The use of covariance structure modeling is explored as a means of moving toward a resolution of the debate over the antecedents of executive compensation. The major strength of this methodology is that it enables researchers to measure the effects of unobserved factors on measured variables. It is suggested that covariance structure modeling is a promising way of studying the effect of institutional isomorphism on executive compensation. The popular business press has questioned repeatedly the justification for and the performance effects of prevailing executive compensation systems (Crystal, 1988; Loomis, 1982; Patton, 1985). These articles argue that executives are more interested in creating wealth for themselves than for stockholders. They also underscore the absence of an obvious link between executive compensation and firm performance. Recent academic research on executive compensation adopts an agency perspective that emphasizes potential conflicts of interest between managers and stockholders. It contends that, in the absence of effective disciplining and monitoring systems, executive compensation plans may direct managers' efforts toward personal wealth enhancement to the detriment of firm value (Baumol, 1958; Berle & Means, 1932). In response, scholars have urged that executive compensation plans contain monetary incentives that only accrue to executives when shareholder wealth is maximized (Kerr, 1985; Rappaport, 1983; Tehranian & Waegelein, 1985). However, designing compensation systems that effectively align the interests of managers and stockholders requires a knowledge of the role and effect of relevant driving forces on compensation. Statistical research on executive compensation has been guided predominantly by a search for tangible, observable determinants (Ciscel & Carroll, 1980), examples of which have been firm size or growth rate (Baumol, 1967; Marris, 1963), inter‐firm and inter‐in‐dustry differences (Coughlan & Schmidt, 1985), and performance (Murphy, 1986). The emphasis on such tangible explanations is not surprising given the overwhelming use of econometric techniques, such as ordinary least squares regression (Ciscel & Carroll, 1980; Finkelstein & Hambrick, 1988), logistic regression (Walking & Long, 1984), time series analysis (Murphy, 1985), and event studies (Brickley, Bhagat & Lease, 1985; Coughlan & Schmidt, 1985; Tehranian & Waegelein, 1985). This paper argues that the focus on tangible, observable variables by compensation researchers is a methodologically ‐ driven practice that constrains theory building and testing. As a result, we may have ignored interesting and relevant theoretical frameworks for the study of executive compensation. We also have overlooked the use of analytical techniques that allow us to examine the role of potentially relevant latent constructs. In this paper, we will describe and illustrate the use of covariance structure modeling for the study of institutional pressures on executive compensation.
Read full abstract