Overinvestment is defined as corporations allocating free cash flow to negative NPV (Net Present Value) projects. [1] The main causes of overinvestment can be classified in four aspects―agency costs, information asymmetry, bad corporate governance and be- havioral finance.

In specific research, Jensen [1] thinks the corporations’ managers have a strong motive to establish “corporate empire”. They would like to choose negative NPV projects to expand corporation scale due to the separation of ownership and management. Pan Min and Jin Yan [2] find if the information asymmetry is not considered, the particularity of the equity system of Listed Companies in our country will lead to the overinvestment behavior due to equity financing. Wei Minghai and Liu Jianhua [3] , Kong Dongmin and Feng Xi [4] support the hypothesis that the low cash dividend policy could promote listed companies’ overinvestment. David, and Valeriy Denis [5] believe that financing constraints could affect the corporations’ cash holdings, which further affect the level of investment. Zhang Huili and Lu Zhengfei [6] find that in group companies, the higher the proportion of cash held by subsidiaries, the more serious of the overall overinvestment is. Yu Meihong and Xu Longbing and others [7] find that the ownership concentration or the existence of controlling shareholders leads to overinvestment. What’s more, the separation of controlling shareholders and cash flow rights further exacerbate the overinvestment level. Tang Xuesong, Zhou Xiaosu and others [8] propose that in our special system, government intervention, political connections and other reasons would lead to overinvestment, as well as Du Xingqiang, CengQuan, et al [9] . In addition, Hua Guiru, Liu Zhiyuan and others [10] find that the investors’ emotional change would also lead to overinvestment.

In terms of how to reduce overinvestment, Myers [11] proposes debt financing caused insufficient investment. Jensen [1] supports Myers’s conclusion. He points out that the cash dividend and debt could effectively reduce overinvestment. Li Weian and Jiang Tao [12] find that shareholders’ behavior governance, board governance, stakeholder governance had positive effect on reducing corporations’ overinvestment. However, the effect of the board of supervisors, managers, information disclosure mechanism is not significant. Wei Minghua and Liu Jianhua [3] point out that the low cash dividend policy of state-owned listed companies could lower the overinvestment level, and the company internal governance structure and external governance environment would improve the phenomenon of investment. Wang Yanchao [13] finds that when corporations have excess cash holdings, the corporations with non-constrained financing policy are prone to excessive investment. On the contrary, the financing constraints of the corporation would reduce overinvestment. Yu Meihong, Xu Longbing and others [7] find that compared with private-owned holding company, government-owned holding company overinvestment phenomenon is more serious, and the improvement of external governance environment can inhibit overinvestment in a certain extent. Zhang Huili and Lu Zhengfei [6] find that the improvement of corporate governance mechanism could reduce the overall overinvestment level which is caused by the subsidiaries’ high cash holding. Kong Dongmin and Feng Xi [4] find that the cash dividends can effectively restrict the overinvestment behavior and play a better inhibitory effect in corporations with a lower free cash flow.

Economic value added is the difference between net profit after tax and capital cost in a certain period of time. It fully takes the opportunity cost of capital into account. And it is a management evaluation system including performance evaluation, incentive compensation and management philosophy. Sun Zheng, Wu Qian [14] , Byrne [15] , Biddle, Bowen and Wallace [16] point out that EVA is better than NOPAT and FCF to measure the firm’s market value and shareholder returns. Bartolomé Deyá Tortella and SandroBrusco [17] find that the managers try to improve the efficiency of investment activities if the corporations use EVA performance evaluation. But it is likely to cause a higher level of debt. Stern [18] points out that the EVA performance evaluation can reduce the operating behavior which couldn’t increase the corporation value and shareholder value. And it is significant in both private corporations and state-owned corporations. Worthington and West [19] put forward that EVA can encourage corporation managers to make efforts to increase shareholders’ wealth and improve corporation financial performance. Rogerson [20] finds that EVA performance evaluation could push the corporation managers to choose effective investment level so as to improve investment efficiency and enhance the value of corporations. Jan Mouritsen [21] compares the difference between EVA and IC (intellectual capital)’s contributions in increasing corporation value. He finds that on the one hand, EVA is more appropriate to measure the cost of capital, on the other hand, EVA is a results-oriented evaluation method. It constrains the behavior of managers.

Putting EVA into the assessment indicators, the purpose of SASAC is to effectively perform their duties, safeguard the interests of shareholders, preserve and increase the value of state-owned assets, establish effective incentive and restraint mechanisms. Since the EVA performance evaluation began to be implemented form 2010, there is not much domestic research on EVA. This paper summarizes researches about the impact of EVA implementation on overinvestment. Wang Xufang, Wang Li [22] and Liu Jing [23] point out that deducting the cost of capital when we calculate EVA is to guide the managers consider the opportunity cost of capital. The decision principle of project investment is that the expected investment income must be greater than the cost of capital, or the investment value is otherwise damaged. BaoXuedong and Xiao Wenjuan [24] express the same view. They point out that equity capital is a kind of opportunity cost, and investment should be questioned whether the project income can make up for the cost of equity capital. Wang Ligang [25] put forward that the implementation of EVA could make the managers strengthen the feasibility analysis of investment projects to ensure investment income. Yuan Kang [26] , Chi Guohua, Wang Zhi, Yang Jin and others [27] think the use of EVA can solve the principal-agent problem and can maximize the interests of shareholders in a certain extent.

