The profitability effect: Insights from international equity markets

• DOI: http://doi.org/10.1111/eufm.12189
• Published date: 01 September 2018
• Date: 01 September 2018

DOI: 10.1111/eufm.12189

ORIGINAL ARTICLE

The profitability effect: Insights from international

equity markets

Te-Feng Chen1Lei Sun2K. C. John Wei1Feixue Xie3

1The Hong Kong Polytechnic University

Emails: tefeng.chen@polyu.edu.hk;

johnwei@ust.hk

2Boston Consulting Group, Shanghai,

China

Email: sunlei.sun@hotmail.com

3The University of Texas at El Paso

Email: fxie@utep.edu

Abstract

Using data from 33 international markets over the period

1990--2017, we reveal that the profitability effect exists in

many countries other than the USA, but the size of this effect

is sensitive to the measure of profitability and portfolio sorts.

The profitability effect is significant in pooled developed

and global markets, though less so in emerging markets. The

evidence lends support to incorporating a profitability fac-

tor in regional/global factor models. Cross-region and cross-

country analyses show mixed evidence for a positive rela-

tionship between profitability effects and market develop-

ments, though the overall findings lean toward supporting

the prediction of the investment model.

KEYWORDS

profitability effect, profitability factor, cross-sectional returns, investment

CAPM, behavioral finance, international equity markets

JEL CLASSIFICATIONS

G12, G14, G15

1INTRODUCTION

The primary goals of this study are to provide large-scale empirical evidence on the profitability effect

based on different profitability measures in international equity markets and to examine how the prof-

itability effect varies across countries with different levels of market development. We first examine

We thank Lu Zhang (the editor) and an anonymous referee for constructive and insightful comments and suggestions that have

significantly improved the paper. We also appreciate the helpful comments from Andrew Karolyi, Hugues Langlois, Sheridan

Titman, Neng Wang, and seminar and conference participants from the 2014 European Finance Association (EFA) conference,

the 2014 Financial Management Association (FMA) conference, the 2014 Asian Finance Association (Asian FA) conference,

Dongbei University of Finance and Economics, Fudan University, National Chengchi University, National Sun Yat-sen Univer-

sity, and National Taiwan University. Previous versions of the paper carried the titles ‘‘On explanations for the gross profitabil-

ity effect: Insights from international equity markets’’ and ‘‘The gross profitability effect: International evidence.’’ John Wei

acknowledges financial support from the Research Grants Council of the Hong Kong Special Administrative Region, China

(project no. 16503514). Any remaining errors are our sole responsibility.

545

Eur Financ Manag. 2018;24:

545--580 wileyonlinelibrary.com/journal/eufm ©2018 JohnWiley & Sons, Ltd.

546 CHEN ET AL.

whether extant findings on the profitability effect transfer beyond the USA and whether the size of this

effect is sensitive to the denominator of profitability measures and portfolio sorts. We focus on the six

measures of profitability commonly used in the literature: gross profitability of Novy-Marx (2013), op-

erating profitability of Fama and French (2015), operating profitability of Ball, Gerakos, Linnainmaa,

and Nikolaev (2015), two lagged gross profitability measures of Zhang (2017), and the profitability

measure of Hou, Xue, and Zhang (2015).

For each profitability measure we construct four measures to quantify the profitability effect in each

country in our sample. The first two measures of the profitability effect are the equal-weighted (EW)

and value-weighted (VW) return spreads between the two extreme profitability quintiles within each

country. The other two measures are the Fama and MacBeth (1973) type of ordinary least squares (OLS)

and weighted least squares (WLS) regression coefficients on profitability. Because profitability mea-

sures may be substantially affected by the variation in accounting standards across countries, we take

a within-country approach by comparing returns on high- and low-profitability firms within the same

country when measuring the profitability effect. This helps alleviate the concern that cross-country

differences in accounting standards may lead to differences in profitability measures and hence the

profitability effect in different countries. We also configure a composite measure of profitability and

an average measure of the profitability effect across the six measures of profitability for each country.

We next examine how the profitability effect varies across countries in the context of predictions

from behavioral finance and the investment capital asset pricing model (CAPM). As elaborated in

Zhang (2017), behavioral finance, which relies on dysfunctional markets, predicts that profitability

should be stronger in emerging markets and weaker in developed markets. The rationale is that in-

vestors in emerging markets are less sophisticated and markets are less efficient, with more severe

limits to arbitrage. The investment CAPM that, by contrast, depends on well-functioning markets, pre-

dicts that the profitability effect should be stronger in developed markets where markets are more effi-

cient and managers are more likely to practice capital budgeting to maximize firm value, and weaker

in emerging markets.1Hence, we investigate the profitability effect at the levels of pooled developed

markets, emerging markets, and global markets based on portfolio sorts and Fama and MacBeth (1973)

regressions. We also apply the Carhart (1997) four-factor model controlling for well-known risk fac-

tors to explore whether there is justification for including a profitability factor in regional and global

factor models. Furthermore, we directly examine cross-country differences in the profitability effect in

relation to capital market development to determine whether evidence exists for the investment CAPM

or behavioral finance explanations.

