مشوق های مالیاتی و فعالیت R & D : شواهد از سطح شرکتی در تایوان

This paper investigates the effect of tax incentives on R&D activities in Taiwanese manufacturing firms. The propensity score matching (PSM) estimates show that recipients of R&D tax credits appear on average to have 53.80% higher R&D expenditures than that they do without receiving tax credits, while there is no significantly higher growth rate of R&D expenditure. This study further employs the panel instrumental variable (IV) and generalized method of moment (GMM) techniques to control for endogeneity of R&D tax credits and firm heterogeneity in determining R&D expenditure. The R&D tax credit is witnessed to exhibit a significantly positive influence on R&D expenditure and its growth, especially for electronics firms. The marginal effect is moderate, ranging from 0.094 to 0.120. Specifically, the R&D elasticity concerning tax credits tends to increase gradually along with the approaching expiration of R&D tax credits measure, lending a supportive view on its efficacy.

Most empirical studies and endogenous growth theories have highlighted the importance of innovation to economic growth.1 Many countries have also attempted to create a favorable innovation environment and protective regularity, aiming to promote R&D in firms and consequently to contribute to sustainable economic growth. Essentially, R&D is uncertain and both time- and money-consuming. It is also recognized to possess public good characteristics, thereby preventing the market from providing sufficient quantities of R&D from the perspective of social return. To bridge the gap between private and social rate of return and foster industrial R&D activity, various policy measures have been launched. Specifically, the R&D tax credit has become increasingly popular in developed countries, such as the U.S., Canada, and some OECD countries since the early 1980s.2 Taiwan, one of the best performers among latecomers, has been very successful in narrowing the technological gap during the past two decades with its counterparts among leading countries, especially in the electronics industry. Her R&D/GDP ratio, a simple measure of a knowledge-intensive economy, rose from 1.62% in 1990 to 2.78% in 2008 gradually, which was a little higher than corresponding ratios of 2.77% and 2.64% in the U.S. and Germany in 2008.3 As for R&D output, Taiwan has recorded extremely fast growth both domestically and in the U.S. Taiwan not only placed 4th in the world in terms of the quantity of its U.S. patents since 2003, but also ranked high in terms of patents per capita, compared with G7 countries and the other “Asian Tigers” (Trajtenberg, 2001). This achievement is rare in the developing world and is almost nonexistent within or outside of Asia. During the technological development process, the Statute for Upgrading Industries (SUI) that applies tax incentives, subsidies, and supporting measures to assist innovative activity is one of Taiwan's key industrial technology policies (Lien et al., 2007). However, economists have been generally skeptical regarding the efficacy of tax incentives. Expiration of the R&D tax credit of the SUI at the end of 2009 raises a legitimate concern: has the R&D-preferential policy induced greater R&D expenditures and a higher R&D growth in Taiwanese firms? From the perspective of public finance, the erosion of the tax base attributed to R&D incentives is possibly one cause of the fiscal shortage. Whether or not limited government resources should be used to encourage R&D depends on the efficacy of these measures to induce far greater R&D and contribute to sustainable growth. However, this important issue is not well examined in Taiwan. High-tech industries are generally more R&D intensive and the primary recipients of R&D tax credits in the U.S. (Wu, 2008). This situation applies to Taiwan, while many so-called traditional industries (typically less R&D-intensive) voice criticism that R&D tax credits work more favorably for high-tech firms. Innovative behavior strongly relates to the technological environment surrounding the location of a firm. A relatively fertile technological environment induces firms to devote more R&D efforts, whereas the appearance of innovation is relatively rare in an infertile technological environment. This implies significant variations in innovative activity patterns between high-tech firms and their non-high-tech counterparts. To enforce a more effective policy of granting R&D tax credits, one possible improvement is to establish various tax credits across industries. This leads to another essential and prominent issue: does R&D-inducement effect differ between high-tech and other industries in Taiwan? Assessing the potential differences in R&D-inducement effects of R&D tax credits across industries can provide useful insights for legislation of new R&D policies. At least two difficulties arise when using firm-level data to evaluate the effectiveness of an R&D tax credit within a country. Firstly, due to their different tax positions and expectations on future R&D spending, the variation between firms in credit effectiveness is highly endogenous (Bloom et al., 2002). Secondly, as indicated in Hall and Van Reenen (2000), there are many ways in which the R&D tax credit gives rise to heterogeneity, and often perverse incentives are a key feature in the debate on the desirability of R&D tax credits. Because the recipients of tax credits may differ in some firm characteristics from non-recipients (Czarnitzki et al., 2011), it is important to correct both selection bias and firm heterogeneity across recipients. This study thus adopts the non-parametric propensity score matching (PSM) method developed by Heckman et al., 1997 and Heckman et al., 1998 to correct possible selection bias. While the PSM approach deals with the selection bias problem to differentiate the treatment effect of R&D tax credit, it does not deal well with the second difficulty of unobservable firm heterogeneity. Fortunately, our dataset contains detailed information concerning the amount of R&D business tax deduction and firm characteristics, thus enabling us to adopt the panel instrumental variable (IV) and generalized method of moment (GMM) techniques to deal with both problems of endogeneity and firm heteroskedasticity. This paper evaluates the effect of tax credits for R&D and its growth in Taiwanese manufacturing firms, to contribute to the empirical literature in the following ways. First, the question of how and to what extent tax credits stimulate industrial R&D has attracted widespread international attention among economists, with only limited empirical studies focused on developed countries such as the U.S., Canada, and France. However, tax incentive policies are worthwhile considerations in not only developed economies, but also for newly industrialized and developing countries (NIEs). Taiwan has successfully achieved substantial technological development over the past two decades, and its outstanding performance in terms of innovation makes it an excellent case for investigating the tax incentives issue. Our firm-level evidence from Taiwan can complement the existing literature that focuses only on advanced countries. Second, we further separate samples into electronics and non electronics firms to examine whether and to what degree potential differences exist in effectiveness of R&D tax credit. This investigation can provide useful insights into ways to further revise R&D tax credits, because a uniform tax credit system for all industries is widely criticized as inappropriate. Third, this study employs the PSM method to correct the selection bias problem to differentiate the treatment effect of R&D tax credit. This enables us to compare the R&D activities of R&D tax credit recipients and non-recipients. This study further examines the marginal effect of R&D tax credits on R&D expenditures of firms. In this context, this research employs the panel IV and GMM techniques to deal with endogeneity and unobservable firm heterogeneity, appropriately assessing the effect of R&D tax credits on the R&D efforts of firms. The remainder of this paper is organized as follows. The next section provides a brief review of the literature on R&D tax credits. Section 3 introduces Taiwanese R&D incentive measures and describes the data used in this study. Section 4 presents the empirical model and examines the R&D-inducement effect of an R&D tax credit using the propensity score matching method. Section 5 presents the findings from a further investigation into the marginal effect of R&D tax credits on R&D across industries. The final section concludes with the main results and their policy implications.

