enterprises. Before dividing the enterprises into those that are in a pilot industry and those in a nonpilot industry, we harmonized the four-digit industrial codes from the Industrial Enterprise Database for 2000–2002 with the national economic industry classification standard (GB/T 4754 − 2002) to ensure the continuity of the industrial classifications. We follow the processing method used for customs documents to process the data from the industrial enterprise database and the customs database.

Matching the data in the Chinese industrial enterprise database with that in the Chinese customs database is a key step in the empirical analysis. Because the company identifiers are different in each database, the Chinese name of the company and year of its founding are used to match observations between the two databases for the sake of accuracy. In other words, if a company has the exact same Chinese name in a specific year in both datasets, it is assumed that it is the same company in each dataset. The products exported by the merged companies account for 81% of all HS 6-digit products. After completing the above processing, we finally obtain an unbalanced panel of 4809 export companies in the three northeastern provinces. The maximum period of observation for a given company is 8 years, and there are a total of 15021 observations.

We draw a line chart to illustrate the direct impact of the VAT reform on the effective VAT rate for the enterprises. The calculation formula for the effective VAT rate is as follows: effective VAT rate = current VAT payable/main business income. Figure 2 and Fig. 3 show the changes in the effective VAT rate for the nonpilot industry companies and the pilot industry companies that are exposed to the VAT reform in the Northeast. From 2000–2007, the effective VAT rate for nonpilot industry enterprises was significantly higher than that for pilot industry enterprises. For pilot companies, the VAT reform policy was implemented in 2004. Near 2004, the slope of the broken line for the pilot companies was steeper than that for nonpilot companies, indicating that the implementation of the policy lightened the tax burden. If the VAT reform policy and other events had no impact on the firms in the nonpilot industries, then the line for the nonpilot industries should not be consistent with that for the pilot industries. The result in Fig. 2 may be due to the adjustment of industry codes from 2000 to 2002, as the adjustment may have affected the codes for high-tech industry enterprises, and so during the categorization of the pilot enterprises, some pilot enterprises may have been omitted, resulting in a drop in the effective VAT rate.

4.3.2. Variables

The explained variable is export sophistication, $fci$. In this paper, the reflection method is used to calculate the sophistication of exports for each enterprise, and after dividing by the standard deviation, the result is logarithmically transformed. The core explanatory variable is the interaction of the policy dummy variable with the time dummy variable, $VAT * year$.

A series of control variables that represent the individual characteristics of the enterprise and various macro factors are included. The control variables for the individual enterprise characteristics include enterprise size (size), the assets-to-liabilities ratio (debt), inputs of intermediate products (mid), employment costs (wage), the output efficiency of labour (apl), the output efficiency of capital (apk), enterprise age (age) and the corporate profit rate (profit). Enterprise size (size) is expressed as the logarithm of the total assets of the enterprise. The assets-to-liabilities ratio (debt) is expressed by dividing the total liabilities of the enterprise by the total assets. Enterprises with low debt ratios have a higher credit rating. Inputs of intermediate products (mid) are determined by the ratio of the enterprise’s industrial intermediate inputs to the total value of its industrial output. The employment cost (wage) is the ratio of the total wages payable by the enterprise to the number of employees at the end of the year and represents the cost of labour as an enterprise production factors. The output efficiency of labour (apl) and of capital (apk) is represented by the ratio of industrial value-added to the number employees at the end of the year and to the annual average balance of net fixed assets, respectively, which represent the output efficiency of the enterprise’s production factors. Enterprise age (age) is calculated as the current year minus the year of establishment plus 1. The company’s profit margin (profit) is expressed by dividing the company’s operating profit by the sales revenue from its main business products. Macro factors include dummy variables for the eight years of observation, i.e., from 2000 to 2007 (time fixed effects) and 38 dummy variables for industries at the two-digit level according to the national economic industry classification code (industry fixed effects). In addition, dummy variables for Liaoning, Heilongjiang, and Jilin Provinces in the Northeast (regional fixed effects) are also included. The summary statistics are reported in Table 3.

