International Price and Earnings Momentum

The following pages discuss the research paper by Markus Leippold and Harald Lohre
(2009).

Momentum is a force that is gained by movement. Price momentum entails the observation
that past winning stocks continue to deliver superior returns in the short run while past
losing stocks subsequently continue to disappoint. Earnings momentum refers to the
observation of momentum in stock prices following the direction of analysts’ earnings
forecast revisions.

Past studies have detected trading costs to be the single most important impediment to
successfully implement momentum strategies in the U.S.. The amount of trading costs is not
only driven by the huge turnover but also by liquidity risk. Hence, Chordia, Goyal, Sadka,
Sadka, and Shivakumar (2007) accordingly find the post-earningsannouncement drift to be
confined to illiquid securities. Arena, Haggard, and Yan (2008) resolve this puzzle for the U.S.
by showing that momentum profits are especially pronounced for stocks with high
idiosyncratic volatility which suggests that the momentum phenomenon persists, since any
arbitrageur wishing to exploit the anomaly is limited by high arbitrage costs. In
substantiating this claim, we provide additional evidence that international momentum
strategies appear to be mostly limited to highly illiquid stocks.

AIM1: To detect as many countries as possible where the momentum anomaly actually
exists.

Some past studies show that data snooping anomaly are not robust to this battery of
teststhis paper finds that both price and earnings momentum are fairly persistent with
regard to data snooping biases. Hence, the phenomenon is even more intriguing and the
need for a sound economic explanation of the origins of momentum is apparent.

Chordia and Shivakumar (2006) claim that price momentum is merely a noisy proxy for
earnings momentum in the U.S.. This explanation is intuitive, since price momentum may
well be rationalized in a model of investors underreacting to fundamental news as
represented by earnings revisions. This paper checks whether this explanation constitutes
a broad pattern in a large sample of 16 European countries.
Findings were as follows:
1. First, while the result of Chordia and Shivakumar (2006) for their sample period
ending in 1999 is replicated, the conjectured pattern has recently become more
subtle. During the market frenzy at the end of the nineties, a decoupling of price and
earnings momentum in the U.S. is observed which suggests that this period may be
dominated by investors’ over- instead of underreaction.
2. Second, considering an aggregate European momentum strategy, European price
momentum appears to be a manifestation of earnings momentum throughout the
whole twentyyear sample period.
 3. Third, while argument cannot be replicated in all European countries, there is
considerable evidence that earnings momentum is a crucial determinant in
explaining price momentum for most countries.

Review of Momentum Strategies

1. Price Momentum

First documented by: Jegadeesh & titman (1993)

Approach to quantify momentum: Consider a portfolio that is long in the winner decile and
short in the loser decile. These decile portfolios arise from several winner and loser
portfolios based on overlapping time periods. The stocks are ranked monthly according to
their performance over the last six months and assigned accordingly to the respective
quintile portfolios. These are held for six months. Hence, the winner or loser decile of the
associated price momentum strategy of a given month is made up of six portfolios.
Jegadeesh and Titman (1993) find such a price momentum strategy to earn more than 1%
above the risk-free rate per month. In explaining the phenomenon of U.S. price momentum,
Jegadeesh and Titman (2001) examine post-holding period return patterns of momentum
portfolios. These patterns favor a behavioral explanation of momentum to be triggered from
market participants’ under- or overreaction to new information. Overreaction will drive
stock prices to levels that are not fundamentally justified, giving rise to a subsequent
reversion back to their initial level. On the other hand, given limited information processing
capabilities, investors may underreact to news which may positively effect a company’s
fundamental value. Since overconfidence likely causes investors to cling to their original
views, this fundamental news may only gradually transmit into the company’s stock price. In
this case, one obtains a flat post-holding period return of a momentum
strategy.

2. Earnings Momentum

First Documented by: Ball and Brown (1968).

