The momentum effect in the European stock market

The Momentum Effect in the European Stock

Market

João van den Heuvel 10628398 Universiteit van Amsterdam

Bachelor Thesis Supervisor: Jan Lemmen

June 29, 2016

Abstract

In this paper the existence of the momentum effect in the European stock market is examined. The momentum effect is a market anomaly where stocks with high past performance continue to outperform stocks with low past performance. By analyzing the profits that are generated through performing momentum strategies, where past winners are bought and past losers are sold, conclusions can be drawn. Data from 87 different companies listed on the Euronext 100 is examined from the period 2005-2015. Results show that there is significant upward momentum, which means that prices continue to move in a positive direction. Due to the opposite and significant movement from the expected downward momentum, which is also positive in this case, the results on the overall momentum effect remain inconclusive.

Statement of Originality

This document is written by João van den Heuvel, who declares to take full responsibility for the contents of this document. I declare that the text and the work presented in this document is original and that no sources other than those mentioned in the text and its references have been used in creating it. The Faculty of Economics and Business is responsible solely for the supervision of completion of the work, not for the contents.

1. Introduction

The efficient market hypothesis states that in an efficient capital market, security prices reflect all available information, and thus should provide unbiased estimates of the underlying values (Basu, 1977, p.663). It should therefore be impossible to outperform the market by skilled stock-selection, since stocks always trade at their true value on the stock market. The efficient market hypothesis is associated with the view that stock prices follow a “random walk”, which was firstly examined by Maurice Kendall in 1953 (Kendall & Hill, 1953, p.11). It was found that stock prices were as likely to go up as to go down on a particular day, regardless of past performance. If prices are bid immediately to fair levels, they only react to new information, which should be unpredictable by definition. If new information could be predicted, it would be reflected in today’s prices.

However, some market anomalies contradict that capital markets are fully efficient, the ‘momentum effect’ being one of them. This effect is a well-known phenomenon in finance; it shows that stocks with high past returns continue to outperform stocks with low past returns. Anticipating on these price movements, where future performance can partly be predicted, these predictable price movements should indicate an arbitrage opportunity. The implication that this opportunity does exist makes this subject an interesting one to examine, since abnormal returns can be obtained by simply buying stocks that performed well, and selling stocks that performed poorly. Indeed, a lot of research on this strategy has been done already, Narasimhan Jegadeesh and Sheridan Titman being the first ones to examine this in 1993. They find that these momentum strategies realized abnormal returns during the period from 1965 to 1989 (Jegadeesh & Titman, 1993, p.89). However, it remains difficult to explain how these opportunities still seem to occur if they are widely known. In fact, the momentum effect is regarded as one of the most puzzling anomalies in finance (Grinblatt & Han, 2002, p.1).

The opposite of the momentum effect is the ‘mean-reversal effect’, where stocks that performed badly in the previous period outperform stocks that performed well. De Bondt and Thaler (1985) present the existence of the mean-reversal effect in their study about the overreaction of the stock market. They find that holding portfolios with stocks that performed poorly in the previous 3 to 5 years and selling portfolios that performed well in the same number of years results in generating significant positive returns (De Bondt & Thaler, 1985, p.801). A study by Jegadeesh and Lehmann (1990) on the short-term price reversal shows that contrarian strategies where stocks are selected on their performance of the previous week or month generate significant positive abnormal returns (Jegadeesh & Titman, 1993, p.66). Although this effect was suggested to be a result of the lack of short-term liquidity, further research by Jegadeesh and Titman (1991) supports this view by providing evidence on the relation of short-term reversal and the bid-ask spreads.

The most common difference between the momentum effect and the mean reversal effect is that they occur in different time periods. In the very short run (1 to 4 weeks) and in the long run (3 to 5 years), contrarian strategies seem to result in positive abnormal returns, while in the medium run momentum strategies seem to generate significant profits.

This paper however elaborates solely the momentum effect, and therefore will examine returns of holding- and formation periods of six months. Since most of the research on this subject has been done more than a decade ago, and the financial markets are

constantly moving and emerging, the findings of earlier research might be outdated. Therefore, this paper examines if there is a significant momentum effect in the European stock market, using the years 2005-2015 as the time span. The research question can be posed as: “Did a momentum effect occur in the European stock market in the period 2005-2015?”