Table 3

Descriptive statistics for the main variables.
Variable	N	Mean	Standard deviation	Minimum	Maximum
fci	15,020	-1.0306	0.5430	-7.9893	0
year	15,021	0.6747	0.4685	0	1
VAT	15,021	0.8591	0.3480	0	1
VAT*year	15,021	0.5833	0.4930	0	1
size	15,021	10.3690	1.5634	7.4955	14.9128
debt	15,021	0.5522	0.2941	0.0152	1.5239
mid	15,007	0.7377	0.3439	0	23.2195
wage	7,628	16.8496	19.2835	522	0.4667
apl	7,628	130.9294	385.3536	-3046.5400	14121
apk	12,767	4.2848	68.0524	-220.2410	7084.3330
age	15,015	10.1826	10.5166	1	64
profit	15,009	-0.6758	80.6899	-9881	4.0040

5.1. Baseline results

The marginal effect of the VAT reform on exports as estimated with a probit model (Eq. (1) ), as shown in Table 4. The results show that for both state-owned enterprises and non-state-owned enterprises, the average probability of enterprises in the Northeast region exporting products increased significantly after the reform. Moreover, the magnitude of the increase in the probability of exporting products was higher for SOEs than for non-SOEs.

Table 4

Results of the probit model.
Export	SOEs		Non-SOEs
	(1)	(2)	(3)	(4)
VAT*year	0.229^***	0.421^***	0.0276^***	0.291^***
	(0.0354)	(0.1163)	(0.0093)	(0.0442)
Controls	No	Yes	No	Yes
Obs	18096	2595	100881	10959

Note: Standard errors are reported in parentheses. ***, **, and * denote the 1%, 5%, and 10% significance levels, respectively.

Next, in this paper, the DID model (Eq. (2)), is used to evaluate the effect of the policy on export sophistication. The preliminary regression results are shown in Table 5. Each column reports results that include controls for firm and time fixed effects. Column (1) does not add any other control variables. Columns (1) and (3) additionally control for industry fixed effects. Column (4) includes regional fixed effects along with the firm and time fixed effects. The key explanatory variable is the interaction term between the dummy variable for the VAT reform and the implementation year, $VAT * year$. As indicated in Table 5, the coefficients on the interaction term are all significantly positive, which indicates that the VAT reform has indeed increased the sophistication of enterprise exports. This is probably because the reduction in the effective VAT rate prompted enterprises to increase their capital and R&D investments, thereby increasing the technical content of their exported products. In columns (3) and (4), which include both industrial and regional fixed effects, the coefficient on the interaction term is larger than that in column (2), indicating that the differences between industries and between provinces explain part of the impact of the VAT reform on the sophistication enterprise exports.

Table 5

Results for the DID model.
Variables	(1)	(2)	(3)	(4)
VAT*year	0.0192^*	0.0367^**	0.0455^***	0.0466^***
	(0.0109)	(0.0164)	(0.0174)	(0.0176)
Controls	-	Yes	Yes	Yes
Firm fixed effects	Yes	Yes	Yes	Yes
Year fixed effects	Yes	Yes	Yes	Yes
Industry fixed effects	Yes	-	Yes	-
Region fixed effects	-	-	-	Yes
Obs	15020	7607	7607	7607

Note: Standard errors are reported in parentheses. ***, **, and * denote the 1%, 5%, and 10% significance levels, respectively.

5.2. Considering the income tax policy: DDD

In addition to the VAT reform, an income tax policy was also implemented as part of the Northeast Revitalization Strategy. The corporate income tax policy was directed at all companies in the Northeast region and has three main components: increasing the depreciation rate for fixed assets, shortening the amortization period for intangible assets, and increasing the pretax standard deduction for taxable wages, which indirectly reduced the tax burden on enterprises. Compared with the VAT reform, the income tax policy targeted a wider range of enterprises. At present, most of the existing literature has studied the VAT reform only, ignoring the impact of the income tax policy implemented during the same period on corporate behaviour. Therefore, the effect of the VAT reform is likely overestimated, which reduces the reliability of any policy evaluations.

The income tax policy and other policies included in the Northeast Revitalization Strategy were wide-ranging and may have promoted the development of enterprises across all industries in the Northeast region. The increase in the sophistication of enterprise exports in the pilot industries may be due to the VAT reform itself or may be due to other Northeast revitalization policies. The abovementioned DID method controls only for industrial and regional differences within and across the three northeastern provinces and cannot account for the impact of other tax policies on the sophistication of enterprise exports. Therefore, a DDD model is constructed to address the potential nonrandom selection of policy pilot areas. In this way, the positive effect of the VAT reform on the sophistication of enterprise exports can be further verified.

