momentum and contrarian strategies in international stock markets: further evidence

J. of Multi. Fin. Manag. 15 (2005) 235–255

Momentum and contrarian strategies in internationalstock markets: Further evidence

Qian Shena,1, Andrew C. Szakmaryb,2, Subhash C. Sharmac,∗a Department of Economics, Finance and OSM, School of Business, P.O. Box 429, Alabama A&M University,

Normal, AL 35762, USAb Department of Finance, Robins School of Business, University of Richmond, Richmond, VA 23173, USA

c Department of Economics, Southern Illinois University at Carbondale, Carbondale, IL 62901, USA

Received 14 April 2004; accepted 14 September 2004Available online 9 April 2005

Abstract

Previous studies have shown that market participants underestimate earnings growth for past win-ner stocks, and that growth stocks are more sensitive to earnings surprises. These findings suggestimplementing momentum strategies with growth stocks. This study investigates linkages betweenvalue versus growth investment styles and momentum strategies in international markets. In addition,we extend Jegadeesh and Titman (2001)-type tests, which attempt to distinguish between competingexplanations of the momentum phenomenon, to international market indices. Our full sample resultsshow that momentum profits are concentrated in the growth indices, and that there is evidence of short-term overreaction in these and other indices that is subsequently corrected. Our subsample resultsare mixed; there is some evidence that the profitability of momentum (but not contrarian) strategiespersists in the post-December 1987 period. However, unlike the earlier period, there is no evidencethat markets overreact and that these overreactions are subsequently corrected.© 2005 Elsevier B.V. All rights reserved.

JEL classification:G11; G15

Keywords:Momentum; Value/glamour; International

∗ Corresponding author. Tel.: +1 618 453 5070.E-mail addresses:[email protected] (Q. Shen), [email protected] (A.C. Szakmary), [email protected]

(S.C. Sharma).1 Tel.: +1 256 372 4885; fax: +1 256 372 5874.2 Tel.: +1 804 289 8251; fax: +1 804 289 8878.

1042-444X/$ – see front matter © 2005 Elsevier B.V. All rights reserved.doi:10.1016/j.mulfin.2004.09.001

236 Q. Shen et al. / J. of Multi. Fin. Manag. 15 (2005) 235–255

1. Introduction

In recent years, practitioners and academic scholars have found that some relatively sim-ple trading strategies based on past cross-sectional stock returns yield significant abnormalprofits. Researchers divide these strategies into two major categories: the contrarian strategythat relies on price reversals and the momentum strategy based on price continuations. Forthe U.S. Stock Market,DeBondt and Thaler (1985)investigate return patterns over extendedperiods of time and find that contrarian strategies perform exceptionally well over 3–5-yearhorizons. In contrast,Jegadeesh and Titman (1993)document that strategies that buy winnerstocks and sell loser stocks (i.e. momentum strategies) generate significant positive returns(about 1% per month) over 3–12-month holding periods. In a comprehensive investigation,Conrad and Kaul (1998)find both momentum and contrarian profits in the U.S. market,depending on the time horizon investigated. Specifically, the contrarian strategy is profitablefor short-term (weekly, monthly) and long-term (2–5 years, or longer) intervals, while themomentum strategy is profitable for medium-term (3–12-month) holding periods.3

Many studies have found that momentum strategies work in international stock markets,although findings have not been uniform.4 Rouwenhorst (1998)examines twelve Europeanmarkets’ stock returns between 1980 and 1995. He finds that an internationally diversifiedportfolio of past medium-term winners outperforms a portfolio of medium-term losers bymore than 1% per month after taking risk factors into consideration. He also presents evi-dence that European and U.S. momentum strategies have a common component suggestingthat a common factor may drive the profitability of momentum strategies in both the U.S.and international markets. Echoing these findings,Schiereck et al. (1999)report substantialintermediate horizon profits to momentum strategies in the German market, and find that theGerman and U.S. markets behave very similarly.Van Dijk and Huibers (2002)link momen-tum profits in European markets to analyst behavior. Specifically, they find that analystssystematically underestimate earnings for strong price-momentum stocks, underestimateautocorrelation in earnings growth between consecutive years, and are in general too slowto adjust their earnings forecasts.5 They suggest that further research should focus on therelative importance of the underreaction explanation for momentum.Hon and Tonks (2003)examine momentum strategies in the U.K. stock market. Interestingly, they report strongevidence for momentum profits out to 24-month horizons, but only for 1977–1996. Theyfind no evidence of momentum profits in the earlier 1955–1976 period in the U.K.

In general, evidence for momentum profits is weaker in Asian markets.Chui et al. (2000)report evidence for momentum profits in these markets, but with the notable exceptionsof those in Japan and Korea.Hameed and Kusnadi (2002)find virtually no evidence ofmomentum profits in the stock markets of six Pacific Basin countries, and althoughKang

3 Our study examines intermediate horizon (1–12 months) and long horizon (2–5 years) strategies. We do notfocus on the very short horizon contrarian strategies (under 1 month) examined byLo and MacKinlay (1990),Jegadeesh and Titman (1995), and others in the U.S. market. Numerous studies have examined these short-termreversal strategies in international markets; a recent article byLee et al. (2003)contains a good review.

4 The 3–5-year contrarian strategies of De Bondt and Thaler (1985) have received virtually no examinationin countries outside the U.S. at the individual firm level. To our knowledge, onlySchiereck et al. (1999)haveexamined such strategies, in the German market. They do not report significant contrarian profits.

5 Chan et al. (1996, 1999)have reported very similar findings for U.S. markets.

Q. Shen et al. / J. of Multi. Fin. Manag. 15 (2005) 235–255 237

et al. (2002)do find significant momentum profits at 3–6-month horizons in the Chinesemarket, their study was confined to the “A” shares traded exclusively by local investors.

Because of the wide availability of country index data, many studies of momentumand contrarian strategies in international equity markets have examined these strategiesat the country (as opposed to individual firm) level.6 These studies attempt to assess ifinvestors under and/or overreact to information not only at the level of the individual stock,but more systematically on a countrywide basis. Evidence for mean reversion in countrystock indices, which is related (albeit imperfectly) to the success ofDeBondt and Thaler(1985)-style contrarian strategies, is reported byRichards (1995)andBalvers et al. (2000).Richards (1997)finds that the DeBondt and Thaler contrarian strategies applied to countryindices earn annualized excess returns of approximately 6% at three and 4-year horizons,and that these excess returns are statistically significant. Richards also reports annualizedmomentum profits of around 3% per annum at 3–12-month horizons, but these are notsignificant. However,Asness et al. (1997)andBalvers and Wu (2002)do report significantmomentum profits in country indices at 6–12-month horizons.Chan et al. (2000)alsouse international equity indices and examine three different phenomena in their study, i.e.country-specific momentum profitability; the relation among momentum profits, interestrate and exchange rate movements; and, the relation between momentum profits and tradingvolume. From these tests, they conclude that the momentum profits arise mainly from time-series predictability in stock market indices, rather than from interest rate or exchangerate movements. In contrast to the U.S. market, where momentum profits generally requireranking and holding periods in excess of 3 months, they also find significant positive profitsfor short holding periods (less than 4 weeks) in the international equity indices.

The causes of momentum and/or contrarian profits have been the subject of considerabledebate. One debate revolves around whether momentum profits can be reconciled withmarket efficiency. Those who argue that such reconciliation is possible stress that observedmomentum profits may be a product of data mining, and/or that momentum profits aresimply fair compensation for risk. While data mining is always a possibility, it is unlikely tobe the major explanation for the momentum anomaly, given that roughly similar strategieshave been shown to work in a variety of global settings, and for different subperiods in theU.S. as documented byConrad and Kaul (1998). Furthermore,Jegadeesh and Titman (2001)find that in U.S. markets, over the 1990–1998 sample period (which was not included intheir original 1993 paper) momentum strategies continue to be profitable, and past winnersoutperform past losers by about the same magnitude as in the earlier period.7

6 Another motivation for examining momentum strategies at the country level is transactions costs. Most of thesecountry indices can be traded using exchange traded funds at modest transactions cost. In contrast, some recentstudies, e.g.Lesmond et al. (2004), Korajczyk and Sadka (2004)find that momentum profits at the individual stocklevel are greatly reduced or eliminated if transactions costs are properly accounted for.

7 In the U.S. market, the long-horizon contrarian strategies do not appear to work as consistently as theintermediate-horizon momentum strategies:Conrad and Kaul (1998)find that the profitability of the formeris essentially limited to the pre-1947 period. In contrast, studies that have found mean reversion, and DeBondt andThaler contrarian profits in international equity indices have all used post-WW2 data.Richards (1997)examinesthe temporal properties of contrarian profits in country indices and finds no firm evidence of any shift in the1972–1995 period.


