Momentum

Narasimhan Jegadeesh1 and Sheridan Titman2'3

1Goizueta Business School, Emory University, Atlanta, Georgia 30322;
email: Narasimhan_Jegadeesh@bus.emory.edu
2Finance Department, University of Texas, Austin, Texas 78712-1179;
email: titman@mail.utexas.edu

3National Bureau of Economic Research, Cambridge, Massachusetts 02138

Annu. Rev. Financ. Econ. 2011. 3:493-509

Keywords
First published online as a Review in Advance on earnings momentum, price momentum, time-varying momentum
August 17, 2011

The Annual Review of Financial Economics is Abstract

online at financial.annualreviews.org
There is substantial evidence that indicates that stocks that perform
This article's doi:
the best (worst) over a three- to 12-month period tend to continue
10.11 46/annure v-financial- 102710-144850
to perform well (poorly) over the subsequent three to 12 months.
Copyright © 201 1 by Annual Reviews. Until recently, trading strategies that exploit this phenomenon
All rights reserved
were consistently profitable in the United States and in most devel-
JEL: G14 oped markets. Similarly, stocks with high earnings momentum
1941-1367/1 1/1205-0493$20.00 outperform stocks with low earnings momentum. This article
reviews the momentum literature and discusses some of the expla-
nations for this phenomenon.

493

This content downloaded from

5.77.207.100 on Sun, 16 Mar 2025 18:57:53 UTC
All use subject to https://about.jstor.org/terms
 A growing body of literature documents evidence of stock return predictability based on a
variety of firm-specific variables. Among these anomalies, the price momentum effect is
probably the most difficult to explain within the context of the traditional risk-based
asset pricing paradigm. Jegadeesh and Titman (JT) document that U.S. stocks that per-
form the best (worst) over a three- to 12-month period tend to continue to perform well
(poorly) over the subsequent three to 12 months (Jegadeesh 8c Titman 1993). In a follow-
up study, JT show that momentum strategies remained profitable in the nineties
(Jegadeesh & Titman 2001); a period subsequent to the sample period in Jegadeesh &
Titman (1993).
As shown in Figure 1, the returns of a zero cost portfolio that consists of a long
position in past winners and a short position in past losers made money in every five-year
period starting in 1965 to 2004. In our opinion, the magnitude and persistence of these
returns are too strong to be explained by risk, so the focus of this review is on the
literature that provides behavioral explanations for this phenomenon. As we discuss
below, this literature considers both the time-series and cross-sectional determinants of
momentum profits.
Figure 1 also reveals that momentum profits in the five-year period starting in 2004
were negative. As we later discuss, the negative returns in this last period were driven
mainly by extremely negative returns in 2009. We discuss the 2009 performance in detail
in Section 9, where we examine the extent to which these returns can be explained by
variables that were previously introduced in the time-series momentum literature.

Figure 1
This figure presents the rolling five-year cumulative returns for a momentum strategy that buys winners and sells losers based on
returns in month t-7 through t- 2 and holds the portfolio for six months.

494 Jegadeesh • Titman

This content downloaded from

5.77.207.100 on Sun, 16 Mar 2025 18:57:53 UTC
All use subject to https://about.jstor.org/terms
1. THE MOMENTUM EVIDENCE

If stock prices either overreact or underreact to information, then profitable trading stra
egies that select stocks based on their past returns will exist. In an influential pape
DeBondt &: Thaler (1985) document that past losers over three- to five-year periods
outperform past winners over the subsequent three to five years. Jegadeesh (1990) an
Lehmann (1990) find that losers over the past one week to one month outperform winner
over the next one week to one month. These studies of very long-term and very short-ter
returns find profitable contrarian strategies and generally led to the conclusion that stoc
prices overreact to information.1
In contrast to these studies, JT focus on the performance of trading strategies with
formation and holding periods between three and 12 months (Jegadeesh & Titman 199
Their strategy selects stocks on the basis of returns over the past J months and holds the
for K months. This /-month/K-month strategy is constructed as follows: At the beginni
of each month ř, securities are ranked in ascending order on the basis of their returns in t
past } months. On the basis of these rankings, JT form 10 equally weighted decile portfo
lios. The portfolio with the highest return is called the winners decile and the portfolio wi
the lowest return is called the losers decile. As they show, each of these strategies ear
positive returns. Moreover, when the strategies skip a week between the portfolio forma
tion period and holding period to avoid the short-term reversals documented in Jegadeesh
(1990) and Lehmann (1990), they generate higher and more significant returns.

1.1. Evidence Around the World

Momentum strategies are profitable in most major markets throughout the world.
Rouwenhorst (1998) replicates JT for 12 European countries and finds profits that are
very close to those in the United States. More recent papers by Griffin et al. (2003) and
Chui et al. (2010) examine momentum profits around the world and find that the
momentum strategy yields positive profits in most large markets, with notable exceptions
in Asia (e.g., Japan).

1.2. Seasonality
Momentum strategies exhibit a unique pattern of seasonality in January. Many of the well-
known strategies such as long-horizon and short-horizon return reversals, the size effect, and
the book-to-market effect are significantly stronger in January than in any other calendar
month. In contrast, JT find that the momentum strategy earns negative returns in January,
but earns significantly positive returns in every calendar month outside of January.

