| Abstract: | Conventional approaches to determining optimal abandonment of a project under uncertainty either assume risk-neutrality or impose a mean-variance criterion. Risk-neutrality is unrealistic while the mean-variance criterion precludes determination of the optimal strategy without consideration of covariances of returns among projects. Further, the use of variance of present value as a risk measure may result in the “optimality” of a time 0 strategy that involves maintaining a position at time t that will be “suboptimal” and would not be maintained. The use of the multiperiod capital asset pricing model (CAPM) as a decision criterion is consistent with contemporary theory of market behavior and remedies the deficiencies of the mean-variance approach noted above. Computationally, the optimal strategy for abandonment, when the commitment must be made at time 0 (a lease, say), can be determined with little difficulty beyond that of mean-variance models. When time of abandonment can remain unspecified, the value of the prospect that abandonment will occur at the optimal time can be determined, though the technique necessary is considerably more complicated. In both cases, the marginal costs of commitments that limit discretion over abandonment can be determined and attributed to those commitments. |