In summary, this paper thinks there are many reasons leading to overinvestment. But the most fundamental reason is the agency problem. Due to shareholders and managers have different objectives, the managers expand the scale of corporations and continue to invest to meet their own ambitions. In order to solve the problem, the most effective way is to establish a reasonable performance evaluation system, which can constraint manager’s investment behavior. The EVA performance evaluation implemented by SASAC takes the cost of capital into account, which to a certain extent can reduce SOEs’ overinvestment phenomenon, especially significant to the low growth corporations. This article conducted an empirical analysis of the implementation of EVA and overinvestment and the research results is expected to provide some support for further research of the EVA performance evaluation system, and to give some references to SASAC and other institutes that are developing or promoting their projects of EVA system for their policy making.

Our research samples consisted of 204 listed SOEs, covering time period from 2007 to 2012.The list of pilot enterprises was accessible from the official website of SASAC. And CSMAR (China Securities Markets and Accounting Research Database) was our data source for all the financial data used in this paper. We excluded three corporations whose shells were borrowed and one financial corporation, leaving us with 1224 observations. And we choose overinvestment model by Richardson to measure overinvesting behaviors of the listed SOEs, and used the outcome (inv) as the dependent variable to observe the effect of the implementation of EVA evaluation. Then, based on Tobin’s Q Value, the samples were divided into two groups with high growth and low growth. OLS regression test was performed on both groups in order to observe different effect of EVA evaluation on corporations’ overinvestment with different growth.

Descriptive statistics for the main variables are as follows. We used the econometric software STATA to conduct all the descriptive statistics and empirical tests in this paper. Overinvestment of all the corporations in our sample ranged from a minimum of −0.5373 to a maximum of 0.4601, with an average of 0, which indicated that there are underinvestment among SOEs as well as overinvestment. Size had an average of 22.5377, a minimum value of 2 and a maximum value of 6 in our sample. Descriptive statistics of other variables are as shown in Table 2, so we don’t repeat the results here. We calculated the average of Tobin’s Q from year 2007 to 2012 for each corporation, after deleting two corporations with missing data. All the corporations were divided into two groups by the median of the average 1.9350, one with high growth and the other with low growth.

Table 3 shows the descriptive statistics of overinvestment among the whole sample and two different growth groups. The overinvestment of SOEs with low growth is bigger than corporations with high growth in 2007 and 2008, while lower than corporations with high growth afterwards, as reflected by the median number. According to Pilot List of SOEs to Implement EVA in 2007-2009 by SASAC, there were 80 pilot participants among in 2007, 89 pilot participants in 2008, and 122 pilot participants in 2009. With the SOEs starting to implement EVA evaluation system one after another, the overinvestment of corporations with low growth began to reduce, and finally became less than the overinvestment of corporations with high growth, which correspond to our hypothesis two in some extent.

Table 4 showed the change in overinvestment of the corporations from the whole sample, high growth group and low growth group after implementing EVA. We found an obvious positive correlation between overinvestment and the implementation of EVA, while the median was less after the implementation of EVA in the low growth group, which is matched with our hypothesis two. Besides, we found that the overin-

vestment level in corporations with low growth is higher than the whole sample and high growth group, yet reduced after implementing EVA and became nearer to the whole sample and high growth group, which supports our hypothesis two as well.

Table 5 is the regression results of our model two, which can be observed the effect of implementing EVA to SOEs’ overinvestment. The results showed the correlation between implementing EVA and overinvestment is positive, but this conclusion didn’t pass the significance test. We run White test afterwards, which proved there was no heteroscedasticity. So our hypothesis one was not supported based on our sample data,

Dependent variable in this model is overinvestment (overinv), as defined in Table 1; The above numbers in each cell are estimated coefficient, the number below in braces are t values; *, **, *** means 10%, 5% and 1% significance level.

which means the implementation of EVA evaluation may not have depressing effect on SOEs’ overinvestment as we expected.

Table 6 shows the regression results of model two, by which we observed the different

effect of implementing EVA to the overinvestment of corporations with different growth. The depressing effect of implementing EVA on low growth corporations was statistically significant, and the significance level became higher with cluster sampling. As for the sample group with a higher growth, the effect was not proved statistically significant. Our empirical results supported our hypothesis two, which means that implementing EVA does have a significant negative effect on overinvesting of SOEs with relatively low growth. So applying EVA evaluation can improve low growth SOEs’ overinvestment.

We selected new sample of overinvestment for a robust test. About 2/3 of our estimated overinvestment were negative, which means these corporations didn’t overinvest. In order to ensure the stability and reliability of our rejection to hypothesis one, we ranked the whole sample according to the order from small to large and select the largest 1/3 of them all to run a new regression for our hypothesis one. The empirical results were as showed in Table 7, indicating that implementing EVA has no statistically significant effect to overinvestment, which is consistent with our former test. So hypothesis one was not accepted.