Similarly, asKorajczyk and Sadka (2004)note, the consensus in the literature is thatrisk factors do not completely explain intermediate horizon momentum profits.Fama andFrench (1996)concede that their three-factor asset-pricing model does not explain returnsto momentum portfolios. Nevertheless, some studies, e.g.Conrad and Kaul (1998), haveindicated momentum profits could be a by-product of certain stocks being riskier thanothers in some unknown way, and thus having higher expected returns. This is becausemomentum strategies take long (short) positions in stocks with high (low) past returns;if these past returns are high (low) because of unknown systematic risk factors, then thesame stocks should continue to earn relatively high (low) returns in future periods. If thisinterpretation is correct, then momentum profits can be consistent with market efficiency.Conrad and Kaul (1998)argue that the cross-sectional variation in the mean returns ofindividual securities plays an important role in their profitability, and could potentially ac-count for the profitability of momentum strategies. However, recent findings reported byJegadeesh and Titman (2001, 2002)cast substantial doubt on the Conrad and Kaul Hy-pothesis. If momentum profits were due primarily to cross-sectional differences in meanreturns, then past winners (losers) should continue to be superior (inferior) performersindefinitely into the future. Instead,Jegadeesh and Titman (2001)find that momentumportfolio (winners minus losers) returns are positive only during the first 12 months of port-folio formation; if anything, momentum portfolio returns beyond this 12-month horizonare negative. As another risk-based explanation,Chordia and Shivakumar (2002)allegethat momentum profits are explained by exposure to macroeconomic risk factors; how-ever,Griffin et al. (2003)demonstrate that this linkage does not exist in an internationalcontext.

In addition toJegadeesh and Titman (2001), other studies have reported results consistentwith behavioral models (e.g.Barberis et al., 1998, Daniel et al., 1998, Hong and Stein, 1999)that suggest that investors initially underreact and, ultimately, overreact to new information.Chan et al. (1996)show that stock prices underreact to earnings news and momentum profitsconcentrate on subsequent earnings announcements. Also, they suggest that investors tendto engage in herding behavior.Lakonishok et al. (1992)find evidence of underreaction forpension fund managers, whileGrinblatt et al. (1995)report similar behavior by mutual funds.However, in a comprehensive study of institutional investors,Badrinath and Wahal (2002)find significant differences in trading practices across different classes of institutions (e.g.banks, pension funds, investment advisors and mutual funds), and caution against makingbroad generalizations concerning institutional trading behavior.

Another anomaly that has received a great deal of attention in the finance literature isthe value/glamour anomaly. Many studies have found that a strategy of investing in valuestocks (i.e. those with relatively low market-to-book ratios, P/E ratios or past earningsgrowth) produces higher returns than investing in growth stocks in the U.S. as well as ininternational markets. Researchers have offered a variety of reasons for this performancedifference.Fama and French (1993)argue that higher average returns on value stocks merelycompensate for the higher risk they bear; value stocks have positive loadings on a factorrelated to relative distress. On the other hand,Lakonishok et al. (1994)and other studiesargue that value strategies earn higher returns because these strategies exploit the suboptimalbehavior of the typical investor. Investors irrationally extrapolate past earnings growth,thereby overvaluing companies that have performed well in the past, and undervaluing


those that have performed poorly. Finally, still another branch of the asset pricing literature(see, for example,Kothari et al., 1995) holds that the superior past performance of valuestocks likely arises from data snooping, which finds accidental patterns in historical datathat are unlikely to be repeated in the future.

The main purposes of this study are to investigate linkages between value/glamour andmomentum strategies in international markets, and to extendJegadeesh and Titman (2001)-type tests, which attempt to distinguish between competing explanations of the momen-tum phenomenon, to international market indices. There are two reasons why momentumstrategies may work better if implemented with growth stocks. First, asChan et al. (1996)show for the U.S. andVan Dijk and Huibers (2002)show for European markets, analystsunderestimate future earnings growth for past winner stocks.Skinner and Sloan (2002)demonstrate that growth stock prices are considerably more sensitive to earnings surprises;consequently, if past winner stocks experience positive earnings surprises on average, thesesurprises will have a greater impact on those winner stocks that are also growth stocks. Asecond issue regarding growth stocks is that there is greater uncertainty in their valuation.As Miller (1977) argues, in an environment in which short selling is at least somewhatrestricted, the most optimistic investors have a disproportionate impact on stock prices. Thegreater the dispersion of opinion as to what a stock is worth, the greater the likelihoodthat a stock will be overvalued. Thus, we conjecture that growth stocks are more likelyto be mispriced. If our conjecture is correct, we would expect much of the momentumprofits generated by holding past winner growth stocks to reverse if these stocks are heldbeyond the 1-year horizon past studies have found optimal for implementing momentumstrategies.

Linkages between the momentum and the value/growth phenomena have been examinedin U.S. markets, but not in a global context.Asness (1997)directly studies this interactionin the U.S. market during the period of 1963–1994 and concludes that both value andmomentum strategies are effective, although value and momentum measures are negativelycorrelated.Grinblatt and Moskowitz (2003)similarly report that, controlling for marketcapitalization, momentum strategies work better if implemented with low book-to-market(i.e. growth) stocks.

In this study, we examine momentum and contrarian strategies at the country indexlevel using the Morgan Stanley Capital International (MSCI) growth, value and blendedindices. If growth stocks are more sensitive to earnings surprises than value stocks, wewould intuitively expect that momentum strategies would work better within the growthindices, and this is exactly what our results show. An examination of performance by monthsafter portfolio formation reveals that somewhere between 7 and 10 months after formation,the momentum portfolios, which initially earn large positive returns, begin earning negativereturns. This underperformance continues out to 36 months after formation. However, thisunderperformance, while highly significant, is qualitatively similar for the growth and valueindices. Thus, our second conjecture that growth stocks are more likely to be mispriced, isnot strongly supported.

In addition to testing for momentum and contrarian profits in our full sample (December1974–November 2000), we also conduct tests for two approximately equal-length subpe-riods, December 1974–November 1987, and December 1987–November 2000. During thelater period, overall equity indices for 13 emerging markets are also available from MSCI,


enabling us to ascertain the profitability of momentum/contrarian strategies in emergingmarket indices, and to compare these with developed markets. Our subperiod results showthat, in developed markets, evidence for contrarian profits is much weaker in the post-December 1987 period than in the previous period. The momentum profits may persistto some degree (depending on the methodology employed); however, contrary to the ear-lier period, there is no evidence of investor overreaction post-1987. Similarly, in emergingmarkets, momentum profits appear to be present post-1987, but at most time horizons thevariance of returns is too large to draw meaningful statistical inferences at conventionallevels.

The balance of this paper is organized as follows: the basic methodology and data aredescribed in Section2. Section3contains more detailed descriptions of the specific empiricaltests and their results. Section4 concludes the paper.

2. Basic methodology and data

In evaluating momentum and contrarian strategies, we rely primarily on the methodologyof Conrad and Kaul (1998), hereafter abbreviated CK, which is in turn an extension of theapproaches ofLehmann (1990)andLo and MacKinlay (1990). Specifically, consider buyingor selling stocks (actually, in our case, entire national stock market indices) at timet− 1based on their performance from timet− 2 to t− 1, where the period{t− 1, t} spans anyfinite time interval. Also, assume that the “performance” of a particular national stock indexis determined relative to the average performance of all indices that are used in the tradingstrategy. Consequently, if the entire universe of assets were included in the strategy, theneach index’s performance would be measured relative to the return on an equal-weighted“global market portfolio,”Rmt.

Let wit−1 denote the fraction of the trading strategy portfolio devoted to indexi. In amomentum strategy, the future weight for each index is directly proportional to its pastrelative performance, that is,

wit−1(k) = ± 1

N[Rit−1(k) − Rmt−1(k)] (1)

whereRit−1(k) is the return on indexi at timet− 1, i = 1, . . ., N, Rmt−1(k) is the return onequal-weighted portfolio of all indices, andk is the length of the time-interval{t− 1, t}.The dollar weights in Eq.(1) lead to an arbitrage (zero-cost) portfolio by construction

N∑

i=1

wit−1(k) = 0 ∀k. (2)

The realized profits at timet, πt(k), to the trading strategies implied by the weights inEq.(1) are given by

πt(k) =N∑

i=1

wit−1(k)Rit(k). (3)


Since all the strategies considered in this paper are zero-cost strategies, only the dollarprofits are defined by Eq.(3). We therefore will largely rely on the sign and statisticalsignificance of the averages of the time series of theπt(k)’s; that is, we examine whetherrealized profits are statistically significantly positive (or negative).

In order to ascertain the robustness of our results, we also conduct some tests usingan adaptation of the approach ofJegadeesh and Titman (1993, 2001), hereafter abbrevi-ated JT. Following their method, the strategies we evaluate include portfolios with over-lapping holding periods. Therefore, in any given montht, the strategies hold a series ofportfolios that are selected in the current month as well as in the previousK− 1 months,whereK is the holding period. Specifically, a strategy that selects indices on the basis ofreturns over the pastJ months and holds them forK months is constructed as follows:each montht the indices are ranked in descending order on the basis of their returns inthe pastJ months. Based on these rankings, we form three equally weighted portfolios.The first group (past winners), T1, holds the top third of the indices based on past perfor-mance. The second group (T2) holds the middle third of indices, while the third group (pastlosers), T3, holds the bottom third. As in JT, we also evaluate a strategy, T1− T3, that ineach montht, buys the past winners and short sells the past losers, holding this positionfor K months. All of the JT portfolios close out the positions initiated in montht−K inmontht. Hence, following JT (1993, 2001) we revise the weights on 1/K of the indices inthe entire portfolio in any given month and carry over the rest from the previous month.For an intuitively easier illustration of how portfolio returns for a given month are calcu-lated based on cohorts identified over previous months, seeJegadeesh and Titman (2001,pp. 702–703).

