We use equation (3.1) to estimate the effect of CEO option-based incentives (vega and delta) on firms’ risky financial policy (book leverage) and risky investment policy (business acquisitions intensity), holding other variables constant:
, , = + , + , + , +
, + & , + , + , + , + +
+ _ + , , (3.1)
where lnLEVERAGE is the natural logarithm of book leverage; ACQUISITION is annual business acquisitions expense normalised by market value of assets. We choose book leverage instead of market leverage as the leverage measure because Coles et al. (2006) argue that book leverage better reflects CEOs’ decision making, whereas share market performance easily affects market leverage and therefore is not a good indicator of active managerial discretion. Further, we choose book leverage because a few sample firms in this study have their debts measured in terms of tradable bonds and some of the firms’ debt are non-tradable debts such as bank loans which are recorded in book value terms. On the other hand, we use Coles et al.’ (2006) method to estimate the market market leverage as a robustness check45.
This study also aims to examine the effects of CEO option incentives on firms’ risky investment policy. We use firms’ business acquisitions intensity as the measure for the CEOs’ risky investment decisions. This is because Hagendorff and Vallascas (2011) argue that merger and acquisition decisions are the most important investment decisions made by CEOs.
We control the following variables in equation (3.1): firm size, firm growth opportunities and R&D expenses. One possible firm debt policy determinant is firm size. Because of the difficulty in monitoring as a firm’s size increases, Dong et al. (2010) argue that larger firms are more likely to adopt high leverage in their capital structure than smaller firms. Consistent with Dong et al. (2010), we use the natural logarithm of sales (lnSALES) as the proxy for firm size. We also control for firms’ growth opportunities, represented by book to market value of assets (BM). Myers (1977) argues that a firm’s assets are categorised into assets in place and future growth opportunities. Firms with great growth opportunities are expected to have a low leverage in their capital structure, because equity financing could better address an underinvestment problem through risky external borrowing
(Myers, 1977; Billett, King & Maucer, 2007). Another control variable is R&D expenditure. Hall (2002) argues that banks are reluctant to lend credit to R&D-intensive firms that do not have enough physical assets as collateral. Further, R&D investments have greater uncertainty to generate stable funds in order to service the debt (Hall, 2002).
Table 3.3 Variable definitions for equations (3.1) and (3.2) Variable name Variable Definition
Dependent variable
LEVERAGE Book leverage. Book value of assets minus book value of equity, over book value of equity ACQUISITION Business acquisitions expense scaled by market
value of assets Independent variable
Delta (A$000’s) Pay-performance sensitivity. Change in the dollar value of a CEO’s option portfolio for a 1% change in share price (Core and Guay, 2002)
DELTA Dollar delta/CEO total compensation. Vega(A$000’s) Change in the dollar value of a CEO’s option
portfolio for a 1% change in volatility of share return (Core and Guay, 2002)
VEGA Dollar vega/CEO total compensation Control variables
AGE Age of CEO
BM Book to market value of assets. Book value of assets over the sum between book value of debt and market value of equity.
CASH (CEO cash salary + bonus + short term cash incentives)/ CEO total compensation CEO_CHANGE CEO change dummy
INDUSTRY Industry dummy
INSIDER Number of executive directors over total number of board directors
PPE Property, plant and equipment scaled by market value of assets
R&D Research and development expenses scaled by market value of assets
SALES Dollar value of gross sales in millions
TENURE Number of years the CEO has been employed holding the current position
VOLATILITY Annualised standard deviation of logarithmic daily share return
YEAR Year dummy
P/E Option Moneyness. Year-end (30 June)
underlying share price over option exercise price at year end (30 June)
Note: CEO total compensation is defined as cash compensation plus option-based compensation fair price, non-cash benefits and superannuation.
In equation (3.1), we also control for CEOs’ cash compensation (CASH), because Guay (1999) suggests that the higher the CEOs’ outside wealth the more the risk-taking CEOs are going to be. If a CEO has
more power over other directors in the boardroom, then the CEO will have more influence in setting his or her compensation incentives. Thus, we control for the CEO power proxies, including CEO age (AGE) and CEO tenure (TENURE),
Further, we control the following dummy variables in equation (3.1): YEAR, CEO_CHANGE and Industry. We use YEAR as a dummy variable to capture the year effects which is equals to 1 for a given year excluding 2003, and zero otherwise. We include a CEO change dummy (CEO_CHANGE), which takes a value of 1 when the firm has a new CEO that year; 0 otherwise, to control for the possible impact of CEO turnover on the firm’s debt policy. Industry is the industry dummy. We code Industry using the GICS sector code. If a firm is in the corresponding industry group, then Industry value is 1; otherwise 0. The cross-sectional error term, E, is composed of two effects: the unobserved firm specific effect and an idiosyncratic error, which varies across time and section. Table 3.3
provides the detailed definitions of the variables used in equation (3.1).
The firms’ financial and investment policy may be influenced by the CEOs’ option incentives (vega and delta). It is also true that the CEOs’ option incentives are affected by the firms’ risky corporate policies. For example, John and John (1993) argue that shareholders of high-leverage firms
discourage high-vega executives’ incentives in order to be less exposed to the bankruptcy risk. Géczy, Minton, and Schrand (2007) find that firms speculating using derivatives are more likely to use option-based compensation scheme in order to encourage the executives to take more risks. Further, Low (2009) finds that the decrease in risk is associated with low vega, which suggests that vega and delta are correlated with each other. Thus, we specify a simultaneous system of equations (three- stage least squares or 3SLS) taking into account the covariances across equation error terms46:
, = , , , , , , (3.2) , = , , , , , , , (3,3)
, = , , , , , . (3.4)