VAT Neutrality and Corporate Cash Holdings —Based on the Research of Uncredited VAT Refund Policy

1 Introduction

The fiscal and tax systems, as the financial basis and policy instruments for high-quality economic growth, not only directly motivate microeconomic entities through top-level design, but also play a vital role by optimizing taxation and business environment. According to the Doing Business 2020 released by the World Bank, China was among the top 10 global improvers of business environment for two years in a row, albeit with great space for improvement in tax indicators. VAT is the largest tax in China. The principle of tax neutrality for VAT requires to avoid distortion of market economic behaviors as much as possible in the taxation process, reduce the excessive tax burden of enterprises, and improve the allocation of market resources. However, under the current VAT shifting law and the design of the deductible range, the gradual backlog of tax credits not only brings financial difficulties to enterprises, undermines production and operation, but also distorts the principle of tax neutrality, which is not conducive to optimizing tax business environment (Ni et al., 2019). To further lighten the burden on business entities and optimize the business environment, the Ministry of Finance and State Tax Administration jointly issued the Notice on Relevant Tax Policies on Uncredited VAT Refund for Certain Sectors in 2018, listing 18 major categories of industries eligible for refunding the closing balance of retained VAT credit in advanced manufacturing sectors like equipment manufacturing, modern service industry like research and development, power grid enterprises, among others. Compared with the ex-post subsidies under the previous preferential tax policies, the uncredited VAT refund policy smoothed the refund channel, thereby improving the corporate cash flow and the VAT neutrality, and even effectively avoiding the risk of “loses by choice” and “institutional arbitrage” by enterprises (Liu et al., 2020). In addition, as an important means of optimizing the business environment, the policy may further reduce capital costs, improve market prospects and optimize business expectations. Has the implementation of the policy effectively reduced the occupation of corporate funds by distorted VAT neutrality? This paper answers this question based on a study from the perspective of corporate cash holdings.

Cash is a key current asset for an enterprise to maintain operations and development, address future risks, repay due debts, and fulfill tax obligations. Cash holdings constitute an integral part of corporate financial decisions. A heavy comprehensive tax burden and a harsh tax environment affect financial conditions and market expectations, while the potential tax pressure could distort the cash holding behavior of enterprises (McGuire et al., 2014). On the one hand, the high tax credits occupy corporate funds, which increases the capital cost and deteriorates the corporate’s financial situation. On the other hand, the market expectations of non-refundable tax credits prompt companies to limit investment and hold a large amount of cash, thereby further distorting the allocation of resources. According to CSMAR data (as shown in Figure 1), ^[1] the mean value of tax credits of listed companies and the ratio of companies with tax credits have increased year by year from 2007 to 2020, with the total amount of tax credits in 2020 approximating RMB 600 billion. Meanwhile, the corporate cash holding ratio has remained elevated and the mean value of corporate cash holdings has kept increasing.^[2] It can be seen that the large backlog of funds at both ends of the government and enterprises may severely distort the allocation of financial resources, which does not meet the requirements of reducing the burden on entity enterprises and revitalizing the market economy in the stage of high-quality development. Therefore, an in-depth study into the impact of uncredited VAT refund policy on corporate cash holdings in line with the policy of “implementing tax and fee cuts to boost market development” is of great theoretical and practical significance to improve the fiscal, tax and financial systems, better the allocation of financial resources, and revitalize corporate development during the 14th Five-Year Plan period.

Figure 1

Cash Holdings and Tax Credits of Listed Companies (2007–2020)

