A COMPARISON BETWEEN FAMA AND FRENCH MODEL AND LIQUIDITY-BASED THREE-FACTOR MODELS IN PREDICTING THE PORTFOLIO RETURNS

AAMJAF, Vol. 2, No. 2, 43–60, 2006

ASIAN ACADEMY of

MANAGEMENT JOURNAL

of ACCOUNTING

and FINANCE

Ruzita Abdul Rahim, and Abu Hassan Shaari Mohd. Nor

Universiti Kebangsaan Malaysia, 43600 Bangi, Selangor, Malaysia

*Corresponding author: ruzitaar@pkrisc.cc.ukm.my

Abstract

The main objective of this paper is to evaluate the forecasting accuracy of two liquidity- based three-factor models, SiLiq and DiLiq, which have been developed as potential improvements on the Fama-French model. Using common stocks of 230 to 480 listed firms, this study constructs 27 test portfolios double-sorted on: (i) size and book-to- market ratio (B/M), (ii) size and share turnover (TURN) and (iii) B/M and TURN. The study sets the periods of January 1987 to December 2000 for estimation and January 2001 to December 2004 as forecast sample. The forecast errors are measured using mean absolute percentage errors and Theil's Inequality Coefficient. The preliminary results clearly document that three-factor models outperform CAPM. While the hypotheses of no significant differences cannot be rejected, the marginal difference in the errors of the competing three-factor models indicate that predicting returns on stocks traded on Bursa Malaysia can be slightly improved by incorporating illiquidity risk in a three-factor model in the form of DiLiq.

Keywords: illiquidity risks, Fama-French Model, liquidity-based model, Multifactor Model

INTRODUCTION

Since its introduction in 1993, Fama-French model has been extensively attended to the extent that it is currently considered the workhorse for risk adjustment in academic circles (Hodrick & Zhang, 2001). While the model performs exceptionally well compared to the capital asset pricing model (CAPM) of Sharpe (1964), Lintner (1965), and Black (1972), its performance against other multifactor models in general is inconclusive. Consistent with Fama and French's (1996) assertion that like any other model, the Fama-French model is not without weakness (Fama & French, 1996), this study finds it of a great contribution to the asset pricing literature if alternative models can be developed as potential improvement on the model. Motivated by (i) Fama and French's (1996) conclusion that 3-factor model suffice to explain stock returns, (ii) the fact that the additional risk factor in the Fama-French model are firm-specific factors, and (iii) Dey's (2005) assertion that the sources and pricing of risk in emerging and developed markets are different, this study plans to achieve the objective by developing variants of 3-factor models that incorporate other firm-specific factor that is of greater concern to the investors in the studied market. Notwithstanding the fact that the concern on liquidity is a universal truth, as an emerging equity market Bursa Malaysia offers "… an ideal setting to examine the impact of liquidity on expected returns" (Bekaert, Harvey, & Lundblad, 2005) because "… liquidity is one firm characteristic that is of particular concern to investors in emerging market" (Rowenhorst, 1999: 1441). Furthermore, because the proposed models in this study are also an implication of Intertemporal CAPM (ICAPM), the choice of liquidity is judicious given that "… liquidity is a natural choice as an asset-pricing factor since it is a state variable in the ICAPM sense" (Chollete, 2004: 1). This hypothesis is supported with substantial evidence on the superior performance of liquidity-adjusted versions of the CAPM (Lo & Wang, 2001; Liu, 2004) and Fama-French model (Bali & Cakici, 2004; Chollete, 2004; Liu, 2004; Chan & Faff, 2005; Miralles & Miralles, 2005).

To test our hypotheses that the proposed liquidity-based models work as potential improvement on the Fama-French model, their forecasting accuracies are assessed against the benchmark model. While re-examination on the Fama- French model naturally adds to existing literature particularly in the sample market where similar studies are limited (Drew & Veeraraghavan, 2002; Drew, Naughton, & Veeraraghavan, 2003), the main contribution of this study is the development of liquidity-based 3-factor models which apparently is an effort that does not seem to have been attempted in any studies before. The remainder of the article is organized as follows. Section 2 reviews related studies, Section 3 describes the data and methodology, Section 4 presents the findings and discusses the results, while Section 5 concludes and discusses the implication.

Read more