As argued by Jensen (1993), the primary tasks of a firm’s board of directors are to advise, hire, fire and determine the level and form of managerial compensation. Managerial pay can be structured as part cash and in part be tied to a performance index, such as corporate earnings or the firm’s stock price. The latter effectively aligns the interest of managers with those of stockholders, which in turn reduces agency problems related to free cash flow, managerial time horizons and effort levels. At the same time, stock-based compensation increases managerial exposure to non-diversifiable risk, which may cause risk-averse managers to underinvest in risky projects. The trade-off between the benefits of managerial incentive alignment and the cost of underinvestment is largely an empirical issue, and the widespread observation that managerial compensation is primarily paid in cash 1 suggests that managerial risk aversion weighs heavily or that boards generally resort to substitute monitoring mechanisms. The paper by Angbazo and Narayanan (1997) is part of a rapidly growing empirical literature attempting to identify important cross-sectional determinants of the form of executive compensation. Motivated in particular by the early work of Fama (1980) and Fama and Jensen (1983), this literature conjectures that executive pay is not only motivated by manager‐shareholder incentive alignment and risk preferences, but also by the workings of the firm’s corporate governance structure. This literature recognizes that the board of directors is an imperfect agent for shareholders, as board members have their own private incentives and struggle with informational asymmetries vis-` the top management, all of which affects the board’s monitoring effectiveness. As discussed below, this view suggests that factors such as board size, board longevity (tenure), the proportion of the board consisting of independent or outside directors, whether the CEO is also chairman of the board, the role of institutional investors as board members, etc., along with 1 See, e.g., Jensen and Murphy (1990), Holthausen and Larcker (1993), and Hwang and Anderson