We use the following equation to estimate the effects of CEO option-based incentives (vega and delta) on the firms’ current market value, holding other variables constant:
46We calculated the cross-equation correlation of error to detect the endogeneity of simultaneous correlations between the error terms. The results show the correlation coefficient between the first equation is 0.302 and the second equation error terms; 0.347 between the second equation and the third equation error terms; 0.412 between the first equation and the third equation error terms.
, = + , + , + , + & , + , +
+ , + _ , + , + , + , +
+ + , , (3.5)
where the dependent variable is the natural logarithm of Tobin’s Q. Tobin’s Q is defined as the book value of debt plus market value of equity, scaled by book value of assets. YEAR is the year dummy;
INDUSTRY is the industry dummy; CEO_CHANGE is the CEO change dummy. The detailed definitions of the variables used in equation (3.5) are also documented in Table 3.2.
There are two measures of a firm’s value or performance as noted in the corporate governance literature. The first are accounting ratios, such as return on assets (ROA) and return on equity (ROE) (Hutchinson 2003; Fahlenbrach & Stulz, 2011). The second measure focuses on the share market price such as Tobin’s Q. Tobin’s Q is equal to the ratio of the sum of the market value of equity and the book value of debt, to the book value of assets (Aggarwal & Samwick, 2006). We chose Tobin’s Q rather than ROA or ROE as the proxy for firm market-based performance because: (1) Tobin’s Q is a better reflection of a firm’s market-based performance (Aggarwal & Samwick, 2006; Dong et al., 2010); and because: (2) the chief dependent variables such as delta and vega are market-based incentives.
We use a squared form of delta ( , ) in equation (3.5) in order to test our hypothesis (2) that
there is an inverted U-shaped relationship between delta and Tobin’s Q. In addition, there is a need to control risk-increasing incentive vega, which offsets delta’s risk-aversion effect. This is because, as the firm’s share price fluctuates, the high-delta CEO experiences equity risk (Guay, 1999). As a result, risk-averse CEOs would require additional compensation to assume an additional non-diversifiable risk (Bulan et al., 2010; Smith & Stultz, 1985).
We control the following firm characteristics in equation (3.5): firm size, R&D, PPE, and board of directors’ insider ratio. We use the logarithmic transformation of gross sales as the proxy for firm size. Larger firms are subject to more public monitoring, resulting in less information asymmetry and less CEO risk-aversion (Jensen & Meckling, 1976). R&D expense (compared with capital expenditure such as PPE) implies higher risk, i.e., more R&D-intensive firms will experience high volatility of share returns (Chan, Lakonishok & Sougiannis, 1999). Coles et al. (2006) argue that high-vega CEOs are more incentivised to reallocate investment funds away from tangible assets, such as capital expenditure, toward intangibles such as R&D expenses. Hence, if shareholders choose value-
maximising investment decisions, then more R&D-intensive firms will be more valued because of the high risk. Further, we control for the proportion of non-executive directors on the board of directors because, if a CEO is a board member, then it is very likely that other executive directors will be
influenced to vote in favour of the CEO when it comes to voting on compensation policies. Moreover, CEOs will be asked to undertake more risk-increasing and value-enhancing projects when there are more non-executive directors monitoring CEOs closely than executive directors on the board (Grabke-Rundell & Gomez-Mejia, 2002).
We also control for the following CEO characteristics: CEO cash compensation, CEO age, and CEO tenure. As cash compensation increases, CEOs’ risk aversion may increase, thus CEOs are more likely to pass up risky, but value-enhancing projects (Coles et al., 2006). A long-tenured CEO, who has been serving a firm for a relatively long period, is assumed to formulate and implement long-term
environmental sustainability decisions. However, as a CEO becomes more experienced , the CEO may have greater power to influence the board of directors to introduce less-risky firm polices, such as high-delta compensation plans, at the expense of reduced shareholders’ value (Bulan et al., 2010; Fahlenbrach & Stulz, 2011). We also control for CEO age; Hallock (1998) suggests that a CEO’s age can serve as a proxy of a CEO’s referent power because it is an estimate of human capital.