2. POTENTIAL SOURCES OF MOMENTUM PROFITS

A natural interpretation of momentum profits is that there is a delayed reaction

information. For example, if stock prices only react partially to good news, then bu

1 As pointed out in a recent paper by Asness et al. (2009), the profitability of momentum or trend-following strateg
exists in several markets. For example, Chan et al. (2000) show that a momentum strategy is profitable when ap
to portfolios of country stock indexes; Shen et al. (2007) show that momentum strategies generate profi
commodity futures markets; and Sweeney (1986), Taylor & Allen (1992) and Okunev & White (2003) show
momentum strategies work in currency markets. Although these papers illustrate the robustness of the mome
strategy, our review considers only the literature on equity markets.

www.annualreviews.org • Momentum 49 j

This content downloaded from

5.77.207.100 on Sun, 16 Mar 2025 18:57:53 UTC
All use subject to https://about.jstor.org/terms
 stocks after the initial release of the news will exploit the delayed reaction and generate
profits. However, such underreaction is not the only possible source of momentum
profits, and JT show that there are several other factors that can also contribute to
momentum profits.
One other potential source of momentum profits is cross-sectional dispersion in
expected returns. Intuitively, given that realized returns contain a component related
to expected returns, securities that experience relatively high returns in one period can be
expected to have higher than average returns in the following period. Momentum strate-
gies can also benefit from positive serial correlation in factor returns. With positive serial
correlation, large factor realizations in one period will be followed by higher than average
factor realizations in the next period. The momentum strategy will tilt toward high beta
stocks following periods of large factor realizations, and hence it will benefit from the
higher expected future factor realizations.
To assess whether the existence of momentum profits implies market inefficiency, it is
important to identify which of these sources contribute to momentum profits. If the profits
are due to either the second or the third component, they may be attributed to compensa-
tion for bearing systematic risk and need not be an indication of market inefficiency.
However, if the superior performance of momentum were due to the first component, then
the results would suggest market inefficiency.
To examine whether cross-sectional differences in risk explain momentum profits,
several studies examine risk-adjusted returns under specific asset pricing models. In
Jegadeesh & Titman (1993), JT adjust for risk using the capital asset pricing model
(CAPM) benchmark, and in Jegadeesh & Titman (2001), they adjust for risk using
the Fama-Franch three-factor model benchmark, as do Fama & French (1996) and
Grundy & Martin (2001). In each of these cases, the alphas of the momentum strategy
are significantly positive, suggesting that cross-sectional differences in risk do not explain
momentum profits.
If the serial covariance of factor related returns were to contribute to momentum
profits, then the factor realizations should be positively serially correlated. JT examine this
implication in the context of a single-factor model and find that the serial covariance of
six-month returns of the equally weighted index is negative (-0.0028), indicating that the
serial correlation of factor returns is unlikely to have a positive effect on momentum
profits.
Momentum profits can also potentially arise if stock prices react to common factors
with some delay. Intuitively, if stock prices react with a delay to common information,
investors will be able to anticipate future price movements based on current factor realiza-
tions and devise profitable trading strategies. JT show that in some situations such delayed
reactions will result in profitable contrarian strategies, but in other situations it will result
in profitable momentum strategies (Jegadeesh & Titman 1995).
The contribution of this lead-lag effect to momentum profits depends on the relation
between contemporaneous betas and lagged betas. Specifically, lead-lag contributes to
momentum if firms with large contemporaneous betas also tend to exhibit large lagged
betas. For example, one can imagine small stocks that have high betas, but actually
underreact to market returns, and hence have large lagged betas as well. Therefore, during
market increases, these firms will tend to outperform, and because of their lagged beta,
they will outperform in the following period as well. Put somewhat differently, the con-
temporaneous betas are less dispersed than they should be given fundamentals, causing

49<> Jegadeesh • Titman

This content downloaded from

5.77.207.100 on Sun, 16 Mar 2025 18:57:53 UTC
All use subject to https://about.jstor.org/terms
stock prices to move together too closely with one another. In other words, if the market
moves up, high beta stocks will increase more than low beta stocks, but not by as much as
they should. It is possible that delayed reactions of this nature may be due to the tendency
of investors to buy and sell stocks in baskets rather than individually. With such delayed
reactions, a momentum strategy will buy high beta stocks following a market increase, and
will profit from the delayed response in the following period.
If lead-lag effects do contribute to momentum profits then we expect the magnitude of
momentum profits to depend on the magnitude of past market returns. To investigate
the importance of this source, JT regress momentum profits on the squared return of the
value-weighted market during the six-month formation period. Over the 1965 to 1989
sample period the coefficient on the squared market return is negative, suggesting that the
lead-lag effect does not contribute to momentum profits in this time period.
In summary, the evidence suggests that momentum profits arise because of a delayed
reaction to firm-specific information. One interpretation of this evidence is that investors
tend to underreact to firm-specific information. An alternative interpretation is that the
delayed reaction is actually an overreaction, by investors who either react to the informa-
tion with a delay or who like to chase past winners. As we discuss below, the latter
interpretation suggests that the momentum portfolio excess returns are eventually
reversed.

3. INDUSTRY MOMENTUM

The results discussed in the last section clearly indicate that the common factor in a s
factor model cannot explain momentum profits. JT therefore conclude that the momen
profits are due to the nonmarket component of returns. Although the nonmarket com
nent is the idiosyncratic component of returns in a single-factor model, it is possible
momentum is related to other factors in a more general multifactor setting. For exam
if we introduce industry factors, serial covariance in industry returns, rather than the
covariance of firm-specific components of returns, may account for the momentum pr
Moskowitz &c Grinblatt (1999) evaluate industry momentum. They form value
weighted industry portfolios and rank stocks based on past industry returns. They
that high momentum industries outperform low momentum industries in the six mon
after portfolio formation. To assess the extent to which the industry return contribu
momentum profits, they examine the performance of a random industry strategy. Sp
cally, they replace each firm in the winner and loser industries with other firms tha
not in these industries, but have the same ranking period returns as the firms that
replace. The random industry portfolios have similar levels of past returns as the w
and loser industry portfolio. However, Moskowitz & Grinblatt find that the profit fo
momentum strategy with the random industry earns close to zero returns. On the ba
this test they conclude that the momentum strategy profits from industry momentum
not from momentum in the firm-specific component of returns.
Grundy & Martin (2001) reexamine the extent to which industry momentum
tributes to momentum profits. Grundy & Martin find that for a six-month ran
period and a contiguous six-month holding period, the actual industry strategy ea
significantly positive return of .78%, whereas the simulated industry strategy earn
returns. Additionally, Grundy & Martin consider a strategy that skips a month bet
the ranking period and holding period to avoid the potential biases due to bi