To resolve this problem, we selected six central provinces for comparison with the provinces in the northeast region for the following reasons: (1) At the beginning of the founding of the People's Republic of China, all of these provinces had a strong industrial base. The northeastern provinces and the six central provinces share certain similarities: the old industrial base cities have similar characteristics, and the share of old industrial base cities in the six central provinces is similar to that in the three northeastern provinces. Moreover, the GDP rankings for the six central provinces and the three northeastern provinces were similar from 2000 to 2007. (2) Although the State Council implemented the policy entitled "Several Opinions on Promoting the Rise of the Central Region" in 2006, the timing of the implementation of that policy does not coincide with the VAT reform, and it did not include any specific tax policy. Therefore, it is unlikely that the central region was affected by other macroregional development policies from 1999 to 2007.

For these reasons, six provinces in the central region are selected as the control group for the DDD model and are compared with the three northeastern provinces. At the industry level, enterprises in the six pilot industries specified in the VAT reform are compared with the enterprises in all other industries, and at the time level, observations from before and after 2004 are compared. The triple-differences model constructed in this paper is as follows:

$$\begin{gathered} fc{i_{it}}={\beta _0}+{\beta _1}(VAT * year * region)+{\beta _2}(VAT * year) \hfill \\ +{\beta _3}(year * region)+{\beta _4}(VAT * region)+{\beta _5}{X_{it}}+{\alpha _{it}}+{\mu _{it}} \hfill \\ \end{gathered}$$

where $VAT$ and $year$ represent the industry dummy variable and the time dummy variable, respectively and $region$ signifies the region dummy variable. It equals one if the company is located in one of the three provinces in Northeast China and zero if the company is located in one of the six provinces in Central China. ${\alpha _{it}}$ represents fixed effects at the province, time, and industry levels. The definitions of the other variables are the same as those in model (2).

The main focus of this article is the coefficient on the triple interaction term, ${\beta _1}$. This coefficient is the DDD estimate, which represents the net effect of the VAT reform. According to Table 6, the coefficients on the triple interaction term are all positively significant, which is consistent with the DID results. This regression eliminates interference from other policies such as the income tax policy included in the Northeast Revitalization Strategy, further demonstrating that the VAT reform improved the sophistication of enterprise exports.

Table 6

Results of the DDD model.
Variables	(1)	(2)	(3)	(4)
VATyearregion	0.0488^***	0.0728^***	0.0735^***	0.0788^***
	(0.0140)	(0.0227)	(0.0233)	(0.0237)
Controls	-	Yes	Yes	Yes
Firm fixed effects	Yes	Yes	Yes	Yes
Year fixed effects	Yes	Yes	Yes	Yes
Industry fixed effects	Yes	-	Yes	-
Region fixed effects	-	-	-	Yes
Obs	29359	15006	15006	15006

Note: Standard errors are reported in parentheses. ***, **, and * denote the 1%, 5%, and 10% significance levels, respectively.

5.3. Placebo test

Given that the innovation capabilities and technological capacity of the enterprises in the six major industries studied in this paper may be greater than those of enterprises in other industries, the enterprises in the six major industries may have been more willing to increase the complexity of their exports. For this reason, a placebo test is conducted with only the observations from 2000 to 2003. The years 2002–2003 are chosen as the pseudo treatment dates, with 2000–2001 used as the preperiod. If the estimated coefficients on the fictitious policy variables are not significant, this means that apart from the impact of the reform, there is no systematic difference in the trends in export sophistication between the treatment group and the control group. The results show that the coefficients on the interaction term are not significantly negative, which is in line with expectations. This result indicates that the effect estimated by the benchmark model is reliable and that the model passes the placebo tests.

Table 7

Results of the placebo test.
Variables	(1)	(2)	(3)	(4)
VAT*year	-0.0213	-0.0207	-0.0112	-0.0110
	(0.0162)	(0.0305)	(0.0320)	(0.0320)
Controls	-	Yes	Yes	Yes
Firm fixed effects	Yes	Yes	Yes	Yes
Year fixed effects	Yes	Yes	Yes	Yes
Industry fixed effects	Yes	-	Yes	-
Region fixed effects	-	-	-	Yes
Obs	4886	2364	2364	2364

Note: Standard errors are reported in parentheses. ***, **, and * denote the 1%, 5%, and 10% significance levels, respectively.

