The State of Intertemporal Asset Pricing: A review, synthesis and empirical application

Nis Bjørnholt Bak

Student thesis: Master thesis


This paper presents a review of contemporary, partial equilibrium, intertemporal asset pricing theory and develops and tests a model based on a combination of several theories in the field. This model performs nearly as well as the standard Fama-French three-factor model, yet unlike it, it is based on economic rationale rather than only empirical fit. Reviewed are three models, APT, CCAPM and the ICAPM. The APT is found to be general to the point of having little economic content, and the theory on limits to arbitrage means that it may not present an accurate description of asset prices. The CCAPM and the ICAPM are based on expected utility theory, the validity of which has recently been doubted due to findings in behavioural finance. Even if little heed in paid to this criticism, the models are fraught with problems, many based on the issues inherent in summing up the budget constraints, preferences and beliefs of the countless individuals that interact to set prices. Proposed solutions to the issues highlighted with the models are used as a point of departure for identifying factors that can be used as input for an amalgamated, multifactor model. Two factors were found particularly promising and subsequently used: 1) The ILLIQ factor of Amihud (2002), which measures illiquidity and information disagreement 2) The SKD factor of Harvey and Siddique (2000), which measures the coskew of an assets return distribution. This makes it possible to account for a non-linear relationship between excess market returns and excess asset returns and thus nuances to linear relationship suggested by the beta factor. The amalgamated model, containing these two factors and a market factor, is tested on 1963:7-2011:12 NYSE and AMEX data. Both the developed model and the FF model fair well over the entire test period, as well as in the first third of the period. In the middle and last first the results are mixed, with the FF model coming out on top in the middle period but the developed model performing best in the most recent period. Overall it is found that combinations of the developed model and the FF model performed the best, yet such combinations goes against the aim of developing a model founded in economic theory.

EducationsMSc in Finance and Accounting, (Graduate Programme) Final Thesis
Publication date2012
Number of pages114