In this context, this paper regards the uncredited VAT refund policy promulgated in 2018 as an exogenous shock to improve the neutrality for VAT. Based on the statistics of listed companies in the non-financial sectors from 2013 to 2020, this paper applies the difference-in-differences (DID) approach to study the impact of uncredited VAT refund policy on corporate cash holdings. The study contributes to the existing research in the following aspects. First, from the perspective of corporate cash holdings, it examines the “resource effect” and “signal effect” of uncredited VAT refund policy on corporate cashing holdings. Based on the samples of advanced manufacturing industry and pilot R&D companies, this paper explains the mystery of high corporate cash holdings, which enriches the literature on corporate cash holdings in China with theoretical contents and empirical data. Second, by associating the policy at the macro level with corporate cash holdings at the micro level, this paper verifies the significant influence of the policy on corporate cash holdings based on the tax neutrality theory, thereby providing new ideas for further optimizing VAT neutrality and devising forward-looking fiscal and tax systems. Third, the phased evaluation of the effects of the policy provides practical experience for reference to apply the policy to all industries and improve the VAT system under the new development pattern during the 14th Five-Year Plan period. Meanwhile, it sends positive policy signals for further improving the allocation of market resources and invigorating market players.

2 Theoretical Analysis and Research Hypotheses

In China, the companies hold cash mainly out of motivations for transaction, prevention and agency business (Wang et al., 2020). In theory, the uncredited VAT refund policy can help reduce the cash holdings of pilot enterprises. For a start, the policy smoothed the channel for refunding VAT credits. Compared with those companies that carry forward the excess input tax, the companies with refunded VAT credits could fully activate the idle funds occupied by VAT credits, thereby increasing the corporate cash flow and reducing capital costs. Meanwhile, the optimized capital conditions may reduce the companies’ motivation to hold cash for transaction. Next, the policy can improve the companies’ expectations for tax burden and financial risks, while optimized fiscal and tax environment may directly improve the market business environment, thereby reducing the companies’ motivation for prevention to hold cash. Lastly, despite the policy’s limited influence on corporate governance, the eased tax burden, better capital conditions, and optimized market may prompt companies to reduce cash holdings due to a high agency cost. In this context, the first hypothesis is put forward.

Hypothesis 1 (H1): The uncredited VAT refund policy is conducive to reducing cash holdings of pilot enterprises.

The smooth refund channel is an important prerequisite for giving play to the tax neutrality (Wu et al., 2021). For one thing, the uncredited VAT refund policy reduces the distortion of resource allocation incurred by non-neutral tax by refunding tax credits, prompting companies to lower cash holdings for transactions. This is the “resource effect” of the policy. For another, given the uncredited VAT refund policy, companies can improve their expectations for VAT neutrality and tax business environment, thereby lowering cash holdings for preventive motivation. This is the “signal effect” of the policy.

As a result of the “resource effect”, the tax endowment of corporate cash holdings may be effectively improved. For a start, a high amount of tax credits occupies the capital and increases tax burden of companies. In this context, the uncredited VAT refund policy directly activates the idle funds and eases the financing constraints for companies. It’s expected that under the influence of the policy, companies with higher amount of tax credits and more financing constraints can adjust cash holdings to a more notable extent. Next, as the VAT is a turnover tax, the effect of increasing operating cash flow generated by the reform to replace business tax with VAT will emerge gradually, and the ability to shift tax burden will exert an increasingly prominent impact on the operating cash flow (Qiao and Chen, 2017). Compared with companies boasting pricing power and great capability of shifting tax burden to upper- and lower-reach manufacturers, the companies with weak shifting capabilities can reduce cash holdings more notably. Third, unlike the ex-post subsidies and tax incentives based on production profits and pre-tax profits, enterprises have difficult access to substantial tax incentives (Liu et al., 2020). The uncredited VAT refund policy has a greater impact on the companies with low profitability and weak competitiveness.

In terms of the “signal effect”, the sustained uncredited VAT refund policy can effectively improve the tax environment for businesses. Compared with regions with weak tax collection and low tax burden, the policy exerts a greater impact on the corporate cash holdings in regions with heavier tax burden. On the other hand, the policy releases the signals of “implementing tax and fee cuts to boost market development”, which plays an important role in reducing environmental uncertainty and stabilizing business expectations. The policy is expected to effectively alleviate the distortion of market economic behavior caused by non-neutral VAT credits which are non-refundable and exert a greater impact on the cash holdings of companies featuring high environmental uncertainty and large operating fluctuation.