5.4. Considering product types

In addition, an alternative method for calculating the explained variable is used as a robustness test. The new variable is calculated as follows: the sophistication of the enterprise exports $fci2$ is calculated by dividing the standardized product complexity index $pci$ by the number of different product types exported by the enterprise. The resulting value is standardized.

$$fci{2_i}=\frac{{pc{i_j}}}{N}$$

The results in Table 8 show that the coefficients on the interaction term are all significantly positive, indicating that the use of an alternative calculation method for the explained variable does not affect the significance of the regression results, although the estimated coefficients are not as significant as those from the benchmark regression. This finding proves that our main conclusion is robust; that is, the VAT reform did indeed increase the sophistication of enterprise exports.

Table 8

Results for an alternatively calculated explained variable.
Variables	(1)	(2)	(3)	(4)
VAT*year	0.0199^*	0.0341^**	0.0338^**	0.0324^**
	(0.0103)	(0.0149)	(0.0158)	(0.0159)
Controls	-	Yes	Yes	Yes
Firm fixed effects	Yes	Yes	Yes	Yes
Year fixed effects	Yes	Yes	Yes	Yes
Industry fixed effects	Yes	-	Yes	-
Region fixed effects	-	-	-	Yes
Obs	15020	7604	7604	7604

Note: Standard errors are reported in parentheses. ***, **, and * denote the 1%, 5%, and 10% significance levels, respectively.

5.5. Accounting only for the time effect: Single difference

Last, we adopt the traditional estimation method for determining the effect of the VAT reform on the sophistication of enterprise exports; that is, we add only the time dummy variable to the model to analyse and account for the influence of the reform on the control group. The results in Table 9 show that the coefficients on the year dummy are all positive and significant, which is highly consistent with the DID regression results. Compared with the coefficient from the baseline model, the coefficient from the single-difference model is larger. It is likely that the estimated coefficient is capturing the fact that the treatment group was not only affected by the VAT reform but also positively affected by other Northeast revitalization policies, indicating that the single-difference method overestimates the true effect. The effect of the VAT reform on the sophistication of enterprise exports is verified, and we have proven that the single-difference method provides a biased evaluation of the effect of the policy.

Table 9

Results for Single difference.
Variables	(1)	(2)	(3)	(4)
year	0.120^***	0.118^***	0.119^***	0.120^***
	(0.0084)	(0.0126)	(0.0128)	(0.0128)
Controls	Yes	Yes	Yes	Yes
Firm fixed effects	Yes	Yes	Yes	Yes
Year fixed effects	Yes	-	Yes	-
Industry fixed effects	-	-	-	Yes
Region fixed effects	Yes	Yes	Yes	Yes
Obs	12903	6533	6533	6533

To examine the differences in the effect of the VAT reform among the various companies affected by the policy, we further conducted a heterogeneity analysis by considering the differences between enterprises, such as the ownership and the intensity with which they use industry factors.

6.1. Different ownership

We divide the enterprises into two groups on the basis of their ownership: state-owned enterprises and non-state-owned enterprises. The group state-owned enterprises includes state-owned enterprises broadly defined, i.e., state-owned enterprises, wholly state-owned companies, and state-owned joint ventures. The nonstate-owned enterprises group includes collective enterprises, joint-stock cooperative enterprises, private enterprises, and other enterprises that are not state-owned enterprises. Table 10 shows that the effect of the policy on the sophistication of the exports of state-owned enterprises is not significant. For non-state-owned enterprises, the results are consistent with the results of the full-sample regression: the VAT reform improves the sophistication of the exports of non-state-owned enterprises.

Table 10

Results of the analysis by enterprise ownership.
Variables	SOEs			Non-SOEs
	(1)	(2)	(3)	(4)	(5)	(6)
VAT*year	0.0429	0.0329	0.0316	0.0354^**	0.0442^***	0.0450^***
	(0.0559)	(0.0580)	(0.0597)	(0.0153)	(0.0164)	(0.0166)
Controls	Yes	Yes	Yes	Yes	Yes	Yes
Firm fixed effects	Yes	Yes	Yes	Yes	Yes	Yes
Year fixed effects	Yes	Yes	Yes	Yes	Yes	Yes
Industry fixed effects	-	Yes	-	-	Yes	-
Region fixed effects	-	-	Yes	-	-	Yes
Obs	296	296	296	7311	7311	7311

Note: Standard errors are reported in parentheses. ***, **, and * denote the 1%, 5%, and 10% significance levels, respectively.

