Optimal dynamic capital structure choice is fundamentally a problem of commitment. In a standard trade‐off setting with shareholder‐debtholder agency conflicts, full commitment counterfactually predicts the firm would rely almost exclusively on debt financing. Conversely, absent commitment a Modigliani‐Miller‐like value irrelevance and policy indeterminacy result holds. Thus, the content of dynamic trade‐off theory must depend on the commitment technology. In this context, collateral is valuable as a low‐cost commitment device. Because ex ante optimal commitments are likely to be suboptimal ex post, observed capital structure dynamics will exhibit hysteresis and depart significantly from standard predictions.