In both the CK and JT approaches, to test the significance of momentum profits in eachportfolio, we uset-statistics that are asymptotically distributed asN (0, 1), under the nullhypothesis that the “true” profits are zero. Since in general we use overlapping data (the datafrequency is monthly but returns are measured over holding periods of up to 36 months), wecorrect our standard errors for heteroskedasticity and autocorrelation using theNewey andWest (1987)adjustment. In every case, the number of lags used in the adjustment equalsthe number of months of overlap.

The data used in this study is freely available from Morgan Stanley Capital Interna-tional (MSCI). From their web page, we downloaded monthly total return indices for largecapitalization value stocks, large cap growth stocks, and overall (blended) large cap equityindices for all eighteen developed markets that had data available from December 1974 toNovember 2000. These markets are Australia, Austria, Belgium, Canada, France, Germany,Hong Kong, Italy, Japan, the Netherlands, Norway, Singapore, Spain, Sweden, Switzerland,the United Kingdom and the United States. In constructing these indices, MSCI apportionsstocks into the value and growth categories based on book-to-market value: for each coun-try, half of the stocks (the ones with the highest book-to-market) are placed in the valuecategory, and the remaining stocks in the growth category. For comparison purposes, wealso downloaded the overall (blended) equity indices for all 13 emerging markets that haddata available from December 1987 to November 2000. These markets were Argentina,Brazil, Chile, Greece, Indonesia, Korea, Malaysia, Mexico, The Philippines, Portugal, Tai-wan, Thailand, and Turkey. We do not examine growth and value indices for emergingmarkets since these only begin in December 1994.


Beyond the easy accessibility, the advantages of the MSCI data set are: (1) data areformatted in both U.S. dollars and local currency; (2) value and growth stock indices areconstructed in the same manner in each developed and emerging market; and (3) MSCI hasalready reconciled survivorship bias problems in forming the data set.

3. Empirical tests and results

Table 1shows average realized profits for trading strategies implemented with eachdifferent style index across countries using various formation periods (i.e. differentj) fordeveloped markets and the entire December 1974–November 2000 sample period usingthe Conrad and Kaul method. We consider strategies implemented with value indicesexclusively, strategies with growth indices exclusively, and strategies with blended equityindices; these are each implemented with 18 indices (i.e.N= 18). We also consider a

Table 1Average profits of momentum strategies by index style (CK method)

F/H perioda Value Growth Blended Value + growth

1 month 0.000102 0.000143 0.000102 0.000138(1.56137) (2.05345)** (1.80967)* (2.28235)**

2 months 0.000089 −0.000131 0.000130 0.000193(0.56573) (−1.41410) (0.88917) (1.18888)

3 months 0.000140 −0.000238 −0.000071 0.000384(0.47626) (−1.34778) (−0.47972) (1.32143)

6 months 0.001253 0.002704 0.001964 0.001883(1.32117) (3.11117)*** (2.52465)** (2.30610)**

9 months 0.000813 0.003738 0.002267 0.002262(0.50574) (3.41351)*** (1.82204)* (1.81917)*

12 months −0.001170 0.002140 0.000538 0.000618(−0.47146) (1.29787) (0.26940) (0.32662)

24 months −0.009954 −0.013015 −0.011048 −0.011590(−2.6148)*** (−1.76058)* (−2.24280)** (−2.32217)**

36 months −0.017748 −0.028404 −0.021768 −0.023001(−2.84005)*** (−2.27453)** (−2.55475)** (−2.50496)**

48 months −0.026034 −0.034600 −0.028242 −0.028805(−2.93101)*** (−2.09375)** (−2.56629)** (−2.40582)**

60 months −0.034041 −0.046917 −0.037837 −0.037964(−2.90915)*** (−2.14866)** (−2.63477)*** (−2.40796)**

This table presents average profits to momentum strategies implemented with value, growth and blended countryindices from 18 developed markets over different formation/holding periods. The value + growth column is for themomentum strategy implemented simultaneously with both value and growth indices from each country (i.e. with36 indices). We calculate average profits according to Eqs.(1)–(3). The numbers in parentheses aret-statisticsbased on the Newey–West (1987) adjustment and are asymptoticallyN (0, 1) under the null hypothesis that “true”profits are zero. Negative profits accruing to momentum strategies indicate that contrarian strategies are profitable.

∗ Significance at the 10% level.∗∗ Significance at the 5% level.

∗∗∗ Significance at the 1% level.a Denotes formation and holding period, which are constrained to be equal.


strategy implemented for both value and growth indices at the same time; here eachcountry contributes two indices to the mix andN= 36. We form ten different holdingperiodsk, wherek ranges from 1 to 60 months. For simplicity, followingConrad andKaul (1998), in Table 1we implement only strategies for which the lengths of the pastperformance evaluation period and the future holding period are identical. For example,if we form portfolios based on the performance of national stock indices over the past3-month period, the holding period of the trading strategy is also 3 months. Since theprofits of momentum (contrarian) strategies are exactly equal to losses of contrarian(momentum) strategies [see Eqs.(1) and (3)], we implement only momentum strategies(i.e., we use the weightswit−1 = +1/N[Rit−1(k) −Rmt−1(k) ∀ k]). Consequently, a positive(negative) estimate inTable 1implies that on average a momentum (contrarian) strategy isprofitable.

Table 1also contains thet-statistics in parentheses to test the statistical significanceof the average profits (losses); these statistics are asymptoticallyN (0, 1) under the nullhypothesis that the “true” profits are zero. As described earlier, we use overlapping monthlyobservations and implement the Newey–West (1987) adjustment to correct the standarderrors for autocorrelation in the monthly time series of momentum profits.

Several interesting findings emerge from the results reported inTable 1. First, we notethat the profitability patterns in the overall blended equity indices agree with findings fromJegadeesh and Titman (1993), Conrad and Kaul (1998), and other studies, i.e. momentumprofits are found for medium time horizons (6–9-month holding periods) while contrarianprofits are observed for longer holding periods of 2–5 years. Unlike individual stocks in theU.S., but consistent with theChan et al. (2000)study using international market indices,momentum profits are also observed at very short (i.e. 1-month) horizons.

What is new is that unlike in previous studies, we examine growth and value indicesseparately. We find that momentum profits are much stronger in the growth indices, partic-ularly for 6- and 9-month formation/holding periods. For instance, for the growth indices,the t-statistic is 3.11 at 6 months, and 3.41 at 9 months; both of these are significant atwell below the 0.01 level. In contrast, for the blended equity indices, thet-statistics are2.52 and 1.82, respectively, at 6 and 9 months, while for the value indices thet-statisticsare only 1.32 and 0.51, respectively, at 6 and 9 months. When the momentum strategiesare implemented with both the growth and value indices simultaneously, the average expost performance is similar to what is obtained using the blended indices. Thus, altogether,our 6 and 9 month results indicate, fairly conclusively, that intermediate horizon momen-tum strategies in international markets are concentrated in the growth indices: adding valueindices into the mix, in any manner we tried, reduces the profitability of the momentumstrategies.

Interestingly, the results inTable 1also indicate that the growth/value distinction isless important for the long horizon (2–5 years) contrarian strategies. At the 24–60-monthhorizons we evaluate, our results show that both growth and value index-based contrarianstrategies appear to be significantly profitable. In general, the dollar contrarian profits aregreater (momentum profits more negative) for the growth indices, but because the standarderrors are more than proportionately larger for the growth indices, thet-statistics are morenegative for the value indices, indicating that (if anything) contrarian strategies applied tothe value indices work more reliably.


To ascertain the robustness of the findings reported inTable 1, we repeat the analysisusing a variant of the JT (1993, 2001) approach as described earlier.8 The JT results arereported inTable 2, wherein T1 represents the past winners portfolio (equally weightedcombination of the top third of indices based on formation period performance), and T3represents the past losers portfolio. T1− T3 is the difference in realized profits betweenwinner and loser portfolios, i.e. profits accruing to a strategy of buying past winners andselling past losers short. Unlike the CK method, which only gives us a single lump sumprofit measure similar (but not identical) to the T1− T3 column inTable 2, the JT approachgives us information on the performance of individual portfolios formed on the basis ofpast returns. This additional information may be useful for evaluating the robustness of thefindings.

The JT findings inTable 2are not in all cases identical with the CK results reportedin Table 1. For example, the JT methodology appears to indicate that momentum profitscontinue to prevail at the 12-month horizon, while the CK methodology does not. On theother hand, the CK results indicate contrarian profits at 24 months, while the JT resultsdo not. Nevertheless, on the whole, there is a strong degree of congruence between theJT results inTable 2and the CK results inTable 1. Similar to our findings using the CKmethod, there exist strong momentum profits in all four (i.e. value, growth, blended, andvalue + growth) index types for 6- and 9-month formation/holding periods, and the profitsare strongest when these momentum strategies are implemented with growth indices. Forexample, in panel B that reports profits for growth indices witht-statistics, for the 9-monthformation/holding (F/H) period, the difference between winner and loser portfolios is 0.057with a t-statistic of 4.02, while for value indices it is 0.032 with at-statistic of 1.81. Againfor the 9-month F/H period, differences in profits from blended equity and value + growthindices from panels C and D are 0.055 and 0.045 with thet-statistics being 3.58 and 3.09,respectively. This finding confirms that momentum strategies implemented with growthindices appear to be more profitable than those implemented with value indices, though theprofitability using growth indices is not markedly greater than those obtained with overall(blended) indices.