www.annualreviews.org • Momentum 497

This content downloaded from

5.77.207.100 on Sun, 16 Mar 2025 18:57:53 UTC
All use subject to https://about.jstor.org/terms
 spreads. When industry portfolios are formed in this manner, a momentum strategy does
not yield significant profits either for the actual industry strategy or for the simulated
industry strategy. In comparison, the momentum strategy with individual stocks earns a
significantly positive profit of .79% during the 1966 to 1995 period.
As JT show, the momentum strategy with individual stocks is more profitable when
the ranking period and holding period are not contiguous than when they are contigu-
ous. When the holding period and the ranking period are contiguous, the profits to the
momentum strategy are attenuated by the negative serial correlation in returns induced
by the bid-ask spreads, and by the short-horizon return reversals. In contrast, industry
momentum profits entirely disappear for the six-month ranking period when the ranking
period and the holding period are not contiguous. The industry momentum seems to
benefit from the positive first-order serial correlation in portfolio returns, whereas the
individual stock momentum is reduced by short-horizon return reversals.

4. BEHAVIORAL EXPLANATIONS

As we mentioned in the introduction, it is very difficult to explain the observed momentum

profits with a risk-based model. Therefore, researchers have turned to behavioral model
to explain this phenomenon. Most of the models assume that the momentum effect is
caused by the serial correlation of individual stock returns, which, as we discussed above
appears to be consistent with the evidence. However, they differ as to whether the seria
correlation is caused by underreaction or delayed overreaction. If the serial correlation i
caused by underreaction, then we expect to see the positive abnormal returns during the
holding period followed by normal returns in the subsequent period. However, if the
abnormal returns are caused by delayed overreaction, then we expect that the abnormal
momentum returns in the holding period will be followed by negative returns given that
the delayed overreaction must be subsequently reversed. Hence, these behavioral model
motivate tests of the long-term profitability of momentum strategies that we discuss below
In addition, the behavioral models have implications about the cross-sectional determi-
nants of momentum, which are also discussed below.
Barberis et al. (1998) discuss how a conservatism bias might lead investors to under-
react to information, giving rise to momentum profits. The conservatism bias, identified in
experiments by Edwards (1968), suggests that investors tend to underweight new informa-
tion when they update their priors. If investors act in this way, prices will slowly adjust to
information, but once the information is fully incorporated in prices there is no further
predictability about stock returns.
More recent explanations that are consistent with underreaction include what is
referred to as the disposition effect, which suggests that loss-averse investors tend to hold
on to their past losers and sell their past winners, and the tendency to anchor on past
prices. Grinblatt & Han (2005) provide evidence that is consistent with the disposition
effect, and George & Hwang (2004), who show that stocks perform well after hitting their
52-week highs, provide evidence of anchoring.
The idea of delayed overreaction was originally introduced by Delong et al. (1990),
who show that positive feedback trading strategies (investment strategies that buy past
winners and sell past losers) cause market prices to deviate from fundamental values. To
a large extent, the subsequent literature presents behavioral models that formalize how
various behavioral biases can lead investors to follow such positive feedback strategies.

498 Jegadeesh • Titman

This content downloaded from

5.77.207.100 on Sun, 16 Mar 2025 18:57:53 UTC
All use subject to https://about.jstor.org/terms
For example, Barberis et al. (1998) hypothesize that investors identify patterns based on
what Tversky & Kahneman (1974) refer to as a "representative heuristic," which is the
tendency of individuals to identify "an uncertain event, or a sample, by the degree to
which it is similar to the parent population." In the context of stock prices, Barberis
et al. argue that the representative heuristic may lead investors to mistakenly conclude
that firms realizing consistent extraordinary earnings growths will continue to experi-
ence similar extraordinary growth in the future. They argue that although the conserva-
tism bias in isolation leads to underreaction, this behavioral tendency in conjunction
with the representative heuristic can lead to prices overshooting their fundamental value
and, eventually, long-horizon negative returns for stocks with consistently high returns
in the past.2
Daniel et al. (1998) and Hong Stein (1999) propose alternative models that are
also consistent with short-term momentum and long-term reversals. Daniel et al. argue
that the behavior of informed traders can be characterized by a self-attribution bias.
In their model, investors observe positive signals about a set of stocks, some of which
perform well after the signal is received. Because of their cognitive biases, the informed
traders attribute the performance of ex-post winners to their stock selection skills and
that of the ex-post losers to bad luck. As a result, these investors become overconfident
about their ability to pick winners and thereby overestimate the precision of their signals
for these stocks. Based on their increased confidence in their signals, they push up the
prices of the winners above their fundamental values. The delayed overreaction in this
model leads to momentum profits that are eventually reversed as prices revert to their
fundamentals.

Hong Sc Stein (1999) do not directly appeal to any behavioral biases on the part of
investors but they consider two groups of investors who trade based on different sets of
information. The informed investors or the news watchers in their model obtain signals
about future cash flows but ignore information in the past history of prices. The other
investors in their model trade based on a limited history of prices and, in addition, do not
observe fundamental information. The information obtained by the informed investors is
transmitted with a delay and hence is only partially incorporated in the prices when first
revealed to the market. This part of the model contributes to underreaction, resulting in
momentum profits. The technical traders extrapolate based on past prices and tend to push
prices of past winners above their fundamental values. Return reversals obtain when prices
eventually revert to their fundamentals. Both groups of investors in this model act ratio-
nally in updating their expectations conditional on their information sets but return pre-
dictability obtains due to the fact that each group uses only partial information in updating
their expectations.