In addition, compared with enterprises with cash holdings at an “average” level, the uncredited VAT refund policy may produce both “resource effect” and “signal effect” to exert a greater impact on the enterprises featuring limited resource endowments and extremely high cash holdings, along with those featuring well-developed operation model and extremely low cash holdings. It gives rise to the Hypothesis 2:

Hypothesis 2 (H2): The uncredited VAT refund policy produces both “resource effect” and “signal effect” on corporate cash holdings. The “resource effect” brings about effective improvement in terms of the differences in tax endowments of corporate cash holdings. The “signal effect” could effectively improve the expectations about market prospect of corporate cash holdings. The two effects could function together to exert a greater impact on pilot enterprises with extremely high or low cash holdings.

Given that companies’ investment and financing decisions are affected by varied development characteristics at different phases of their lifecycle, the impact of uncredited VAT refund policy on corporate cash holdings varies at different phases of lifecycle due to the factors of “resource effect” and “signal effect”. Theoretically, companies in the growth phase are characterized by greater financing constraints, higher capital expenditure, and limited profitability, making them less risk-resistant and difficult to enjoy substantive tax incentives, coupled with a large amount of tax credits (Liu et al., 2020). The uncredited VAT refund policy not only optimizes companies’ financial conditions through direct refund, but also raises market expectations by improving fiscal and tax environment. Compared with companies in the growth phase, the companies in the maturity phase boast well-developed business model and stable profitability. In this context, the refunded tax credits and promising expectations may further lower their motivations to hold cash. Unlike companies in the growth or mature phase, those companies in the decline phase face decreasing sales and market shares, financial difficulties, threats of delisting, and risks of mergers & acquisitions. For these companies, the policy may alleviate their financial struggles. In this context, the Hypothesis 3 is put forward:

Hypothesis 3 (H3): Compared with the companies in the decline phase, the uncredited VAT refund policy exerts a greater impact on the cash holdings of companies in the growth or maturity phase.

The uncredited VAT refund policy constitutes an integral part of the measures to optimize the tax environment for businesses against the backdrop of “tax and fee cuts”. A favorable tax environment is of great significance for invigorating the microeconomic entities and enhancing the taxpayers’ sense of fulfillment (Liu and Fan, 2019). Due to the “signal effect”, the policy produces varied impact on the cash holdings of taxpayers at different levels of credit ratings. The taxpayers rated at Level A or B for tax credit will be eligible for refund of end-of-period tax credits. Among the pilot enterprises, those with level-A tax credit rating boast better financial conditions, and thus the refunded tax credits may further improve its financial conditions and reduce its cash holdings. On the other hand, the better market expectations due to the policy also help the enterprises with level-A tax credit ratings reduce their cash holdings. In this context, Hypothesis 4 is put forward.

Hypothesis 4 (H4): Compared with enterprises at other levels of tax credit rating, the uncredited VAT refund policy exerts a larger impact on the cash holdings of enterprises with level-A tax credit rating.

3 Research Design

4 Empirical Rest Results

4.3 Parallel Trend, Placebo, and Robustness Test

4.3.1 Parallel Trend Test

With a view to further testing the parallel trend before the implementation of the uncredited VAT refund policy and the possible time lag of the policy, we take inspiration from Wu et al. (2021). Taking 2017, one year prior to the introduction of the policy, as the benchmark, the multiplication items generated by treat and year in each year during the sample period is added in the Formula a (1), shown as follows:

(3)  cashhold  i , t = β 0 + ∑ t = 2013 t = 2020 , t ≠ 2017 β t  treat  i ×  year  t + ∑  controls  i , t + ∑  firm  i + ∑  year  t + ε i , t

Wherein, β_t is the key coefficient of concern, and the others are the same as in the Formula (1). As shown on the left side of Figure 2, the estimated values of the coefficient β_t before the policy implementation are not significant, in consistent with the assumption about parallel trend. The regression coefficients in the years of and after the policy implementation significantly reduced corporate cash holdings. In addition, the statistical significance and economic significance of the policy effect gradually decreased over time.