Similarly, the JT results inTable 2confirm that contrarian strategies are profitable for36–60-month formation/holding periods. Just as in the case of the CK method inTable 1,profits appear to be greater in magnitude when these contrarian strategies are implementedwith growth indices, but they are more statistically significant when implemented with thevalue indices.

Following Jegadeesh and Titman (2001), but using the CK methodology, inTable 3weinvestigate how far into the future momentum strategies using different style indices can earnabnormal returns. We focus on the 6- and 9-month formation periods that were found to yieldsignificantly positive post-formation profits for the growth and blended indices. Panel A ofthe table reports month-by-month profits, for the first 12 post-formation months, for value,

8 As an initial robustness check, we tested whether the CK results reported inTable 1would differ if the nationalstock indices were converted to a currency other than U.S. dollars, i.e. Deutschemarks or Japanese Yen. Theexchange rate data is from the Federal Reserve Bank of St. Louis’ FRED®. We find that momentum profits whenthe indices are converted to DM or Yen are very comparable to those obtained when the indices are denominatedin U.S. dollars. Therefore, we omit reporting these results.


Table 2Average profits of momentum strategies by index style (JT method)Panel A: value indices Panel B: growth indices

F/H period T1 T2 T3 T1− T3 T1 T2 T3 T1− T3

1 month 0.012918 0.011834 0.009091 0.003827 0.011966 0.007890 0.006989 0.004977(4.69327)*** (4.63160)*** (3.47666)*** (1.81045)* (4.03725)*** (2.95194)** (2.56567)** (2.03551)**

2 months 0.024032 0.024621 0.018179 0.005853 0.021493 0.016881 0.014566 0.006928(4.78723)*** (5.72997)*** (3.66948)*** (1.44611) (4.05431)*** (3.70184)*** (2.96579)*** (1.64457)*

3 months 0.036598 0.036742 0.027362 0.009235 0.036798 0.023806 0.019380 0.017418(4.97897)*** (5.65218)*** (4.13818)*** (1.58914) (4.60822)*** (3.51350)*** (2.98759)*** (2.62760)***

6 months 0.082721 0.075360 0.051651 0.031069 0.083826 0.053940 0.033782 0.050044(6.05965)*** (6.18353)*** (3.89901)*** (2.43611)** (5.39781)*** (4.37582)*** (2.71801)*** (4.14341)***

9 months 0.114356 0.116058 0.082725 0.031631 0.115771 0.087542 0.058673 0.057098(6.03999)*** (6.40559)*** (3.93004)*** (1.80970)* (5.25842)*** (4.62706)*** (3.00177)*** (4.01971)***

12 months 0.141037 0.154891 0.125693 0.015344 0.142724 0.121357 0.095483 0.047241(5.48696)*** (6.23433)*** (4.95015)*** (0.81940) (4.95253)*** (4.57541)*** (3.61082)*** (2.48895)**

24 months 0.274817 0.289914 0.313472 −0.038655 0.218021 0.267917 0.278891 −0.060870(5.37285)*** (6.16978)*** (6.52712)*** (−1.46465) (3.57292)*** (6.19962)*** (4.67571)*** (−1.15818)

36 months 0.377263 0.439261 0.486286 −0.109022 0.286243 0.401874 0.433791 −0.147548(4.84790)*** (5.40850)*** (6.51419)*** (−2.50529)** (3.44291)*** (5.90612)*** (4.52316)*** (−1.97841)**

48 months 0.523450 0.618729 0.667378 −0.143928 0.419321 0.560018 0.563516 −0.144195(5.67008)*** (6.56525)*** (6.23172)*** (−2.60371)** (4.74147)*** (6.37320) (4.41357)*** (−1.50517)

60 months 0.693745 0.801913 0.858627 −0.164882 0.529816 0.741851 0.764447 −0.234630(7.26756)*** (6.81643)*** (6.37860)*** (−2.76510)*** (6.09623)*** (8.84551)*** (4.69249)*** (−2.02409)**

Panel C: blended indices Panel D: value + growth indices

F/H period T1 T2 T3 T1− T3 T1 T2 T3 T1− T3

1 month 0.012662 0.008858 0.009477 0.003185 0.012306 0.010727 0.007311 0.004995(4.52095)*** (3.42507)*** (3.75585)*** (1.53657) (4.52164)*** (4.25221)*** (2.83833)*** (2.49979)**

2 months 0.021078 0.023500 0.016582 0.004496 0.023294 0.020672 0.015921 0.007374(4.10210)*** (5.48510)*** (3.51246)*** (1.15921) (4.64099)*** (4.92690)*** (3.33347)*** (1.94543)*

3 months 0.037861 0.030833 0.023509 0.014352 0.036995 0.029918 0.023430 0.013565(5.17530)*** (4.60804)*** (3.71874)*** (2.45534)** (5.07853)*** (4.65107)*** (3.66251)*** (2.40428)**

6 months 0.084982 0.066720 0.042619 0.042363 0.081467 0.065190 0.043984 0.037483(6.06411)*** (5.57182)*** (3.32948)*** (3.70076)*** (5.86412)*** (5.49655)*** (3.51078)*** (3.37946)***

9 months 0.122556 0.103555 0.067062 0.055493 0.114761 0.103431 0.069370 0.045391(6.13791)*** (5.60290)*** (3.32828)*** (3.58244)*** (5.80782)*** (5.62706)*** (3.49120)*** (3.09405)***

12 months 0.147076 0.142042 0.108798 0.038278 0.144055 0.134044 0.112493 0.031562(5.41919)*** (5.54658)*** (4.28417)*** (1.98470)** (5.43708)*** (5.29373)*** (4.42189)*** (1.80374)*

24 months 0.249035 0.288299 0.298140 −0.049105 0.243865 0.285306 0.292346 −0.048481(4.33563)*** (6.66611)*** (5.72137)*** (−1.18436) (4.48590)*** (6.26355)*** (5.69592)*** (−1.35312)

36 months 0.328572 0.434246 0.467796 −0.139224 0.325150 0.434620 0.452588 −0.127438(4.08944)*** (6.26755)*** (5.51115)*** (−2.71121)*** (4.01762)*** (5.80613)*** (5.68092)*** (−2.29613)**

48 months 0.462172 0.600035 0.633679 −0.171507 0.467166 0.597496 0.611543 −0.144377(5.03246)*** (6.79071)*** (5.43112)*** (−2.50607)** (5.26860)*** (6.41583)*** (5.36402)*** (−1.94490)*

60 months 0.611890 0.782679 0.817907 −0.206017 0.605366 0.787429 0.802404 −0.197039(6.21476)*** (7.81817)*** (5.57544)*** (−2.81054)*** (6.81254)*** (7.88009)*** (5.38620)*** (−2.26356)**

These panels show average profits to momentum strategies implemented with value, growth and blended countryindices from 18 developed markets over different formation/holding period using the Jegadeesh and Titmanmethodology. The value + growth panel is for the momentum strategy implemented simultaneously with bothvalue and growth indices from each country (i.e. with 36 indices). At the beginning of each montht the indicesare ranked in descending order on the basis of their returns in the pastJmonths. Based on these rankings, we formthree equal-weighted portfolios. The T1 portfolios (past winners) consist of the 6 top performing country indicesover the formation period; T2 contains the middle 6, and T3 (past losers) consists of the bottom 6.


∗∗∗ Significance at the 1% level.

