Examples of Factors in the ICAPM

Question

The intertemporal capital asset pricing model (ICAPM) is different from the CAPM in that in the ICAPM, utility is conditioned on some set of state variables. The ICAPM results in a multifactor pricing model of the market if investor care about hedging against changing investment opportunities and exposure to these factors. Are there any canonical examples of these factors?

I know that the Fama-French multifactor models (e.g., the three factor model) are often justified on grounds of the ICAPM (or APT -- arbitrage pricing theory). What are the underlying risk factors that justify this model? Are there any other examples of underlying risk factors that show up a lot?

Answer 1

ICAPM Factors

People have chosen different ways to pick factors. Chen, Roll and Ross are a classic example of attempts to find reasonable ICAPM factors. Fama-French factors are often explained as correlated with underlying ICAPM factors. Other researchers have chosen to look for factors without assuming outwardly observable exposures by analyzing returns using factor analysis or principal components.

Explanations of Fama-French

I don't know if there are "commonly accepted" explanations for the Fama French factors. This is not to say that there aren't explanations, there are a lot of them. Just no one seems to agree on which ones are "best". Candidates include both consumption based explanations AND production based models, each of which relates Fama-French factors to underlying economic variables. Cochrane has a nice summary of the performance of both in pricing Fama-French portfolios here.

Additional factors commonly used

Fama and French are now talking about a five-factor asset pricing model which also includes profitability and investment, similar to Novy-Marx's paper and work by Chen, Novy-Marx and Zhang which more directly relates to production asset pricing. Momentum is also a consistent candidate, although it's a real pain to put any kind of explanation to. There are tons of others. This paper lists many of them, as well as calling the significance of some of them into question.

Answer 2

It seems important to pay attention to data, not just theory. In particular, the entire size effect, represented by SMB in the FF3FM, occurs in January. This has been known for decades (e.g., Keim 1983) and has continued. There is also a substantial January effect in value, represented by HML. This is inconsistent with APT, ICAPM, production-based or consumption-based asset pricing, etc. The FF3FM is best viewed as controlling for some existing anomalies, robustness tests, etc., but it is plainly not how the world trades off any notion of traditional risk and expected return and no amount of investigation or study can make it so. Move on.