The paper takes Chinese listed companies from 2010 to 2016 as samples, to examine the relationship between executive overconfidence and equity incen-tive. Results show that executive overconfidence has a significant weakening effect on equity compensation incentives (including stock options and re-stricted stocks), that is, compared with rational executives, the company will reduce the equity incentives for overconfident executives.
In modern enterprises where ownership and control are separated, due to the inconsistency of the utility function and information asymmetry, there is inevitably a proxy conflict between the principal (shareholder) and the agent (the executive). Fama (1980) [
However, the “rational person” as the premise of the traditional optimal salary contract theory is not fully satisfied under realistic conditions. Roll (1986) [
Shareholders in modern enterprises can diversify the company’s non-systematic risks through diversified investments in the capital market, and thus have the characteristics of risk neutrality and even risk hobbies. However, it is risk aversion for executives whose personal wealth and human resources are closely related to the companies they serve and cannot achieve diversified investments. The inconsistency between the two parties’ risk appetites causes executives to act against the goal of maximizing shareholder value. This is a typical principal-agent problem that is common in modern enterprises.
According to the optimal contract theory, the shareholders or the board of directors who are faithful representatives of their interests can solve the agency problem by designing an effective executive compensation contract. The equity compensation is highly praised and widely application because it can link the interests of shareholders and executives closely (Jensen and Meckling, 1976) [
With the rise of behavioral finance, the “rational man” hypothesis of the traditional principal-agent model was relaxed. The objective existence of irrational factors of overconfidence and its impact on executive risk exposure have also been confirmed. In fact, companies with overconfident executives have more R & D investment (Hirshleifer, 2012) [
Gervais, Heaton and Odean (2011) [
Hypothesis: Executive overconfidence has a significant weakening effect on equity compensation incentives, that is, companies will reduce incentives for over-confident executives’ equity compensation compared to rational executives.
As mentioned above, foreign scholars usually measure the incentive intensity of equity compensation based on the vega and the slope of the salary-performance relationship. However, since the equity incentive system in China started late, the correlation between Vega and Delta is calculated. The data is difficult to obtain, so domestic scholars mostly use the value of equity compensation or the proportion of equity incentives to the total share capital as the proxy variable of the incentive intensity of equity compensation (Xiao Shufang, 2013) [
Incentive i,t = 0.01 ∗ Price i,t ∗ ( Rstock i,t + Options i,t ) 0.01 ∗ Price i,t ∗ ( Rstock i,t + Options i,t ) + CashComp i,t (1)
among them, Price i,t For the closing price of the stock of i company at the end of the year, Rstock i,t with Options i,t The number of restricted stocks and stock options held by i company executives at the end of the year, CashComp i,t The total amount of cash compensation earned by i company executives that year.
Despite the prevalence of executive overconfidence, the measure of overconfidence is very difficult. The more commonly used method is the stock option method (Humphery-Jenner, 2016) [
In addition to the above explanatory variables, we also control other factors that are considered important in the existing compensation literature. At the executive level, we control the average age and tenure of executives. At the company level, we control the separation of roles, corporate growth, R & D intensity, return on assets, market-to-book ratio, stock return, company market value, corporate age, fixed assets, financial leverage, property rights and equity concentration.