The VAT reform was part of the Northeast Revitalization Strategy. The key recipients of support under the Northeast Revitalization Strategy were state-owned enterprises, which may have weakened the enthusiasm of state-owned enterprises for development (Ren et al.,2020). In this context, it was easier for state-owned enterprises to obtain bank loans and finance large-scale investment projects through special treasury bonds, so they were relatively less subject to financial constraints. Therefore, the state-owned enterprises did not innovate and failed to improve their product quality and technical complexity. In contrast, non-state-owned enterprises can operate flexibly, have wide scopes of operation, and have relatively complete incentives to innovate. In addition, most private enterprises cannot obtain government assistance when they experience financial difficulties due to their weak credit ratings. Therefore, they actively seized the opportunity provided by the policy to transform and upgrade the technical content of their products and fully utilize the role of the market.

This result is a further expansion of the regression results from the probit model, presented earlier in the article. The reform reduced the tax burden on enterprises and increased the exports of state-owned and non-state-owned enterprises. However, due to the focus of the policy, the different types of enterprises had different motivations for development, which explains the difference in the impact of the policy on export sophistication.

6.2. Different factor intensities

We further studied whether differences in the factor intensity of the industries in which the enterprises are located generated differences in the effect of the policy. Following Xu and Zhang (2008), the two industries with the most missing data from among the 37 two-digit industries in the Chinese industrial database, namely, handicrafts and other manufacturing industries and water production and supply industries, are dropped from the analysis. Next, based on UNCTAD (2002), the 35 industrial sectors are divided into primary product and labour-intensive sectors, medium and low-tech sectors, and high-tech sectors³ by technological ability, factor intensity, and enterprise scale.

Table 11

Results of the analysis by industry factor intensity.
Variables	Primary product and labour-intensive sectors		Medium and low-tech sectors		High-tech sectors
	(1)	(2)	(3)	(4)	(5)	(6)
VAT*year	0.0767^***	0.0788^***	0.0039	0.0043	0.107^***	0.107^***
	(0.0260)	(0.0262)	(0.0219)	(0.0219)	(0.0239)	(0.0240)
Controls	Yes	Yes	Yes	Yes	Yes	Yes
Firm fixed effects	Yes	Yes	Yes	Yes	Yes	Yes
Year fixed effects	Yes	Yes	Yes	Yes	Yes	Yes
Industry fixed effects	Yes	-	Yes	-	Yes	-
Region fixed effects	-	Yes	-	Yes	-	Yes
Obs	3983	3983	2301	2301	1119	1119

Note: Standard errors are reported in parentheses. ***, **, and * denote the 1%, 5%, and 10% significance levels, respectively.

The regression results are shown in Table 11. It is obvious that the results for the primary product and labour-intensive sectors and for the high-tech sectors are highly positive, indicating that the VAT reform effectively increased the sophistication of enterprise exports in these two groups of sectors, while the sophistication of companies in the medium- and low-tech sectors was not significantly impacted. The abovementioned heterogeneity may be because the primary product and labour-intensive sectors are more dependent on labour. Because of the low cost of labour, these firms have a substantial advantage when exporting labour-intensive products. However, due to the small amount of added value from their production technology, such firms may face more intense competition on the international market when exporting. The implementation of the VAT reform may have encouraged such capital-nonintensive companies to invest more in fixed assets in order to receive policy dividends (Yu and Qi, 2022), to stimulate the vitality of their corporate technological innovation, and to increase the sophistication of their product exports to gain an advantage in international competition. Firms that make large fixed asset purchases under the consumption VAT benefited the most from the VAT reform. Firms in the high-tech sectors are highly dependent on advanced machinery and equipment and focus more on equipment upgrades and R&D investment, which further increases the complexity of their exported products. The VAT reform reduced the cost of using fixed assets and thus increased the use of fixed assets. This in turn increased the complexity of firm exports. However, in the low- and medium-tech sectors, fixed assets are not used frequently, and market competition is lower. Therefore, firms in those industries have fewer incentives to upgrade their equipment and export products.

