Sequential Investment Decisions with Bayesian Learning

This paper analyzes investment decisions that can be made in a modular form. It is motivated by the empirical observation that managements are particularly worried about “downside” risk. With a sequential approach this risk is minimized. An investment in a module produces information as well as profits or losses. In our model a larger investment produces more information in addition to larger profits or losses. Costs for changing the level of the investment from period to period are introduced. The optimal sequential investment policy is studied for a two-period problem. Conditions are presented under which no investment, a partial investment, or a full investment in the first period is optimal.