To validate the hypothesis, this paper draws on Humphery-Jenner (2016) and builds a multiple regression model as needed:
Incentive = β 0 + β 1 OC + β 2 Separation + β 3 Research + β 4 ROA + β 5 Return + β 6 Value + β 7 Fixassent + β 8 PB + β 9 Lev + β 10 Firmage + β 11 Growth + β 12 State + β 13 COCEN + β 14 Age + β 15 Tenur e + ∑ Industry + ∑ Year + ε (2)
The variables involved are shown in
This paper takes the senior management team of China’s A-share listed companies from 2010 to 2016 as the research object. The senior management team is
Variable nature | Variable name | Variable symbol | Variable definitions |
---|---|---|---|
Explained variable | Equity compensation incentive | Incentive | See formula (1) |
Explanatory variables | Executive overconfidence | OC | Profit forecasting method |
Company level Control variable | Separation of two posts | Separation | Take 1 when the chairman and general manager are not the same person, otherwise take 0 |
Growth | Growth | (Main income of the current period − main income of the previous period)/main income of the previous period | |
R & D intensity | Research | R & D expenses/main business income | |
Return on total assets | ROA | Return on total assets, total profit/total assets | |
Market ratio | PB | Total market value/net assets | |
Stock return | Return | Annual return on individual stocks | |
Company market value | Value | Natural logarithm of the company’s market capitalization | |
Company age | Firm age | Number of years from the company’s listing to the statistical deadline | |
Fixed assets | Fixasset | The natural logarithm of the company’s fixed assets | |
Financial leverage | Lev | Asset-liability ratio, total liabilities/total assets | |
Nature of property | State | The value is 1 when the company is a central/local state-owned enterprise, otherwise it is 0. | |
Equity concentration | COCEN | Proportion of the top ten shareholders of the company | |
Executive level Control variable | Average age of executives | Age | Average age of all senior managers |
Executive average term | Tenure | Average term of all senior management |
the company’s manager, deputy manager, financial controller, board secretary of the listed company and other senior management personnel as stipulated in the company’s articles of association. The relevant data on equity compensation (that is, the number of stock options and restricted stock held by company executives at the end of the year) was collected manually from the announcement of the company’s relevant equity incentives. Other data were obtained from the wind database and the Guotaian database. In order to ensure the accuracy and effectiveness of the data, the following screening process is adopted: 1) Excluding companies that have not disclosed their performance announcements or performance forecasts in the following year; 2) Excluding companies that are ST, PT and delisted; 3) Excluding Financial listed companies; 4) Excluding companies with large changes in executives; 5) Excluding companies with abnormal data; 6) Ending observations outside the 1% quantile of all consecutive variables deal with. In the end, 1686 listed companies received a total of 9979 annual observations for 7 years. In this paper, descriptive statistics and multiple regression analysis of data are carried out by Stata.
Whole sample | Rationality (OC = 0) (1) | Overconfidence (OC = 1) (2) | Difference (3) = (1) − (2) | ||||
---|---|---|---|---|---|---|---|
Max | Min | STD | Mean | ||||
Incentive | 8.347 | 0 | 1.685 | 0.507 | 0.537 | 0.356 | 0.181*** |
separation | 1.000 | 0 | 0.443 | 0.731 | 0.7410 | 0.619 | 0.122*** |
Research | 25.300 | 0 | 4.077 | 3.921 | 3.952 | 3.616 | 0.337** |
Growth | 234.300 | −100.000 | 40.830 | 16.686 | 16.892 | 14.376 | 2.517* |
ROA | 23.250 | −22.460 | 6.383 | 3.795 | 4.034 | 1.117 | 2.917*** |
PB | 51.060 | −1.246 | 6.438 | 5.016 | 5.038 | 4.766 | 0.272 |
Return | 15.210 | −0.838 | 0.633 | 0.218 | 0.218 | 0.211 | 0.007 |
Value | 19.110 | 9.398 | 0.944 | 13.226 | 13.248 | 12.979 | 0.269*** |
Firm age | 26.050 | 0.997 | 6.227 | 9.3983 | 9.692 | 6.105 | 3.586*** |
Fixasset | 18.040 | −2.013 | 1.739 | 10.772 | 10.773 | 10.769 | 0.004 |
Lev | 102.800 | 4.635 | 22.460 | 43.958 | 43.829 | 45.412 | −1.583* |
State | 1.000 | 0 | 0.473 | 0.337 | 0.355 | 0.136 | 0.219*** |
COCEN | 98.550 | 1.320 | 15.670 | 57.416 | 57.394 | 57.668 | −0.275 |
Age | 60.630 | 32.330 | 3.611 | 46.417 | 46.493 | 45.569 | 0.924*** |
Tenure | 8.743 | −0.266 | 1.842 | 3.419 | 3.428 | 3.317 | 0.111* |
Note: *p < 0.1, **p < 0.05, ***p < 0.01 (two-tailed test).
