Why Invest in Emerging Markets? The Role of Conditional Return Asymmetry

• DOI: http://doi.org/10.1111/jofi.12420
• Date: 01 October 2016
• Published date: 01 October 2016
• Author: ALBERTO PLAZZI,ERIC GHYSELS,ROSSEN VALKANOV

THE JOURNAL OF FINANCE •VOL. LXXI, NO. 5 •OCTOBER 2016

Why Invest in Emerging Markets? The Role

of Conditional Return Asymmetry

ERIC GHYSELS, ALBERTO PLAZZI, and ROSSEN VALKANOV∗

ABSTRACT

We propose a quantile-based measure of conditional skewness, particularly suitable

for handling recalcitrant emerging market (EM) returns. The skewness of inter-

national stock market returns varies significantly across countries over time, and

persists at long horizons. In EMs, skewness is mostly positive and idiosyncratic,

and significantly relates to a country’s financial and trade openness and balance of

payments. In an international portfolio setting, return asymmetry leads to sizeable

certainty-equivalent gains and increases the weight on emerging countries to about

30%. Investing in EMs seems to be about expectations of a higher upside than down-

side, consistent with recent theories.

EMERGING STOCK MARKETS HAVE GROWN significantly, in numbers and trading

volume, over the last 20 years. Now more than ever, investors seeking diversi-

fication are able to invest with relative ease in emerging economies. However, a

larger number of liquid stock markets does not, in and of itself, imply that the

prospects of international diversification have improved.1Paradoxically, while

interest and liquidity in emerging stock markets are trending upward, inter-

national diversification prospects are declining. The question then arises as

to whether there exist reasons beyond standard mean-variance diversification

that can justify investing in emerging economies.

∗Eric Ghysels is with the Department of Finance, Kenan-Flagler Business School, and Depart-

ment of Economics, University of North Carolina at Chapel Hill, and CEPR. Alberto Plazzi is

with the Institute of Finance, Universit`

a della Svizzera Italiana, and the Swiss Finance Institute.

Rossen Valkanovis with the Rady School of Management, UCSD. We thank Geert Bekaert, Robert

Engle, Ren´

e Garcia, Peter Hansen, Roger Koenker,Jun Liu, Eric Renault, Allan Timmermann, and

Hal White for useful discussions. We also thank two anonymous referees, the Associate Editor,and

Kenneth Singleton (the Editor) for very insightful comments that helped us improve our paper. We

have also benefitted from comments at the European Finance Association, NYU VolatilityInstitute,

Saint Louis Federal Reserve Bank, and Society for Financial Econometrics (SoFiE) conferences,

and at seminars at the European Central Bank, University of Brussels, University of Houston,

University of Lausanne, University of Luxembourg, University of Zurich, and Vienna University

of Economics and Business. The authors gratefully acknowledge financial support from Inquire

Europe for this project. The authors have no conflicts of interest with respect to The Journal of

Finance disclosure policy.

1The opposite seems to be true, in fact, if we consider the return correlation between emerging

markets (EMs) and developed markets (DMs): it has steadily increased over the years. See, for

instance, Harvey (1995), Fama and French (1998), and most recently, Christoffersen et al. (2012).

This fact has often been explained by the increase in international market integration.

DOI: 10.1111/jofi.12420

2145

2146 The Journal of Finance R

In this paper, we ask whether it is profitable to invest in emerging stock mar-

kets in the quest for (conditional) skewness. More specifically, we investigate

the economic gains from exploiting predictable asymmetries in the distribu-

tion of returns across international markets. Skewness has a long history in

finance. Early contributions focus on the coskewness of returns with the market

portfolio.2These theoretical contributions continue to guide our fundamental

understanding about the role of skewness in a portfolio setting. More recent

empirical studies show that, in the U.S. stock market, own skewness of re-

turns (rather than coskewness) plays an important role in driving expected

returns. These papers are motivated by novel theoretical work arguing that

investors are willing to trade off diversification benefits for skewness.3These

recent advances notwithstanding, our understanding of skewness, on both the

theoretical and the empirical fronts, is far from complete.

