Polynomial Goal Programming and the Implicit Higher Moment Preferences of U.S. Institutional Investors in Hedge Funds
Financial Markets and Portfolio Management, Forthcoming
43 Pages Posted: 2 Jan 2014
Date Written: January 1, 2014
Polynomial goal programming (PGP) is a flexible method that allows investor preferences for different moments of the return distribution of financial assets to be included in the portfolio optimization. The method is intuitive and particularly suitable for incorporating investor preferences in higher moments of the return distribution. However, until now, PGP has not been able to meet its full potential because it requires quantification of “real” preference parameters toward those moments. To date, the chosen preference parameters have been selected somewhat “arbitrarily”. Our goal is to calculate implied sets of preference parameters by using investors’ choices of and the importance they attribute to risk and performance measures. We use three groups of institutional investors — pension funds, insurance companies, and endowments — and derive implied sets of preference parameters in the context of a hedge fund portfolio optimization. To determine “real” preferences for the higher moments of the portfolio return distribution, we first fit implied preference parameters so that the PGP optimal portfolio is identical to the desired hedge fund portfolio. With the obtained economically justified sets of preference parameters, the well-established PGP framework can be employed more efficiently to derive allocations that satisfy institutional investor expectations for hedge fund investments. Furthermore, the implied preference parameters enable fund of hedge fund managers and other investment managers to derive optimal portfolio allocations based on specific investor expectations. Moreover, the importance of individual moments, as well as their marginal rates of substitution, can be assessed.
Keywords: Polynomial Goal Programming, Fund of Hedge Funds, Higher Moments, Investor Preferences
JEL Classification: G11, G23
Suggested Citation: Suggested Citation