6.3. The effect of the reform on innovation and the labour market

Finally, to test hypotheses H2a and H2b, the DID model is used to conduct a regression analysis of two potential mechanisms: innovation and the labour market. The models are as follows:

$$r{d_{it}}={\beta _0}+{\beta _1}(VAT * year)+{\beta _2}{X_{it}}+{\alpha _{it}}+{\varepsilon _{it}}$$

$$employmen{t_{it}}={\beta _0}+{\beta _1}(VAT * year)+{\beta _2}{X_{it}}+{\alpha _{it}}+{\varepsilon _{it}}$$

$$employmmen{t_{it}}/capita{l_{it}}={\beta _0}+{\beta _1}(VAT * year)+{\beta _2}{X_{it}}+{\alpha _{it}}+{\varepsilon _{it}}$$

where $r{d_{it}}$is the R&D investment of the enterprise, which is calculated by dividing the R&D expenses of the enterprise by the sales revenue from its main business products. This measure represents the innovation induced by the VAT reform. The variable$employmen{t_{it}}$is measured as the logarithm of the number of employees at the end of the year. It represents the labour market scale effect of the reform. The variable $employmmen{t_{it}}/capita{l_{it}}$ represents the assets owned per labourer and is calculated by taking the logarithm of the ratio of the average annual balance of net fixed assets to the number of employees at the end of the year. This variable represents the substitution effect. The remaining variables are consistent with those in the benchmark regression model, and the regression results are shown in Table 12.

Table 12

Results of the mechanism analysis.
Variables	Innovation Effect		Labour Market Effect
	R&D		Employment		Employment/capital
VAT*year	0.0008^**	0.0007^**	0.669^*	1.019^***	0.274^***	0.208^***
	(0.0003)	(0.0003)	(0.3665)	(0.2590)	(4.5627)	(5.1385)
Controls		Yes		Yes		Yes
Firm fixed effects	Yes	Yes	Yes	Yes	Yes	Yes
Year fixed effects	Yes	Yes	Yes	Yes	Yes	Yes
Obs	7912	5243	14988	7594	13517	6135

Note: Standard errors are reported in parentheses. ***, **, and * denote the 1%, 5%, and 10% significance levels, respectively.

The results show that the policy encouraged and motivated enterprise to invest in R&D. Enterprises used the funds saved from the VAT reform to further increase their investment in product R&D, to intelligently transform their R&D, to increase the monetary incentives for R&D personnel, and finally improve their production technology. We have verified H2a. Moreover, the reform indeed caused a significant increase in the number of employed people. The scale effect is believed to be greater than the substitution effect, and the VAT reform promoted an increase in the amount of capital possessed per labourer. H2b holds. The above results prove that the VAT reform did increase the sophistication of enterprise exports through its effects on innovation and the labour market.

³The primary product and labour-intensive sectors include coal mining and washing, oil and gas mining, ferrous metal mining and dressing, nonferrous metal mining and dressing, nonmetallic mining and dressing, other mining, agricultural and sideline food processing, food manufacturing, beverage manufacturing, tobacco products, textiles, clothing, footwear and hat manufacturing, leather, fur, feathers (velvet) and related products, wood and bamboo processing, rattan and palm grass products, furniture manufacturing, paper and paper products, petroleum processing, coking and nuclear fuel processing, chemical fibre manufacturing, nonmetallic mineral products, and nonferrous metal smelting and rolling . The medium and low-tech sectors include printing and the reproduction of recording media, the manufacturing of cultural, educational and sporting goods, rubber products, ferrous metal smelting and rolling, metal products, general equipment manufacturing, special equipment manufacturing, transportation equipment manufacturing, electrical machinery and equipment manufacturing, electricity and heat production and supply. The high-tech sector includes chemical raw materials and chemical products manufacturing, pharmaceutical manufacturing, plastic products, communications equipment, computers and other electronic equipment manufacturing, and instrumentation and cultural office machinery manufacturing.

To mobilize the enthusiasm of enterprises for production and eliminate the adverse effects of redundant taxation, the Chinese government launched an important tax system reform in 2004, changing from a production-oriented VAT to a consumption-oriented VAT, thus encouraging enterprises to accelerate their upgrades and to invest in new equipment. This paper studies the first pilot area for the policy—Northeast China— and uses enterprise-level panel data from 2000 to 2007 to study the impact of the 2004 VAT reform pilot on the sophistication of enterprise exports. We also discuss the mechanisms underlying the reform’s impact.

First, we find that the implementation of the VAT reform encouraged enterprises in Northeast China to export and that it played a significant role in increasing the sophistication of enterprise exports. The regression results are also robust when excluding other policy influences during the same period, forwarding policy time, replacing explanatory variables, and only considering time effects. Second, under the VAT reform, the sophistication of the exports of non-state-owned enterprises significantly increased, and the influence of reform on the export sophistication is found to be insignificant not only on state-owned enterprises but also on medium and low-tech sectors’ enterprises. Third, the mechanism analysis shows that the VAT reform stimulated the vitality of enterprise innovation and increased investment in R&D, thereby increasing the sophistication of exported products. On the other hand, the VAT reform increased the number of labourers employed by enterprises and improved the sophistication of enterprise exports through its effect on the labour market.