Figure 2

Parallel Trend Test and Placebo Test

4.3.2 Placebo Test

Taking the Column (2) in Table 4 as the benchmark, based on the year of implementing the uncredited VAT refund policy, a corresponding number of enterprises are randomly selected to join the treatment group for indirect analysis. The coefficient estimates are:

(4) β 1 ∗ = β 1 + θ cov  treat  i ×  year  t , ε i , t ∣ ∑  controls  i , t var  treat  i ×  year  t ∣ ∑  controls  i , t

Wherein, controls represent all observable control variables. If θ is 0, an unbiased estimate is obtained. As shown on the right side of Figure 2, after taking 500 random samples to simulate the random impact of the uncredited VAT refund policy, it can be found that the randomly distributed estimated values are concentrated around 0, far above the benchmark coefficient estimate of –0.009. By means of inverse method, θ equals 0, indicating that other unobserved random factors do not affect the basic conclusion.

4.3.3 Robustness Test

Replacement of explained variables and core explanatory variables. With reference to Faulkender (2002) and Yang and Yin (2018), the explained variables are standardized by total assets, non-cash assets, and sales revenue respectively, and adjusted through year-industry median value and standard deviation, as shown in Columns (1)~(8) in Table 3. As for the core explanatory variables, the policy timing is based on the unit of month, and the calculations are based on the actual months of policy implementation. The variable treat equals 1/2 in 2018 and 1 in other years.

Multi-dimensional control of missing variables. First, control the impact of the actual VAT burden and the amount of tax credits to be deducted.^[1] Second, control the differences in the tax environment for businesses at the provincial level. Third, control the high-order joint fixed effects of “time × industry” and “time × province”. Fourth, adjust potential influencing factors like heteroscedasticity and autocorrelation by industry clustering, time, and company-level dual clustering.

The observed values may have problems related to time series. First, to alleviate the problem of reverse causality in regression, all explanatory variables are processed using one-phase lag. Second, all sample observations are averaged before and after policy implementation for two-period DID estimations.

Subsample regression. First, given the high cash holdings of start-ups, the balanced panel estimation is adopted. Second, since the uncredited VAT refund policy was implemented in June 2018, the year 2019 is set as the policy timing and the samples in 2018 are deleted, in a bid to exclude the impact prior to the policy implementation in June 2018. Third, given the impact of the reform “to replace business tax with VAT” and the industry-wide application of the uncredited VAT refund policy, 2016–2020 and 2016–2019 are selected as the sample periods for analysis. Fourth, given the accelerated depreciation of fixed assets in the same period and weighted deduction of R&D expenses, the samples of enterprises in the manufacturing industry are selected for analysis, while excluding the samples of high-tech enterprises.

Self-selection sampling test. First, PSM-DID estimation with 1:3 nearest neighbor matching is applied to solve the selection bias of observable variables and ease the impact of time-varying and non-time-varying unobserved variables. Second, the method of entropy balancing—DID is used for estimation.

The results of the robustness test do not change the core conclusion that “the uncredited VAT refund policy reduces corporate cash holdings”.^[1]

5 Extended Research

5.1 Heterogeneity Test

5.1.1 Heterogeneity Analysis from the Dimension of “Resource Effect”

In the analysis, samples are grouped by the scale of retained tax credit, the financing constraints, the tax burden shifting, and the market competitiveness,^[2] with the results shown in Panel A of Table 5. It can be seen that in terms of the scale of retained tax credit and the financing constraints, enterprises with a larger amount of retained tax credits and tighter financing constraints reduced cash holdings more significantly, reflecting better corporate capital conditions as a result of improving VAT refund channel and activating idle capital. In terms of tax burden shifting and the market competitiveness, enterprises with poorer tax burden shifting and market competitiveness reduced cash holdings more notably, indicating that the incentive of uncredited VAT refund policy could ease the tax burden for enterprises more effectively.

