减税激励、研发操纵与研发绩效 杨国超 刘静 廉鹏 芮萌 中南财经政法大学会计学院 首都经济贸易大学会计学院 华东政法大学国际金融法律学院 中欧国际工商学院 摘 要： 张维迎和林毅夫就“政府是否应该制定产业政策”展开了激烈争论。与他们从较为宏观的视角分析不同, 本文利用高新技术企业认定管理办法确立的公司研发投入占销售收入之比这一“一刀切”的微观认定门槛, 研究公司是否会操纵研发投入以满足相关标准。研究发现, 研发收入之比的分布在法规门槛附近不连续, 这确认了公司研发操纵行为的存在。基于分布不连续特征定义的研发操纵公司获得了更多的税收优惠和政府补助。进一步地, 研发操纵现象在民营企业、盈利企业以及位于税收征管力度较弱地区的企业中更为显著, 且研发操纵公司的研发投入与公司未来发明专利申请、发明专利授权数量间的正相关关系更弱。研究表明, 与政策制定者的初衷不同, 研发激励政策会激励公司进行研发操纵, 最终导致公司研发绩效下降。本文不仅为产业政策引发的激励扭曲效应提供了微观证据, 还为更好地制定和执行高新技术企业认定政策提供了理论参考。关键词： 高新技术企业认定; 产业政策; 法规门槛; 研发操纵; 研发绩效; 作者简介：杨国超, 电子信箱:yang.guochaooutlook.com;作者简介：刘静, 电子信箱:wenmu9163.com;作者简介：廉鹏, 电子信箱:lianpeng82gmail.com;作者简介：芮萌, 电子信箱:oliverceibs.edu。基金：国家自然科学基金青年项目 (71402054) 、国家自然科学基金项目 (71372203) Tax-reducing Incentives, R&D Manipulation and R&D PerformanceYANG Guochao LIU Jing LIAN Peng RUI Meng Zhongnan University of Economics and Law; Capital University of Economics and Business; East China University of Political Science and Law; China Europe International Business School; Abstract： At the end of 2016, Professors Weiying Zhang and Justin Yifu Lin launched a heated debate about whether the government should adopt industrial policy. Although Professor Zhang emphasized the market s role as the “invisible hand”, Professor Lin argued for the active role of the government. Considering that R&D investment is of great importance to economic growth and the potential risks of corporate R&D activities, the government often provides tax incentives to enhance R&D investment. Thus, this paper attempts to shed light on industrial policy disputes from the perspective of R&D stimulus policies. In China, to encourage companies to invest in R&D, the Ministry of Science and Technology, the Ministry of Finance and the State Administration of Taxation jointly issued the Administrative Measures for the Identification of High-tech Enterprises ( hereafter referred to as Administrative Measures) on April 14, 2008.As Krugman ( 1983) suggests, industrial policies supporting specific firms often require certain selection criteria. The Administrative Measures use a nearly one-size-fits-all R&D intensity standard to classify a firm as a high-tech enterprise: if a firms sales in the preceding year are less than RMB50 million, the ratio of R&D investment to sales must not be less than 6%; if a firms sales in the preceding year are between RMB50 million and RMB200 million, the ratio of R&D investment to sales must not be less than 4%; and if a firms sales in the preceding year are more than RMB200 million, the ratio of R&D investment to sales must not be less than 3%. Thus, the ratio of R&D investment to sales is an important threshold.Only when R&D intensity exceeds a threshold may a company be referred to as a high-tech enterprise and thus enjoy a 15% corporate tax rate, compared with the 25% tax rate for non-high-tech companies, along with a wide range of government subsidies. This paper studies whether companies manipulate R&D to meet the relevant standards using firm-level data.Following Burgstahler & Dichev ( 1997) , this paper identifies companies that may engage in R&D manipulation, as their R&D intensities just meet or beat the threshold defined by the Administrative Measures. Its sample includes companies listed on the Chinese A-share market after excluding financial companies, companies that demonstrate abnormal conditions, and companies with missing values from 2008 to 2014. This paper applies the methodology proposed by Beaver et al. ( 2007) to test R&D intensity discontinuity around the threshold. It finds that the distribution of R&D income ratio is discontinued near the regulatory threshold, which confirms the existence of corporate R&D manipulation. R&D manipulation is positively correlated with the tax benefits and government subsidies received by a firm. In addition, it is more pronounced for private firms, profit-making firms, and firms located in weak tax enforcement provinces. Furthermore, if a firms R&D intensity just meets or beats the threshold specified by the Administrative Measures, the positive relationship between R&D investment and the number of patent applications or patents granted decreases.This paper may contribute to the following strands of literature. First, it provides micro-evidence for the “industrial policy dispute ”. As there is information asymmetry in the process of industrial policy implementation and the cost of violating regulation is relatively low in China, companies may engage in opportunistic behavior to meet the “threshold”, making industrial policies less effective. Second, in contrast to research that has mainly focused on companies earnings thresholds, this paper studies the threshold of a specific accounting item ( i.e., the R&D intensity threshold) . Third, this paper studies the relationship between R&D investment and R&D output from the perspective of R&D manipulation. It has long been recognized that R&D investment may improve innovation levels. However, this paper em