Estimating the speed of adjustment toward target leverage using the standard partial adjustment model assumes that all firms within the sample adjust at the same (average) pace. Dynamic capital structure theory predicts heterogeneity in adjustment speed due to firm-specific adjustment costs. Applying an estimator designed to be unbiased in the context of unbalanced dynamic panel data with a fractional dependent variable (DPF estimator), we conduct an extensive analysis of cross-sectional heterogeneity in the speed of adjustment of firms. We find evidence for pronounced heterogeneity, where speed of adjustment is the highest for firms with high default risk or expected bankruptcy costs, and if opportunity costs of deviating from a target are high. Our evidence is consistent with the general relevance of the trade-off theory.

Researchers in empirical corporate finance often use bounded ratios (e.g., debt ratios) as dependent variables in their regressions. Using the example of estimating the speed of adjustment toward target leverage, we show by Monte Carlo and resampling experiments that commonly applied estimators yield severely biased estimates, as they ignore that debt ratios are fractional (i.e., bounded between 0 and 1). We propose a new unbiased estimator for adjustment speed in the presence of fractional dependent variables that also controls for unobserved heterogeneity and unbalanced panel data. This new estimator is suitable for corporate finance applications beyond capital structure research.

Researchers in empirical corporate finance often use bounded ratios (e.g., debt ratios) as dependent variables in their regressions. Using the example of estimating the speed of adjustment toward target leverage, we show by Monte Carlo and resampling experiments that commonly applied estimators yield severely biased estimates, as they ignore that debt ratios are fractional (i.e., bounded between 0 and 1). We propose a new unbiased estimator for adjustment speed in the presence of fractional dependent variables that also controls for unobserved heterogeneity and unbalanced panel data. This new estimator is suitable for corporate finance applications beyond capital structure research.

No abstract