2. Literature review

Because of its economic relevance, much research on this subject has been done across different markets and different time spans. Jegadeesh and Titman first questioned whether a strategy based on past performance results in significant returns. They select NYSE and AMEX stocks from the period 1965 and 1989, with varying holding- and formation periods from 1 to 4 quarters. Adding a second panel with a set of strategies that skip a week between the portfolio formation period and the holding period, this leads to a total of 32 different strategies. After that, portfolios subsisting of 10 stocks are ranked from best performance to worst, with the most extreme ones called the ‘winners’ and the ‘losers’. In each month, the winners are bought and the losers are sold. Their results show that the returns of all zero-cost portfolios are positive, and with the exception of one of the 32 strategies, these returns are also statistically significant (Jegadeesh & Titman, 1993, p.69). The return of the zero-cost portfolio is in this case the momentum return, since short selling a loser portfolio and buying a winner portfolio is, with neglecting the transaction costs, approximately costless to the investor.

The most general finding is that strategies with long formation periods of 9 to 12 months, combined with short holding periods of 3 to 6 months perform slightly better than the remaining strategies. They also find that the beta of the loser portfolio is higher than the beta of the winner portfolio (with values of 1.38 and 1.28, respectively), which implies a negative beta for the zero-cost portfolio (1993, p. 73). Jegadeesh and Titman therefore conclude that the momentum effect cannot be explained in terms of market risk. If the profits would stem from taking more risk, the zero-cost portfolio would rather require a positive beta. When examining the market capitalizations of the different stocks, it is found that the size of the stocks in the winner portfolios turns out to be twice as large as the size of the stocks in the loser portfolio. Research from Bantz (1981) shows that small NYSE firms have significant

higher risk adjusted returns than large NYSE firms, in the period 1936 to 1975 (Bantz, 1981, p.16). This results in a so-called ‘small minus big factor’, which is included in the Fama & French three-factor model, to account for the spread in returns between small- and large-sized firms. Since the findings of Jegadeesh and Titman regarding the firms’ market capitalizations contradict the expectations that the winner portfolios should consist of smaller firms, the small minus big factor could also not explain the momentum effect.

In 1998, Conrad and Kaul perform comparable research by evaluating eight strategies in different periods, using stocks from the NYSE and AMEX from 1926 to 1989. The

difference between their research and that of Jegadeesh and Titman is that there is not a distinction made between the length of the formation and the holding period. They choose strategies with holding periods from different time spans, varying from 1 week to 36 months (Conrad & Kaul, 1998, p.491). They find that in 11 of the 36 strategies, the momentum strategy results in significant positive abnormal returns, all in a medium horizon of 3 to 12 months (1998, p.496). Hence, their results confirm the documented momentum effect of Jegadeesh and Titman.

Research from Chan, Jegadeesh and Lakonishok on momentum strategies, using stocks listed on the NYSE, AMEX and NASDAQ from the period 1977 to 1993, also confirms that a significant momentum effect exists. Strategies where stocks were selected by prior six-months returns results in earning positive returns of 8.8% over the subsequent six months (Chan et al., 1996, p.1709). A possible explanation for this outcome is that the market responds sequentially to new information. If the market is surprised by good or bad earnings news, on average the market continues to be surprised in the same direction over the

subsequent announcements. Returns however are not always directly related to recent-earnings; other events such as stock buybacks, insider trading and new equity issues may affect the returns and thus the news as well.

In their paper of under- and overreaction, and the momentum strategy in asset markets (1999), Hong and Stein develop a unified behavioral model to explain the interaction between heterogeneous agents (Hong & Stein, 1999, p. 2144). They make a distinction between two types of agents; one is called the ‘newswatcher’ and the other is called the ‘momentum trader’, neither of them acting fully rational. Rather, each type of agent is boundedly rational, meaning that it is only able to process some subset of the available public information. The newswatchers in this case are limited in the sense that they do not condition on current or past prices; they make forecasts based on signals that they privately observe about future fundamentals. Momentum traders however, do condition on past prices; their limitation is that their forecasts must be simple functions of the history of past prices (1999, p.2145). The third assumption that is made in the model is that private information is diffused gradually across the newswatcher population. It follows that when only newswatchers are

active, there is underreaction but never overreaction, since prices adjust slowly to new information. Subsequently, momentum traders take advantage of this underreaction, and the profit opportunities are arbitraged away; this implies that the market becomes approximately efficient. But since momentum traders are limited to simple strategies, this intuition is incomplete. On the contrary, the attempt of exploiting the underreaction caused by the newswatchers, a reverse effect is visible. Eventually, even when momentum traders are risk-neutral, an overreaction is created to any news. The insight from this model is that momentum buyers in an earlier stage impose a negative externality on ‘late’ momentum buyers, and that each of them does not exactly know where they are in the cycle.

