Abstract:
The theory of portfolio selection has often been applied to help improving decisions on the environment. The information required to apply this theory includes data on covariance of the uncertain returns between all combinations of the economic options and normally distributed returns. As it may be problematic to fulfil all data requirements and assumptions, the paper proposes a variant of robust portfolio optimization as an alternative. It considers future uncer-tainties in a non-stochastic fashion by means of possible deviations from the nominal return of land-use alternatives. The maximization of the economic return of the land-use portfolio is conditional to meeting an inclusive set of constraints. These demand that the same, whenever possible high, proportion of a required return threshold is achieved by means of the robust solution for each deviation scenario considered. The paper compares the land-use portfolios derived non-stochastically with portfolios generated by classical stochastic mean-variance op-timization. Based on data for eight agricultural crops typical for the Ecuadorian lowlands the results show that, depending on the deviation scenario assumed, the robust portfolios show a greater Shannon index compared to classical portfolios. For the same standard deviation of returns (derived through the classical sum of all covariances) the robust portfolios show no more than 2-5% loss in economic return in most cases. Occasionally, the loss has been higher, up to 20%. In this case the Shannon index was about 2.5 times higher compared with that of the conventional portfolio. The highly diverse portfolio achieved a much better protection against low relative performance. The results obtained show that the non-stochastic derivation of land-use portfolios is a good alternative to the classical stochastic model, whenever eco-nomic information is scarce.
