Optimal money burning: Theory and application to corporate dividends

We explore signaling behavior in settings with a discriminating activity and several costly nondiscriminating ("money-burning") activities. Existing theory provides no basis for selecting one method of burning money over another. When senders have better information about activity costs than receivers, each sender's indifference is resolved, the taxation of a money-burning signal is potentially Pareto-improving, and the use of the taxed activity becomes more widespread as the tax rate rises. We apply this theory to dividend signaling. Its central testable implication-that an increase in the dividend tax increases the likelihood of dividend payout-is verified empirically.