5. LONG-HORIZON RETURNS OF MOMENTUM PORTFOLIOS

As we discussed earlier, the momentum-effect is consistent with both investors

underreacting to information, as well as with investors overreacting to past informatio

2The time horizon over which various biases come into play in the Barberis et al. model (and in other behavior
models) is unspecified. One could argue that the six-month ranking period used in Jegadeesh & Titman (1993) and
others may not be long enough for delayed overreaction due to the representative heuristic effect. In such an event we
would only observe underreaction due to the conservatism bias.

www.annualreviews.org • Momentum 499

This content downloaded from

5.77.207.100 on Sun, 16 Mar 2025 18:57:53 UTC
All use subject to https://about.jstor.org/terms
 with a delay, perhaps due to positive feedback trading. The positive feedback effect, which
is consistent with some of the behavioral models described in Section 3, implies that the
momentum portfolio should generate negative returns in the periods following the holding
periods considered in previous sections.
JT examine the long-horizon performance of momentum strategies to examine whether
the evidence suggests returns reversals in the postholding periods (Jegadeesh & Titman
1993, 2001). Over the 1965 to 1998 sample period, the results reveal a dramatic reversal
of returns in the second through fifth years. Cumulative momentum profit increases mono-
tonically until it reaches 12.17% at the end of Month 12. From Month 13 to Month 60
the momentum profits are on average negative. By the end of Month 60 the cumulative
momentum profit declines to -.44%.
The robustness of long-horizon return reversals can be evaluated by examining the
performance of momentum portfolios in two separate time periods, the 1965 to 1981
and 1982 to 1998 subperiods. In addition to being the halfway point, 1981 represents
somewhat of a break point for the Fama and French factor returns. The Fama-French size
(SMB) and value (HML) factors have higher returns in the pre-1981 period (the monthly
returns of the SMB and HML factors average .53% and .48%, respectively) than in the
post- 1981 period (the monthly returns of the SMB and HML factors average -.18% and
.33%, respectively).
The evidence indicates that the momentum strategy is significantly profitable, and quite
similar in both subperiods, in the first 12 months following the formation date. The returns
in the postholding periods, however, are quite different in the two subperiods. In the 1965
to 1981 subperiod, the cumulative momentum profit declines from 12.10% at the end of
Month 12 to 5.25% at the end of Month 36 and then declines further to -6.29% at the end
of Month 60. Hence, the evidence in this subperiod supports the behavioral models that
suggest that positive feedback traders generate momentum. In the 1982 to 1998 subperiod
the cumulative profit decreases insignificantly from 12.24% at the end of Month 12 to
6.68% at the end of Month 36 and then stays at approximately the same level for the next
24 months. Hence, the evidence in the second subperiod does not support the behavioral
models.

6. CROSS-SECTIONAL DETERMINANTS OF MOMENTUM

The insights provided by the behavioral models also suggest that stocks with differen
characteristics should exhibit different degrees of momentum. For example, given that
the momentum effect is due to inefficient stock price reaction to firm-specific information
it is likely to be related to various proxies for the quality and type of information tha
is generated about the firm, the relative amounts of information disclosed publicly and
generated privately, and to the cost associated with arbitraging away the momentum
profits.
Hong et al. (2000) find that even after controlling for size, firms that are followed
by fewer stock analysts exhibit greater momentum. This finding is consistent with the
Hong & Stein (1999) prediction that slow dissemination of public information increases
momentum profits. Because there is less public information about stocks with low analyst
coverage, information about the companies may be incorporated into their stock prices
more slowly. In addition, given that there is less public information available about these

500 Jegadeesh • Titman

This content downloaded from

5.77.207.100 on Sun, 16 Mar 2025 18:57:53 UTC
All use subject to https://about.jstor.org/terms
stocks, one might expect relatively more private information to be produced, which Daniel
et al. (1998) suggest will increase price momentum.
Daniel & Titman (1999) find that momentum profits are significantly higher when
the strategy is implemented on growth (low book-to-market) stocks rather than value
(high book-to-market) stocks. They suggest that this result may be due to the fact that it is
harder to evaluate growth stocks than to evaluate value stocks. Psychologists report that
individuals tend to be more overconfident about their ability to do more ambiguous tasks.
So, the overconfidence hypothesis suggests that momentum is likely to be greater for
growth stocks.
Zhang (2006) examines this issue more broadly and finds that higher information
uncertainty, as measured by dispersion in analyst forecasts,3 return volatility, and cash
flow volatility predict higher momentum profits. Sagi & Seasholes (2007) empirically
document similar results: Momentum is stronger in stocks with higher revenue volatility
and lower costs of goods sold, and these results suggest that momentum profits arise from
the fact that firms that performed well in the recent past have new growth options to
exploit.
Lee & Swaminathan (2000) examine the relation between momentum profits and
turnover, and find that momentum is higher for stocks with greater turnover. This finding
is somewhat surprising when viewed from the transaction cost perspective. Stocks with
higher turnover can be traded more easily, and generally, there is more public information
generated for high turnover stocks than for low turnover stocks. One potential explanation
for their findings may be that there are larger differences in opinion about higher turnover,
and larger differences of opinion may arise from difficulties in evaluating the fundamental
values of these stocks. Hence, the Daniel & Titman explanation for why growth stocks
exhibit greater momentum may also apply to high turnover stocks. Another explanation is
that turnover is related to the amount of attention that a stock attracts. Hence, high
turnover stocks may be more exposed to positive feedback trading strategies proposed by
Delong et al. (1990).
Avramov et al. (2007) find that momentum is profitable only among firms with low
credit ratings. Extreme winner and loser portfolios are comprised of high credit risk stocks.
For stocks with a credit rating between AAA and BB, momentum profits are insignifi-
cant. These stocks account for 96.6% of the market capitalization and 78.8% of the total
number of rated firms. Several other papers, however, find that the momentum effect is far
more pervasive. For example, JT find momentum effect for small, medium, and large
stocks (Jegadeesh & Titman 1993). Also, Fama & French (2008) find that "the relation
between momentum (the center-stage anomaly of recent years) and average returns is
similar for small and big stocks."
Finally, Chui et al. (2010) examine the determinants of the profitability of momentum
strategies across countries. They hypothesize that cultural differences may be related to
behavioral biases, and hence, cross-country cultural differences may explain cross-country
differences in the profitability of momentum strategies. To measure cross-country differ-
ences in culture they use the individualism index developed by Hofstede (2001), which they
argue is related to overconfidence and self-attribution biases, and find that it is positively
correlated with momentum profits.