observations of overconfidence of executives accounted for 8.2% of the total sample. It can be seen that the average value of equity compensation incentive intensity is only 0.51%, the minimum value is 0, the maximum value is 8.35%, and there are significant differences between the two subsamples. Rational executives have greater equity compensation incentive intensity than overconfident executives. This is consistent with our motivational assumptions. In addition, it can be found that, compared with overconfident executives, companies with rational executives have more R & D investment, higher growth, higher profitability, larger enterprise scale, and smaller financial leverage; Executives in state-owned enterprises are relatively more rational.
Before the empirical analysis, the Pearson correlation analysis is used to test whether there is a multi-collinearity problem in the model. The results are shown in
The model (2) is tested by full-sample regression.
Incentive | OC | Research | ROA | Return | Value | Fixasset | PB | lev | Firm age | Growth | State | COCEN | Separation | Age | |
---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
Incentive | 1.000 | ||||||||||||||
OC | −0.040*** | 1.000 | |||||||||||||
Research | 0.092*** | −0.051*** | 1.000 | ||||||||||||
ROA | 0.110*** | −0.112*** | 0.039*** | 1.000 | |||||||||||
Return | 0.111*** | −0.023** | 0.051*** | 0.077*** | 1.000 | ||||||||||
Value | 0.100*** | −0.004 | 0.010 | 0.255*** | 0.220*** | 1.000 | |||||||||
Fixasset | −0.027*** | 0.034*** | −0.283*** | −0.046*** | −0.092*** | 0.472*** | 1.000 |
PB | 0.003 | −0.015 | 0.054*** | −0.024** | 0.280*** | 0.025** | −0.306*** | 1.000 | |||||||
---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
lev | −0.075*** | 0.098*** | −0.338*** | −0.382*** | −0.031*** | 0.027*** | 0.245*** | 0.137*** | 1.000 | ||||||
Firm age | −0.124*** | 0.097*** | −0.214*** | −0.187*** | −0.021** | 0.114*** | 0.084*** | 0.147*** | 0.396*** | 1.000 | |||||
Growth | 0.098*** | −0.034*** | −0.007 | 0.256*** | 0.080*** | 0.107*** | −0.088*** | 0.055*** | 0 | −0.081*** | 1.000 | ||||
State | −0.192*** | 0.046*** | −0.175*** | −0.129*** | −0.108*** | 0.111*** | 0.296*** | −0.058*** | 0.304*** | 0.415*** | −0.103*** | 1.000 | |||
COCEN | 0.036*** | −0.043*** | −0.012 | 0.243*** | 0.030*** | 0.239*** | 0.124*** | −0.133*** | −0.174*** | −0.383*** | 0.078*** | −0.046*** | 1.000 | ||
Separation | −0.042*** | 0.003 | −0.120*** | −0.059*** | −0.050*** | 0.039*** | 0.117*** | −0.021** | 0.140*** | 0.204*** | −0.033*** | 0.262*** | −0.041*** | 1.000 | |
Age | −0.081*** | 0.004 | −0.121*** | −0.061*** | −0.018* | 0.186*** | 0.288*** | −0.035*** | 0.121*** | 0.214*** | −0.117*** | 0.303*** | −0.020** | 0.059*** | 1.000 |
Tenure | 0.048*** | −0.050*** | 0.059*** | −0.025** | 0.123*** | 0.143*** | 0.140*** | −0.041*** | −0.051*** | 0.107*** | −0.112*** | 0.011 | −0.081*** | −0.052*** | 0.310*** |
Note: *p < 0.1, **p < 0.05, ***p < 0.01 (two-tailed test).