246Q.S

henetal./J.o

fMulti.F

in.M

anag.1

5(2005)235–255

Table 3Month-by-month and long run profits accruing to 6 and 9 month formation period momentum strategies (CK method)

Post-formationmonth(s)

6-month formation period 9-month formation period

Value Growth Blended Value + growth Value Growth Blended Value + growth

Panel A: average monthly profits by post-formation month1st 0.000164 0.000458 0.000277 0.000339 0.000308 0.000652 0.000458 0.0004922nd 0.000198 0.000456 0.000326 0.000331 0.000234 0.000586 0.000412 0.0004023rd 0.000199 0.000403 0.000323 0.000293 0.000313 0.000675 0.000516 0.0004774th 0.000208 0.000411 0.000313 0.000296 0.000260 0.000571 0.000417 0.0004015th 0.000192 0.000409 0.000304 0.000280 0.000133 0.000354 0.000261 0.0002336th 0.000300 0.000491 0.000405 0.000372 0.000107 0.000276 0.000208 0.0001847th 0.000199 0.000348 0.000286 0.000257 −0.000046 0.000117 0.000078 0.0000328th −0.000018 0.000075 0.000049 0.000026 −0.000243 −0.000113 −0.000147 −0.0001779th −0.000065 0.000064 0.000009 0.000007 −0.000291 −0.000142 −0.000191 −0.00020910th −0.000178 −0.000058 −0.000088 −0.000107 −0.000323 −0.000285 −0.000265 −0.00028911th −0.000260 −0.000222 −0.000218 −0.000228 −0.000355 −0.000374 −0.000335 −0.00034412th −0.000369 −0.000394 −0.000361 −0.000366 −0.000367 −0.000413 −0.000370 −0.000370

Panel B: aggregated profits by post-formation months1–6 0.001253 0.002704 0.001964 0.001883 0.001355 0.003434 0.002368 0.002325

(1.32117) (3.11117)*** (2.52465)** (2.30610)** (1.09274) (3.46996)*** (2.44700)** (2.31445)**

7–12 −0.000678 0.000192 −0.000209 −0.000154 −0.001627 −0.000755 −0.001106 −0.001076(−0.77253) (0.33688) (−0.30425) (−0.22982) (−1.39100) (−0.85207) (−1.12207) (−1.14672)

13–18 −0.001007 −0.001292 −0.001094 −0.001151 −0.001016 −0.001670 −0.001303 −0.001351(−1.41417) (−1.52496) (−1.47360) (−1.60507) (−1.02409) (−1.41250) (−1.25800) (−1.31902)

19–24 −0.000349 −0.000880 −0.000661 −0.000634 −0.000819 −0.001263 −0.001066 −0.001072(−0.60971) (−1.34185) (−1.15663) (−1.08806) (−1.23940) (−1.47467) (−1.53539) (−1.55115)

25–30 −0.000907 −0.001162 −0.000986 −0.001050 −0.001458 −0.002114 −0.001810 −0.001790(−2.23633)** (−1.90777)* (−2.35249)** (−2.43322)** (−2.72288)*** (−2.35476)** (−2.99812)*** (−2.89046)***

31–36 −0.000812 −0.001493 −0.001285 −0.001142 −0.000790 −0.001907 −0.001401 −0.001354(−1.68924)* (−2.07408)** (−2.39538)** (−2.09161)** (−1.33630) (−2.25371)** (−2.29757)** (−2.15928)**

7–36 −0.002491 −0.004236 −0.003249 −0.003342 −0.004118 −0.007405 −0.005496 −0.005785(−1.82435)* (−1.34330) (−1.55527) (−1.55071) (−2.14201)** (−1.87352)* (−2.14484)** (−2.07700)**

13–36 −0.002625 −0.004808 −0.003686 −0.003770 −0.003651 −0.007228 −0.005344 −0.005508(−2.11422)** (−1.73529)* (−2.02628)** −(1.97829)** (−2.14030)** (−1.87352)* (−2.14484)** (−2.07700)**




growth, blended, and value + growth index strategies. Because monthly profits should becomparable across styles and across time relative to formation, to conserve space, we reportonly the profits themselves and omitt-statistics. These results show that the momentumstrategies earn positive profits for 6–9 months after portfolio formation, but negative profitsthereafter. Consistent with the results inTable 1, regardless of whether the formation periodis 6 or 9 months, momentum strategies implemented with growth indices only outperformthose implemented with any type of blended index in each of the first six post-formationmonths, while strategies implemented with value indices underperform those using blendedindices.

Panel B inTable 3reports momentum profits (with associatedt-statistics) for 6-monthholding periods, up to 36 months after portfolio formation. These results show that all of thestrategies (i.e. value, growth, blended and value + growth), regardless of formation period,earn negative profits in every 6-month period that begins in the 13th post-formation monthor later. These profits are all at least marginally significantly negative when aggregatedover 13–36 months post-formation, and they appear to be more negative for the growthstrategies. Because of the substantially greater standard errors associated with the growthstrategies, however, the profit reversals for the growth strategy are less significant than forthe value strategy. Nevertheless, these results confirm and extend those ofJegadeesh andTitman (2001), who report similar long-term reversals for momentum strategies based onindividual stocks in the U.S. market. Thus, our results indicate that in international marketsas well, cross-sectional differences in mean returns across indices cannot account for thenear-term profitability of momentum strategies.9

As a further investigation, we examine the profits accruing to momentum/contrarianstrategies across developed markets by subperiod, and (insofar as possible) compare theprofitability of these strategies across developed and emerging market indices. Since theearliest MSCI data for emerging markets starts in December 1987, we divide our data fordeveloped markets into two subperiods to match the emerging markets data and to obtaina precise comparison. The first subperiod is from December 1974 to November 1987, andthe second subperiod is from December 1987 to November 2000. Because the total dataduration within each subperiod is relatively short, we only examine formation and holdingperiods of up to 24 months in both CK and JT methods.

We report the basic subperiod/emerging market results, using the CK methodology,in Table 4. Again, we constrain the formation and holding periods to be equal. For thedeveloped markets, these results indicate that momentum profits are much stronger duringthe first subperiod that extends from December 1974 to November 1987.10 For example,during this sub-period, the growth indices generate a mean profit of 0.44 for a 6-month

9 Jegadeesh and Titman (1993)find that in contrast to what happens normally, loser stocks significantly out-perform winners in January. In their 2001 study, they reexamine this January seasonality and reach the sameconclusion. To our knowledge, seasonality has not been examined for momentum profits based on internationalmarket indices. Consequently, for each of the strategies implemented inTable 1, we tested whether profits in Jan-uary were significantly different from other months. These results, available from the authors on request, indicatedthat seasonality was not a factor in momentum/contrarian strategies based on international market indices.10 Given the global stock market crash in October 1987, we also examined momentum/contrarian profitability for

the first subperiod with October–November 1987 excluded. These results generally indicated higher momentumprofits for the first subperiod than we report inTable 3.

248Q.S

henetal./J.o

fMulti.F

in.M

anag.1

5(2005)235–255

Table 4Average profits of momentum strategies by index style: two subperiods and emerging markets (CK method)

F/H period Developed markets value Developed markets growth Developed markets blended Developed markets value + growth Emerging marketsblended

Period 1 Period 2 Period 1 Period 2 Period 1 Period 2 Period 1 Period 2 Period 2

1 month 0.000174 0.000025 0.000232 0.000051 0.000194 0.000006 0.000211 0.000062 0.000304(1.91422) (0.26452) (2.28362)** (0.53434) (2.29791)** (0.08356) (2.40630)** (0.73896) (0.71557)

2 months 0.000193 0.000050 0.000342 0.000197 0.000253 0.000074 0.000284 0.000167 0.000375(0.87692) (0.22446) (1.12324) (0.81221) (1.10184) (0.42193) (1.15509) (0.79853) (0.36363)

3 months 0.000370 0.000039 0.000983 0.000250 0.000693 0.000127 0.000710 0.000180 0.000409(0.86681) (0.09734) (1.78061)* (0.66059) (1.56574) (0.43700) (1.54078) (0.52442) (0.23973)

6months 0.001629 0.001485 0.004401 0.001623 0.003029 0.001513 0.003047 0.001300 0.008413(1.41789) (0.97128) (3.30885)*** (1.54292) (2.68369)*** (1.46004) (2.75682)*** (1.10151) (1.52969)

9months 0.001335 0.001214 0.004554 0.003254 0.003006 0.002112 0.003132 0.001978 0.005792(0.66628) (0.45310) (2.89977)*** (1.93869)* (1.73448)* (1.08889) (2.01725) (0.94886) (0.54688)

12months −0.000017 −0.001713 0.000330 0.003489 0.000314 0.000840 0.000435 0.000870 −0.016268(−0.00558) (−0.39068) (0.13671) (1.33649) (0.11823) (0.24120) (0.17974) (0.25729) (−1.02575)

24months −0.008999 −0.007445 −0.027692 0.009043 −0.017502 0.000806 −0.018464 0.000660 −0.030952(−1.98482)** (−0.95656) (−2.88288)*** (1.83583)* (−2.69970)*** (0.13527) (−2.67002)*** (0.11938) (−1.68346)*

This table presents average profits to momentum strategies implemented with different style indices for two subperiods and for emerging markets. Thefirst period isDecember 1974–November 1987, and the second period is December 1987–November 2000. Only overall country indices (blended style) are available for emergingmarkets, and only for the second period. We calculate average profits according to Eqs. (1)–(3). The numbers in parentheses aret-statistics based on the Newey-West(1987) adjustment and are asymptoticallyN (0,1) under the null hypothesis that “true” profits are zero.




formation/holding period, with at-statistic of 3.31. For a 9-month period, the average profitis 0.46 with at-statistic of 2.90. In contrast, out to a 9-month horizon, the momentumprofits obtained in the second sub-period (i.e. from December 1987 to November 2000) areall smaller and with one exception insignificantly different from zero (the profits from usingthe growth indices at the 9-month horizon are marginally significant). Surprisingly, we donot find any significant momentum profits in the (blended) emerging market indices. Thisresult, however, may be due primarily to the volatility of these indices and the relativelyshort period over which they are examined. At 6- and 9-month horizons, the momentumstrategies applied to emerging markets appear to earn larger positive profits than similarstrategies in developed markets (in either subperiod), but the standard errors are so largethat we cannot establish statistical significance.