Our study makes several contributions. First, it broadens current understandings of the profitability

effect beyond the US market. Using monthly stock returns from 33 countries over 1990--2017, we find

a pervasive and significant profitability effect outside the US market, especially for results based on

EW return spreads or OLS regression slopes. However, the evidence of a significant profitability effect

is weaker based on VW return spreads or WLS regression slopes. In addition, by averaging the prof-

itability effects across countries into the aggregated levels of developed markets, developing markets,

and global markets, we find a significant profitability effect for every measure of profitability in every

aggregated category of markets based on EW returns. Moreover, the profitability effect on average

is about 40% greater in developed markets than in developing markets in most cases. In contrast, the

evidence based on VW returns is again weaker, and for some profitability measures the profitability ef-

fect tends to be larger in developing markets than in developed markets, although the difference seems

economically small.

1Titman, Wei, and Xie (2013) also put forward a similar argument for the investment effect (proxied by total asset growth)

regarding the comparison between behavioral and rational (i.e. 𝑞theory) explanations.

CHEN ET AL.547

Second, we find that portfolio sorts and deflators matter for the profitability effect. Take the gross

profitability of Novy-Marx (2013) that is based on annual gross profit-to-assets with annual sorts,

for example. The EW return spread on the high-minus-low quintile portfolio is significant at the 5%

level in 16 countries. Meanwhile, the EW return spread on annual gross profit-to-lagged assets is

statistically significant only in six economies. By contrast, it is statistically significant in 25 economies

with monthly sorts based on quarterly earnings updates.

Third, both portfolio and regression analyses indicate that quarterly earnings with monthly sorts

provide evidence of a strong profitability effect in more economies than any other measures of prof-

itability. This finding supports Hou et al.’s (2015) choice of earnings as the profitability measure for a

workhorse factor in their 𝑞-factor model.

Fourth, based on analysis at the aggregated levels of pooled regions, the evidence from EW or VW

return spreads either with or without controlling for risk shows that the profitability effect exists at the

global level and in developed markets, but not in emerging markets. The difference in the profitability

effect between developed and emerging markets is economically large. These findings are consistent

with the prediction of the investment CAPM and inconsistent with the behavioral explanation. These

findings also lend support for the addition of a profitability factor to regional and global factor models to

explain regional/global equity anomalies as in Fama and French (2017). The results from OLS or WLS

regressions are in general consistent with those from portfolios analysis, except for emerging markets.

The profitability slopes from WLS are mostly significant and in some cases larger for the pooled

emerging markets than for the pooled developed markets, inconsistent with the investment CAPM

prediction. Our country-level and regional/global-level results are robust to alternative measures of

profitability and alternative measures of the profitability effect, in local currency or in US dollars. Our

global profitability effect is not driven by large economies such as the USA and Japan, or by small

firms.

Lastly, our evidence generally fails to support the behavioral finance explanation for the cross-

sectional variation in the profitability effect based on cross-country and cross-region analyses. When

profitability effects are measured by average EW or VW return spreads in cross-region analysis as

well as by average EW return spreads or average OLS profitability slopes in cross-country analysis,

the results tend to support the investment model. That is, profitability effects are significantly stronger

in more developed economies where capital markets tend to be more efficient, managers appear to

be more likely to practice capital budgeting to maximize firm value, and limits to arbitrage are less

severe. However, we find no support for the investment model when profitability effects are measured

by Fama and MacBeth (1973) regression slopes in cross-region analysis as well as by average VW

return spreads or WLS profitability slopes in cross-country analysis. Collectively, although our results

lean toward supporting the investment CAPM, the evidence is overall mixed, which calls for further

future work on this topic.

2LITERATURE REVIEW

The predictive power of firm earnings or profitability on the cross-section of average stock returns

has attracted considerable attention over the last decade. For example, Fama and French (2006)test

the positive relation between expected profitability and expected returns predicted by valuation theory.

Using the fitted value from the first-stage cross-sectional regression for expected profitability as one

of the explanatory variables in the second-stage cross-sectional regressions, Fama and French (2006)

show that the profitability effect is not significant. (Fama and French,2008, p. 1663) further indicate

that hedge returns on profitability sorts do not ‘‘provide much basis for the conclusion that there is