Verardo (2009) finds similar results with dispersion in analyst forecasts but interprets her evidence to signify
differences of opinion.

www.annualreviews.org • Momentum joi

This content downloaded from

5.77.207.100 on Sun, 16 Mar 2025 18:57:53 UTC
All use subject to https://about.jstor.org/terms
 7. TIME-SERIES DETERMINANTS OF MOMENTUM PROFITS

To test whether momentum profits are dependent on the state of the economy, severa
conditioning variables have been proposed to predict time-series variations in momentum
profits. These studies estimate monthly time-series regressions of the momentum profits
(MOMti) on a conditioning state variable (STATE) of the following form.

MOMttt = To + ßi*STATEt_i + et.

Chordia & Shivakumar (2002), using standard macro variables, find that the momentum
strategy is only profitable during times of economic expansion. However, Griffin et al
(2003) find that macroeconomic variables cannot predict momentum profits in interna
tional markets. Cooper et al. (2004) also find that macroeconomic multifactor models tha
Chordia & Shivakumar use are not robust to standard price screens and skip-a-month
returns. However, they find that the lagged three-year market return does predict momen
tum profits. Specifically, the momentum strategy generates significantly positive returns
(0.93% average monthly returns) following positive market returns, but insignificantly
negative returns (-0.37% average monthly returns) following negative market returns.
Stivers & Sun (2010) find that higher return dispersion predicts lower future momen
tum profits. Return dispersion is measured as the standard deviation of 100 size and book-
to-market monthly portfolio returns over the prior three months. They suggest that retur
dispersion may act as a state variable that has information about subsequent market
volatility. Their regression results indicate that the inclusion of return dispersion subsumes
the predictive power of the market state in Cooper et al. (2004) and macro factors in
Chordia & Shivakumar (2002).
Wang & Xu (2010) find that recent market volatility in combination with market state
(Cooper et al. 2004) predicts momentum profits. Momentum profits tend to be higher
following periods of low market volatility. In particular, the momentum strategy generates
especially low average monthly returns (-3.01%, ř-stat -1.94) during down market/hig
volatility states.
Antoniou et al. (2010) find that investor sentiment predicts momentum profits. Investo
sentiment is estimated by taking the residual of a regression of the Conference Board
Consumer Confidence Index on a set of macroeconomic variables following the approach
used in Baker & Wurgler (2006, 2007). During optimistic states, momentum strategie
generate significant average monthly profits of 1.64%, but during pessimistic states yield
insignificant average monthly profits of 0.56%. Their results remain with the inclusion of
market state variables. Momentum profits are particularly high in up/optimistic state
generating 1.8% average monthly profits but only averaging 0.8% for up/pessimisti
states. Unlike the previous studies, Antoniou et al. (2010) explicitly test the subsequent
long-run reversal effect to momentum strategies and find that momentum profits revers
only after optimist periods.

8. EARNINGS MOMENTUM

The results so far have focused on the profitability of momentum strategies based on
past returns. Naturally, returns are driven by changes in underlying fundamentals. Stock
returns tend to be high, for example, when earnings growth exceeds expectations or
when consensus forecasts of future earnings are revised upward. An extensive literature

502 Jegadeesh • Titman

This content downloaded from

5.77.207.100 on Sun, 16 Mar 2025 18:57:53 UTC
All use subject to https://about.jstor.org/terms
examines return predictability based on momentum in past earnings and momentum in
expectations of future earnings as proxied by revisions in analyst forecasts. This section
reviews the evidence from the earnings momentum literature and presents the interaction
between earnings momentum and return momentum.
A partial list of papers that investigate the relation between past earnings momentum
and futures returns includes Jones & Litzenberger (1970), Latane & Jones (1979), Foster
et al. (1984), Bernard &c Thomas (1989), and Chan et al. (1996). These papers typically
measure earnings momentum using a measure of standardized unexpected earnings (SUE).
SUE is defined as: SUE = earnings - Expected quarterly earnings
Standard deviation or quarterly earnings
These papers use variations of time-series models to determine earnings expectations.
Typically, the papers either assume that quarterly earnings follow a seasonal random walk
with drift or use changes in analyst earnings forecast to measure earnings momentum.
A study by Givoly &c Lakonishok (1979), which examines a sample of 67 firms from
1967 to 1974 using earnings forecast data from Standard and Poors Earnings forecaster,
finds that stocks with upward revisions outperform stocks with downward revisions by
approximately 5%. Stickel (1991) finds similar results using the Zacks Investment
Research database over the 1981 to 1984 sample period. Chan et al. (1996) use IBES, the
earnings forecast database, over the 1977 to 1993 sample period and find that Up revision
portfolios earn 7.7% higher return than the Down revision portfolios over the six months
after portfolio formation.
The collective evidence in the literature indicates that the analyst forecast revision
strategy is remarkably robust. The profitability of this strategy is not sensitive to the
specific definition of forecast revisions, nor is it sensitive to the data source for analysts'
forecasts. Also, both the SUE strategy and the forecast revision strategy persisted for a
fairly long period of time after the initial publication of the evidence.