Because we feel we can reliably go out to only 24 months given the short span of the data,it is difficult to draw firm conclusions for the contrarian strategies by subperiod.11 However,what results we do obtain (for 2-year formation/holding periods) indicate that, like the mo-mentum strategies, the contrarian strategies work better in the first period than in the morerecent one. In the pre-December 1987 period, all of the 2-year momentum strategies pro-duce significantly negative profits (indicating that contrarian strategies are profitable); thissignificance completely disappears in the post-December 1987 period. Indeed, in the lattersubperiod, momentum strategies actually produce marginally significant positive profits at2-year horizons.

We show subperiod results employing the JT methodology inTable 5. Panel A showsrealized profits for momentum strategies implemented with value indices in developedmarkets during two periods. The left partition is the first period’s (i.e. from December 1974to November 1987) profits for different formation/holding periods, and the right partitionis the second period’s (i.e. from December 1987 to November 2000) profits. Panels B, C,and D report realized profits for growth, equity blended, and value + growth indices, whilepanel E shows profits from emerging markets for the second period.

Here, the JT results differ in important ways from the CK subperiod results inTable 4.As expected, for the developed markets, for the 6- and 9-month holding periods that areof primary interest, profits (as measured by the T1− T3 columns) are stronger during thefirst subperiod; however, the profitability of these 6–9-month momentum strategies remainssignificantly positive in the second period, except in the case of the value indices. Indeed, atthe 12-month horizon for the growth indices, profitability is greater in the second period. Incontrast, there is strong evidence that contrarian strategies work at 24-month horizons forall index styles in developed markets during the first period, but this profitability completelydisappears in the second period. For the emerging market indices, the JT results are similarto the earlier CK results. The only noteworthy difference is that the JT method indicatessignificant momentum profits (at the 5% level) at the 6-month horizon. Thus, overall, the JTresults inTable 5indicate a greater degree of persistence of intermediate horizon momentumprofits into the post-December 1987 period than do the CK results inTable 4.

11 A 24-month formation period combined with a similar length holding period results in 4 years of lost datawithin subperiods that are approximately 13 years long. Later, when we examine long-term returns based on 6-and 9-month formation periods, we report future holding period returns out to 36 months because the total loss ofdata (combined with the formation period) is under 4 years.


Table 5Average profits of momentum strategies by index style: two subperiods and emerging markets (JT method)

Panel A: developed markets—value indices

F/H period Period 1 Period 2

T1 T2 T3 T1− T3 T1 T2 T3 T1− T3

1 month 0.013275 0.013812 0.008904 0.004371 0.012566 0.009881 0.009276 0.003290(3.19378)*** (3.41303)*** (2.49930)** (1.43829) (3.47253)*** (3.15524)*** (2.42213)** (1.11991)

2 months 0.026744 0.026342 0.017243 0.009501 0.021390 0.022945 0.019092 0.002299(3.20659)*** (3.92814)*** (2.43123)** (1.61606) (3.80271)*** (4.26743)*** (2.74264)*** (0.41744)

3 months 0.043837 0.040093 0.027036 0.016801 0.029636 0.033519 0.027675 0.001961(3.76935)*** (3.93936)*** (2.80296)*** (2.02165)** (3.61813)*** (4.35034)*** (3.09855)*** (0.25510)

6 months 0.097541 0.089600 0.058579 0.038961 0.069040 0.062215 0.045256 0.023784(4.45211)*** (4.26200)*** (2.76363)*** (2.16856)** (4.40838)*** (5.15719)*** (2.79364)*** (1.34633)

9 months 0.128722 0.133975 0.105306 0.023416 0.101648 0.100208 0.062750 0.038898(4.04517)*** (3.92705)*** (3.00347)*** (1.07679) (4.70277)*** (’6.62097)*** (2.62605)*** (1.47822)

12 months 0.164715 0.183537 0.147978 0.016738 0.121001 0.130651 0.106837 0.014164(3.52550)*** (3.85693)*** (3.23408)*** (0.68680) (4.83963)*** (6.58124)*** (4.03795)*** (0.51027)

24 months 0.317171 0.345799 0.375901 −0.058730 0.245495 0.251225 0.270252 −0.024756(3.03808)*** (3.33846)*** (3.97986)*** (−2.44804)*** (5.32353)*** (12.02029)*** (6.96282)*** (−0.60902)

Panel B: developed markets—growth indices


T1 T2 T3 T1− T3 T1 T2 T3 T1− T3

1 month 0.013867 0.008585 0.007030 0.006837 0.010090 0.007204 0.006950 0.003140(2.95698)*** (2.08000)** (1.82468)* (2.01549)** (2.77812)*** (2.11155)** (1.80041)* (0.89549)

2 months 0.023987 0.018718 0.014049 0.009938 0.019064 0.015092 0.015070 0.003994(2.84754)*** (2.56020)** (1.86770)* (1.60117) (2.94473)*** (2.77712)*** (2.35803)** (0.70536)

3 months 0.046795 0.024872 0.018494 0.028301 0.036798 0.023806 0.019380 0.017418(3.73047)*** (2.31432)** (1.85317)* (2.87531)*** (3.02714)*** (2.92198)*** (2.47498)** (0.83559)

6 months 0.104682 0.063053 0.037353 0.067329 0.064574 0.045528 0.030486 0.034088(4.10309)*** (2.84046)*** (1.72916)* (3.87506)*** (3.80315)*** (4.09168)*** (2.39239)** (2.11099)**

9 months 0.134126 0.105576 0.074168 0.059957 0.099534 0.071589 0.044965 0.054569(3.25229)*** (2.97598)*** (2.05959)** (2.77863)*** (5.19865)*** (4.52433)*** (2.45962)** (2.90978)***

12 months 0.156014 0.144050 0.130307 0.025707 0.131478 0.102155 0.066017 0.065462(2.70342)*** (2.75405)*** (2.58693)** (0.88789) (6.19962)*** (5.20900)*** (2.94706)*** (2.69621)***

24 months 0.221467 0.312595 0.396910 −0.175443 0.215635 0.236986 0.197185 0.018450(1.60006) (3.21396)*** (3.66958)*** (−4.03043)*** (5.34008)*** (16.14659)*** (4.12419)*** (0.28012)

Panel C: developed markets—blended indices


T1 T2 T3 T1− T3 T1 T2 T3 T1− T3

1 month 0.014899 0.008818 0.009843 0.005057 0.010454 0.008897 0.009116 0.001337(3.37421)*** (2.15874)** (2.78045)*** (1.61003) (3.03431)*** (2.78036)*** (2.53031)** (0.49543)

2 months 0.024416 0.024761 0.015999 0.008417 0.017825 0.022272 0.017149 0.000676(2.88792)*** (3.65313)*** (2.20935)** (1.39800) (3.04655)*** (4.23845)*** (2.79851)*** (0.13964)

3 months 0.047256 0.033573 0.022222 0.025034 0.028828 0.028198 0.024747 0.004081(4.05067)*** (3.19117)*** (2.26952)** (2.79788)*** (3.62166)*** (3.66881)*** (3.12197)*** (0.58989)

6 months 0.103682 0.080957 0.046097 0.057584 0.067721 0.053579 0.039409 0.028312(4.47661)*** (3.91103)*** (2.12933)** (3.57862)*** (4.53731)*** (4.50301)*** (2.79831)*** (1.81378)*

9 months 0.141131 0.127143 0.081094 0.060036 0.106124 0.082688 0.054649 0.051474(3.93182)*** (3.65879)*** (2.26523)** (2.76302)*** (5.40593)*** (5.72922)*** (2.60621)*** (2.35766)**

12 months 0.169883 0.169431 0.135223 0.034659 0.127778 0.118868 0.086438 0.041340(3.29521)*** (3.40195)*** (2.89135)*** (1.35277) (5.38184)*** (6.14193)*** (3.45945)*** (1.46123)

24 months 0.283642 0.333441 0.394461 −0.110819 0.225077 0.257048 0.231457 −0.006380(2.26825)** (3.43571)*** (4.18877)*** (−2.73325)*** (5.15079)*** (14.29682)*** (5.33060)*** (−0.11116)


Table 5 (Continued)Panel D: developed markets—value + growth indices


T1 T2 T3 T1− T3 T1 T2 T3 T1− T31 month 0.013517 0.012103 0.007116 0.006401 0.011110 0.009369 0.007504 0.003606

(3.15478)*** (3.03371)*** (1.97507)** (2.20903)** (3.29893)*** (3.01959)*** (2.03368)** (1.31265)2 months 0.026206 0.022601 0.014734 0.011472 0.020457 0.018792 0.017077 0.003380

(3.12364)*** (3.44272)*** (2.04057)** (1.97963)** (3.68043)*** (3.61838)*** (2.69659)*** (0.70165)3 months 0.045450 0.032114 0.023000 0.022451 0.028864 0.027807 0.023844 0.005021

(3.86846)*** (3.15145)*** (2.39545)** (2.63693)*** (3.71145)*** (3.78501)*** (2.85859)*** (0.72605)6 months 0.099412 0.077173 0.048818 0.050594 0.064902 0.054128 0.039521 0.025381

(4.35196)*** (3.66261)*** (2.29031)** (3.21494)*** (4.28727)*** (4.97940)*** (2.89101)*** (1.68153)*