Although the effectiveness of fiscal incentives for private R&D has attracted widespread interest in many developed countries, it has received much less attention in newly industrialized economies (NIEs) and developing countries. Taiwan has been one of the most successful NIEs in the world in innovative performance over the past two decades, particularly in the technological field of electronics. One possible reason is that the government has adopted long-standing tax incentives to foster private R&D. Economists have traditionally been skeptical of the efficacy of any fiscal provisions. Is the R&D tax credit really an effective mechanism for encouraging firms to invest more in R&D? This issue is particularly important to Taiwan, because the tax credit policy that forms part of the SUI will expire at the end of 2009. The government must decide in the near future whether to extend the tax credit policy or change it in part, if not completely. This paper evaluates the effect of tax credits on R&D activity in Taiwan that differs from previous studies focused on estimating tax price elasticity of R&D based on R&D cost. We first adopt the propensity score matching method to simulate the scenario of how the treatment of R&D tax credits affects R&D activity of firms. The PSM approach helps to correct the selection bias that previous studies inadequately address. Secondly, to control for unobservable firm heterogeneity between treated and control groups which may affect R&D activity, this study employs both techniques of panel instrumental variable and GMM to estimate the influences of tax credits on R&D expenditure and its growth. Using a panel dataset of 576 enterprises listed on the Taiwan Stock Exchange over the 2000–2005 period, our empirical findings are summarized as follows. First, the PSM estimate shows that R&D tax credits induce a higher (53.8%) R&D expenditure, on average. This suggests that there is no crowding-out effect of public R&D support on private R&D, whereas there is a positive effect of R&D tax credits on private R&D. On the other hand, the growth level estimation of PSM demonstrates also a positive influence of tax credits on R&D expenditure growth, while this enhancing effect is not statistically significant. Secondly, the R&D-inducement effect of tax credits was found to differ between electronics and non-electronics firms. Specifically, this treatment effect is much stronger for electronics firms such that the recipients of R&D tax credit experience a higher (16.9%) growth of R&D investment and a higher (71.8%) R&D expenditure than those they do lacking this policy instrument. Alternatively, we do not find significant treatment effects on R&D expenditure and growth brought about by the R&D tax credit for non-electronics recipients. Third, estimates obtained using IV technique involving the panel data model show that the elasticity of R&D with respect to the taxation remit is 0.302 on average. The induced R&D expenditure effectuated by tax credits accounts for 9.4% to 12.0% of R&D for the sample firms. Moreover, estimates obtained using GMM lend the similar finding that tax credits do have a R&D growth-enhancing effect, particularly for electronics firms. This suggests that tax credits have a moderate effect on private R&D in Taiwan. Finally, the R&D elasticity of tax credit tends to increase slightly along with the approaching expiration of SUI. It suggests that firms prefer to appropriate the benefit of policy measure and lends supportive evidence on the effectiveness of R&D tax credits. Based on the above analyses, this study derives several policy implications. First, as this study overall finds a significantly positive R&D-inducement effect of R&D tax credits, the R&D preferential policy is deemed an effective policy tool for fostering private R&D activity, thereby supporting previous aggregate evidence. Secondly, as the marginal R&D-inducement effect facilitated by tax credit is relatively limited, whether this policy tool continues to be implemented after the SUI expires is worth careful evaluation, as the problem of fiscal deficit has seriously worsened in Taiwan in recent years.21 Third, because the effect of R&D tax credits differs between electronics and non-electronics firms, the government should devise ways to adjust the taxation reduction rate for R&D investment across industries. Finally, due to the limitation of using LMEs, this study cannot examine the criticism that SMEs have difficulty qualifying for tax credit. This issue is worth further investigation.