Tax regulation is a way for the state to participate in the process of national income distribution by means of taxation, and to regulate the economic interests of all parties in a compulsory and gratuitous way, so as to regulate the social and economic life. It affects production and circulation by regulating consumer demand, and is characterized by a wide range of regulation. Taxation can not only penetrate into the various fields and links of social reproduction, but also taxation can reach all sectors and units of the national economy, from the state down to enterprises, so as to realize the regulation of the speed and direction of economic operation. Tax regulation can also be purposeful to an object for special regulation, set up different taxes, tax items, tax rates to distinguish between different tax objects, plus adjusting the tax rate, tax exemptions, or additive levies to regulate the way. At the same time, the tax is based on the law, has the mandatory, similarly, the tax regulation also has the unquestionable authority. Tax regulation regulates the relationship between supply and demand and the income level by changing the tax burden and tax rate structure. The implementation of the VAT additional deduction policy can realize tax and fee reduction, encourage enterprises to invest the refunded amount in R&D and production, accelerate transformation and upgrading, and improve the vitality of enterprises.

Schumpeter believes that innovation is the fundamental driving force of economic development, which not only refers to technological innovation, but also includes the new combination of production factors and production conditions ( De Sao Pedro Filho, Lima, Moreira Da Silva Neto, Andre Da Silva Muller, & Batista Da Costa, 2017 ). This new combination can be introduced into the production system and form a new production function, thus promoting the development of the economy. With the development of global economic integration and the continuous progress of society, innovation occupies a core position in the economic development of all countries, and is taken as the basic national policy to promote the economic development of all countries. As an important indicator to measure the economic development of a country, all countries in the world need to systemize the corresponding innovation strategy and build a reasonable institutional structure to vigorously develop their own innovation activities, so as to promote the further development of the economy. In the chain of enterprise innovation activities, government tax incentives are also essential.

First, the VAT additional deduction policy directly reduces the actual VAT paid by enterprises by increasing their input VAT, thus lowering their tax burden, saving their tax cash outflows, and increasing their free cash flow. Research has shown that corporate cash flow significantly promotes innovation. Studies have found ( Cincera, Ravet, & Veugelers, 2016 ) that adequate cash flow can reduce the risk of R&D, thereby positively impacting corporate R&D innovation. The additional cash flow brought by the VAT additional deduction policy makes enterprises more willing to invest in innovation activities that benefit their long-term growth and maximize corporate value. The increased cash flow can substitute for some high-cost external financing, serving as internal financing to reduce the overall capital cost of enterprises. Alleviating financing constraints reduces the investment risk of innovation activities, making the management more inclined to decide on pursuing innovation activities ( Li, Liu, Jia, & Ma, 2024 ; Wang, Li, & Wang, 2022 ; Zhao & Wang, 2022 ).

Secondly, some scholars believe ( Li & Zhang, 2009 ) that according to information transmission theory, when the government conveys favorable information about the overall development trend of enterprises to the market, market investors will shift their investment decisions toward increased investment. The implementation of the VAT additional deduction policy in 2019 positioned the production and life services sectors benefiting from the policy as favorable prospects in the eyes of market investors. Therefore, to continue reaping the benefits of the policy, enterprises will pay more attention to their market competitiveness, which will drive them to intensify their innovation efforts.

Based on the above analysis, this paper proposes the following hypothesis:

H1: The VAT additional deduction policy significantly promotes innovation output in the production and life services sectors.

This paper evaluates the impact of the VAT additional deduction policy on innovation in the production and life services sectors by analyzing changes in innovation output of listed companies before and after the policy’s implementation. The sample selection follows these steps: Industry classification is based on the “Annotations on Sales of Services, Intangible Assets, and Real Estate” (Cai Shui [2016] No. 36), with the productive services sector including postal services, telecommunications services, and modern service industries. The life services sector includes cultural and sports services, education and medical services, catering, and accommodation services. This study selects listed companies that are not simultaneously affected by the additional deduction policy and changes in VAT rates, using listed financial companies as the control group.

The sales behavior of most service sector enterprises in the first five months of 2016 was subject to business tax, not the first full year of full VAT collection in the service sector, and more relevant data are missing in 2023, so this paper adopts the data of enterprises in 2017-2022 as the experimental interval. The data for listed companies mainly come from the Cathay Pacific database (abbreviation CSMAR) and the iFinD enterprise database, covering the period from 2017 to 2022, and are processed and analyzed using STATA. To ensure data accuracy and sample representativeness, the raw data were processed as follows: 1) Samples with missing key variables were excluded; 2) Companies that went bankrupt before 2022 and those listed after 2017 were excluded; 3) Companies with ST or *ST status during the study period were excluded. Due to a small amount of missing data during the experimental period for some firms, in order to retain more observations and improve the precision of the model estimation results, this paper uses unbalanced panel data for regression. Ultimately, 444 research samples and 2411 observations were obtained.