9 months 0.132036 0.122161 0.086740 0.045296 0.099480 0.086863 0.054005 0.045475(3.75820)*** (3.49548)*** (2.46430)** (2.33447)** (4.99426)*** (6.00279)*** (2.63210)*** (2.11674)**

12 months 0.165220 0.156120 0.141961 0.023259 0.126147 0.115364 0.087559 0.038588(3.26113)*** (3.12929)*** (2.97873)*** (0.93992) (5.55000)*** (6.06961)*** (3.76115)*** (1.56921)

24 months 0.265370 0.346350 0.373202 −0.107832 0.228977 0.243045 0.236368 −0.007391(2.20141)** (3.49410)*** (3.74277)*** (−3.43530)*** (5.69532)*** (12.80268)*** (5.89022)*** (−0.14925)

Panel E: emerging markets indices (period 2)

F/H period T1 T2 T3 T1− T31 month 0.011922 0.009190 0.001310 0.010612

(1.72814)* (1.48571) (0.17167) (1.59782)2 months 0.015246 0.021439 0.008043 0.007203

(1.34277) (1.75429)* (0.51384) (0.54443)3 months 0.020181 0.030379 0.016300 0.003881

(1.11373) (1.97981)** (0.64869) (0.19205)6 months 0.094380 0.050400 −0.004837 0.099216

(2.97419)*** (1.62984) (−0.09016) (2.06388)**

9 months 0.121601 0.047907 0.044698 0.076903(2.70353)*** (1.03625) (0.56147) (1.02726)

12 months 0.090862 0.047066 0.115832 −0.024969(2.00721)** (0.81224) (1.11609) (−0.23840)

24 months 0.071426 0.128687 0.145854 −0.074428(1.15583) (0.86404) (1.65083)* (−1.03647)



In Table 6, using the CK methodology, for each of the subperiods for the developedmarket indices as well as for the emerging market indices in the second subperiod, weinvestigate how momentum strategies perform further into the future, that is, beyond aninitial 6-month holding period. We report profitability (andt-statistics) for both the 6- and 9-month formation periods by 6-month holding period, ranging from the first 6 post-formationmonths all the way up to months 31–36. In one respect, these results agree with our earlierfinding inTable 3for the whole sample period: momentum profits are generally largest (andsometimes significant) for the first 6 post-formation months. TheTable 6results, however,underscore a crucial difference between the first and second subperiods, most notably for thegrowth indices. In the pre-November 1987 period, positive profits in the first 6 months arefollowed by negative profits later, indicating markets initially overreact by pushing up thebest-performing countries growth indices too much, with this overvaluation subsequentlycorrected in later periods. In the December 1987–November 2000 period, however, we donot observe negative profits in later periods (except, insignificantly so, in the value indexstrategy). This finding, coupled with the smaller initial momentum profits, is consistent withno initial overreactions among the growth indices (or any of the others, for that matter) in

252Q.S

henetal./J.o

fMulti.F

in.M

anag.1

5(2005)235–255

Table 6Long-run profits accruing to 6- and 9-month formation period momentum strategies: by subperiod and by developed vs. emerging markets (CK method)Post-formation months Developed markets value Developed markets growth Developed markets blended Developed markets value + growth Emerging markets

blendedPeriod 1 Period 2 Period 1 Period 2 Period 1 Period 2 Period 1 Period 2 Period 2

6-month formation period1–6 0.001629 0.001485 0.004401 0.001623 0.003029 0.001513 0.003047 0.001300 0.008413

(1.41789) (0.97128) (3.30885)*** (1.54292) (2.68369)*** (1.46004) (2.75682)*** (1.10151) (1.12296)7–12 −0.000358 −0.000987 −0.000550 0.000462 −0.000397 −0.000257 −0.000337 −0.000202 −0.006249

(−0.31352) (−0.68109) (−0.83799) (0.48463) (−0.45277) (−0.22476) (−0.41310) (−0.17684) (−1.93451)*

13–18 −0.000925 −0.001542 −0.003131 −0.000042 −0.001973 −0.000707 −0.002042 −0.000771 −0.012585(−0.87795) (−1.43567) (−2.12431)** (−0.05347) (−1.65853)* (−0.72585) (−1.72723)* (−0.87349) (−2.35955)**

19–24 −0.000549 −0.000361 −0.002115 0.000742 −0.001186 0.000038 −0.001324 0.000130 −0.003734(−0.54195) (−0.59412) (−1.84316)* (1.13778) (−1.17327) (0.06066) (−1.24580) (0.22672) (−0.59083)

25–30 −0.000737 −0.000970 −0.001707 0.000553 −0.001092 −0.000343 −0.001276 −0.000181 0.002008(−1.08924) (−1.96765)* (−1.76402)* (0.94550) (−1.63909) (−0.73528) (−1.89902)* (−0.36638) (0.94318)

31–36 −0.00166 0.00016 −0.00208 0.000688 −0.00209 0.000311 −0.00186 0.000454 0.005918(−2.09868)** (0.28929) (−1.84624)* (1.29958) (−2.51688)** (0.62040) (−2.20242)** (0.91670) (1.06106)

7–36 −0.002241 −0.000864 −0.008421 0.003861 −0.005342 0.001534 −0.005237 0.001470 −0.010628(−0.99470) (−0.52938) (−1.71894)* (2.16750)** (−1.64350) (0.96797) (−1.48862) (0.99238) (−1.11811)

13–36 −0.003064 −0.001736 −0.008172 0.002444 −0.005650 0.000270 −0.005669 0.000305 −0.010287(−1.26175) (−1.08941) (−1.72964)* (1.80486)* (−1.73574)* (0.18579) (−1.59784) (0.23228) (−1.18328)

9-month formation period1–6 0.001862 0.001495 0.004976 0.002329 0.003435 0.001840 0.003530 0.001625 0.008615

(2.60934)** (0.74659) (3.35153)*** (1.69636)* (2.37684)** (1.35783) (2.60934)** (1.04916) (1.12296)7–12 −0.001466 −0.001798 −0.002498 0.000328 −0.001884 −0.000653 −0.001882 −0.000597 −0.013048

(−1.60586) (−0.90542) (−2.12088)** (0.25188) (−1.49564) (−0.39846) (−1.60586) (−0.37854) (−1.93451)*

13–18 −0.001121 −0.001690 −0.004371 0.000323 −0.002581 −0.000696 −0.002729 −0.000714 −0.014695(−1.54419) (−1.24224) (−2.07989)** (0.31574) (−1.47717) (−0.55967) (−1.54419) (−0.63036) (−2.35955)**

19–24 −0.000809 −0.000892 −0.002694 0.001130 −0.001555 −0.000070 −0.001793 0.000079 −0.002863(−1.43245) (−1.25801) (−1.83567)* (1.34280) (−1.26747) (−0.09534) (−1.43245) (0.11688) (−0.59083)

25–30 −0.001781 −0.000929 −0.002997 0.000789 −0.002367 −0.000276 −0.002415 −0.000031 0.005374(−2.64926)*** (−1.46261) (−2.19507)** (1.09490) (−2.65659)*** (−0.44851) (−2.64926)*** (−0.05059) (0.94318)

31–36 −0.002124 0.000472 −0.003218 0.000881 −0.002677 0.000508 −0.002730 0.000726 0.005082(−3.10526)*** (0.62418) (−2.51890)** (1.20235) (−3.06938)*** (0.77373) (−3.10526)*** (1.08992) (1.06106)

7–36 −0.004909 −0.001132 −0.014530 0.005155 −0.009583 0.002061 −0.009702 0.001954 −0.016902(−1.59498) (−0.45290) (−2.09669)** (1.87631)* (−2.12353)** (0.82952) (−1.94285)* (0.83440) (−1.22877)

13–36 −0.004907 −0.002089 −0.012384 0.003202 −0.008534 0.000321 −0.008753 0.000526 −0.012208(−1.54690) (−0.88682) (−1.97569)** (1.60775) (−2.00866)** (0.15152) (−1.84662)* (0.26995) (−1.26254)




developed markets. In contrast, for the emerging markets, the results inTable 6are weaklyconsistent with continuing overreactions, even in the post-December 1987 period, as theinitial momentum profits are positive (albeit insignificant as explained earlier) while profitsin months 7–12 and 13–18 are negative and significant.

4. Conclusion

This study investigates linkages between value/glamour and momentum strategies ininternational markets, and extendsJegadeesh and Titman (2001)-type tests, which attemptto distinguish between competing explanations of the momentum phenomenon, to interna-tional market indices. To accomplish the former, we examine momentum and contrarianstrategies separately in the Morgan Stanley Capital International (MSCI) growth, value andblended indices. If market participants underestimate short-run earnings growth for pastwinner stocks, and if growth stocks are more sensitive to earnings surprises than valuestocks, we would intuitively expect that momentum strategies would work better within thegrowth indices. This is precisely what our results show. Specifically, for our full sampleperiod inTables 1 and 2, we find that momentum profits are much stronger in the growth in-dices, particularly for 6- and 9-month formation/holding periods, than in the value indices,or in blended overall country equity indices. When we implement momentum strategieswith both the growth and value indices simultaneously, the average ex post performanceis similar to what is obtained using the blended indices. Thus, our 6- and 9-month resultstogether indicate, fairly conclusively, that intermediate horizon momentum strategies ininternational markets are concentrated in the growth indices, because adding value indicesinto the mix reduces the profitability of the momentum strategies.

