Abstract:    Due to high risks and long period of innovation project, manager is reluctant on R&D expenditures, which may lead to the problem of management myopia. Literatures propose that there are two opposite effects of media report on management myopia. These two effects are reputation effect and pressure effect respectively. Although media report can regulate management myopia through the channel of reputation effect, empirical evidences indicate that market pressure caused by media results in the decrease of innovation investment. Compared to inside governance mechanism such as proper right and contract, media report, as well as law and financial market, is treated as outside governance mechanism. Under the background of transition economy, with the incomplete inside governance mechanism, researches indicate that outside governance mechanism plays an important role on agency problems. Compared with other outside governance mechanism, media report is more convenient for application. Thus, it is meaningful to investigate how media coverage can improve R&D expenses through the channel of reputation effect.
   We adopt the new performance appraisals (Economic Value Added, short for "EVA") of top management in "The Revision of Interim Measures for Business Performance Appraisals of Persons-in-Charge at Central Government-Owned Enterprises (CGOE)" in 2009 as a natural experiment to test the relationship between media coverage and R&D expenditures. With this experiment, we test the assumption that whether new EVA performance appraisal adopted in CGOE lessens short-term performance pressure of top management and increases the concerns of management on reputation, thus leads to the replacement of pressure effect by reputation effect on the relationship of media coverage and innovation investment. 
   The reform of performance appraisal is proposed by central government and out of control of corporate, which provides a good setting of exogenous experiment to test the assumption. The new appraisal is applied on CGOEs, which means that non-CGOEs can be treated as controlled group. With the adoption of the appraisal, net profit is replaced by EVA as the performance measure for top management of CGOEs. While R&D expenses are deducted from the calculation of net profit, in the calculation of EVA, R&D expenditures are added into the evaluation of corporate performance. Therefore, the adoption of new performance appraisal on CGOEs means that the expenditures in R&D will increase corporate performance. Accordingly, the change of performance appraisal decreases the short-term pressure on top management. 
    We obtain financial data of Chinese listed firms from CSMAR data base, and the data of R&D expenditures from WIND data base. The number of media coverage for each company is obtained with Baidu news searching machine. The data covers the period from 2007 to 2012. The statistics shows that the mean of R&D expenditures is significantly higher after 2010, that is, the year for CGOEs to adopt new appraisal. We also find that the improvement of the mean of R&D expenditures is driven by CGOEs, while there is no such an increase of the mean for non-CGOEs. The statistical numbers indicate that the adoption of EVA appraisal is positively related with R&D expenditures.
    Then we run regressions to test whether the positive relation between EVA appraisal and R&D expenditures is due to the reputation effect of media. The regression result indicates that before 2010, the number of media coverage is significantly negative with R&D expenditures, which is consistent with the pressure effect that media coverage decreases innovation. What′s more, there is no significant difference of the effect between CGOEs and non-CGOEs. However, after CGOEs adopts new EVA appraisal from 2010, media coverage is significantly positive with R&D expenditures, whereas for non-CGOEs, the relationship between media and R&D remains negative. 
    We then take a Chow test to investigate whether new appraisal on CGOEs causes the change of the relationship between media coverage and R&D expenditures. We find that for CGOEs, from 2010 there is a change of this relationship from significantly negative to significantly positive. However, there is no such a change for non-CGOEs. This result indicates that the adoption of EVA appraisal for CGOEs in 2010 lead to the change of the effect of media on R&D expenditures from negative pressure effect to positive reputation effect.
   We then use difference-in-difference approach to test the above change of media effect on innovation. We find that new appraisal on CGOEs improves R&D expenditures significantly. We also find that the positive effect of new appraisal on R&D expenditures is due to the positive effect of media coverage. The result indicates EVA appraisal on CGOE leads to the replacement of pressure effect by reputation effect of media. The result exhibits robustness with the concerns of omitted exogenous event at the same year, change of controlled group, placebo test and reversal causality.
   Finally, we divide the observations of CGOEs into two subsamples with the degree of marketization index. We find that with the adoption of new appraisal, the positive relationship between media coverage and innovation expenditures is more significant in the area with higher marketization. Because management reputation is more valuable in these areas, this finding is consistent with our assumption that the new appraisal on CGOEs reduces the short-term pressure of performance and thus leads to more concern of management on reputation.  
   Our finding reveals that as an outside governance mechanism, the role of media on management behavior is related with proper management incentives, such as EVA evaluation. On the other side, the effect of EVA evaluation on management agency problem is through the channel of informal mechanisms, which is media coverage in our analysis. Our finding is also helpful to further the understanding about media coverage and corporate governance. According to our finding, formal and informal mechanism should be jointly adopted to regulate agency problem. Our suggestions for the problem of agency behavior are as followings: (1) Deepen the reform of state-owned-enterprise and adopt right stimulus for management; (2) Strengthen the role of informal mechanism on corporate governance.

Key words: media coverage, innovation investment, reputation effect, pressure effect, EVA appraisal