Hong and Stein (1999) therefore explain the continuation effect as a consequence of the gradual diffusion of private information, and the inability of newswatchers to extract this information prom prices (1999, 2163). As a consequence, prices tend to ‘overshoot’ in the longer run, since momentum traders try to exploit this effect, which explains the long-term price reversal. Empirical research of Hong et al. (2000) that tests this model with NYSE and AMEX data confirms the earlier developed theory (Hong et al., 2000, p. 293).

In 1998 Geert Rouwenhorst performs research on the momentum effect, and examines stocks that are not listed on the American stock exchange. His sample consists of the monthly returns of 2190 firms from 12 European countries, from 1978 to 1995. The returns are eventually converted to the Deutsche mark. The methodology however, is the same as the one performed by Jegadeesh & Titman, where stocks are selected with formation- and holding periods differing from 1 to 4 quarters. Similarly, the portfolios are ranked based on their results, with the best and worst performing portfolios called the ‘winners’ and the ‘losers’, respectively.

It is found that internationally diversified portfolios of past winners outperform the portfolios of past losers for about 1% per month (Rouwenhorst, 1998, p. 283). After

controlling for size and market risk effects, the continuation effect could not be contributed to traditional views that small firms tend to outperform large firms, and that the better

performing firms were exposed to additional market risk. As an extension to the earlier findings, Rouwenhorst considers if the continuation effect is due to country-specific momentum. Controlling for country composition somewhat decreases the monthly return from 1.16% to 0.93% per month (1998, p.275). He therefore suggests that country momentum is relatively unimportant for explaining the overall momentum effect. When testing the momentum effect in the individual countries, it is found that in all of the 12 European countries the momentum effect is present, and with the exception of Sweden, the results are all statistically significant (1998, p.275). The strongest momentum effect is found in Spain, followed by The Netherlands, Belgium and Denmark. However, the standard deviations in the individual countries are two to three times higher than those of the internationally diversified

strategy, which implies that the excess return variance can be diversified since it is country specific. In conclusion, Rouwenhorst findings support the research that had been done on the American stock market, and the additional study on country-specific momentum could not explain the overall effect.

Further research done by Rouwenhorst (1999) reveals a significant momentum effect in less developed countries as well. Research on cross-sectional returns of 1750 individual stocks in 20 different emerging markets, illustrates that the return factors in emerging markets are quite similar to those in developed markets (Rouwenhorst, 1999, p. 1462). In 17 of the 20 countries, the winner portfolios outperform the loser portfolios by 0.39% per month if stocks are equally weighted, and 0.58% per month if countries are equally weighted. Rouwenhorst attributes the lower documented effect to the earlier findings that return continuation

increases with past return, and that the momentum portfolios in the emerging markets include stocks of the top and bottom 30%, instead of the 10% that was used in earlier research on emerged markets (1999, p. 1449).

Chan, Hameed and Tong (1999) perform research on the profitability of momentum strategies in international equity markets, and question whether exchange rate movements affect this profitability. They further investigate whether trading volume information affects the profitability of momentum strategies, coming from the belief that ‘it takes volume to move prices’ (1999, p. 2) and that without sufficient trading, stock prices may underreact to information. Anticipating on this situation would yield profits in the country in question. Chan et al. (1999) use data from 23 different countries in the period 1980 to 1995 from the Asia-Pacific, Europe, North America and South Africa. They also collect data on the trading volume, except for the countries Belgium, Denmark, Italy, South Africa, Spain and

Switzerland, in terms of turnover or dollar volume.

