Date of Award

Spring 5-2019

Document Type

Honors Thesis

Degree Name

Bachelor of Science

Department

Economics

Advisor/Committee Chair

Lewis Segal, Ph . D.

Abstract

The Capital Asset Pricing Model (CAPM) and its extensions are a family of empirical asset pricing models which partition risk as either "systematic" (market-wide) or "idiosyncratic" (stock-specific). Examples of systematic risk-factors include the market return, company size, and company value. Within the framework of the CAPM-family of models, it is assumed that the effects of these systematic risk-factors are homogenous among sectors. This paper develops an extension to the CAPM relaxing this assumption, by directly comparing these systematic risk-factors at the sector-level. Utilizing CRSP and Compustat data, systematic risk-factor premiums are estimated for each sector, which demonstrates heterogeneity, with respect to sector. An analysis of means and statistical significance reveals that a separate stock-picking strategy is necessary within each individual sector, and that there exist factors that are irrelevant to some sectors altogether. The estimated sector premiums are utilized to develop a GICS Ten-Factor Model, which has superior explanatory power amongst the CAPM-family. The GICS Model has an average Adjusted-R2 of 27%, compared to the CAPM which has a value of 15.5%. It is then demonstrated that the GICS Model is superior to the CAPM-family in regard to high-Beta Portfolio construction-with a Sharpe Ratio of 0.61 compared to the CAPM which has a value of 0.42. This paper demonstrates that systematic risk-factors are heterogeneous among sectors, and details how this information is materially useful to investors.

Download

Included in

Finance Commons

Share

COinS