It indicates the drift which encompasses the tendency of stock prices to drift in the direction
suggested by recent earnings surprises. This observation is most likely caused by investors
with bounded rationality failing to fully appreciate the earnings information, which results in
a delayed price response.
The implementation of the earnings momentum strategy is similar to the one of price
momentum. As in Chan, Jegadeesh, and Lakonishok (1996), we build a moving average of
cumulated revisions over the prior six months to capture the change in earnings
expectations and We go long in the highest earnings revisions quintile and short in the
lowest quintile in any given month.
Linking Price and Earnings Momentum

Do price and earnings momentum may reflect the very same mispricing or behavioral bias.
Prior studies like Chan, Jegadeesh, and Lakonishok (1996) find that the U.S. momentum
effect is concentrated around subsequent earnings announcements and show that price
momentum may partially be explained by underreaction to earnings information. However,
they contend that price momentum is not subsumed by earnings momentum, since each
ranking variable has some incremental predictive power for future returns. Given that Hong
and Swaminathan (2003) only detect price momentum in countries that also exhibit
earnings momentum nevertheless makes the case for a closer relation of the two anomalies.
Indeed, Chordia and Shivakumar (2006) show that U.S. price momentum appears to be a
manifestation of earnings momentum for the period from 1972 to 1999.

DATA USED

Sample: Comprehensive sample of companies domiciled in 17 equity markets, 16 European

markets and the U.S., covering the period from 1987 to 2007. All data has been gathered
from Datastream. After pruning the initial list for double counting (secondary issue, cross
listing), non- equity issues (ADR’s), regionally listed stock; a sample of 16,662 companies is
arrived at. further exclude those having market capitalization below 10 million USD, which
leaves us with a final sample of 13,291 companies.

Return Data: Monthly stock returns in local currency inclusive of dividends by employing
total return figures are considered. To represent the respective markets, broad market
indices as compiled by Datastream are chosen and 3-month-T-bills serve as a proxy for the
risk-free rate. For curing data issues, two major adjustments are proposed. One is to remove
non-common equity from the respective country research lists and the other is to screen for
irregular return patterns. Since the former has already been dealt with when deleting
secondary issues, only the issue to address the quality of return data remains.

Detecting Price and Earnings Momentum

The descriptive statistics of momentum-based quintile portfolios by country are found. In

computing momentum portfolio returns, the standard approach of Jegadeesh and Titman is
followed.

Profitability of the price momentum hedge strategy is assessed by considering the return
differential along with its t-statistic. For the U.S., a monthly hedge return of 79 basis points
at a monthly volatility of 4.4% is obtained. This gives rise to a t-statistic of 2.80. The latter is
even higher for the European hedge strategy providing a return of 119 basis points per
month but at a lower volatility. Further, using the t-statistic metric, it is found that 12
European countries that have anomalous returns on a 5% level or better.

While the loser quintile is sometimes contributing to the return spread, the lion’s share is
due to the winner quintile. This finding confirms prior evidence that a long-only investor
may well benefit from an according momentum strategy. However, the extreme quintile
portfolios are the riskiest across all countries, since the winner and loser portfolios prove to
be more volatile than the portfolios with less extreme price momentum. To judge a
systematic risk bias of these portfolios, the betas are cmputed according to the classical
regression:

Where, Rit denotes the return of quintile i, R Ft is the risk-free rate and RMt is the market
return of the respective country. For more than half of the countries, the extreme quintile
portfolios exhibit high betas, while the remaining portfolios appear to be homogeneous in
terms of beta. Moreover, in 14 countries the highest betas are for the loser quintile. Also,
there is a size bias for the two extreme quintile portfolios. Size is measured in terms of the
logarithm of market value, and it is found that the two extreme portfolios are mostly
populated by smaller companies. Concerning the price momentum strategy, it is observed
betas that are slightly negative suggesting that one may partially hedge against downside
moves of the market.

Assessing profitability of the earnings momentum hedge strategy, it is found that the U.S.
strategy earns 58 basis points per month at a volatility that is only half the size of the price
momentum volatility. Thus, the according t- statistic of 4.11 is more convincing. This
observation of improved risk-adjusted performance also applies to the European earnings
momentum strategy with a return of 83 basis points per month at 1.71% volatility, giving a
t-statistic of 7.52. Further, price momentum outperforms earnings momentum in terms of
return at the cost of higher volatility. Again, the extreme quintile portfolios are more risky
than the middle portfolios. However, in contrast to price momentum the long leg has less
beta exposure while the short leg of the earnings momentum strategy has a large exposure
to this factor. Also, the earnings momentum strategy exhibits negative betas that are usually
smaller in absolute terms than those of the according price momentum strategy.