Their implemented momentum strategies yield weekly returns of at least 0.25%, all statistically significant with z-levels higher than 2, except for the 12-weeks holding period (1999, p.22). They therefore conclude that using active momentum strategies that reallocates investment from loser countries to winner countries for every 2 to 4 weeks will eventually outperform a passive buy-and-hold strategy by at least 1% per month. Considering the trading volume, except for the 26-week holding period, holding portfolios of countries with high lagged trading volume seem to generate higher returns than those of countries with low lagged trading volume (1999, p.13). This indicates that the momentum effect is stronger in cases where there is higher trading volume, and contradicts the view profits arise from underreaction to information due to insufficient trading (1999, p.14). Lagged exchange rate returns however show a negative relationship with equity returns, which indicates that taking into account exchange rate fluctuation does not add much to momentum profits (1999, p.9).

The major source of abnormal returns arises from the continuation effect in individual stock indices.

In their paper on European momentum strategies, Doukas and McKnight (2005) test whether the empirical findings of Hong et al. (2000) are also valid in the European stock market, and try to fill the gap in literature, which is only gives results on the American stock market. They use data on returns over the period from 1988 to 2001, from 3,084 firms in 13 European countries. Their findings confirm those of Rouwenhorst, and support the view that average stock returns are related to past performance (Doukas & McKnight, 2005, p. 315). Their evidence is also consistent with the gradual-information-diffusion model of Hong and Stein. They find that momentum strategies work better in stocks that have low analyst coverage (stocks that are less observed by the market). Moreover, they uncover that the profitability of momentum strategies is inversely related to the forecast dispersion of analysts. This is consistent with the view that investors do not update their beliefs adequately, and do not place sufficient emphasis on the statistical weight of new information.

In conclusion, most research on different markets in different time spans confirms that a clear momentum (and reversal) effect exists. It is widely agreed that the momentum effect is mostly visible in the medium run, while there is more evidence on the reversal effect in the long run. Hong and Stein (1999) try to clarify this with their gradual-information-diffusion model, explaining the deviations from the fundamental value of stocks via both under- and overreaction of agents. As Rouwenhorst confirms that the captured momentum effect in the American stock market is also present in the European stock market (1998), and in emerging markets (1999), Douklas and McKneight prove that the results from the gradual-information-diffusion model are also valid in the European stock market. Furthermore, a global research by Chan et al. (1999) proves that in different continents besides Europe and the US, a momentum effect exists as well. Although their findings all agree on the existence of the momentum effect, their explanations on why it occurs are widely different.

3. Methodology & data

To see if there is momentum in the European stock market, data from 87 companies listed on the Euronext 100 is used, from the period 2005-2015. Since the stocks have to be ranked based on their performance, the returns have to be determined first. Daily total return indices that represent the aggregate value growth of the companies are therefore retrieved from Datastream. In these indices, the aggregate daily dividend is included as an incremental amount to the daily change in price index. The total return index is calculated as follows: 𝑅𝐼! =   𝑅𝐼!!!∗_!"!"!

!!!∗   1 +

!"

!""∗! , where 𝑅𝐼! is the return index on day t and 𝑅𝐼!!! the return

the number of days in a financial year, which is 260 in this case. To calculate the returns in percentages, the formula (𝑅𝐼!− 𝑅𝐼!!!)/𝑅𝐼!!! is used.

Before ranking the companies, in order to construct ‘winner’ and ‘loser’ portfolios similar to the method used by Jegadeesh and Titman, the company’s performance has to be measured. This can be done by measuring ‘Jensen’s alpha’, which is the historical

performance of the stock over and above that is predicted by the Capital Asset Pricing Model (Berk & DeMarzo, 2014, p.410). Instead of using the total daily returns that were calculated, the CAPM regression is done by using excess daily returns, which can be found by

subtracting the risk-free rate from the daily returns1_{. In order to determine the market risk}

premium, the risk-free rate is also subtracted from the daily market return. The equation for the CAPM is as follows: 𝑅!− 𝑅! =  𝛼 + 𝛽(𝑅!− 𝑅!), with 𝑅!− 𝑅! being the company’s

excess return, 𝛼 being Jensen’s alpha, 𝛽 the sensitivity of the company with regard to the market, and (𝑅!− 𝑅!) the market risk premium.

