A Comparison of the Efficacy of Liquidity, Momentum, Size and Book‐to‐Market Value Factors in Equity Pricing on a Heterogeneous Sample: Evidence from Asia

• Date: 01 November 2016
• Published date: 01 November 2016
• DOI: http://doi.org/10.1111/fmii.12078

A Comparison of the Efficacy of Liquidity,

Momentum, Size and Book-to-Market Value Factors

in Equity Pricing on a Heterogeneous Sample:

Evidence from Asia

BYBRUCE HEARN

This paper compares the size and book-to-market value factors of Fama and French (1993)

alongside Momentum of Jagadeesh and Titman(1993) with two Liu (2006) liquidity factors

formed from 1 year rebalancing and 1 month rebalancing respectively.A heterogeneous and

comprehensive sample of the top blue chip stocks of all national Asian equity markets with

further differentiation undertaken between sub samples formed for Japan only and Asia

excluding Japan for period January 2000 to August 2014. Our empirical results suggest

that multifactor time invariant pricing models based on augmented capital asset pricing

model (CAPM) framework are ineffective in explaining the cross section of stock returns

in the presence of significant inter and intra-market segmentation. However an alternative

model specification based on a time varying parameter specificationand using same sets of

factors yields significant enhancements in explaining cross section of stock returns across

universe. We find that momentum factor largely lacks significance while a time varying

two factor model, based on CAPM plus liquidity factor, is optimal. The liquidity factor

being that of Liu (2006) and annually rebalanced. Our findings are important for investment

managers seeking appropriate factors and modelling techniques to hedge against risks as

well as firm’s financialmanagers seeking to reduce costs of equity capital.

Keywords: Liquidity, CAPM, Emerging Financial Markets, Asia.

JEL Classification: G11, G12, G15, O55.

I. INTRODUCTION

The capital asset pricing model (CAPM) has generated a considerable literature

since its initial formulation by Sharpe (1964) and Lintner (1965) with the ba-

sic single factor model being extended through its augmentation with additional

valuation factors purporting to enhance its ability to explain the cross section of

stock returns in a given universe. These typically include returns-based factors

capturing cross sectional differences in size and accounting book to market value

(Fama and French, 1993), momentum (Jegadeesh and Titman, 1993, 2001) and

more recently liquidity (Pastor and Stambaugh, 2003; Liu, 2006). An individual

stock or portfolio’s association with these aggregate factors is then simply inter-

preted as their coefficients in a time series regression (Black, Jensen and Scholes,

1972) where the model intercepts or alphas are expected to be zero (Merton,

1973). The inclusion of such additional factors within an extended CAPM based

framework is subject to debate with adherents arguing of the impact on investor

welfare arising from these necessitating compensatory premiums to be attributed

Corresponding author: Bruce Hearn, School of Management, Business and Economics, University of Sussex.

E-mail: b.a.hearn@sussex.ac.uk.

C2016 New YorkUniversity Salomon Center and Wiley Periodicals, Inc.

254 Bruce Hearn

to these underlying factors (see Liu, 2006 for detailed discussion). On the contrary

authors such as Lakonishok et al. (1994), Daniel and Titman (1997) and Daniel

et al. (2001) assert that such pricing anomalies associated with these factors are

related to inefficiencies in the way markets incorporate information into equity

prices. Despite this controversy there is a general consensus regarding the impor-

tance of additional factors within a multifactor format – be this expressed through

an intertemporal capital asset pricing model (ICAPM) of Merton (1973) or the

arbitrage pricing model format of Ross (1976). However the question of whether

equities priced with these factors are better priced using local domestic or inter-

national market universes is enduring (see Karolyi and Stulz, 2003). This is due

to unresolved quandaries over segmentation given the fundamental importance of

notions of asset market integration that underscore asset pricing theory. As such

we are motivated to differentiate between contrasting multifactor asset pricing

models yielding varying explanatory power of the cross section of stock returns

while taking account of inherent segmentation within multi-country universes.