(1) Incentive | (2) Incentive | (3) Incentive | |
---|---|---|---|
OC | −0.151** | −0.225*** | −0.234*** |
(−2.46) | (−3.13) | (−3.26) | |
Separation | 0.107** | 0.107** | |
(2.32) | (2.31) | ||
Research | 0.028*** | 0.028*** | |
(4.83) | (4.70) | ||
ROA | 0.019*** | 0.019*** | |
(4.46) | (4.37) | ||
Return | 0.148*** | 0.147*** | |
(3.61) | (3.58) |
Value | 0.202*** | 0.206*** | |
---|---|---|---|
(5.61) | (5.70) | ||
Fixasset | −0.070*** | −0.065*** | |
(−3.15) | (−2.90) | ||
PB | −0.012** | −0.012** | |
(−2.49) | (−2.44) | ||
Lev | 0.007*** | 0.007*** | |
(5.26) | (5.29) | ||
Firm age | −0.015*** | −0.014*** | |
(−3.17) | (−3.04) | ||
Growth | 0.004*** | 0.003*** | |
(5.96) | (5.78) | ||
State | −0.690*** | −0.656*** | |
(−12.61) | (−11.74) | ||
COCEN | −0.003* | −0.003* | |
(−1.92) | (−1.74) | ||
Age | −0.047*** | −0.022*** | |
(−9.48) | (−3.48) | ||
Tenure | 0.019* | 0.026* | |
(1.68) | (1.88) | ||
cons | 2.285*** | −1.551*** | −0.707 |
(8.73) | (−3.82) | (−1.48) | |
Industry/Annual Fixed Effect | Yes | Yes | Yes |
N | 9977 | 7962 | 7961 |
Adjust-R2 | 0.031 | 0.081 | 0.082 |
Note: *p < 0.1, **p < 0.05, ***p < 0.01 (two-tailed test).
regression. The first (1) column is the regression result when the characteristic variables at the senior management level are controlled, and the second (2) column is used to control the company-level characteristic variables. As a result, item (3) is the result of controlling both the executive level and the company level characteristic variables. In addition to the control variables, this paper also controls the annual fixed effect and the industry fixed effect.
The regression results show that the coefficient of executive overconfidence (OC) is always negative (−0.234) and significant at the 1% level. This result shows that over-confident executives receive lower-intensity equity compensation than rational executives, which is consistent with our assumptions. That is, relative to rational executives, less equity compensation (convex compensation) is enough to motivate overconfident executives to take risk. Excessive equity compensation can make overconfident executives make risky decisions, which is damaged to shareholder value.
The results related to the control variables are also consistent with the existing literature. The size of the company (coefficient of Value is 0.206), the term of the executives (coefficient of Tenure is 0.026) and the incentive intensity of the equity compensation are positively correlated. The age of the company (coefficient of Firm age is −0.014) and the age of the executives (coefficient of Age is −0.022) are negatively correlated with the incentive intensity of the equity compensation, profitability. Companies with higher (coefficient of Return is 0.147 and coefficient of ROA is 0.019) are more likely to use equity compensation, and the greater the company’s risk (coefficient of Research is 0.028, coefficient of lev is 0.007, and coefficient of Growth is 0.003), the more equity compensation is needed to attract executives, and the results also indicate shareholders. Supervision (coefficient of COCEN is −0.003) and equity incentives can be substituted for each other. In addition, non-state-owned enterprises have higher incentives for equity compensation (coefficient of Separation is 0.107).
To further verify the correctness of the hypothesis, the author also used multiple methods to test the robustness of the regression results.
