Abstract: Government support for innovation is economically justified to address market failures and has the potential to stimulate firm innovation in terms of both R&D inputs and innovation outputs. While it has been widely acknowledged that government incentives can effectively promote corporate innovation, there are growing concerns on issues like input-output mismatches, imbalanced subsidy structure and so on. To address those concerns, this paper attempts to investigate if weighted tax deduction for R&D costs can effectively stimulate firms’ innovation activities and improve their innovative capability from the the full perspective of innovation chain. 
Despite the economic justification of the use of R&D tax incentives as an innovation policy, the empirical evidences on the effectiveness of the policy remains mixed. Specifically, profit-driven firms may use the capital surplus from tax incentives to maximize private benefits or focus on short-term projects, which will not lead to higher social benefits. In this case, the R&D tax incentives do not close the gap between private and social benefits. On the other hand, the R&D tax incentives may also face potential moral hazard due to the information asymmetry between the firms and government. That is, firms may replace private R&D expenditures with tax incentives and do not necessarily spend more on R&D. Therefore, as a mere investment-oriented incentive policy, whether tax incentives can effectively stimulate firms’ innovation investments and thereby further promote the R&D outputs and economic profits, becomes the main concern of the current studies.
Based on the perspective of innovation chain, this paper divides the innovation process into two stages, namely, the technological research and development (R&D) stage and the technological transformation stage, to further analyze the effects of the typical policy tools of the R&D tax incentives -- tax deduction for R&D costs on input, output and profit sides. This paper also introduces the concept of innovative capability, which is decomposed into technological R&D capability and technological transformation capability correspondingly, to assess the effectiveness of the tax deductions for R&D costs comprehensively.
To deal with the endogeneity problem arising from selection bias, this paper combines the Propensity Score Matching (PSM) and Difference-in-differences (DID) models to evaluate the effectiveness of the tax deduction policy. The PSM-DID method combines the advantages of PSM and DID models and, to some extent, eliminates their respective shortcomings. DID model controls the influence of unobservable time-invariant factors. and PSM model controls the selection bias of cross-sectional data. The use of PSM-DID model also better satisfied the common trend assumption that is required by the DID approach.
The data used in this paper are about high- and new- technology enterprises (HNTEs) operating in the Beijing Zhongguancun science park during the period from 2005 to 2015. The HNTEs in Zhongguancun have rich innovation resources, and they are often treated as pilot enterprises for tax incentives, so that they can directly and quickly respond to the tax deduction policy. 
This paper investigates the impact of the tax deduction policy on firms’ R&D inputs, R&D outputs and economic profits to infer the impacts on firms’ innovation capability. In the input, output and profit sides, both scale and intensity measures are employed to measure the firms’ innovation. From the perspective of R&D inputs, this paper selects the amount of R&D investments of enterprises as the scale indicator, and the intensity indicator is measured by the ratio of firms’ R&D investments to the total revenues. At the output side, considering the time lag in the process of patent authorization and the influence of some uncontrollable subjective factors, this paper selects the number of patent applications as the scale indicator, and the ratio of the number of patent applications to the number of employees as the intensity indicator. From the perspective of profit side, the scale indicator is characterized by the sales revenues of new products, and the intensity indicator selected is the ratio of sales revenues of new products to total sales revenues. In addition, in order to make the estimation results more accurate, this paper chooses firm size, return on assets (ROA), rate of return on common stockholders’ equity (ROE), intangible assets, exports, and age as the control variables.
Our study provides empirical evidence on the effectiveness of tax deduction policy in stimulating firm’s innovation. On the input side, empirical results suggest that tax deductions for R&D costs can effectively promote private R&D investments as the intended goal of the government. On the output side, the tax deductions for R&D costs only increases the scale of R&D outputs, while cannot increase the output intensity. Furthermore, on the profit side, the tax deduction policy have a significant effect on the scale and intensity of economic profits. The findings of this paper affirm the guiding role of tax deduction policy and its positive effectiveness in the improvement of firm’s innovative capabilities. 
This paper provides a new perspective for the study of tax deduction policy by introducing the concept of innovation chain. Revealing the influence of tax deduction from the perspective of the innovation chain, this research not only deepens the understanding of the tax deduction policy, but also expands the research framework of the existing researches on innovation policy evaluation, and lays a preliminary theoretical foundation for the further exploration of various innovation policies from the perspective of innovation chain and innovative capability. In addition to theoretical contributions, this research provides empirical evidence  for the debate of the effectiveness of tax incentives at the firm level. Through the advantages of firms’ data, this paper chooses the typical tax incentives—the tax deduction for R&D costs as the research object, instead of confusing the differences between the different tax incentive tools. At the same time, different from the existing researches that have focused more on the tax deductions for R&D costs from the perspective of input side, this paper divides the innovation process into the technological R&D stage and the technological transformation stage and comprehensively evaluates the influence of tax deductions for R&D costs from firm’s innovation inputs, outputs and economic profits. The findings of this paper may help to enhanced management and implementation of tax incentives from the full perspective of innovation chain and firms’ innovative capability. Finally, we proposed some corresponding policy recommendations for better policy design and implementation.

Key words:  tax deduction for R&D costs, innovation chain, firm’s innovative capability, tax incentives, PSM-DID model