Variable Selection for Portfolio Choice

• Activity Number: 621 - Portfolio Choice, Stock Returns, Bankrupcty, and Default
• Type: Contributed
• Date/Time: Thursday, August 3, 2017 : 8:30 AM to 10:20 AM
• Sponsor: Business and Economic Statistics Section
• Author(s): Jin Liu*
• Companies: Peking University
• Keywords: Portfolio Choice ; Variable Selection ; Index Model ; Sparsity ; Oracle Properties

Abstract

We study variable selection in asset allocation: determine the portfolio weights and select the corresponding predictors. First, we construct an index to combine predictors, and then maximize the conditional expectation of the investor's utility function based on the index with two penalties on the portfolio weights and index coefficients. The shrinkage estimation of the portfolio weights and index coefficients are obtained, and the corresponding asymptotic properties (consistency, sparsity and oracle property) are established. Second, we obtain the optimal portfolio weights based on above maximization with every predictor in the first-stage, and then combine these weights by model averaging with a penalty on the averaging coefficients to select predictors. We also obtain the corresponding shrinkage estimation and asymptotic properties. An empirical application in portfolio allocation illustrate the superiority of the proposed estimation procedure.