8.1. Relation Between Earnings and Return Momentum Strategies

Chan et al. (1996) present a detailed analysis of the interactions among various momen-
tum strategies and this subsection closely follows that paper. Not surprisingly, the price
momentum and earnings momentum measures are positively correlated with one
another.

8.2. Two-Way Analysis

Earnings and return momentum strategies are individually useful for predicting stock
returns six to 12 months in the future. Because these variables tend to move together, it is
possible that the findings may reflect not separate effects but different manifestations of a
single effect.
Chan et al. (1996, 2000) examine this issue with predictability tests based on two-way
classifications. At the beginning of each month, they sort the stocks in their sample on the
basis of their past six-month returns and assign them to one of three equal-sized portfolios.
Independently, they sort stocks into three equal-sized portfolios on the basis of SUE and
analyst forecast revisions. Each stock, therefore, falls into one of nine portfolios for each
two-way sort.
Their evidence indicates that past six-month returns and SUE each independently
predict returns in the subsequent period. In particular, the two-way sort generated large

www.annualreviews.org • Momentum 503

This content downloaded from

5.77.207.100 on Sun, 16 Mar 2025 18:57:53 UTC
All use subject to https://about.jstor.org/terms
 differences in returns between stocks that were jointly ranked highest and stocks
jointly ranked lowest. For example, the highest ranked portfolio outperformed the
lowest ranked portfolio by 8.1% in the first six months and 11.5% in the first year.
Chan et al. also find similar results using two-way analysis based on price momentum
and earnings forecast revisions, and based on price momentum and past earnings
announcement window returns. Overall, none of the momentum variables considered
here subsumes any of the others. Instead, they each exploit underreaction to different
pieces of information.

9. RECENT PERFORMANCE AND DETERMINANTS OF

MOMENTUM PROFITS

This section examines the performance of the momentum strategy over the past 20 yea
(1990-2009). This period starts after the end of the sample period in Jegadeesh & Titm
(1993), and hence provides a perspective on the performance of the strategy after t
original published period. As in Jegadeesh & Titman (2001), we find evidence that
momentum effect continued after the publication of the earlier paper, but has diminis
over time and did extremely poorly in the most recent period.
The particular strategy that we examine is the six-month ranking period/six-mon
holding period momentum strategy where we skip a month between the ranking pe
and the holding period to avoid the effect of one-month return reversals that Jegad
(1990) reports. We follow the approach in Jegadeesh & Titman (1993), and in each mo
construct six sets of equally weighed extreme decile portfolios based on returns in
period t-7 to t- 2, t- 8 to t- 3, etc. The winner (loser) portfolio return in month t is the ave
return of the six winner (loser) portfolios based on these ranking periods. The moment
strategy return is the difference between the winner and loser portfolio returns. Our sa
excludes all stocks that would be ranked among the smallest New York Stock Excha
market cap decile and stocks priced less than $5 at the end of the month prior to t
holding period.
Table 1, which presents the annual momentum profits from 1990 to 2009, reveals t
the momentum strategy is profitable in 16 out of the 20 years. The average annual prof
13.5% with a ř-statistic of 2.9. Although the profits are significant over the entire per
the strategy experiences a severe loss of 36.5% in 2009. The poor performance in 20
in particular, and the overall variation in momentum profits over the more recent per
in general, offer an opportunity to examine the extent to which the various source
momentum suggested in the literature explain variation in the profitability of momen
strategies.
For example, the third term of the decomposition described earlier suggests that
momentum is expected to generate negative returns in periods where the market returns
exhibit negative serial correlation. This is because winners tend to have low betas and
losers tend to have high betas following periods when the market does especially poorly.
Hence, if the negative market returns are followed by very strong positive returns, the
momentum portfolio will do poorly. As discussed in Jegadeesh & Titman (1993), this is
exactly what happened in the 1930s, which was the only decade in which the momentum
portfolio exhibited negative returns.
The performance of the market in 2009 is somewhat similar to what was observed
in 1933. The strong market recovery in 2009 followed severe market declines in late

504 Jegadeesh • Titman

This content downloaded from

5.77.207.100 on Sun, 16 Mar 2025 18:57:53 UTC
All use subject to https://about.jstor.org/terms
 Table 1 Momentum strategy - annual returns3

Raw return Alpha Beta

1990 21.61 19.96 -0.01

1991 22.44 12.15 0.36

1992 1.42 -1.06 0.53

1993 22.17 11.56 1.13

1994 -0.32 0.96 0.24

1995 15.08 6.93 0.31

1996 4.16 3.46 0.09

1997 9.05 3.18 0.27

1998 41.50 34.94 0.09

1999 67.26 36.25 1.02

2000 36.01 69.70 1.40

2001 2.56 -8.63 -1.27

2002 16.93 -6.94 -1.15

2003 -3.50 12.27 -0.54

2004 3.63 -1.54 0.50

2005 16.18 13.90 0.36

2006 5.44 -7.57 1.29

2007 25.45 23.31 -0.02

2008 -0.32 -0.14 -0.06

2009 -36.50 -18.84 -0.79

Average 13.51 10.19

ř-statistics (2.90) (2.30)

aThis table presents the annual raw returns and annualized CAPM
momentum strategy that buys winners and sells losers based on
through t- 2 and holds the portfolio for six months. We estimate CAPM
the market model within each calendar year. The sample period is
December 2009.