When the the cumulative returns of the winner and loser quintiles of the earnings and price
momentum strategies were plotted together with the evolution of an equallyweighted
market portfolio a strong support for the findings in Chordia and Shivakumar (2006), namely
that price and earnings momentum are closely related is found. For both earnings and price
momentum, the loser and the winner quintile portfolios move almost in sync. In addition,
the loser portfolio stays well below the market portfolio and the winner portfolio stays well
above it. a price momentum investor may experience very turbulent times with volatility
well in excess of common market levels.

Time-Series Regressions

Since most of the hedge strategies are highly volatile, it is to be determined whether their
high returns are solely compensating for risk. To further examine the performance of
strategies, it is checked if the long-short portfolio returns can be attributed to common risk
factors. The standard approach of Fama and French (1993) is adopted and estimate a
regression model is found.

RLt − RSt would be the return difference of the respective hedge strategy, i.e., the long leg
minus the short leg. Regarding the common risk factor portfolios, country-specific factors
are computed. The market return RMt is represented by some broad market index, the size
factor RSMBt is mimicked by a small cap index minus the risk-free rate, RSCt − RFt, and the
value factor RHMLt is the difference between a value index and the corresponding growth
index, RV t − RGt. The alpha generated by the hedge strategy is net of common risk factors.

Results of a Fama-French regression for price momentum that uses 240 monthly returns
spanning the period from July 1987 to June 2007 revealed that the risk factors explain most
of the variation of the loser and winner quintiles’ excess returns. However, concerning the
long-short strategies, the model’s explanatory power is generally low. The resulting alphas
are positive and significant at the 5%-level for 15 out of 17 countries. Across countries, the
alphas are mostly driven equally by the long and the short leg. However, the U.S. alpha of
101 basis points is almost entirely due to the short leg.

Results of the Fama-French regression for earnings momentum is not captured by common
risk factors as well. All countries exhibit positive alphas that are significant on a 5%-level in
16 cases—the odd one out is Greece. Hence, this analysis significantly hardens the pure
return diagnostics. As for the sources to the earnings momentum alphas, we note that long
and short legs contribute in equal shares.

To ensure that the alphas are not spurious (since they arise from single hypothesis tests
performed for each country) both momentum strategies are subject to recent econometric
methods that additionally account for multiple testing. These testing procedures either
control for the familywise error rate (FWE) or the false discovery proportion (FDP).

Accounting for Multiple Testing

When simultaneously testing several, say S, trading strategies against a common

benchmark, some strategies may outperform others by chance alone. For instance,
extensive re-use of a given database or testing one investment idea on various markets of
similar nature are prime examples. If many countries are tested at the same time, it may
become more likely that some countries’ momentum will be wrongly identified as
anomalies. Therefore, individual hypotheses are combnined into multiple testing
procedures that control for the possibility of data-snooping biases.

A) Methods Based on the FWE

The traditional way to account for multiple testing is to control the familywise error rate,
defined as the probability of rejecting at least one of the true null hypotheses. If this
objective is achieved, one can be confident that all hypotheses that have been rejected are
indeed false. Many methods that control the FWE exist, the simplest one being the well-
known Bonferroni (1936) method, which consists of a plain p-value adjustment, i.e., the
initial significance level “alpha” is divided by the number of hypotheses under test. It is
considered that the extreme case of all hedge strategies yielding the very same alpha, then
individual tests should be carried out at the level “alpha”, which obviously is more powerful
than the Bonferroni (1936) method. The aim is to detect as many countries as possible in
which the momentum anomaly actually exists.

B) Method Based on the False Discovery Proportion (FDP)

When the number of hypotheses under test is very large, the error control may be based on
the false discovery proportion rather than on the familywise error rate. Let F be the number
of false rejections arising from a multiple testing method and let R be the total number of
rejections.
The FDP is defined as the fraction F/R, given that R > 0. Otherwise, the FDP is zero. A
multiple testing method controls the FDP at level “aalpha” if:

, for any P, at least asymptotically. Typical values of “gamma are 0.05 and 0.1.

Is Momentum Due to Data Snooping?

Multiple testing results using the Fama-French price momentum alphas as test statistics
indicate that price momentum is found to be overwhelmingly robust to data snooping. The
detected price and earnings momentum anomalies are confirmed by our battery of tests
that account for multiple testing issues. Having ruled out data snooping biases as possible
explanations to the momentum effects, the economic nature of these phenomena is
determined. In fact, one may wonder whether both price and earnings momentum may be
traced back to similar sources, be it a behavioral bias or a compensation for risk.

