What does high-dimensional factor analysis tell us about risk factors in the Australian stock market?

Abstract

Estimates of the cost of equity are often sensitive to the specification of the linear factor model used in their construction. In this article, we use techniques developed for high-dimensional factor models to consider the identity of systematic risk factors in the Australian equities market. Our results support the use of neither the Capital Asset Pricing Model (CAPM) nor the Fama and French model, although they provide an explanation for the empirical performance of these models. Many other model specifications are also rejected. We find that a single-factor model with an equal-weighted market index is the best model for estimating the cost of equity in the Australian context.

Keywords:

• asset returns
• arbitrage pricing theory
• capital asset pricing model
• factor analysis
• principal components
• cost of equity

JEL Classification::

• G10
• G12
• G31

Notes

¹ See Chen et al. (1986), Antoniou et al. (1998), Beenstock and Chan (1988), Martikainen et al. (1991), Fama and French (1992), Ferson and Harvey (1994), McElroy and Burmeister (1988), Kaneko and Lee (1995), Carhart (1997), Clare and Priestley (1998), Faff and Brailsford (2000), Bilson et al. (2001), Di Iorio and Faff (2002), Ammann and Steiner (2008), Girard and Kiymaz (2009), Kubota et al. (2009) and Mouna et al. (2009).

² i.e. inclusive of dividends.

³ http://www.datastream.com/.

⁴ In a survey of chief financial officers (Graham and Harvey, 2001) find that 73.5% ‘always or almost always’ use the CAPM to estimate the cost of equity capital.