In particular, the role of skewness in a cross-section of international markets

has been largely unexplored, for two reasons. First, the third moment is hard

to estimate as it is especially sensitive to outliers, more so than the first two

moments (Kim and White (2004), Neuberger (2012)). In the context of U.S.

stock returns, the literature addresses this issue by turning to options as a

way of obtaining more precise skewness estimates. In an international setting,

this approach is not feasible as option markets in most countries are illiq-

uid or simply nonexistent. An investigation of skewness from an international

portfolio perspective requires robust skewness estimates that rely on the un-

derlying asset return data alone. Second, incorporating conditional skewness

in a portfolio choice problem that involves a large cross-section of countries

is a challenging endeavor (Guidolin and Timmermann (2008), Brandt (2010),

Harvey et al. (2010)). Beyond addressing the purely technical difficulty, under-

standing the economic forces that drive the potential gains from skewness in a

set of developed and emerging economies that are fundamentally different in

their economic growth, financial openness, trade globalization, and balance of

payments is an intriguing and unanswered question.

Weoffer t hree contributions. First, we propose a robust method for estimating

conditional asymmetry using potentially noisy EM returns. In particular, we

use an asymmetry measure based on conditional quantiles, which by definition

are not sensitive to outliers. The emphasis on conditional asymmetry is driven

by the conjecture that changes in economic and political conditions, financial

regulations, etc. are likely to be associated with changes in the distribution of

returns (e.g., Bekaert and Harvey (1995) and Martin and Rey (2006)).

Second, we explore several empirical decompositions of the estimated coun-

try skewness to gain insight into the economic sources underlying its time-

series fluctuations and cross-sectional differences. Specifically, we decompose

a country’s skewness into a systematic component, driven by the conditional

2See Rubinstein (1973) and Kraus and Litzenberger (1976,1983).

3For empirical evidence, see Boyer, Mitton, and Vorkink (2010) and Conrad, Dittmar, and

Ghysels (2013). For theory, see Hong and Stein (2003), Brunnermeier, Gollier, and Parker (2007),

and Barberis and Huang (2008).

Why Invest in Emerging Markets? 2147

skewness of the world portfolio, and a residual idiosyncratic component. In

a second decomposition, we regress a country’s skewness onto financial and

economic variables previously shown to proxy for variation in the investment

opportunity set. The idea here is to see how much of the skewness is spanned

by these commonly used variables and whether the unspanned fluctuations

play a role in the portfolio analysis. We also look at whether financial open-

ness, goods market globalization trends, and balance of payments differences

across countries—and EMs, in particular—can account for the differences in

skewness.

Third, we incorporate the skewness measure in the parametric portfolio

weights framework of Brandt, Santa-Clara, and Valkanov (2009). This port-

folio approach allows us to introduce asymmetry in a tractable fashion without

having to specify the joint distribution of returns, as is usually done in the

international portfolio allocation literature (Jondeau and Rockinger (2006),

Guidolin and Timmermann (2008), Christoffersen et al. (2012)). The formu-

lation of the problem allows us to isolate the role conditional skewness plays

in international asset allocation. More importantly, we are able to trace out

the effect of emerging economies’ skewness in the optimal portfolio. The sys-

tematic/idiosyncratic and spanned/unspanned decompositions as well as panel

predictive regressions allow us to further narrow down the economic prove-

nance of the portfolio gains.

Turning to the specifics, the conditional asymmetry measures that we con-

sider, denoted by SK

α,t−1, are based on whether the interval between condi-

tional return quantiles 1 −αand αis centered at the conditional median. For

example, for α=0.75, if at time t−1 the interquartile range is not centered at

the median, then the return distribution is asymmetric. While we emphasize

the fact that this statistic is different from estimating directly the third moment

of returns, we show that our conditional asymmetry measures can be viewed

as approximate scaled versions of the conditional third moment—to the best of

our knowledge, this connection is new to the literature. We also consider a mea-

sure denoted SKINT,t−1that is obtained by integrating over all quantiles α. The

SK

α,t−1and SKINT,t−1statistics have three features of practical importance:

they (i) are robust to outliers, (ii) do not involve the use of options data, and (iii)

can be computed at various horizons. While we have addressed the importance

of the first two points, item (iii) is equally important as we focus our portfolio

analysis on monthly and quarterly (rather than daily) holding-period returns.

Recent asset allocation literature emphasizes long-term investing strategies

(Viceira (2001), Campbell and Viceira (2002)). Rebalancing is costly in EMs

and a portfolio strategy that relies on frequent trading is unlikely to yield

good after-transaction-costs performance. Given the short time span of many

international series, the quarterly horizon strikes a balance between rebalanc-

ing frequency and estimation error. We rely on a novel mixed-data sampling

(MIDAS) approach for the modeling of conditional quantiles that exploits the

richness of daily returns to form long-horizon conditional skewness forecasts.

We estimate SKINT,t−1and SK

α,t−1for a total of 43 countries at monthly and

quarterly horizons. The SK statistics exhibit significant variation over time.