An information-theoretic approach to estimating risk premia
Name
1051300223-MIT.pdf
Description
Full printable version
Size
2.48 MB
Format
Adobe PDF
Checksum (MD5)
3f1a948956b82ebab91cdea2339ff502
Author(s)
Kazemi, Maziar Mahdavi
Advisor(s)
Hui Chen.
Date Issued
2018
Publisher
Massachusetts Institute of Technology
Abstract
Evaluation of linear factor models in asset pricing requires estimation of two unknown quantities: the factor loadings and the factor risk premia. Using relative entropy minimization, this paper estimates factor risk premia with only no-arbitrage economic assumptions and without needing to estimate the factor loadings. The method proposed here is particularly useful when the factor model suffers from omitted variable bias, rendering classic Fama-MacBeth/GMM estimation infeasible. Asymptotics are derived and simulation exercises show that the accuracy of the method is comparable to, and frequently is higher than, leading techniques, even those designed explicitly to deal with omitted variables. Empirically, we find estimates of risk premia that are closer to those expected by financial economic theory, relative to estimates from classical estimation techniques. For example, we find that the risk premia on size, book-to-market, and momentum sorted portfolios are very close to the observed average excess returns of these portfolios. An exciting application of our methodology is to performance evaluation for active fund managers. We show that we are able to estimate a manager's "alpha" without specifying the manager's factor exposures.
Description
Thesis: S.M. in Management Research, Massachusetts Institute of Technology, Sloan School of Management, 2018.
Cataloged from PDF version of thesis.
Includes bibliographical references (pages 31-35).
Subjects
Sloan School of Management.
MIT Department
Sloan School of Management
Terms of Use
MIT theses are protected by copyright. They may be viewed, downloaded, or printed from this source but further reproduction or distribution in any format is prohibited without written permission.
Persistent DSpace Link
