Debt, Incentives and Performance: Evidence from UK Panel Data

Abstract

A large body of theoretical literature suggests that capital structure plays an important role as a managerial incentive mechanism. What of the evidence for the agency approach? Crosssectional empirical studies have identified a positive effect of leverage on expected performance (measured by Q) for firms with low growth opportunities. However, this evidence does not take into account the endogeneity of capital structure decisions. Our paper investigates the determinants of capital structure and performance, allowing for endogeneity and dynamics. Our results suggest that conclusions reached by previous studies which did not take into account the endogeneity issue should be treated with caution. What determines firms' capital structure decisions? What are the implications for firm performance? These questions have attracted a great deal of attention in the theoretical literature on firms' financial structure; see Harris and Raviv (1991) for an excellent survey, as well as Hart (1995) and Shleifer and Vishny (1997) for more recent discussions. Our paper examines the same questions empirically, in the light of existing theories and notably the agency literature which highlights the potential role of debt as a mechanism to mitigate conflicts between management and shareholders (Jensen and Meckling, 1976; Grossman and Hart, 1982; Jensen, 1986; Harris and Raviv, 1991; Stulz, 1990; Hart and Moore, 1995; Zwiebel, 1996).1 We identify some key econometric problems which arise when translating such theories into an empirical model that can be estimated with available data. Our main contribution lies in the analysis of panel data, which enables us to deal with these econometric problems. In so doing, we are able to shed some light on the value of the underlying theoretical approach. At the same time, we show that results obtained in the existing empirical literature, which has largely overlooked the econometric issues we address, are not robust; the implications of these findings are then discussed. The argument is essentially as follows. In most contributions to the agency