Portfolio Diversification and Complementarity in Asset Demand Systems

Abstract

Investors evaluate their entire portfolio, not individual assets, striving to balance returns and risks through effective diversification. This paper introduces a flexible demand system accommodating heterogeneous substitution, cross-asset complementarities, and diverse investment strategies. By relaxing multinomial logit assumptions, our model better captures portfolio allocation decisions, linking portfolio weights to both individual asset and portfolio-wide characteristics. We propose a demandinverse approach for the identification of structural parameters. This approach implies a Generalized Method of Moments estimation procedure with novel instruments addressing cross-asset dependencies. Monte Carlo simulations validate the model, demonstrating improved finite-sample properties over standard multinomial logit frameworks.

Keywords: asset demand system, flexible substitution, cross-asset complementarity