Abstract:

In most existing models of capital structure， the dynamics of cash flow/asset value are known to bondinvestors. However， there are three good reasons for us to think about departures from this assumption. First， in practice， it is typically difficult for bond investors to observe a firm's assets directly due to delayed accounting reports. Second， the Ellsberg paradox and related experimental evidence demonstrate that people deal with risk and ambiguity in different ways. Risk refers to the case where the probability distribution over the state of the world is known， while ambiguity refers to the situation where the distribution itself may be unknown to the economic agents. Finally， as Hansen and Sargent （2001） pointed out， economic agents believe that the observed economic data come from a set of unspecified models. Concerns about model misspecification make a decision maker desire robust decision rules.How model uncertainty distorts a firm's leverage decision is investigated. The robust control method is adopted and an alternative explanation is provided for the debt conservation puzzle. The standard capital structure model implies the optimal leverage ratio should be 70%， while the average leverage ratio in the data is only 25%. To resolve this puzzle， it is assumed the bond investors are ambiguous averse about the firm's cash flow dynamics. Since the existence of model uncertainty reducesthe debt value and increases the cost of debt financing， the firm would issue less debt and choose a lower leverage. Quantitatively， the optimal leverage ratio in our model is 25.1%， which is consistent with the data.

Key words: model uncertainty, capital structure, cash flowvolatility, debt conservation, endogenous bankruptcy