We construct the equity compensation incentive dummy variable (Incentive_dum) instead of the equity compensation incentive strength variable. When the company implements the equity incentive for the executive, Incentive_dum takes 1; otherwise, it takes 0. The dummy variable is replaced by the equity compensation incentive intensity variable to perform logit regression on the above model to test whether the result is robust. The results are shown in column (1) of
When we performed the White test, we found that the model has heteroscedasticity. Therefore, the model was modified by White’s robust standard error. The results are shown in column (2) of
In order to further eliminate the effects of heteroscedasticity and model error setting on the results, the regression coefficients are more effective. Glesjer’s test is used to determine the possible heteroscedastic form, and the hypothesis model is re-estimated by generalized least squares (FGLS). The results are shown in
(1) Incentive_dum | (2) Incentive | (3) Incentive | |
---|---|---|---|
OC | −0.529*** | −0.234*** | −0.263*** |
(−4.13) | (−3.47) | (−28.43) | |
Separation | −0.081 | 0.107** | 0.111*** |
(−1.16) | (2.16) | (20.36) | |
Research | 0.059*** | 0.026*** | 0.025*** |
(6.95) | (3.88) | (38.05) | |
ROA | 0.060*** | 0.019*** | 0.019*** |
(7.72) | (4.23) | (44.16) | |
Return | 0.006 | 0.147*** | 0.070*** |
(0.09) | (2.74) | (13.80) | |
Value | 0.417*** | 0.206*** | 0.218*** |
(6.65) | (5.64) | (58.95) | |
Fixasset | −0.082** | −0.065*** | −0.067*** |
(−2.16) | (−2.89) | (−31.15) | |
PB | −0.065*** | −0.012*** | −0.012*** |
(−4.99) | (−3.10) | (−30.79) | |
Lev | 0.013*** | 0.007*** | 0.007*** |
(5.55) | (5.52) | (52.23) | |
Firm age | −0.045*** | −0.014*** | −0.014*** |
(−5.38) | (−3.34) | (−30.44) | |
Growth | 0.002*** | 0.003*** | 0.003*** |
(2.63) | (4.61) | (40.14) | |
State | −2.251*** | −0.656*** | −0.617*** |
(−14.79) | (−14.50) | (−92.32) | |
COCEN | −0.010*** | −0.003* | −0.003*** |
(−3.66) | (−1.82) | (−20.76) | |
Age | −0.044*** | −0.022*** | −0.026*** |
(−4.23) | (−3.24) | (−33.54) | |
Tenure | 0.030 | 0.026* | 0.016*** |
(1.35) | (1.78) | (9.86) | |
cons | −5.275*** | −0.707 | −0.747*** |
(−6.13) | (−1.58) | (−11.29) | |
Industry/Annual Fixed Effect | Yes | Yes | Yes |
N | 7940 | 7961 | 7961 |
R2 | 0.187 | - | - |
Count R2 | 83.12% | 0.082 | 0.979 |
Note: *p < 0.1, **p < 0.05, ***p < 0.01 (two-tailed test).
There may be sample selection bias in measuring earnings overconfidence using earnings forecasts. That is to say, when there is at least one annual profit forecast between 2010 and 2016, the variable can be constructed. However, China’s profit forecast system is semi-mandatory, and executives have the space to choose whether to publish profit forecasts. Executives provide earnings forecasts and executive compensation, so the previous analysis may have sample selection bias. To solve the biased estimates caused by sample selection bias, we estimated a Heckman two-stage model.
