Liquidity and Asset prices: An Empirical Investigation of the Nordic Stock Markets

AuthorNader Shahzad Virk,Hilal Anwar Butt
DOIhttp://doi.org/10.1111/eufm.12041
Published date01 September 2015
Date01 September 2015
Hilal Anwar Butt and Nader Shahzad Virk
© 2014 John Wiley & Sons Ltd
Liquidity and Asset prices:
An Empirical Investigation of
the Nordic Stock Markets
Hilal Anwar Butt and Nader Shahzad Virk
Department of Finance and Statistics, Hanken School Economics, Vasa, Finland
E-mails: hilal.butt@hanken.; nvirk@hanken.
Abstract
This paper presents a simplied single period asset-pricing model adjusted for
liquidity and tests it for the Nordic markets. The detailed empirical evidence is
presented from Finnish test case. Empirical testing of small yet developed markets
is motivated by the increased relevance of the illiquidity effect for illiquid assets/
markets. The main evidence reports liquidity risk makes sufciently larger part of
predicted factor risk premium than the market risk, contrary to comparable US
evidence. This highlights the ability of liquidity related model betas in capturing
the time variation in expected returns across illiquid (Nordic) markets than market
beta.
Keywords: Asset-pricing model, illiquidity effect, predicted factor risk premium,
model betas
JEL classification: G10, G12, G15
1. Introduction
Studies connecting liquidity to asset pricing have evolved over time and are currently
based on a twofold proposition that the level of illiquidity and illiquidity risk are priced
and both are mutually reinforcing. One of the initial studies pioneering the former aspect
of liquidity is of Amihud and Mendelson (1986), they showed a positive relationship
We are grateful to the anonymous referee and John Doukas, the editor of the journal, for
their comments on earlier version of this paper. The authors also wish to thank conference
participants at the Hanken School of Economics, the IFABS, 2012 conference, the SFA,
2012 Annual Meetings and MFS conference, 2013. Discussions with Johan Knif, Kenneth
Högholm and Peter Nyberg are gratefully acknowledged. We are also grateful to Waleed
Ahmad for help with writing codes for data construction. The financial support of Hanken
Foundation and Wallenberg CEFIR is greatly acknowledged. Correspondence: Hilal
Anwar Butt.
European Financial Management, Vol. 21, No. 4, 2015, 672–705
doi: 10.1111/eufm.12041
© 2014 John Wiley & Sons Ltd
Liquidity and Asset prices 673
between an assets level of illiquidity and expected returns. Then Pastor and Stambaugh
(2003) elaborated upon later aspect of illiquidity and demonstrated a link between the
asset returns and liquidity risk. Furthermore, Amihud (2002) investigated systematic
illiquidity risk and proposed that expected market illiquidity is priced positively, while
shocks to market illiquidity lower contemporaneous returns. Amihud (2002) provided
this evidence for the US market whereas, Bekaert et al. (2007) tested these hypotheses
for emerging markets and broaden the evidence.
Around this general conclusion that liquidity risk is priced; there is another associated
aspect of liquidity risk, that is, its various dimensions. Historically, a ight to liquidity/
quality risk
1
(Amihud (2002), Pastor and Stambaugh (2003), Sadka (2006), Fujimoto
and Watanabe (2005) and others) has been explored more, in which an impact of
changing market-wide liquidity is seen on different asset classes. Another dimension of
illiquidity risk is commonality in liquidity as proposed by Chordia et al. (2002),
Hasbrouck and Seppi (2001) among others. Karolyi et al. (2012) extended this evidence
of commonality in liquidity for the stocks in 40 developed and emerging countries. All
the above, along with an additional dimension of systematic liquidity risk (depressed
wealth effect), are neatly presented in the theoretical equilibrium model of Acharya and
Pedersen (2005). Besides, another dimension of modeling systematic liquidity risk
follows Fama and French (1993) type construction of risk mimicking liquidity factor
under the equilibrium assumption of no arbitrage (Liu, 2006).
We simplify the model developed by Acharya and Pedersen (2005) by making the
liquidity adjustment in a single period model rather than using an overlapping generation
model (OLG). The model proposes that a market index, spanned from a mean-variance
efcient asset space net of asset-specic liquidity costs, is a better candidate to reduce the
reported mispricing associated with standard mean-variance CAPM. The generalisation
of the model allows for the determination of asset prices inside the model and accounts
for the total cost of trade rather than exogenously determined prices and model agents
confronting the cost of selling, as in Acharya and Pederson (2005).
2
Importantly, the
adjusted theoretical model and its tested specication in this study both are one period,
whereas in Acharya and Pedersen model in each generation agents are effectively
assumed to have one period utility maximisation for the specication testing of the
model.
As is generally acknowledged, illiquidity effects are pronounced for illiquid assets/
