In the paper, Managers Who Lack Style: Evidence from Exogenous CEO Changes, which was recently made publicly available on SSRN, we study managerial style effects in firm decisions by examining exogenous CEO changes in a panel of 8,615 Compustat firms from 1990 to 2007. The hypothesis that managers have varying preferences or traits that affect their corporate decisions has a great deal of intuitive appeal and is implicit in many discussions of leadership. Prior empirical evidence lends support to this general hypothesis and suggests that managerial style effects play a substantive role in firms’ investment and financing choices. This raises the possibility that much of the unexplained variation in these and related choices is driven by the identities of a firm’s leaders rather than more traditional factors such as various economic tradeoffs.
While prior research on this issue has generated many interesting findings, a major weakness arises from the fact that endogenous leadership changes are used to identify style effects. In this paper we overcome this weakness by identifying a large set of exogenous CEO changes arising from deaths, health concerns, and, in some parts of the analysis, natural retirements. Quite surprisingly, we find no significant evidence of abnormal changes in asset growth, investment intensity, leverage, or profitability subsequent to exogenous CEO changes.
If CEOs manage with style, they certainly do not appear to be present in our sample of managers who assume the top position in an exogenous leadership transition. Given the surprising contrast between our findings and prior evidence, we turn to examining the endogenous job changes in our sample. Here we find that CEO changes that are either overtly forced or likely forced are in fact followed by abnormally large changes in corporate policies. While our evidence based on exogenous changes cautions against any type of causal interpretation of this evidence, it is nonetheless informative. In particular, these findings add to large body of evidence suggesting that firms in crisis tend to simultaneously make numerous operational, governance, and managerial changes.
To complete the picture, we revisit the evidence of Bertrand and Schoar (2003) regarding style effects detected from endogenous events in which executives move across employers. While we do estimate significant F-statistics on manager fixed effects, we also find that these test statistics are highly flawed indicators of the presence of managerial style effects in this setting. We consider more robust tests, but these tests reveal no presence of managerial style effects that executives bring from one employer to the next. Thus, while endogenous management changes are interesting, we find little additional evidence on style effects within the subset of job movers.
Taken as a whole, our evidence offers a much more cautionary view on the importance of managerial style effects in firm policies than prior work would suggest. In particular, for the types of firms that are in our exogenous change sample, namely a typical cross-section of firms operating in relatively stable environments with no apparent crisis, large style effects are nowhere to be seen or detected. This suggests a cautious perspective on the view that much of the unexplained variation in firm policy choices can be attributed to manager-specific traits or idiosyncrasies. It remains possible that style effects are important in a much more limited set of situations, namely when firms are in crisis and choose to force out their incumbent leader.
However, it will be exceedingly difficult to convincingly identify causality in these situations, since the effect of the actual management change will be very hard to separate from the economic impetus underlying the decision to make the change. It is also possible that managerial style manifests itself in other, more subtle, areas than the ones we consider. For example managerial style may affect the type of investment undertaken, the type of debt a firm issues, the communication style of the CEO, the motivational abilities of the CEO, etc. Future research will need to investigate these types of possibilities. It is also possible that managers’ styles change over time (a managerial mood effect), and thus the within-manager variation in style varies as much as the across-manager variation.
Finally, a secondary contribution of our paper is a methodological point. Our analysis indicates, in a fairly dramatic fashion, that inferences from traditional tests on the joint significance of fixed-effects dummy variables with multiple observations for each unit is highly suspect in almost any setting typical encountered in corporate finance. In many cases there will be no information content in these tests, even when reported p-values are well below .0001.
Traditional estimation approaches, such as those discussed by Petersen (2009), do not solve this problem. These approaches are intended to generate asymptotically valid inferences on other explanatory variables, not the fixed-effects themselves. We exploit a simple solution based on computing changes and deriving an empirical distribution for comparison from a set of matching firms, but certainly other non-parametric empirical approaches would also be valid.
The full paper is available for download here.