Endogeneity and the Dynamics of Internal Corporate Governance

The following post comes to us from M. Babajide Wintoki of the Department of Finance at the University of Kansas, and James Linck and Jeffry Netter, both of the Department of Banking and Finance at the University of Georgia.

In our forthcoming Journal of Financial Economics paper, Endogeneity and the Dynamics of Internal Corporate Governance, we use a well-developed dynamic panel generalized method of moments (GMM) estimator to alleviate endogeneity concerns in two aspects of corporate governance research: the effect of board structure on firm performance and the determinants of board structure. It is well known that theoretical and empirical research in corporate finance is complicated by the endogenous relation that exists between the control forces operating on a firm and its decisions. Jensen (1993) broadly classifies these control forces (i.e., governance in a broad sense) as capital markets, the regulatory system, product and factor markets, and internal governance. In much of the extant corporate finance research, researchers attempt to either explain the causes or examine the effects of corporate finance decisions as related to one or more of these control forces. Empirical research often involves determining the causal effect, if any, of a firm characteristic (X) on some measure of firm profits or value (Y). This is usually done using the inference from a regression of Y on X along with several control variables (Z). The question is often framed as: holding Z constant, does X have an economically and statistically significant causal effect on Y?

To date, most empirical research in corporate finance has explicitly recognized at least two sources of endogeneity that may bias estimates of how X affects Y: unobservable heterogeneity (which arises if there are unobservable factors that affect both the dependent and explanatory variables) and simultaneity (which arises if the independent variables are a function of the dependent variable or expected values of the dependent variable). However, we argue that empirical research often overlooks an important source of endogeneity that arises because the relations among a firm’s observable characteristics are likely to be dynamic. That is, a firm’s current actions will affect its control environment and future performance, which will in turn affect its future actions. For example, in the context of board structure, current firm performance will affect future governance choices and these may, in turn, affect future firm performance. We note our model fits well with the theoretical model developed by Harris and Raviv (2008), who show that board structure is not exogenous and not a determinant of performance, but both are functions of other variables. They suggest that finding a relation between board structure and performance may simply be spurious.

As in Himmelberg, Hubbard, and Palia (1999), we do not intend to minimize or ignore the importance of agency conflicts or suggest that governance is irrelevant; rather, we argue that the cross-sectional variation in observed governance structures is driven by both unobservable heterogeneity and the firm’s history. As such, any attempt to explain the determinants of governance or its effect on performance that does not recognize these sources of endogeneity may be biased.

We first discuss the theory behind the GMM estimator and explain why it is appropriate for estimating the governance/performance relation in a dynamic framework. We show the advantage over fixed-effects estimators which are biased when the dynamic relation between the variable of interest and the explanatory variables is important. We specifically apply this technique to estimate the determinants of board structure and the effect of board structure on performance in a panel of 6000 firms from 1991–2003. We find that board structure is, in part, determined by past performance, and after accounting for this, we find no causal relation between board size or independence, and firm performance. We show that bias may explain the results of earlier studies that do not consider dynamics when estimating the board structure-performance relationship. We find that the broad conclusions of existing research that examine the relation between firm characteristics and board structure are relatively unaffected even after we account for any potential effects of past governance on current values of the determinants. This suggests that dynamics are less important in this setting.

While our research concentrates on board structure and performance, others have applied  dynamic panel GMM estimators in other areas (e.g., financial development and growth literature, determinants of capital structure, etc.). However, it is likely to be particularly important in corporate governance since much of this research seeks to determine the effect of governance on performance, an aspect of research that is particularly susceptible to biases that may arise by ignoring the effect of historical performance on current governance.

The full paper is available for download here.

Both comments and trackbacks are currently closed.
  • Subscribe or Follow

  • Supported By:

  • Program on Corporate Governance Advisory Board

  • Programs Faculty & Senior Fellows