markets. However, with few exceptions, most liquidity-related studies are conducted for
the US markets. Arguably, the US market is the most liquid equity market (Bekaert
et al., 2007) and therefore may not be as suitable for empirical testing as other illiquid
markets. The reported diminishing illiquidity premium in the US stock returns over time
(Ben-Rephael et al., 2010) then should come as no surprise. We argue that it is more
appropriate to test liquidity-related models in markets that are sufciently illiquid to
diagnose the level and strength of bearing such risks in comparison to other pertinent
1
The liquidity crises affect the illiquid asset the most.
2
Selling costs can only be deduced when data for each trade, which allows buy and sell orders
to be distinguished, is available, but this method requires substantial microstructure data.
However, even the distinction between buy orders and sell orders is at times obscured in low
frequency data. Thus no exact procedure exists to approximate the cost of selling using any
illiquidity measure.
© 2014 John Wiley & Sons Ltd
Hilal Anwar Butt and Nader Shahzad Virk
674
risks, such as market risk even if they are developed. Li et al. (2011) report liquidity risks
are priced over and above market risk for the 2nd largest global equity market (Japan)
after the equity markets in the USA.
In order to stress the importance of illiquid markets for the relevance of liquidity
premium, we test the adjusted model for four Nordic capital markets namely, Denmark,
Finland, Norway and Sweden.
3
The main test is the quantication of premium related
with liquidity risk(s) when compared with diminishing liquidity premium for US stock
returns. Furthermore, we select a representative market that is, Finland. The test case
selection is to make generalisations across Nordic (illiquid) markets: 1) how liquidity
evolves over time; 2) success of different measures in capturing effect of liquidity and 3)
importance of a particular liquidity risk across different liquidity risks. The Finnish stock
market is a small, developed market that, over the course of a decade, transformed from
an illiquid to a liquid market; yet, the prospect of nding illiquidity remains a possibility
due to the peculiar setting in which the market operates.
4
The cross-section of 25 test portfolios, based on ve different (three) liquidity and
(two) non-liquidity stock characteristics, tests the real strength of the proposed model, as
suggested in Lewellen et al. (2010). Furthermore, we calculate the measure of illiquidity
for the stocks listed in the Finnish market in two distinct ways: the measures proposed in
Lesmond et al. (1999) and Amihud (2002). Both of these illiquidity measures are highly
correlated with ner spread and price impact proxies estimated from low frequency data
(Goyenko et al., 2009). The primary purpose of measuring illiquidity in different ways is
to report which illiquidity proxy better captures the unobserved illiquidity effect that
explains the cross-sectional return differences. Finally, we also check for time variation
in illiquidity and test all of the models, excluding the periods with high illiquidity, using
the reduced sample.
The results show that the illiquidity portfoliosreturns are more related to the
systematic illiquidity risks than the systematic market risk. The impact of this
relationship is substantial; the percentage of the illiquidity premium in the total modeled
risk compensation is approximately 92 percent for a model with highest R
2
and
restrictions on model intercept. In comparison, only 17 percent of the total risk premium
is attributed to illiquidity risks for US stock returns (Acharya and Pederson, 2005).
The remainder of the factor risk premium is attributed to CAPM risk in both markets. The
stronger association of model-predicted factor risk premia with liquidity risks remains
3
We especially note the suggestions from the Editor John A. Doukas in the developments in
this direction.
4
The empirical literature on the Finnish market studying the relationship between stock
returns and different illiquidity proxies is extensive. For instance, Swan and Westerholm
(2002) found that the level of illiquidity has a positive and strong effect on the cross-section
of stock returns from 19931998. Vaihekoski (2009) tested for market specic and asset
specic liquidity risks for a cross-section of six size portfolios and found that asset specic
risk is not priced, whereas the portfolio risk sensitivities reveal a at relationship across the
size of the portfolios. In the latter study, the illiquidity risk was captured by one factor that
only accounts for Amihuds (2002) ight to liquidity notion. As noted previously, other
illiquidity risks, such as commonality effect and depressed wealth effect on assets illiquidity,
exist; therefore, the results contribute supplemental empirical evidence for other risk types
affecting the Finnish stock return variations.

To continue reading

Request your trial

VLEX uses login cookies to provide you with a better browsing experience. If you click on 'Accept' or continue browsing this site we consider that you accept our cookie policy. ACCEPT