آیا مشوق های مالیاتی R & D منجر به افزایش دستمزدها برای کارگران R & D می شود ؟ مدارک و شواهد از هلند
|کد مقاله||سال انتشار||مقاله انگلیسی||ترجمه فارسی||تعداد کلمات|
|5259||2013||8 صفحه PDF||سفارش دهید||7816 کلمه|
Publisher : Elsevier - Science Direct (الزویر - ساینس دایرکت)
Journal : Research Policy, Volume 42, Issue 3, April 2013, Pages 823–830
This paper examines the impact of the Dutch R&D tax incentive scheme on the wages of R&D workers. We construct firm specific R&D tax credit rates that vary over time following variations in the Dutch R&D tax incentive program. Using instrumental variables we estimate a wage-sharing model with an unbalanced firm-level panel data covering the period 1997–2004. The elasticity of the R&D wage with respect to the fraction of the wage supported by the fiscal incentives scheme is estimated at 0.2 in the short run and 0.24 in the long run.
Most OECD countries dedicate significant resources to support private R&D. Two main policy instruments are used for this: direct subsidies and tax incentives. R&D tax incentives seem to have gained in popularity recently. For instance, in 1996, 12 OECD countries offered R&D tax incentives, while in 2008 this number increased to 21 (OECD, 2009). This increase is partly due to the fact that the policy of fiscal incentives is more neutral compared to direct subsidies, in principle offering a tax relief to any eligible R&D expenditure. The usefulness of R&D support programs in general has been traditionally challenged for two reasons: government supported R&D may crowd out private R&D funding or get dissipated in higher R&D wages instead of stimulating real private R&D spending. Many empirical studies have examined the first question and often concluded that there is some additionality, in the sense that firms increase their R&D spending by more than the money they get from government in support of R&D (see Arundel et al., 2008, David et al., 2000, Hall, 2002 and Hall and Van Reenen, 2000 for reviews of empirical studies). Surprisingly, the other key question, whether and to what extent government R&D support efforts dissipate into higher R&D wages as opposed to creating a R&D quantity effect, has remained largely unexplored. A number of papers on the effectiveness of R&D support programs have included a price effect in their analysis (e.g., Aerts, 2008, Lelarge, 2009, Haegeland and Møen, 2007 and Wolff and Reinthaler, 2008), but most of these studies are concerned with direct R&D subsidies and not R&D fiscal incentives. Studies in labor economics have examined the effect of firm and employee characteristics on wages in general (e.g., Dobbelaere, 2004 and Veugelers, 1989) but not specifically on the wages of R&D workers. The contribution of this paper is to examine the magnitude of the effect of the Dutch R&D tax incentive program, known as WBSO,1 on the wages of R&D workers. By studying the price effect of the R&D fiscal incentive program, we seek to contribute to the policy discussion on the effectiveness of R&D support. We borrow from the stream of literature in labor economics to inform us on the specification of the wage equation. In our empirical analysis we use an unbalanced firm-level panel dataset constructed from the annual R&D surveys and production statistics from the Central Bureau of Statistics of the Netherlands. The richness of the merged dataset allows us to construct detailed R&D tax disbursement as a function of R&D tax incentives, providing sufficient variation in both the cross-section and time dimensions to identify the effects of the tax incentive program on R&D wages. Our main empirical finding is that there is a significant price effect of the R&D tax incentive program on the wages of R&D workers in Dutch firms. After allowing for individual heterogeneity, business cycle fluctuations and the endogeneity of R&D tax credits we obtain estimates of the elasticity of R&D wages with respect to tax disbursements that are significantly different from zero. The point estimates of this elasticity range from 19% to 24% depending on whether static or dynamic models, short-run or long-run estimates are considered. The remainder of the paper is organized as follows. Section 2 provides an overview of the theoretical and empirical literature on wage determination and government R&D support. Section 3 lays out our empirical model relating wages to R&D tax incentives, describes our data set and explains how we constructed the variables used in the empirical analysis. Section 4 presents the empirical results and Section 5 summarizes and concludes.