2008 and early 2009, which is similar to the strong market recovery in 1933 follow-
ing market declines during the Great Depression. As JT discuss, winners tend to be
low beta stocks and losers tend to be high beta stocks following market declines, and
hence any sharp market reversals will result in significant losses for momentum
strategies. Indeed, although the beta of the momentum portfolio is close to zero on
average over the entire 1990 to 2009 sample period, the beta in 2009 is -.79. When
we account for the negative beta in 2009, the momentum portfolio return is -1.56%
per month, compared with a raw monthly return of -3.4% per month. Therefore,

www.annualreviews.org • Momentum joj

This content downloaded from

5.77.207.100 on Sun, 16 Mar 2025 18:57:53 UTC
All use subject to https://about.jstor.org/terms
 more than half the losses in 2009 are explained by the beta of the momentum
portfolio.4
Next, we examine the extent to which the variables in the literature that predict time-
series variations in momentum profits anticipated the sharp loss in 2009. We consider the
past three-year returns suggested by Cooper et al. (2004); the negative momentum profits
in 2009 are consistent with this evidence given that 2009 was preceded by strongly nega-
tive market returns in the previous three years. We also consider RD, the cross-sectional
dispersion in stock returns signal of Stivers &c Sun (2010). Average RD in 2009 was 4.84%,
which exceeds 3.20%, the measure in the rest of the sample period. The result that RD in
2009 was bigger than that in the rest of the sample period is directionally consistent with
the low momentum profits in 2009.
We fit the following multivariate regression over the January 1990 to December 2009
period to examine the out-of-sample performance of these signals and the extent to which
they anticipated the poor performance of momentum strategies in 2009:

MOM, = 1.12 - .42 X RD,_i + .90 x MktRetř_36,ř-i.

(-1.03) (2.21)

In this regression, MOMř is the return on the momentum

is Stivers & Sun's (2010) RD variable, and MktRetř_36,ř-i i
weighted index over the previous 36 months. We standardi
by subtracting the mean and dividing this difference by t
corresponding variables. The equation reports the parameter es
The regression estimates indicate that although the sign
the results in Stivers & Sun (2010), the slope coefficient is
The slope coefficient on MktRetř_36,ř_i, however, is statistica
20-year period as well. To examine the extent to which th
momentum profits in 2009, we computed adjusted momentum

adi-MOMt = MOM, - (-.42 x RD,_i + .90 x MktRe

The average adjusted momentum profit for 2009 was -1.5%

tude than the raw profit of -3.4%. Therefore, these signals an
explain a large part of the 2009 losses.
To examine the extent to which these variables, along with d
for the negative momentum portfolio returns in 2009, w
momentum profits against these independent variables. Spec
regression:

MOM^-^ = MOM, - ßt x MktRetf = .85 - .30 x RD^ + .76 x MktRetř_36,ř-i,

(-.85) (2.16)

where ßt is the momentum beta fit within each calendar year. The
we account for both CAPM beta and the other signals for 2009 is .
of the negative momentum returns in 2009 can in fact be explaine

4Daniel (2011) also examines the relation between portfolio betas and momentum pro
Similar to our findings, Daniel also reports that the beta effect provides a partial but
the negative momentum returns in 2009.

506 Jegadeesh • Titman

This content downloaded from

5.77.207.100 on Sun, 16 Mar 2025 18:57:53 UTC
All use subject to https://about.jstor.org/terms
 Of course, there is considerable estimation error in these regressions, and hence one
should not put too much weight on one observation. Nevertheless, the evidence indi-
cates that investors who use momentum signals should pay attention to the market
exposure of the portfolio and they should heed signals that are related to the strategy's
performance.

10. CONCLUSION

Underlying the efficient market hypothesis is the notion that if any predictable patte
exist in returns, investors will quickly act to exploit them, until the source of p
dictability is eliminated. However, this does not seem to be the case for either stock re
or earnings based momentum strategies. Both strategies have been well known and w
well publicized by at least the early 1990s, but both continued to generate excess pr
in the subsequent years.
We would argue that the momentum effect represents perhaps the strongest evid
against the efficient markets hypothesis. For this reason it has attracted substan
research, which documents more details about the anomaly, for example, the exten
which momentum profits are correlated with stock characteristics, as well as attemp
provide behavioral explanations for the phenomena. At this point, we have several in
esting facts to explain as well as possible theoretical explanations. However, finan
economists are far from reaching a consensus on what generates momentum pro
making this an interesting area for future research.

DISCLOSURE STATEMENT

The authors are not aware of any affiliations, memberships, funding, or financial hol
that might be perceived as affecting the objectivity of this review.

LITERATURE CITED

Antoniou C, Doukas JA, Subrahmanyam A. 2010. Investor sentiment and price momentu
Work. Pap., Cent. Empir. Res. Finance, Durham Univ./Old Dominion Univ./Judge Bus. Sch
Cambridge Univ.
Asness C, Moskowitz TJ, Pedersen LH. 2009. Value and momentum everywhere. Work. Pap., A
Cap. Manag./Grad. Sch. Bus., Univ. Chic./NBER/Stern. Sch. Bus., N. Y. Univ./CEPR
Avramov D, Chordia T, Jostova G, Philipov A. 2007. Momentum and credit rating. J. Fin
62:2503-20

Baker M, Wurgler J. 2006. Investor sentiment and the cross-section of stock returns. ]. Finan
61:1645-80

