We empirically evaluate 20 prominent contributions to a broad range of areas in the empirical corporate finance literature. We assemble the necessary data and then apply a single, simple econometric method, the connected-groups approach of Abowd, Karmarz, and Margolis (1999), to appraise the extent to which prevailing empirical specifications explain variation of the dependent variable, differ in composition of fit arising from various classes of independent variables, and exhibit resistance to omitted variable bias and other endogeneity problems.
In particular, we identify and estimate the role of observed and unobserved firm- and manager-specific characteristics in determining primary features of corporate governance, financial policy, payout policy, investment policy, and performance. Observed firm characteristics do best in explaining market leverage and CEO pay level and worst for takeover defenses and outcomes. Observed manager characteristics have relatively high power to explain CEO contract design and low power for firm focus and investment policy. Estimated specifications without firm and manager fixed effects do poorly in explaining variation in CEO duality, corporate control variables, and capital expenditures, and best in explaining executive pay level, board size, market leverage, corporate cash holdings, and firm risk. Including manager and firm fixed effects, along with firm and manager observables, delivers the best fit for dividend payout, the propensity to adopt antitakeover defenses, firm risk, board size, and firm focus. In terms of source, unobserved manager attributes deliver a high proportion of explained variation in the dependent variable for executive wealth-performance sensitivity, board independence, board size, and sensitivity of expected executive compensation to firm risk. In contrast, unobserved firm attributes provide a high proportion of variation explained for dividend payout, antitakeover defenses, book and market leverage, and corporate cash holdings. In part, these results suggest where empiricists could look for better proxies for what current theory identifies as important and where theorists could focus in building new models that encompass economic forces not contained in existing models.
Finally, we assess the relevance of omitted variables and endogeneity for conventional empirical designs in the various subfields. Including manager and firm fixed effects significantly alters inference on primary explanatory variables in 17 of the 20 representative subfield specifications.