An examination of the long-run post-formation performance of momentum strategiesbased on 6- and 9-month rankings (Table 3) shows that for all indices to which thesestrategies are applied (i.e. value, growth, blended and value + growth), negative profits areearned in every 6-month period that begins in the 13th post-formation month or later. Theseprofits are all significantly negative when aggregated over 13–36 months post-formation, andthey appear to be more negative (though not more significant) for the growth strategies. Theseresults confirm and extend those ofJegadeesh and Titman (2001), who report similar findingsfor momentum strategies based on individual stocks in the U.S. market. They indicatethat in international markets as well, cross-sectional differences in mean returns acrossindices cannot account for the near-term profitability of momentum strategies. However,these results do not strongly support the conjecture that growth stocks are more subject tomispricing than value stocks.

The results we report inTables 4–6for international market index momentum strate-gies by subperiod are not entirely conclusive in rendering a verdict on the durability ofthe momentum anomaly over time. Both the CK results inTable 4and the JT results inTable 5indicate, for the most part, that the profitability of momentum strategies is greater inthe December 1974–November 1987 period than in the December 1987–November 2000period. However, the CK results show momentum profits greatly diminishing in the lat-ter period, while the JT results inTable 5show only a small reduction, with 6-, 9- and12-month momentum strategies being significantly profitable in the post-December 1987


period. In contrast, what little evidence we are able to glean from our data regarding con-trarian strategies by subperiod indicates that these strategies work exceedingly well in thefirst subperiod, but completely fail to produce profits post-December 1987. These findingsagree withConrad and Kaul (1998), who show that contrarian profits in the U.S. market arevery dependent on the period examined.

Like Jegadeesh and Titman (2001), we find that in the most recent period we examine(at least for developed markets) there is no evidence of negative returns to momentumportfolios subsequent to the 13th month after portfolio formation. Thus, while pre-December1987 momentum profits are consistent with the hypothesis that investors overreact to newinformation and push up individual country stock indices to unreasonable levels (with theseovervaluations subsequently corrected), there is no evidence of investor overreaction in thepost-December 1987 period. This finding indicates that if the profitability of momentumstrategies persists in the latter period (as the JT findings inTable 5show), the explanationof what produces these momentum profits is not likely to be the same in both periods. Alltold, then, while there is some evidence that momentum strategies applied to internationalgrowth stock indices have worked in the past, and that the profitability of these strategiespersists to some degree, we would be somewhat hesitant in recommending these strategiesto investors today.

References

Asness, C., 1997. The interaction of value and momentum strategies. Financial Analysts Journal 53, 29–36.Asness, C., Liew, J.M., Stevens, R.L., 1997. Parallels between the cross-sectional predictability of stock and

country returns. Journal of Portfolio Management 23 (3, Spring), 79–87.Badrinath, S.G., Wahal, S., 2002. Momentum trading by institutions. Journal of Finance 57 (6), 2449–2478.Balvers, R., Wu, Y., Gilliland, E., 2000. Mean reversion across national stock markets and parametric contrarian

investment strategies. Journal of Finance 55 (2), 745–772.Balvers, R., Wu, Y., 2002. Momentum and mean reversion across national equity markets. Working Paper.Barberis, N., Shleifer, A., Vishny, R., 1998. A model of investor sentiment. Journal of Financial Economics 49,

307–343.Chan, K., Hameed, A., Tong, W., 2000. Profitability of momentum strategies in the international equity markets.

Journal of Financial and Quantitative Analysis 35, 153–172.Chan, L.K.C., Jegadeesh, N., Lakonishok, J., 1996. Momentum strategies. Journal of Finance 51, 1681–1713.Chan, L.K.C., Jegadeesh, N., Lakonishok, J., 1999. The profitability of momentum strategies. Financial Analysts

Journal 55 (6), 80–90.Chordia, T., Shivakumar, L., 2002. Momentum, business cycle, and time-varying expected returns. Journal of

Finance 57 (2), 985–1019.Chui, A., Titman, S., Wei, K.C.J., 2000. Momentum, ownership structure, and financial crises: an analysis of asian

stock markets. Working Paper, University of Texas at Austin.Conrad, J., Kaul, G., 1998. An anatomy of trading strategies. Review of Financial Studies 11, 489–519.Daniel, K., Hirshleifer, D., Subrahmanyam, A., 1998. Investor psychology and security market under- and over-

reactions. Journal of Finance 53, 1839–1886.DeBondt, W., Thaler, R., 1985. Does the stock market overreact? Journal of Finance 40, 793–808.Fama, E., French, K., 1993. Common risk factors in the returns on stocks and bonds. Journal of Financial Economics

33, 3–56.Fama, Eugene F., French, Kenneth R., 1996. Multifactor explanation of asset pricing anomalies. Journal of Finance

51 (1), 55–84.


Griffin, J.M., Ji, Xiuqing, Spencer Martin, J., 2003. Momentum investing and business cycle risk: evidence frompole to pole. Journal of Finance 58 (6), 2515–2547.

Grinblatt, M., Titman, S., Wermers, R., 1995. Momentum strategies, portfolio performance, and herding: a studyof mutual fund behavior. American Economic Review 85, 1088–1105.

Grinblatt, M., Moskowitz, T.J., 2003. The cross section of expected returns and its relation to past returns: newevidence. Working Paper, University of Chicago.

Hameed, A., Kusnadi, Y., 2002. Momentum strategies: evidence from Pacific Basin stock markets. Journal ofFinancial Research 25 (3, Fall), 383–397.

Hon, M.T., Tonks, I., 2003. Momentum in the UK stock market. Journal of Multinational Financial Management13 (1), 43–70.

Hong, H., Stein, J.C., 1999. A unified theory of underreaction, momentum trading, and overreaction in assetmarkets. Journal of Finance 54, 2143–2184.

Jegadeesh, N., Titman, S., 1993. Returns to buying winners and selling losers: implications for stock marketefficiency. Journal of Finance 48, 65–91.

Jegadeesh, N., Titman, S., 1995. Overreaction, delayed reaction, and contrarian profits. Review of Financial Studies8 (4, Winter), 973–993.

Jegadeesh, N., Titman, S., 2001. Profitability of momentum strategies: an evaluation of alternative explanations.Journal of Finance 56, 699–720.

Jegadeesh, N., Titman, S., 2002. Cross-sectional and time-series determinants of momentum returns. Review ofFinancial Studies 15 (1), 143–157.

Kang, J., Liu, M., Xiaoyan Ni, S., 2002. Contrarian and momentum strategies in the China stock market: 1993–2000.Pacific Basin Finance Journal 10 (3), 243–265.

Korajczyk, R.A., Sadka, R., 2004. Are momentum profits robust to trading costs? Journal of Finance 59 (3),139–182.

Kothari, S., Shanken, J., Sloan, R., 1995. Another look at the cross-section of expected stock returns. Journal ofFinance 50, 185–224.

Lakonishok, J., Shleifer, A., Vishny, R., 1992. The impact of institutional trading on stock prices. Journal ofFinancial Economics 32, 23–43.

Lakonishok, J., Shleifer, A., Vishny, R., 1994. Contrarian investment, extrapolation, and risk. Journal of Finance49, 1541–1578.

Lee, D.D., Chan, H., Faff, R.W., Kalev, P.S., 2003. Short-term contrarian investing—is it profitable? Yes and no.Journal of Multinational Financial Management 13 (4/5), 385–404.

Lehmann, B., 1990. Fads, martingales and market efficiency. Quarterly Journal of Economics 105, 1–28.Lesmond, D.A., Schill, M.J., Zhou, C., 2004. The illusory nature of momentum profits. Journal of Financial

Economics 71, 349–380.Lo, A., MacKinlay, C., 1990. When are contrarian profits due to stock market overreaction? Review of Financial

Studies 3, 175–206.Miller, Edward M., 1977. Risk, uncertainty, and divergence of opinion. Journal of Finance 32 (4), 1151–1168.Newey, W., West, K., 1987. A simple positive definite, heteroscedastocity and autocorrelation consistent covariance

matrix. Econometrica 55, 703–705.Richards, Anthony J., 1995. Comovements in national stock market returns: evidence of predictability, but not

cointegrated. Journal of Monetary Economics 36 (3), 631–654.Richards, Anthony J., 1997. Winner-loser reversals in national stock market indices: can they be explained? Journal

of Finance 52 (5), 2129–2144.Rouwenhorst, G., 1998. International momentum strategies. Journal of Finance 53, 267–284.Schiereck, D., De Bondt, W., Weber, M., 1999. Contrarian and momentum strategies in Germany. Financial

Analysts Journal 55 (6), 104–116.Skinner, D.J., Sloan, R.G., 2002. Earnings surprises, growth expectations, and stock returns or don’t let an earnings

torpedo sink your portfolio. Review of Accounting Studies 7 (2/3), 289–312.Van Dijk, R., Huibers, F., 2002. European price momentum and analysts behavior. Financial Analysts Journal 58

(2), 96–105.

momentum and contrarian strategies in international stock markets: further evidence

Documents