Correlation of Price and Earnings Momentum

Price and earnings momentum do follow very similar return paths. To quantify this
similarity, the correlation of selected price and earnings momentum portfolios is computed.
On comparing portfolios with identical price and earnings momentum ranking, it is found, in
the U.S there is a correlation of 0.933 between the loser portfolio and the portfolio with the
lowest earnings revisions. The winner portfolio is also highly correlated with the highest
earnings revision portfolio, exhibiting a correlation of 0.902. This relation holds in the
remaining countries with the same order of magnitude. Thus, price and earnings
momentum hedge strategies are positively correlated as well.

Does Earnings Momentum Subsume Price Momentum?

Chordia and Shivakumar (2006) show that the U.S. price momentum alpha vanishes when
additionally controlling for earnings momentum, while the U.S. earnings momentum alpha
is robust when vice versa controlling for price momentum. Chordia and Shivakumar (2006)
thus reason that price momentum is just a noisy proxy for earnings momentum. While this
reasoning is quite persuasive, does the observation carry over to other markets? To find
this, the Fama-French setting is extended to a four-factor model by adding an earnings
momentum factor:
where RPMNt refers to the returns of the earnings momentum strategy.

Contrasting the Fama-French results to those of the above four-factor model for all
countries’ respective hedge, it is found that while the returns of the quintile portfolios are
reasonably captured by the Fama-French factors, the returns of the price momentum
strategies are not. Even though these strategies sometimes load to onecommon factor or
another, the adjusted R2s are typically quite low. Only for the U.K., France, and Germany
have two-digit adjusted R2s. Considering the alphas of quintile portfolios, there is a
monotonic increase from loser to winner portfolios. For instance, the monthly U.S. price
momentum alpha of 101 basis points results from -90 basis points for the loser quintile and
from 11 basis points from the winner quintile.
However, this huge spread is fairly persistent when controlling for the earnings momentum
factor. The loser quintile’s alpha is -80 basis points and the winner quintile’s alpha reduces
to 1 basis point. As a consequence, the U.S. price momentum is still significant under the
four-factor model, contrasting with the results of Chordia and Shivakumar (2006).
In Europe, while the Fama-French model attains an adjusted R 2 of 9.4%, the fourfactor
model explains 42.9% of the variation in European price momentum returns, cutting down
the Fama-French alpha of 146 basis points to insignificant 16 basis points. Across all
countries, the addition of the earnings momentum strategy seems reasonable. The adjusted
R2 of the hedge strategies usually increases by a considerable amount. In this sense, all
countries’ price momentum alphas are clearly reduced in the four-factor model and so are
the corresponding t-statistics. The price momentum alphas of Germany, Switzerland,
France, Spain, Portugal, the Netherlands, and Finland are subsumed by the respective
earnings momentum factor.

According to Chordia and Shivakumar (2006), for earnings momentum to be the crucial
driver of price momentum, the former should be robust when controlling for the latter.
Hence, the earnings momentum alphas arising from the following four-factor model are
found:

where the original Fama-French model is augmented by the return of the price momentum
strategy, RWMLt (winner minus loser).

Contrasting the Fama-French results to those of the above four-factor model for all
countries’ respective hedge strategies, it is found that the additional factor leads to a
considerable increase in statistical fit. In fact, the adjusted R2 of the Fama-French model and
the four-factor model almost resemble the figures obtained in the price momentum case.
Consistent with Chordia and Shivakumar (2006), the U.S. earnings momentum alpha
remains large at 72 basis points with a highly significant t-statistic of 5.14. Given that the
European earnings momentum alpha has a t-statistic of 6.76, we suspect that this
observation carries over to other countries. Indeed, 13 of 15 original European anomalies
remain significant after controlling for price momentum. Only Italy and Norway do cease to
have significant earnings momentum alphas.

To summarize, among 17 countries we initially find 15 countries exhibiting significant price

momentum alphas in a classical Fama-French setting. Among these 15 countries, seven
countries follow the explanation offered by Chordia and Shivakumar (2006), i.e., earnings
momentum subsumes price momentum. These countries include Germany, Switzerland,
France, Spain, Portugal, the Netherlands, and Finland. Among the eight remaining four-
factor price momentum anomalies, five countries also have four-factor earnings momentum
anomalies (the U.S., U.K., Belgium, Sweden, and Denmark). Two countries’ earnings
momentum alphas cease to be significant (Italy and Norway) and Greece exhibits no
earnings momentum at all. In summary, we obtain an aggregate European pattern that
suggests a translation of Chordia and Shivakumar (2006)’s argument to European equity
markets. Thus, it is all the more surprising why we are v refuting their rationale for the U.S..