The paper treats the implementation of the VAT additional deduction policy as an event shock and uses the difference-in-differences method to evaluate and analyze the policy’s impact on corporate innovation output. To accurately estimate the policy’s effects, control variables such as asset-liability ratio were introduced, and the basic regression model constructed is as follows.

$lnpaten{t}_{it}=\alpha +{\beta }_{0}T{P}_{it}+{\beta }_1{X}_{it}+{\gamma }_i+{\tau }_t+{\epsilon }_{it}$ (1)

Here, the subscript $i$ denotes the enterprise and $t$ denotes the year. The dependent variable $lnpaten{t}_{it}$ measures the level of corporate innovation output. ${\beta }_0$ is the coefficient of primary interest, indicating the effect of the additional deduction policy on corporate innovation output. Its sign and significance reflect whether and how the additional deduction policy impacts corporate R&D output. $X_{it}$ represents control variables, with ${\beta }_1$ as their regression coefficient. ${\gamma }_i$ and ${\tau }_t$ denote individual and year fixed effects, respectively, while ${\epsilon }_{it}$ is the random error term.

Patents are important outcomes and direct manifestations of innovation. The number of patents is often used in academia to measure corporate innovation output. Given the long approval time for patents, patent grants are somewhat lagged. Therefore, this study uses the number of patent applications to measure innovation output. Patent application types include invention patents, utility model patents, and design patents. Considering that invention and utility model patents have higher technical content and better represent corporate innovation levels, this study uses the natural logarithm of the total number of utility model and invention patent applications plus one, following Q. Li, Ma and Shevlin ( Li, Ma, & Shevlin, 2021 ).

The explanatory variable is the VAT additional deduction policy (TP), essentially the interaction of the dummy variables for time (Policy) and sample (Treat). The dummy variable for the enterprise sample is Treat, which equals 1 if the enterprise belongs to the experimental group, otherwise 0. The time dummy variable is Policy, which equals 1 after the VAT additional deduction policy implementation, otherwise 0.

The definitions of the control variables are detailed in Table 1 .

After the data of variables were aggregated and filtered, we obtained data from 444 listed companies in the production and life services sectors, resulting in a total of 2411 observations. The descriptive statistics are shown in Table 2 . From these results, it is evident that from 2017 to 2022, the innovation output (lnpatent) of Chinese listed companies in the sample had a minimum value of 0 and a maximum value of 7.04, with an average innovation output of 2.41. This indicates that the overall level of innovation among Chinese listed companies is relatively low, demonstrating insufficient innovation motivation. Regarding control variables, the standard deviations of the return on total assets, net cash flow from operating activities, and total asset growth rate all exceed their means, indicating significant differences in profitability, internal financing capacity, and growth rates among the sample enterprises. Overall, the descriptive statistics of the variables are consistent with expectations.

Table 3 presents the baseline regression results. Column (1) shows the regression analysis without control variables, where the coefficient of TP is positive and significant at the 1% level, indicating that the additional deduction policy has a significant positive impact on corporate innovation output. Column (2) includes financial characteristic control variables in the econometric model, and the coefficient of TP remains significantly positive at the 1% level. Column (3) includes both financial characteristic and corporate governance control variables, and the coefficient of TP remains significantly positive at the 1% level, confirming the stable positive impact of the additional deduction policy on corporate innovation output. Since the implementation of the VAT additional deduction policy pilot in 2019, the innovation output of treatment group enterprises has increased by approximately 11.3% compared to the control group enterprise(0.274/2.417 × 100%), indicating that the incentive effect of the VAT

additional deduction policy on corporate innovation output is also economically significant. In summary, the baseline regression empirical results support Hypothesis H1, i.e., the VAT additional deduction policy significantly promotes innovation output in the production and life services sectors.

