An application of portfolio decision heuristics to support the 2 selection of research grant proposals

Abstract

Portfolio decisions involve selecting a subset of alternatives that together maximize some measure of value, subject to resource constraints. Exact methods are available to solve portfolio decision problems, but these require time, expertise and effort that may not always be available. In response, recent research has proposed a number of computationally simple, psychologically plausible rule-based heuristics for portfolio decision making. Simulation studies have shown that these portfolio heuristics perform well relative to exact approaches, but portfolio selection heuristics have yet to be applied in a real-world setting. Our study addresses this gap by using portfolio heuristics to support the selection of research grant proposals at a research institute in South Africa. We compare results obtained with portfolio heuristics to those obtained using two more traditional form of decision support, the standard linear-additive portfolio model, and robust portfolio modelling. We found that portfolios constructed using portfolio heuristics yielded over 90% of the value of optimal portfolios, selected slightly different portfolios potentially useful in a sensitivity analysis role, and were experienced as providing transparent and easy to understand decision support. Heuristic portfolios were slighly but consistently outperformed by those generated by robust portfolio modelling. Collectively, our study contributes to the growing body of evidence supporting the use of psychological heuristics in the realm of portfolio decision-making.