For the U.S., we see that the large Fama-French alpha is substantially reduced when
additionally controlling for earnings momentum. However, by the end of 1999, which
coincides with the end of the sample period in Chordia and Shivakumar (2006), this relation
breaks down for some years. Obviously, price and earnings momentum have decoupled
following the burst of the tech bubble. This reasoning supports the general view that
earnings momentum typically will be a result of investors’ underreaction to fundamental
news, while the market frenzy at the end of the nineties is more likely the result from
overreaction. In addition, our finding suggests that U.S. investors will most likely have put
less weight on earnings information following several accounting scandals at the beginning
of the century.

Origins of Momentum: Risk or Behavioral Bias?

1. Momentum and the Macroeconomy

Momentum may simply reflect future macroeconomic activity or the mispricing of certain
macroeconomic variables. To test for such a relation, the methodology of Liew and Vassalou
(2000) and Chordia and Shivakumar (2006) in regressing future GDP growth (12-month
ahead growth in real GDP) on lagged values of the Fama-French factors (MKT, SMB, and
HM) and one of the two momentum factors (12- month compounded momentum, either
price momentum WML or earnings momentum PMN). GDP growth is measured as the
change in the log of GDP.

Findings:

1. Market factor is a leading indicator of future economic growth in some of the

countries,
2. Report SMB and HML to also be negatively related to future GDP growth in major
equity markets. Hence, small cap or value stocks suffer prior to periods of economic
growth, whereas they thrive before an economic slowdown.
3. The link between earnings momentum and macroeconomy appears to be strongest
in the U.S. and the European aggregate. Given a positive relation instead of a
negative one suggests that earnings momentum is a proxy for a macroeconomic risk
factor. Hence, there appears to be no definite pattern in linking earnings momentum
to the macroeconomy, an observation that carries over to the regression results
 obtained using the price momentum factor. The findings sharply contrast with the
U.S. result of Chordia and Shivakumar (2006), who detect a negative relation but for
a different time period, they are by and large affirmative of the international study of
Liew and Vassalou. They fail to find a link between WML and GDP growth.

To conclude, failing to find a definite relation between momentum and the macroeconomy
may suggest that momentum is rather due to a behavioral bias, an idea we will explore in
the following section.

2. Momentum and Information Uncertainty

Hong and Stein (1999) states that firm-specific information only gradually spreads across
investors resulting in underreaction and, as a consequence, short-term return continuation.
If momentum is due to investors’ underreaction to fundamental news, the respective price
drift should be higher in more opaque information environments for which information
diffusion is slowest.
U.S. momentum strategies are more effective in companies of small size or in companies
with low analyst coverage. U.S. price momentum strategy is also more effective when
limited to high uncertainty stocks as measured by firm age, dispersion in analysts’ earnings
forecasts, stock volatility, and cash flow volatility.

To examine price and earnings momentum profits for different degrees of information
uncertainty, consider four measures to proxy for information uncertainty: Analyst coverage,
size, total stock volatility, and idiosyncratic volatility. Idiosyncratic volatility arises from a
standard Fama-French regression and total stock volatility is estimated using stock returns.
In particular, first sort stocks into five quintiles based on past returns. For each quintile, sort
the stocks into three terciles based on one of the four information uncertainty proxies.

FINDINGS

First, the empirical evidence for the price momentum in the U.S. is confirmed. The
momentum effect is more pronounced for stocks with low analyst coverage, smaller size or
higher volatility, be it total or idiosyncratic volatility.
Second, the latter findings do not only translate to the aggregate European momentum
strategy, but also to most of the European country strategies. In fact, only Denmark does
refute the underreaction rationale.
Third, while the earnings momentum results are quite similar among the major equity
markets the results for some smaller countries are somewhat muted.
Thus, having gathered substantial support for the underreaction theory, one may wonder as
to why the momentum anomaly is not arbitraged away. Recent research for the U.S.
contends that high arbitrage costs prevent rational investors from exploiting the momentum
anomaly

3. Momentum and Liquidity

Exploiting U.S. price momentum is costly. In fact, trading costs appear to erode all of the
potential profits rendering the momentum arbitrage opportunity an illusion. There is a close
relation between liquidity risk and U.S. momentum strategies. Thus, liquidity should play a
crucial role in inhibiting profitable execution of the European momentum strategies.