Table 5

Heterogeneity Test

	Panel A: Tests for direct resource effect
	Scale of credits retained tax		Financing constraints		Tax burden shifting		competitiveness Market
Variable	(1)	(2)	(3)	(4)	(5)	(6)	(7)	(8)
	High	Low	High	Low	High	Low	High	Low
treat×post	−0.007* (−1.936)	−0.007 (−1.540)	−0.016*** (−3.883)	0.001 (0.394)	−0.004 (−1.125)	−0.010** (−2.342)	−0.000 (−0.006)	−0.018*** (−3.811)
cons	−0.047 (−0.396)	0.469*** (4.071)	0.354*** (3.327)	0.065 (0.590)	0.198** (2.037)	0.193* (1.832)	−0.045 (−0.426)	0.377*** (2.884)
Control variable	Y	Y	Y	Y	Y	Y	Y	Y
Company/Year	Y	Y	Y	Y	Y	Y	Y	Y
within R2	0.192	0.235	0.254	0.191	0.219	0.244	0.200	0.252
N	8189	8488	13192	8341	9881	11652	10617	10916
Test dif(P)	−0.001(0.312)		−0.017***(0.000)		−0.006**(0.022)		0.018***(0.002)
Panel B: Tests for indirect signal effect
	Tax administration		Environmental uncertainty		Environmental dynamics		Environmental diversity
Variable	(1)	(2)	(3)	(4)	(5)	(6)	(7)	(8)
	Strong	Weak	High	Low	High	Low	High	Low
treat×post	−0.010** (−2.486)	−0.004 (−1.056)	−0.012*** (−2.984)	−0.002 (−0.579)	−0.010*** (−2.637)	−0.003 (−0.712)	−0.006* (−1.784)	−0.004 (−0.875)
cons	0.192 (1.558)	0.187** (2.227)	0.315*** (3.745)	−0.111 (−0.988)	0.275*** (3.076)	−0.074 (−0.677)	0.171* (1.774)	0.131 (1.013)
Control variable	Y	Y	Y	Y	Y	Y	Y	Y
Company/Year	Y	Y	Y	Y	Y	Y	Y	Y
within R2	0.223	0.234	0.231	0.244	0.228	0.245	0.252	0.213
N	9258	12275	12818	8715	13074	8459	13074	8447
Test dif(P)	−0.007***(0.002)		−0.010***(0.000)		−0.008*(0.060)		−0.003(0.206)

5.1.2 Heterogeneity Analysis from the Dimension of “Signal Effect”

Grouped tests are performed in terms of the tax administration of at the regional level and the uncertainty of business environment at the corporate level, with the results shown in Panel B of Table 5.^[1] It can be seen that enterprises subject to tougher tax administration, higher uncertainty of business environment, more dynamic environment, and more diverse environment reduced cash holdings more significantly. It reflects that with the implementation of the uncredited VAT refund policy, enterprises would raise their expectations for VAT neutrality and tax environment, thereby reducing cash holdings out of precautionary motives.

5.1.3 The Impact of the Uncredited VAT Refund Policy on Extreme Corporate Cash Holdings

In view of the right-skewed thick-tailed distribution of corporate cash holdings, as shown on the left part of Figure 3, quantile regression is used to ease the setting limitations for general mean regression, in a bid to explore the impact of the uncredited VAT refund policy on the tail characteristics of the distribution of corporate cash holdings. Specifically, it’s shown as follows:

Figure 3

The Distribution of Cash Holding Ratios and the Impact of the Uncredited VAT Refund Policy on Extreme Corporate Cash Holdings

(5)  Quant  θ  cashhold  i , t ∣  treat  i ×  post  t ,  et. al.  ≡  inf   cashhold  i , t : F i , t  cashhold  i , t ∣  treat  i ×  post  t ,  et. al.  = θ = β 0 ( θ ) + β 1 ( θ )  treat  i ×  post  t + ∑  controls  i , t + ∑  firm  i + ∑  year  t + ε i , t

The above formula shows the conditional quantile of corporate cash holdings. The function with θ as coefficient of explanatory variable changes with the quantile θ. Fit is the function of conditional distributions. θ represents the quantile level of the conditional distribution of explained variables, with the value range of [0, 1]. The others are the same as in Formula (1).