The estimates for each company’s 𝛼 and 𝛽 are found by using simple OLS regression, with its common form Yi = β0 + β1Xi +

ε

variable (Yi) is the stock’s excess return, the intercept (β0) it’s alpha, the coefficient (β1) it’s

beta, the independent variable (Xi) the market’s excess return, and

ε

regressions are done on a semi-annual basis, so for each company 22 different alphas are found.

After that, the companies are ranked based on their alphas (i.e. the company’s performance) and for each six months, portfolios are constructed of both the best- and worst performing stocks. The portfolios with the 10 best performing stocks are called the ‘winner portfolios’, and the ones with the 10 worst performing stocks are called the ‘loser portfolios’. Next, the average six-month returns and the corresponding standard deviations in the

subsequent six months are calculated for both the previous winner- and the loser portfolio. For example, for constructing a momentum strategy in the second half of 2007, the average returns of the second half of 2007 were calculated for the companies from the winner- and loser portfolio of the first half of 2007. To generate momentum revenues, the companies from the winner portfolio from the previous year are bought, and the companies from the loser portfolio from the previous year are sold. Both the long- and the short position together construct the zero-cost position, and its returns can be seen as the momentum returns. By using the corresponding standard deviations, a t-test is done to see if the returns are

significantly higher than zero. These t-tests were also done on a semi-annual basis for three

1 _{To find the risk-free rate, the yields on German 10-year coupon bonds were used, retrieved}

from https://www.bundesbank.de

different positions, which results in 63 different outcomes (a strategy for the first half of 2005 cannot be constructed since it requires data from 2004). For both the long- and the short position, a one-sample t-test is performed, using the following formula: 𝑡 =_!!!!_!, with 𝜒 being the firm’s average six-month return, 𝜇 being zero in this case, 𝑆 the sample’s standard deviation and 𝑛 the number of days (which is 130). Since the short position means selling the stocks of the portfolio, the returns are multiplied by minus 1. For the zero-cost portfolio, the difference in average returns of the winner- and loser portfolio is tested by a two-sample t-test with the formula: 𝑡 = !!!!!! !!!!!

!!! !!!!!

! !!

, where 𝜒! is the average return of the winner portfolio,

𝜒! is the average return of the loser portfolio and both 𝜇! and 𝜇! are zero. Both samples have

different variances, so it would be inappropriate to make use of a pooled variance. The results are shown and discussed in the next section.

4. Results

The table shows the returns and their corresponding t-values of the three different strategies performed on a semiannual basis. The first column shows the concerning period, A being the first six months and B the second six months. The second column shows the returns of taking a long position (L) on the previous winner portfolio, the fourth column shows the returns of taking a short position (S) on the previous loser portfolio and the sixth column shows the returns of the zero-cost portfolio (L-S). The holding periods are of the same size as the formation periods, both being six months. The parentheses imply negative values.

**Significant at the 1% level *Significant at the 5% level

Period Return L T Return S T Return L-S T

2005 B 0.154 8.85** (0.2407) (14.89)** (0.0868) (3.66)** 2006 A 0.137 7.18** (0.0191) (1.25) 0.1179 4.82** 2006 B 0.190 14.56** (0.2023) (13.29)** (0.0123) (0.61) 2007 A 0.119 6.91** (0.1204) (9.34)** (0.0017) (0.08) 2007 B (0.026) -1.35 0.0835 4.06** 0.0579 2.07* 2008 A (0.260) (10.76)** 0.2459 8.88** (0.0143) (0.39) 2008 B (0.437) (9.80)** 0.3251 8.44** (0.1119) (1.90)* 2009 A 0.135 4.05** (0.2202) (6.28)** (0.0856) (1.77)* 2009 B 0.352 16.31** (0.3206) (15.63)** 0.0312 1.05 2010 A 0.045 1.83* 0.0240 1.03 0.0685 2.03* 2010 B 0.289 15.40** (0.1347) (7.48)** 0.1546 5.94** 2011 A 0.124 7.63** (0.0318) (1.94)* 0.0922 3.99** 2011 B (0.303) (8.89)** 0.2756 8.24** (0.0270) (0.57) 2012 A 0.141 8.66** (0.1500) (5.88)** (0.0094) (0.31) 2012 B 0.120 6.19** (0.1277) (5.56)** (0.0076) (0.25) 2013 A 0.057 3.30** (0.0009) (0.05) 0.0557 2.18* 2013 B 0.152 9.98** (0.1949) (12.79)** (0.0426) (1.98)* 2014 A 0.104 6.85** (0.0621) (3.66)** 0.0420 1.85* 2014 B 0.022 1.25 0.0295 1.79* 0.0517 2.14 2015 A 0.144 8.80** (0.2164) (13.71)** (0.0720) (3.16)** 2015 B (0.018) (0.85) (0.0066) (0.33) (0.0241) (0.85)