Recent research testing the efficacy of asset pricing models within the contextof

segmentation has been undertaken by Hou et al. (2010). This focussed on the dif-

ferentiation between contrasting multifactor asset pricing models formed through

the augmentation of single factor CAPM with Fama and French (1993) size and

book-to-market value alongside a variety of cash-flow factors between individual

domestic universes and a global counterpart formed from the 49 constituent mar-

kets. A key limitation in the employment of an array of factors based on firm’s

cash flow balance sheet items is the availability of data within a broader emerging

economy sample, such as across a wider Asian regional universe. Furthermore a

body of recent research suggests liquidity to be a likely candidate factor for consid-

eration in standard asset pricing models. Pastor and Stambaugh (2003) develop a

multifactor pricing model including a simple volume-based liquidity metric while

Liu (2006) finds evidence that a two factor liquidity augmented CAPM better ex-

plains the cross section of stock returns than either the unitary CAPM or the Fama

and French (1993) three factor (henceforth FF3F) models. Furthermore Jagadeesh

and Titman (1993, 2001) find evidence of price-based momentum performance

factor in yielding robust explanation of cross section of stock returns while Carhart

(1997) further developed this into the joint inclusion of momentum factor on top

of FF3F. A principal limitation in all these studies is their almost exclusive focus

on the single country setting of US equity market. Following on from these de-

velopments our first contribution to the literature on application of pricing models

within a multi-country setting is in the differentiation between single factor CAPM

to its augmented counterparts where these are based on FF3F, Carhart (1997) four

factor (4F) augmenting FF3F with momentum, and two factor liquidity models

of Liu (2006) where these use two rival liquidity factors: one based on an annual

rebalancing and holding period, and the other based on monthly rebalancing.

Our second contribution to the literature arises from our relaxing the

time-invariant constraints on parameters estimated within standard multi-factor

augmented CAPM frameworks and allowing for parameter coefficients to

Asset Pricing Evidence from Asia 255

stochastically vary over the duration of sample period. This follows a number

of applied studies using such time varying parameter asset pricing models based

on Kalman filter such as Brooks et al. (1998) in studying a sample of Australian

industry portfolios, and Hearn (2010) studying four South Asian equity markets,

namely India, Pakistan, Bangladesh and Sri Lanka. These studies are generally

constrained in terms of sample time frames and geographic scope. Consequently

our use of Kalman filter time varying parameter methods is an attempt in im-

plicitly taking account of intra-sample segmentation (both intra and inter-market)

within a universe and structural breaks in underlying data series. Furthermore

our application of such methods across a broad and comprehensive Asian sample

alongside sub-samples based on single country (multiple market) of Japan and

Asia-excluding Japan is an explicit attempt in taking account of segmentation.

Our geographic focus on the Asian region is largely justified by this region be-

ing a centre to a large, well developed investmentmanagement industry primarily

centred on the developed markets of Singapore, Hong Kong, Japan and Australia

while the region itself is broad with national markets ranging from amongst the

smallest worldwide (e.g., Maldives, Laos and Cambodia) to some of the largest

(such as Tokyo, Singapore and Hong Kong). It is also institutionally diverse with

prominent religions including Orthodox Christianity (Armenia), Islam (Maldives,

Pakistan, Bangladesh and Indonesia), Hindu (India), and Buddhist (Thailand,

Laos, Cambodia) while many societies informal institutional structure is com-

munitarian and feudal in nature centring on extended familial groups (Claessens

et al., 2000). Formal institutions are almost invariably inherited from European

colonial metropoles where these provide the institutional frameworks supporting

the establishment and sustainability of stock markets – including their regulation,

accounting and reporting standards and the enforcement of these. Thus this region

provides an excellent laboratory for the study of segmentation and optimal choice

of valuation factors for inclusion in multifactor pricing models.

Our findings reveal very little statistical support for time-invariant asset pricing

models in general. The expectation of regression intercepts equalling zero is

violated to a high degree as is evident from extremely high F-statistics arising from

application of Gibbons, Ross and Shanken (1989) (henceforth GRS) statistical

test to ascertain likelihood of the intercepts from a number of test asset portfolios

jointly being equal to zero. The extremely high value of F-statistics also precludes

the use of such GRS methods as a means to differentiate between rival multifactor

models. This very high level of rejection of time invariantparameter models is also

a prominent feature of literature with similar results reported by Fama and French

(1993) in US market and Hou et al. (2011) in a global study of 49 countries.

However in contrast the application of Kalman filter time varying parameter

methodology to augmented multifactor CAPM framework points towards the

efficacy in information criterion terms (Aikaike Information Criterion, AIC) of

two factor liquidity augmented CAPM – with the liquidity factor being that of the

1 year rebalanced and 1 year holding factor. Finally we find verylittle evidence of

a priced momentum factor with this almost wholly lacking statistical significance