Based on the theoretical analysis and empirical results in this paper, we make the following suggestions: on the one hand, innovation should be encouraged. In terms of system innovation, state-owned enterprises should shatter the rigid thinking within their management and stimulate enthusiasm for production and creation among the enterprise talent through strategies such as target and material incentives. In terms of technological innovation, enterprises should make full use of their technical and personnel resources. It is possible to implement an apprenticeship system through school–enterprise cooperation and build industry–university research centres to train talent. Meanwhile, cross-domain and cross-regional innovation cooperation should be established among enterprises to expand the spillovers from technology and transition away from a system of single export products. On the other hand, the business environment should be optimized. It is important that the decisive role of the market in allocating resources be utilized. The government of the Northeast region should delegate power and appropriately reduce taxes, thus attracting participation from additional social capital. Additionally, the government can accelerate the establishment of government–enterprise service platforms and transition into a modern service government supported by the people. Finally, relevant policies regarding intellectual property rights and the awareness of independent competition among northeastern enterprises should be strengthened, thus creating a level playing field.

We declare that we have no financial and personal relationships with other people or organizations that can inappropriately influence our work, there is no professional or other personal interest of any nature or kind in any product, service and/or company that could be construed as influencing the position presented in, or the review of, the manuscript entitled “Journal Of Asian Economics”.

Acknowledgments

This study was supported by the National Natural Science Foundation of China (Grant No.72003018), the Humanities and Social Sciences Youth Foundation, the Ministry of Education of the People's Republic of China (Grant No.20YJC790037), and the Postdoctoral Science Foundation of China (Grant No.2021M690505).

