A panel data robust instrumental variable approach: a test of the new Fama-French five-factor model

ABSTRACT

Fama and French (FF, 2015) propose a new five-factor asset pricing model that adds profitability and investment patterns to the market, size and value variables used in FF (1992). Our purpose is to investigate this new model using an improved generalized method of moments (GMM)-based robust instrumental variables technique in a fixed-effects panel data framework. To test for measurement errors, we use a modified Hausman artificial regression. We also examine an augmented FF six-factor model that includes the Pástor–Stambaugh (PS, 2003) liquidity factor. Using the FF dataset, our GMM-based panel data approach leads us to conclude that the only consistently significant factor is the market factor.

KEYWORDS:

• Fama–French five-factor model
• GMM
• fixed-effects model
• Pástor–Stambaugh liquidity
• robust instruments

JEL CLASSIFICATION:

• C23
• G12

Disclosure statement

No potential conflict of interest was reported by the authors.

Notes

¹ Here we follow our previously developed approach (Racicot and Rentz 2016).

² The data for the five FF factors and the market and sector returns are available from http://mba.tuck.dartmouth.edu/pages/faculty/ken.french/data_library.html

³ The marginal Q is the NPV of future cash flows generated from an additional unit of assets. Note that Equation 3 is derived equating the marginal benefit to marginal cost.

⁴ The LIQ factor is available from Pastor’s website: http://faculty.chicagobooth.edu/lubos.pastor/research/ . We use the tradable LIQ factor and multiply it by 100 to put it in percentage form.

⁵ For more information, see Racicot (2015). Note that the dummy variable is in Equation 5 is excluded from Equation 6.

⁶ Note that W can be replaced by the White (1980) or the Newey and West (1987) HAC asymptotically consistent variance–covariance matrix. In this article, we use the HAC matrix.

⁷ For a discussion of the calculation of F, see Greene (2012, 358).