Optimal dynamic portfolio selection for projects under a competence development model

A model for optimizing investments of human resources over time into a given set of project classes with known competence requirements and known returns is presented. The model considers increments of competencies by learning as well as decrements by knowledge depreciation. We derive analytical results addressing questions such as whether an optimal policy will specialize on a single project class or rather keep a mixed project portfolio. In absence of uncertainty, gradual specialization turns out as advisable, but mixed portfolios in the intermediate term can be better than instantaneous specialization. Algorithms for optimizing the project portfolio selection are outlined.