The right part of Figure 3 shows the impact of the uncredited VAT refund policy on the 1%~99% quantile of corporate cash holdings. It can be seen that the policy’s overall impact coefficient shows a significant downward trend, which is consistent with the above descriptions. However, the degree of influence presents an “inverted U-shaped” nonlinear trend with the quantile change. Compared with enterprises with relatively “average” amount of cash holdings, the policy exerts a greater impact on enterprises with extremely high or low cash holdings. A possible explanation is that since the enterprises with extremely high cash holdings feature limited ability for financing and risk resistance, the “resource effect” of the policy could significantly improve their financial conditions, prompting them to reduce cash holdings. Since the enterprises with extremely high cash holdings boast well-developed production and operation model and strong capability for risk resistance, the “signal effect” of the policy could greatly improve their market expectations, prompting them to reduce cash holdings. In this way, the hypothesis H2 is verified.

6 Conclusions and Policy Recommendations

This paper regards the uncredited VAT refund policy issued in 2018 as an exogenous shock to improve the neutrality for VAT. Based on the statistics of listed companies in the non-financial sectors from 2013 to 2020, this paper applies the DID approach to study the impact of the policy on corporate cash holdings. The study finds that the uncredited VAT refund policy has eased the impact of non-neutral tax credits that cannot be refunded and lowered the corporate cash holdings. Furthermore, the analysis of capital sources verifies that the policy has optimized the corporate capital conditions. These findings have proved to be robust after a series of tests. At the level of cross-sectional differences, the uncredited VAT refund policy exerts both “resource effect” and “signal effect”, enabling companies to reduce cash holdings in terms of both tax endowment and market expectation. Meanwhile, the effects are more significant on enterprises with extreme levels of cash holdings. At the level of policy effects, the policy has improved the corporate behavior of high cash holdings for the production and operation cycle, with more notable impact on enterprises with higher tax credit ratings. At the level of economic consequences, the adjustment of corporate cash holdings induced by the policy has significantly improved business performance and reduced performance fluctuations, with a greater impact on enterprises with high capital intensity. Based on these findings, the following recommendations are proposed.

For a start, it’s necessary to keep advancing and deepening the reform of the uncredited VAT refund system. Currently, the uncredited VAT refund policy has effectively improved the effect of non-neutral VAT. Meanwhile, a smooth VAT refund channel exerts a positive impact on lightening tax burdens, stabilizing market expectations and improving business performance of enterprises. Nonetheless, there still exist the problems like small coverage, rigorous criteria for eligibility, and limited amount of refund. Therefore, in the future reform, it’s advisable to focus on expanding the scope, lowering the threshold and increasing the ratio of tax credits for refund, so as to invigorate enterprises and achieve high-quality development.

Next, it’s necessary to provide reasonable guidance to help optimize the tax environment for businesses. The all-round application of the uncredited VAT refund policy is bound to impose new requirements on tax administration and anti-tax avoidance. Potential problems such as yet-to-be-improved refund efficiency, false invoices, and tax fraud will not only damage the tax environment for businesses and affect companies’ market expectations, but also result in a massive loss of VAT for the government and disrupt the market order of taxation. Therefore, strengthening the supervision of tax refund processes at the institutional and incentive levels, and integrating tax credit with tax refund system have significant policy implications for improving the uncredited VAT refund policy, tax compliance, and tax environment under the new development pattern.

Finally, it’s necessary to give play to the coordinated role of finance and tax, in a bid to perfect the modern fiscal and tax systems. Against the backdrop of “implementing tax and fee cuts to boost market development”, the large-scale refund of VAT credits will increase the fiscal burden, thereby forcing a reform of the fiscal system. On the other hand, the reform of fiscal system could give better play to the role of tax cuts. Given the domino effect of the uncredited VAT refund policy as a systematic program, an overall planning should be made for the top-level design. Specifically, the policy’s heterogeneo us impact on enterprises should be taken into account before adopting targeted measures on key industries and areas at both incremental and stock levels. Meanwhile, the stress should be laid on the integrity of the fiscal and tax systems, so as to avoid potential complex tax system and institutional costs induced by fragmented reforms.