(1) In sample | (2) Incentive | (3) Incentive_dum | |
---|---|---|---|
OC | −0.185*** | −0.502*** | |
(−2.59) | (−3.91) | ||
Separation | 0.064** | 0.222*** | 0.058 |
(2.14) | (4.76) | (0.80) | |
Research | −0.489 | 0.020*** | 0.049*** |
(−1.32) | (3.38) | (5.61) | |
ROA | −0.010*** | 0.056*** | 0.100*** |
(−3.23) | (10.93) | (10.69) | |
Return | 0.037 | 0.143*** | −0.028 |
(1.19) | (3.54) | (−0.40) | |
Value | −0.049** | 0.338*** | 0.567*** |
(−2.26) | (9.11) | (8.49) | |
Fixasset | 0.022* | −0.123*** | −0.145*** |
(1.72) | (−5.43) | (−3.58) | |
PB | 0.001 | -0.015*** | -0.060*** |
(1.22) | (−3.04) | (−4.61) | |
LEV | 0.003*** | −0.001 | 0.004* |
(3.40) | (−0.45) | (1.67) | |
Firmage | −0.041*** | 0.076*** | 0.052*** |
(−15.34) | (9.11) | (3.36) |
Growth | 0.001*** | 0.001 | −0.001 |
---|---|---|---|
(3.21) | (1.12) | (−0.88) | |
State | −0.279*** | 0.075 | −1.365*** |
(−8.60) | (0.96) | (−7.36) | |
COCEN | 0.001 | −0.005*** | −0.013*** |
(0.73) | (−3.28) | (−4.87) | |
Age | −0.014*** | 0.008 | −0.011 |
(−3.58) | (1.22) | (−0.99) | |
Tenure | −0.001 | 0.025* | 0.031 |
(−0.13) | (1.84) | (1.36) | |
EPS | −0.107*** | - | - |
(−3.01) | - | - | |
IMR | - | −4.390*** | −5.095*** |
- | (−13.13) | (−7.93) | |
cons | 2.036*** | −1.503*** | −6.088*** |
(6.79) | (−3.15) | (−6.95) | |
Industry/Annual Fixed Effect | Yes | Yes | Yes |
N | 11,910 | 7961 | 7940 |
Adjust-R2 | - | 0.102 | - |
Note: *p < 0.1, **p < 0.05, ***p < 0.01 (two-tailed test).
Columns (2) and (3) are the results of the second phase of regression. Only the company that publishes at least one profit forecast constitutes a sample of this stage. The dependent variables are the equity compensation incentive intensity and the equity compensation dummy variable. The model is set as before, but the inverse mir calculated according to the results of the first stage is added. The ratio (IMR) is used as an additional control variable. It can be found that in both regressions, the coefficients of the IMR are significantly negative, indicating that there is a sample selection bias in the previous analysis. But after we control this deviation, the coefficient of the overconfidence variable is still significantly negative, except that the absolute value of the coefficient is slightly smaller than the result before controlling the sample selection deviation (
Through the above three methods for robustness testing, it can be verified that the empirical results in the previous paragraph are indeed robust and effective.
This paper examines the relationship between executive overconfidence and equity compensation incentives by taking samples of Chinese listed companies from 2010 to 2016. The study found that the company would give overconfident executives a lower-intensity equity compensation incentive. Because overconfident executives have an upward bias toward the company’s prospects and equity compensation, lower-intensity equity compensation incentives are enough to encourage executives to work hard, so the company can take this into account when designing executive compensation contracts.
This paper incorporates the psychological factors of executives into the research framework of compensation contract, and considers the influencing factors of executive compensation from the perspective of behavioral finance, and broadens the research perspective of executive compensation. The results of this paper show that when the company signs a compensation contract with the executive, in addition to considering the company’s characteristics and executive capacity, it also considers the psychological characteristics of the executive. To a certain extent, overconfidence will alleviate the agency costs brought about by the risk aversion of executives, and will play a certain role in the substitution of equity compensation. Therefore, the company will reduce the incentives for equity compensation granted. In short, whether executives are overconfident is an important consideration for the company to develop an optimal executive compensation contract.
The limitation of this paper is that using simple method instead of Vega to measure the incentive intensity of equity remuneration, which is not accurate measured directly by Vega, and it is what we need further study in the future.
The authors declare no conflicts of interest regarding the publication of this paper.
Chen, S.Y. (2019) Weakening Effect of Executive Overconfidence on Equity Incentive―The Empirical Evidence from Chinese Listed Companies. Open Journal of Business and Management, 7, 151-166. https://doi.org/10.4236/ojbm.2019.71011