Adhikari B. Does a Value‐Added Tax Increase Economic Efficiency?[J]. Economic Inquiry, 2020, 58(1): 496-517.
Yang Y, Zhang H. The value-added tax reform and labor market outcomes: Firm-level evidence from China[J]. China Economic Review, 2021, 69: 101678.
Peng F, Wang L, Shi X, et al. Does value-added tax reform in services accelerate enterprise upgrading? Micro evidence from China[J]. Applied Economics, 2022, 54(9): 1093-1109.
Hidalgo C A, Hausmann R. The building blocks of economic complexity[J]. Proceedings of the national academy of sciences, 2009, 106(26): 10570-10575.
Dale Jorgenson. Capital Theory and Investment Behavior [J].American Economic Review, 1963:247-259.
Wang, Dehua. The impact of the 2009 value-added tax reform on enterprise investment and employment‐Empirical analysis based on Chinese tax survey data. Maastricht University, 2013.
Auerbach, Hassett K, Sodersten J. Taxation and Corporate Investment: The Impact of the 1991 Swedish Tax Reform [J].Swedish Economic Policy Review,1995,2(2):361-383.
Da Rin, Marco, Giovanna Nicodano, and Alessandro Sembenelli. "Public policy and the creation of active venture capital markets." Journal of Public Economics 90.8-9 (2006): 1699-1723.
Liu, Yongzheng, and Jie Mao. "How do tax incentives affect investment and productivity? Firm-level evidence from China." American Economic Journal: Economic Policy 11.3 (2019): 261-91.
Hutton J P, Ruocco A. Tax reform and employment in Europe[J]. International Tax and Public Finance, 1999, 6(3): 263-287.
Nie H, Fang M, Li T. China’s value-added tax reform, firm behavior and performance[J]. Frontiers of Economics in China, 2010, 5(3): 445-463.
Chirinko R S. Corporate taxation, capital formation, and the substitution elasticity between labor and capital[J]. National Tax Journal, 2002, 55(2): 339-355.
Lora E, Fajardo-González J. Employment and taxes in Latin America: An empirical study of the effects of payroll, corporate income and value-added taxes on labor outcomes[J]. Cuadernos de Economía, 2016, 35(67): 75-117.
Garrett D G, Ohrn E, Suárez Serrato J C. Tax policy and local labor market behavior[J]. American Economic Review: Insights, 2020, 2(1): 83-100.
Hobday M. Product complexity, innovation and industrial organisation[J]. Research policy, 1998, 26(6): 689-710.
Ding K, Xu H, Yang R. Taxation and enterprise innovation: evidence from China’s value-added tax reform[J]. Sustainability, 2021, 13(10): 5700.
Aneja A, Kulkarni N, Ritadhi S K. Consumption Tax Reform and the Real Economy: Evidence From India's Adoption of a Value‐Added Tax[J]. Journal of Empirical Legal Studies, 2021, 18(3): 569-602.
Holmstrom B. Agency costs and innovation[J]. Journal of Economic Behavior & Organization, 1989, 12(3): 305-327.
Song M, Ai H, Li X. Political connections, financing constraints, and the optimization of innovation efficiency among China's private enterprises[J]. Technological Forecasting and Social Change, 2015, 92: 290-299.
Howell A. Firm R&D, innovation and easing financial constraints in China: Does corporate tax reform matter?[J]. Research Policy, 2016, 45(10): 1996-2007.
Fortunato P, Razo C. Export sophistication, growth and the middle-income trap[J]. Transforming Economies–Making industrial policy work for growth, jobs and development, 2014: 267-287.
Arrow K J. The economic implications of learning by doing[M]. Readings in the Theory of Growth. Palgrave Macmillan, London, 1971: 131-149.
Wang T, Yin X. Study on the determinants of export sophistication of China’s manufacturing subdivided sectors[M]. Recent trends in decision science and management. Springer, Singapore, 2020: 769-778.
Zhang L, Chen Y, He Z. The effect of investment tax incentives: Evidence from China’s value-added tax reform[J]. International tax and public finance, 2018, 25(4): 913-945.
Johnson, R.C. Trade and Prices with Heterogeneous Firms[J]. Journal of International Economics,2012,86(1):43-56.
Atkinson R D. Competitiveness, innovation and productivity[J]. The Information Technology & Innovation Foundation.–August, 2013: 2-7.
Ketels C. Structural Transformation: A competitiveness-based view[J]. Handbook on Structural Transformation, Oxford University Press: Oxford, 2017.
Hidalgo, C.A., Hausmann, R., Dasgupta, P.S., 2009. The building blocks of economic complexity. Proceed. Natl. Acad. Sci. U. S. Am. 106, 10570–10575.
Guan S, Cheng L W. Does product complexity matter for firms' TFP? [J]. Technological Forecasting and Social Change, 2020,167.
Qi S, Zhou C, Li K, et al. The impact of a carbon trading pilot policy on the low-carbon international competitiveness of industry in China: An empirical analysis based on a DDD model[J]. Journal of Cleaner Production, 2021, 281: 125361.
Ren W, Xue B, Yang J, et al. Effects of the Northeast China revitalization strategy on regional economic growth and social development[J]. Chinese Geographical Science, 2020, 30(5): 791-809.
Xu Yi, Zhang Erzhen. Outsourcing and Productivity: An Empirical Study Based on Industrial Industry Data[J]. Economic Research,2008(01):103-113.
UNCTAD.Trade and Development Report, Annex 1 to Chapter Ⅲ: "Growth and Classification of World Merchandise Exports”,2002.
Yu J, Qi Y. BT-to-VAT reform and firm productivity: Evidence from a quasi-experiment in China[J]. China Economic Review, 2022, 71: 101740.

No competing interests reported.

Download PDF

Version 1

posted

You are reading this latest preprint version

Does value-added tax reform increase the sophistication of enterprise exports? Evidence from China

Status:

Version 1

Abstract

Figures

1. Introduction

2. International Vat Model And China Vat Reform

2.1 International VAT model comparison: EU model and New Zealand model

2.2 China VAT system evolution

3. Hypothesis

3.1. Theory on the impact of a tax reform on technological sophistication

3.2. Mechanism of influence and hypothesis

3.2.1. VAT reform and R&D

3.2.2. The VAT reform and its labour market effects

4. Method

4.1. Research model

4.2. Export sophistication

4.3. Data and independent variables

4.3.1. Samples and data

4.3.2. Variables

5. Analysis And Results

5.1. Baseline results

5.2. Considering the income tax policy: DDD

5.3. Placebo test

5.4. Considering product types

5.5. Accounting only for the time effect: Single difference

6. Further Discussion

6.1. Different ownership

6.2. Different factor intensities

6.3. The effect of the reform on innovation and the labour market

7. Conclusions

Declarations

Acknowledgments

References

Additional Declarations

Status:

Version 1