Given the significant differences in financing resources and innovation motivations between different ownership types, the impact of the VAT additional deduction policy on innovation output may vary. Referring to the classifications of some scholars ( Yu, Shen, & Jiang, 2022 ), we divided the sample into state-owned and non-state-owned enterprises based on the nature of the actual controller, then conducted subsample regressions. Columns (1) and (2) of Table 4 show the regression results. The policy variable coefficient for non-state-owned enterprises is much larger than that for state-owned enterprises, indicating that the incentive effect on innovation output is more significant for non-state-owned enterprises under the VAT additional deduction policy. The estimated coefficient for non-state-owned enterprises is 0.306 and significant at the 1% level, while the impact for state-owned enterprises is only 0.199 and significant at the 10% level. State-owned enterprises generally have better financing conditions and more abundant collateralizable assets, so the tax reduction brought by the VAT additional deduction policy has a smaller impact on the relatively ample internal cash flow of state-owned enterprises. Compared to state-owned enterprises, non-state-owned enterprises have disadvantages in collateral and financing resources, face financing difficulties and high financing costs, and thus the promotion effect of the additional deduction policy on innovation output is greater at the margin.

Scholars believe ( Lee & Pati, 2017 ) that industry differences also affect the policy implementation effects, with different industries facing varying policy environments, input combinations, and R&D processes. Therefore, it is necessary to study the heterogeneity of the VAT additional deduction policy implementation effects from an industry perspective. Referring to the high-tech industry classifications issued by the National Bureau of Statistics (manufacturing and services), we categorized industries such as chemical raw materials and chemical products manufacturing, railway, shipbuilding, aerospace, and other transportation equipment manufacturing, electrical machinery and equipment manufacturing, computer, communications, and other electronic equipment manufacturing, instrumentation manufacturing, telecommunications, radio, television, and satellite transmission services, internet and related services, software and information technology services, research and experimental development, and professional technical services as high-tech industries, with others being non-high-tech industries, and conducted regressions separately. Columns (1) and (2) of Table 5 show the regression results, where the policy effect for high-tech industries is 0.237, higher than 0.211 for non-high-tech industries, indicating that the VAT additional deduction policy has a more significant promotion effect on innovation output in high-tech industries. Compared to non-high-tech industries, high-tech industries are more willing to innovate due to multiple factors, reflecting their unique advantages in technological innovation, market demand, policy support, and industry characteristics. High-tech enterprises usually have professional technical teams and R&D institutions, possessing strong capabilities in technology development and transformation. These teams can stay at the forefront of technology, continually exploring new technologies and applications, providing strong support for corporate innovation. Governments around the world generally prioritize the development of high-tech industries by formulating various policy measures to support corporate innovation activities. These policies include tax incentives, funding subsidies, and R&D support,

providing strong guarantees for high-tech enterprises’ innovation. High-tech products usually have high added value, bringing higher profits to enterprises, prompting them to increase R&D investment and engage in technological innovation, leading to more innovation output.

According to the research ( Chiang, Lee, & Anandarajan, 2013 ), enterprises at different lifecycle stages show significant differences in R&D capabilities, financing situations, investment strategies, and innovation needs. Therefore, the VAT additional deduction policy may have heterogeneous effects on innovation output at different corporate lifecycle stages. Using the median age of the sample enterprises as the standard, we divided them into growth-stage and mature-stage enterprises. Columns (1) and (2) of Table 6 show the subsample regression results, indicating that the incentive effect of the VAT additional deduction policy on innovation output is stronger for growth-stage enterprises. The regression coefficient for mature-stage enterprises in column (1) is 0.226 and significant at the 5% level, while the coefficient for growth-stage enterprises in column (2) is 0.306 and significant at the 1% level. This means that the implementation of the additional deduction policy has a greater impact on growth-stage enterprises compared to mature-stage enterprises, which aligns with the above expectations.

To use the Difference-in-Differences (DID) model, it is necessary that the treatment and control groups exhibit the same development trends before the policy intervention. Therefore, we need to verify whether the parallel trend assumption holds to ensure robust results. Figure 1 illustrates the parallel trend test results for the VAT reform policy on the performance of production and life services enterprises. Before the policy implementation (t = 2019), the differences in innovation output between the treatment and control groups were not significant, indicating that both groups followed the same development trend prior to the policy implementation. Thus, the conditions for using the DID model are met.

To verify the impact of other random factors on the empirical results, we conducted a placebo test by randomly selecting 123 samples from the 444 sample companies as a “pseudo-treatment group” and repeated this selection process 500 times for regression analysis. Figure 2 shows the distribution of policy regression coefficients, where the points are concentrated around 0, suggesting no significant omitted variables in the experimental model setup. Additionally,

using the 2019 additional deduction policy as the explanatory variable and corporate innovation output as the dependent variable, the distribution of regression coefficients approximately follows a normal distribution, indicating that the promoting effect of the 2019 VAT additional deduction policy on innovation output in the production and life services sectors is relatively robust.