Estimating the Speed of Adjustment of Leverage in the Presence of Interactive Effects*

Dynamic panel data regression models with fixed effects to account for unobserved heterogeneity are standard econometric tools. It is not until recently, however, that the problems involved when fitting such regressions to leverage data have been investigated. The main problem is that models of leverage are extremely noisy, much more so than what can be accommodated using fixed effects. The present article can be seen as a reaction to this. The purpose is to consider a more general interactive effects model in which there are multiple time effects, each with their own firm-specific sensitivities. Our empirical results suggest that proper accounting for the interactive effects and the bias that they cause leads to a marked increase in the estimated speed of adjustment to target leverage.