Hagendorff and Vallascas (2011) suggest that older CEOs are less subject to monitoring by regulators, which encourages risky investment.
A firm’s performance may be influenced by the CEO’s option incentives (vega and delta), but it is also likely that high-performance firms may grant their CEOs more option-based compensation. Demsetz and Villalonga (2001) suggest that there is a reverse causation in which the firm performance affects the management compensation in the form of share options. Demsetz and Villalonga (2001) attribute this reverse causation to the possible discrepancy between the investors and insiders expectations for the firm’s performance. This divergence of expectations generates an incentive for executives to alter their equity-holdings to meet their expectations regarding the firm’s future market performance. To address this potential endogeneity issue47, we specify a simultaneous
system of equations (3SLS) as follows:
′ , = , , , , , , (3.6)
, = , , , , , , (3.7) , = , , , , , , (3.8)
47We calculated the cross-equation correlation of error to detect the endogeneity of simultaneous correlations between error terms. We have found that the correlation coefficient between the first equation and the second equation error terms is 0.412; 0.302 between the second equation and the third equation error terms; 0.262 between the first equation and the third equation error terms. These large correlations show the necessity of applying the 3SLS method to our equation system.
3.6.3
M&M firms’ CER and CFP relationship
Both Waddock and Graves (1997) and Jo and Harjoto (2012) suggest that CER engagement may increase firm value, vice versa may also hold, i.e, higher-market-performance firms may invest more in CER related activities. In order to disentangle the unclear direction of causation between the endogenous CER engagement (as measured by REHAB) and firm financial performance (as measured by lnTobin’s Q), we use a 2SLS estimation method with the following equation system:
′
, = , , , , , , , , , , (3.9) , = , , , , , , , , , . (3.10)
Table 3.4 Variable definitions for equations (3.9) and (3.10). Variable Definition
CFP proxy
Tobin’s Q (Book value of debt + market value of equity)/book value of asset
CER proxy
REHAB Provision for site rehabilitation/sales CEO and firm characteristics
AGE Age of a CEO
CASH CEO cash salary + bonuses + short term cash incentives DELTA Change in the dollar value of a CEO’s options portfolio
for a 1% change in the share price/total compensation DERIVATIVE Year-end fair value of foreign exchange derivative from
a firm's financial report/sales
E&D Investment in exploration and mine development/sales LEVERAGE (Book value of assets - book value of equity)/book value
of equity
SALES Sales in AUD$ millions
TENURE Number of years a CEO has been employed in the current position
YEAR Year dummy
CEO_CHANGE CEO change dummy: the value of 1 when the firm has a new CEO that year; 0 otherwise
FIRM_CASH Cash balance in balance sheet
We take the natural logarithm of Tobin’s Q, sales, and book leverage in order to improve the skewed variables to be more normally distributed. In addition, we use CEO total compensation to scale delta, because both Hagendorff and Vallascas (2011) and Liu and Mauer (2011) argue that a CEO’s large absolute dollar values of delta may be relatively small compared with the CEO’s wealth; hence, scaled delta provides a better representation of the magnitude of economic incentives in CEO incentive plans. We insert another control variable FIRM_CASH (the firm’s cash at bank and in hand
in the firm’s balance sheet) to control for the firm’s resources48. The detailed definitions of the
variables used in equations (3.9) and (3.10) are documented in Table 3.4.
The IV for provisions for REHAB is E&D expenses. Mudd (2010) argues that for the M&M firms, true environmental sustainability commitment depends on technology innovation and scientific
knowledge, additional to social and environmental factors. Similarly, Filippou and King (2011) argue that the biggest exploration and development expense in the M&M industry is in improving extraction and metallurgical processes such as remote control systems and robots for underground mines. Thus, we use the investment in exploration and mine development (the sum of exploration and mine development expenses, divided by sales) as an IV for REHAB. We report the OLS first-stage partial F-statistics and Sargan Hansen over-identifying restriction test results in the estimation results section to show that our IVs are valid.
3.7
Conclusion
This chapter outlines the hypotheses and the methods for computing vega, delta and LTIP values, the relevant sample selection procedure, research design and estimation methods. The next chapter presents descriptive summaries and discusses the regression results of the study models.