نتیجه گیری انگلیسی
This paper examines the price effect of the Dutch R&D tax incentive program aimed at stimulating R&D expenditures in business firms. Fiscal incentives are the main policy instrument in the Netherlands to promote R&D by granting firms deductions from their social security contributions proportionately to their annual R&D wage bill. A recent evaluation of the fiscal incentives program concluded that it was effective in stimulating business R&D (Lokshin and Mohnen, 2012). This paper provides micro-level econometric evidence that there is also a wage effect related to the R&D tax incentives program. Part of the R&D tax credits get transmitted into higher R&D wages because of inelastic labor supply, search costs for scientists and engineers, incentives given to R&D employees, or bargaining power of R&D employees. To estimate the magnitude of the price elasticity we exploit a rich unbalanced firm-level panel data set covering the years 1997–2004. We estimate a model sharing-in model where the average R&D wage depends on the effective rate of R&D tax credit, the alternative industry wage, and several firm and industry specific control variables. We allow for individual heterogeneity, and we correct for the endogeneity of the effective rate of R&D tax credit and for the dynamic panel bias when we estimate a dynamic version of the model. The estimated elasticity of the average R&D wage with respect to the effective rate of R&D tax credit is significantly different from zero and of the order of 0.2, with little difference between the short-run and the long-run effect because the average wage adjusts quickly to its optimal value. Our estimates of the wage effects are somewhat smaller than those found by Goolsbee (1998) for the U.S. and by Marey and Borghans (2000) for the Netherlands. This difference could be due to two reasons. First, the level of aggregation is different. Goolsbee (1998) works at the aggregate level and Marey and Borghans (2000) at the sector level. At a higher level of aggregation, spillovers are more likely to be present and produce extra pressure on the wage rates of R&D workers. The other reason is that we only consider R&D tax incentives and not total government R&D or total private and public R&D. Tax incentives are more neutral compared to direct R&D subsidies or R&D performed in government labs, which are more focused and may require a certain type of R&D workers with a relatively more inelastic supply. It is worth mentioning that our results of the price effect of R&D tax incentives are in line with the findings reported in the labor literature on wage determination. Hildreth and Oswald (1997) estimated an elasticity of wages with respect to profits per worker of about 4% using firm-level data. Abowd and Lemieux, 1993 report an elasticity of about 20%. Van Reenen (1996) applied three different measures of rents created through product innovation (profits per head, Tobin's Q and the difference between real sales per worker and average industry wage) to examine the impact of innovation induced rents on wages on a panel of British firms. He reports an elasticity of wages to innovation rents of about 0.29 for the quasi-rent measure, 0.05 for the profits per head measure and 0.04 for Tobin's Q rent measure. Our findings have a number of policy implications. The existence of a wage effect of R&D tax credits suggests that the efficiency of the R&D tax incentive program could be enhanced if the wage effect could be avoided. What goes into higher wages for scientists and engineers could go into more real expenditures on research and development. We estimate the wage effect to reduce the quantity effect and the effectiveness of R&D tax incentives by some 25 percent. On the other hand, higher wages might be the price to pay to retain high-skilled researchers with promising returns in the future. There are a number of limitations in our approach. The model presented in this paper assumes a constant disbursement parameter across all R&D workers. A more refined analysis, which would allow this parameter to differ across R&D workers, could not be performed because of the lack of data on individual-specific wages. This issue could be addressed in the future when the Central Bureau of Statistics makes it possible to link firm and individual-specific data. Relying on aggregated firm data does not allow us to control for differences in R&D workers’ characteristics such as seniority and schooling. Inability to control for individual worker characteristics amounts to an omitted variables bias (see Abowd et al., 1999, for discussion). The aggregation bias is mitigated if R&D incentives affect all R&D workers of a firm, which is the case in our data. Other interesting avenues for future research are the comparison of the wage effects of different instruments for stimulating R&D, such as direct and indirect measures of support and the competition for R&D talent between countries. Other instruments, such as educational policies and open-emigration policies as a way to decrease the inelastic supply of talents are worth investigating but remain outside the scope of this paper.