Analyze the profitability of the momentum strategies when restricting to winner and loser
stocks characterized by different degrees of liquidity. A stock’s turnover allow to capture the
trading quantity dimension. As for the price impact dimension, we resort to the ILLIQ
measure of Amihud (2002), which is the absolute daily return divided by the associated
dollar volume. To obtain an aggregate monthly value of ILLIQ, we simply compute its mean
over the corresponding daily values. The fourth measure is the one introduced by Liu (2006),
which has been designed to capture multiple dimensions of liquidity such as trading speed
and trading quantity.

First sort stocks into five quintiles based on past returns or earnings revisions. For each
quintile the stocks are further sorted into three terciles based on one of the four liquidity
measures.

FINDINGS:
Across most countries and liquidity metrics, the general pattern is that the least momentum
profits occur for the most liquid stocks and that profitability is increasing with illiquidity.
Thus pattern of momentum profitability decreasing with liquidity.
When using the share turnover as liquidity metric, the relation between liquidity and
momentum profitability is sometimes reversed. In US, we complement the findings of
Chordia, Goyal, Sadka, Sadka, and Shivakumar (2007), who show the post-earnings
announcement drift to be equally useless among illiquid stocks as measured by ILLIQ. Our
findings on earnings momentum profits for the aggregate Europe sample are quite different.
Across all liquidity measures, the strategy earns at least 74 basis points with t-statistics in
excess of five. Thus, liquidity appears to be a more severe impediment to implementing
earnings momentum strategies as opposed to price momentum strategies.

CONCLUSION

The investigation of a given security mispricing typically addresses two questions. Is the
anomaly simply a compensation for risk or is the anomaly real and, if yes, what behavioral
bias is driving it? Of course, these questions are only meaningful if the security mispricing is
not spurious in the first place. Hence, one needs to safeguard against data snooping biases.
We find that both price and earnings momentum are robust with respect to multiple testing
issues.

The investigation of a given security mispricing typically addresses two questions. Is the
anomaly simply a compensation for risk or is the anomaly real and, if yes, what behavioral
bias is driving it? Of course, these questions are only meaningful if the security mispricing is
not spurious in the first place. Hence, one needs to safeguard against data snooping biases.
We find that both price and earnings momentum are robust with respect to multiple testing
issues.
Given that momentum does not appear to proxy for macroeconomic risk, we narrow the
search in favor of a behavioral-based explanation of the momentum anomaly. In particular,
winner and loser portfolios characterized by high information uncertainty give rise to even
larger momentum profits. Thus, given that price momentum largely is earnings momentum
in disguise, our evidence supports the rationale of momentum being driven by investors’
underreaction to fundamental news. Moreover, we attribute the persistence of the
momentum anomaly to the fact that significant arbitrage costs prevent investors from its
exploitation. We find liquidity to be a crucial driver in governing the momentum effects.
However, while the U.S. momentum effects clearly are most pronounced among illiquid
winner and loser stocks, there are some European markets that exhibit very profitable
momentum strategies even for highly liquid stocks.

Key findings of this paper include:

1. Price and earnings momentum are pervasive features of international equity

markets even when controlling for data snooping biases- ie price and earnings
momentum are constantly defying capital market efficiency in international equity
markets. For European countries, price momentum is subsumed by earnings
momentum on an aggregate level. However earnings momentum appears to be a
crucial driver of the price momentum anomaly in many markets.

2. A relation is not established between momentum and macroeconomic risks, hence, a

behavioral-based explanation is expected to be at work.

3. Momentum profits are found to be more pronounced for portfolios characterized by

higher information uncertainty. Hence, the momentum anomaly may well be
rationalized in a model of investors underreacting to fundamental news. Find
momentum strategies to be most profitable when restricted to winner and loser
portfolios characterized by proxies of high information uncertainty. In other words,
the noisier the fundamental information, the slower its incorporation into prices,
which is in accordance with underreaction of investors.

4. Momentum works better when limited to stocks with high idiosyncratic risk or higher
illiquidity, suggesting that limits to arbitrage deter rational investors from exploiting
the anomaly.