Baker M, Wurgler J. 2007. Investor sentiment in the stock market. J. Econ. Perspect. 21:129-52
Barberis N, Shleifer A, Vishny R. 1998. A model of investor sentiment. J. Financ. Econ. 49:307-43
Bernard VL, Thomas JK. 1989. Post-earnings-announcement drift: Delayed price response or risk
premium? /. Account. Res. 27:1-36
Chan LKC, Jegadeesh N, Lakonishok J. 1996. Momentum strategies. J. Finance 51:1681-713
Chan K, Hameed A, Tong W. 2000. Profitability of momentum strategies in the international equity
markets./. Financ. Quant. Anal. 35:153-72
Chui A, Titman S, Wei KCJ. 2010. Individualism and momentum around the world. J. Finance
65:361-92

www.annualreviews.org • Momentum 507

This content downloaded from

5.77.207.100 on Sun, 16 Mar 2025 18:57:53 UTC
All use subject to https://about.jstor.org/terms
 Chordia T, Shivakumar L. 2002. Momentum, business cycle and time- varying expected returns.
/. Finance 57:985-1019
Cooper M, Gutierrez R, Hameed A. 2004. Market states and momentum. J. Finance 59:1345-65
Daniel K, Hirshleifer D, Subrahmanyam A. 1998. Investor psychology and security market under and
overreactions. /. Finance 53:1839-86
Daniel K, Titman S. 1999. Market efficiency in an irrational world. Financ. Anal. J. 55:28-40
Daniel K. 2011. Momentum crashes. Work. Pap., Columbia Univ.
DeBondt WFM, Thaler RH. 1985. Does the stock market overreact? J. Finance 40:793-805
DeLong JB, Shleifer A, Summers LH, Waldmann RJ. 1990. Positive feedback investment strategies and
destabilizing rational speculation. J. Finance 45:379-95
Edwards W. 1968. Conservatism in human information processing. In Formal Representation of
Human Judgement , ed. B Kleimutz. New York: Wiley
Fama E, French K. 1996. Multifactor explanations of asset pricing anomalies. J. Financ. Econ. 51:55-84
Fama E, French K. 2008. Dissecting anomalies./. Finance 63:1653-77
Foster G, Olsen C, Shevlin T. 1984. Earnings releases, anomalies, and the behavior of security returns.
Account. Rev. 59:574-603

George T, Hwang C-Y. 2004. The 52-week high and momentum investing. J. Finance 59:2145-76
Givoly D, Lakonishok J. 1979. The information content of financial analysts' forecasts of earnings:
some evidence on semi-strong inefficiency. J. Account. Econ. 1:165-85
Griffin JM, Ji X, Martin JS. 2003. Momentum investing and business cycle risk: Evidence from pole to
pole./. Financ. 58:2515-47
Grinblatt M, Han B. 2005. Prospect theory, mental accounting and momentum. J. Financ. Econ.
78:311-39

Grundy BD, Martin SJ. 2001. Understanding the nature of risks and the sources of rewards to
momentum investing. Rev. Financ. Stud. 14:29-78
Hofstede G. 2001. Culture's Consequences : Comparing Values , Behaviors, Institutions , and Organi-
zations across Nations. Beverly Hills, CA: Sage Pubi. 2nd ed.
Hong H, Stein J. 1999. A unified theory of underreaction, momentum trading and overreaction in
asset markets. /. Finance 54:2143-84
Hong H, Lim T, Stein JC. 2000. Bad news travels slowly: size, analyst coverage, and the profitability of
momentum strategies ./. Finance 55:265-95
Jegadeesh N. 1990. Evidence of predictable behavior of security returns. J. Finance 45:881-98
Jegadeesh N, Titman S. 1993. Returns to buying winners and selling losers: implications for stock
market efficiency. J. Finance 48:65-91
Jegadeesh N, Titman S. 1995. Overreaction, delayed reaction and contrarian profits. Rev. Financ.
Stud. 8:973-93
Jegadeesh N, Titman S. 2001. Profitability of momentum strategies: an evaluation of alternative
explanations. J. Finance 56:699- 720
Jones CP, Litzenberger RH. 1970. Quarterly earnings reports and intermediate stock price trends.
J. Finance 25:143-48
Latane HA, Jones CP. 1979. Standardized unexpected earnings - 1971-1977./. Finance 34:717-24
Lee C, Swaminathan B. 2000. Price momentum and trading volume. J. Finance 55:1217-69
Lehmann B. 1990. Fads, martingales and market efficiency. Q. J. Econ. 105:1-28
Moskowitz TJ, Grinblatt M. 1999. Does industry explain momentum? /. Finance 54:1249-90
Okunev J, White D. 2003. Do momentum-based strategies still work in foreign currency markets?
J. Financ. Quant. Anal. 38:425-47
Rouwenhorst KG. 1998. International momentum strategies. J. Finance 53:267-84
Sagi JS, Seasholes MS. 2007. Firm-specific attributes and the cross-section of momentum. /. Financ.
Econ. 84:389-434

Shen Q, Szakmary AC, Sharma SC. 2007. An examination of momentum strategies in commodity
futures markets./. Futur. Mark. 27(3):227 -56

jo8 Jegadeesh • Titman

This content downloaded from

5.77.207.100 on Sun, 16 Mar 2025 18:57:53 UTC
All use subject to https://about.jstor.org/terms
Stickel S. 1991. Common stock returns surrounding earnings forecast revisions: more puzzling
evidence. Account. Rev. 66:402-16

Stivers C, Sun L. 2010. Cross-sectional return dispersion and time-variation in value and momentum
premia./. Financ. Quant. Anal. 45:987-1014
Sweeney RJ. 1986. Beating the foreign exchange market. J. Finance 41:163-82
Taylor MP, Allen H. 1992. The use of technical analysis in the foreign exchange market. J. Int. Money
Finance 11:304-14

Tversky A, Kahneman D. 1974. Judgment under uncertainty: heuristics and biases. Science 185:1 124-31
Verardo M. 2009. Heterogeneous beliefs and momentum profits. J. Financ. Quant. Anal. 44:795-822
Wang KQ, Xu J. 2010. Time-varying momentum profitability. SSRN Work. Pap.
Zhang XF. 2006. Information uncertainty and stock returns. J. Finance 61:105-37

www.annualreviews.org • Momentum 509

This content downloaded from

5.77.207.100 on Sun, 16 Mar 2025 18:57:53 UTC
All use subject to https://about.jstor.org/terms