By looking specifically at the returns of the long- and short positions, a certain trend can be spotted throughout the sample period. In 15 of the 21 periods, taking a long position on the winner portfolios (which means buying them in the subsequent period) results in positive returns at a 5% significance level. In 3 of the 21 periods, the results remain insignificant. On the contrary, taking a short position on loser portfolios results in significant negative returns in 12 of the 21 periods, which opposes downward momentum. In 4 of the 21 cases, the results remain insignificant. Depending on which effect is stronger, conclusions can be drawn on the overall momentum effect. Looking at the zero-cost position, 8 periods show positive

momentum returns whereas 5 periods show negative momentum returns. The results of the 8 remaining periods were insignificant. This suggests ambiguous conclusions, with neither of the outcomes being highly remarkable.

The existence of upward momentum, which means that stocks that performed well are also more likely to do so in the subsequent period, appears to be the most evident finding. This is consistent with the existing theory and earlier findings on stock price momentum. However, in 3 periods (2008A, 2008B and 2011B) the opposite is shown, which indicates a reversal effect where buying the winner portfolios resulted in significant negative returns. A possible explanation for this could be the recent financial crisis, with its first disruptions dated in the third quarter of 2007 (Miskin, 2011, p.2). This could also explain the negative (but insignificant) returns of the long position in the second half of 2007. Moreover, looking at the short position on the loser portfolios in these periods, it shows that selling the worst

performing portfolios all result in positive and returns on both the 5% and 1% significance level. This indicates that all stocks performed poorly, and that there is no distinction between their performances in the previous period. The sample period between the second half of 2007 and the second half of 2008 could therefore be seen as an abnormal period, and negative returns could be expected beforehand. This leaves the second half of 2011 as the only period with significant differing outcomes.

Anticipating on downward momentum however, results in negative returns in 15 periods, with 12 of them being significant at the 5% level. It seems that selling stocks that performed poorly in the previous period does not generate the expected profits under the momentum theorem. This is due to the fact that the worst performing stocks did not experience great losses in the subsequent period, and moreover, did not experience great losses in the period in which the alphas were determined. When examining the worst

performing stocks, their returns turn out to be rather positive than negative, which could be an explanation of why the total momentum profits appear to be insignificant. Given the idea that the lowest alphas are based on returns that aren’t so bad after all, or at least, are non-negative

most of the time, it should not be a surprise that the returns aren’t so bad either in the following period.

This raises the question whether the outcomes are due to sample selection, instead of to their true underlying values. A possible reason for this could be that a sample of 87

companies is too small to identify true differences in performance. Another explanation could be that there is an overall upward bias that concerns all companies. This would justify why taking a short position on the worst performing stocks would not be profitable, since most of them aren’t likely to experience losses. Yet, this also brings up the idea that the outcomes for taking a long position on the best performing stocks are overrated, and should therefore be undermined up to a certain point. Further research has to be done on the characteristics of stock indices and the Euronext 100 in particular, to examine if a certain upward bias exists. In that case, it could explain the empirical deviations from what was expected from the existing theory.

4. Summary and conclusion

The momentum effect, which was first documented by Jegadeesh and Titman in 1993, is regarded as one of the most puzzling anomalies in finance. Numerous researchers have since then examined the phenomenon, across different countries and in different time periods, resulting in mostly the same significant outcomes. It is widely agreed that price momentum exists in the medium run, which is about 3 to 12 months. In the very short run and in the long run however, a mean-reversal effect seems to occur, where stocks tend to move in opposite directions, deviating from their fundamental values. Jegadeesh and Titman (1993) could not explain price momentum by means of market risk, since the zero-cost portfolio showed to have a negative beta. Since the stocks in the loser portfolio are mostly smaller than the stocks in the winner portfolio, market capitalization could neither explain the documented effect.

Different insights from Hong and Stein (1999) illustrate momentum by virtue of under- and overreaction in asset markets. In their behavior model on gradual-information-diffusion, the interaction of economic agents causes medium-term momentum and will eventually lead to long-term price reversal. Empirical research of Hong et al. (2000) confirms this theory. Research markets other than the NYSE and AMEX exhibit that price momentum exists as well in other countries, both in developed and in emerging ones. Rouwenhorst (1998) documented evident findings on the European stock market, and examined the momentum effect in the countries individually as well.

The empirical findings of this paper on the European stock market are not as explicit as those of Rouwenhorst, but several conclusions can be drawn. Upward momentum seems to be present and persistent in time; in 12 of the 21 examined periods momentum strategies resulted in positive and significant returns. This is consistent with the existing theory, and

could partly explain the findings on the overall momentum effect. The negative returns in 2007 and 2008 could be explained by the recent financial crisis, leaving the second half of 2011 as the only period with significant different outcomes.

The findings on downward momentum are in contrast with the expectations; momentum strategies resulted more often in negative returns instead of in profits, and in 12 periods these results were statistically significant. This leaves the results on the overall momentum effect being inconclusive, since the findings on both upward- and downward momentum rather contradict each other. The outcomes of upward momentum seem to be stronger than those of downward momentum, as in 8 of the 21 periods a positive overall effect is documented. In 5 periods the opposite is true, which makes it interesting to perform further research on this topic. These outcomes could partly be due to sample selection, and more information about the characteristics of stock markets and in particular the Euronext 100 is needed. In the 8 remaining periods the results seem to be statistically insignificant, and therefore no conclusions can be drawn from this. Coming back to the research question, whether a momentum effect is exists in the European stock market in the period 2005-2015, remains the answer being ambiguous. Statements on both upward- and downward momentum could be made with a decent level of certainty, but findings on the total momentum effect remain however inconclusive.

5. References

Basu, S. (June 1977). Investment Performance of Common Stocks in Relation to Their Price-Earnings Ratios: A Test of the Efficient Market Hypothesis. The Journal of

Finance, Vol. 32(3), pp. 663-682.

Berk, J. B., DeMarzo, P. M. (2014). Corporate finance (Third Edition). Boston: Pearson.

Frederic S. Mishkin (2011). Over the Cliff: From the Subprime to the Global Financial Crisis. Journal of Economic Perspectives, Vol. 25(1), pp. 49-70.

Geert K. Rouwenhorst (February 1998). International Momentum Strategies. The Journal of Finance. Vol. 53(1), pp. 267–284.

Geert K. Rouwenhorst (August 1999). Local Return Factors and Turnover in Emerging Stock Markets. The Journal of Finance, Vol. 54(4), pp. 1439–1464.

Harrison Hong, Jeremy C. Stein (December 1999). A Unified Theory Of Underreaction, Momentum Trading, and Overreaction in Assets Markets. Journal of Finance, Vol. 54(6), pp. 2143-2184.

Harrison Hong, Terrence Lim, Jeremy C. Stein (February 2000). Bad News Travels Slowly: Size, Analyst Coverage, and the Profitability of Momentum Strategies. The Journal of Finance, Vol. 55(1), pp. 265–295.

Jennifer Conrad, Gautam Kaul (1998). An Anatomy of Trading Strategies. The Review of Financial Studies, Vol. 11(3), pp. 489-519.

John A. Doukas, Phillip J. McKnight (June 2005). European Momentum Strategies,

Information Diffusion, and Investor Conservatism. European Financial Management, Vol. 11(3), pp. 313–338.

Kendall, M., Hill, A. (1953). The Analysis of Economic Time-Series-Part I: Prices. Journal of the Royal Statistical Society. Series A (General), Vol. 116(1), pp. 11-34.

Mark Grinblatt, Bing Han (January 2002). The Disposition Effect and Momentum. National Bureau Of Economic Research, Working Paper No. w8734. Retrieved 3 June 2016 from http://www.nber.org/papers/w8734.

Narasimhan Jegadeesh, Sheridan Titman (March 1993). Returns to Buying Winners and Selling Losers: Implications for Stock Market Efficiency. The Journal of Finance, Vol. 48(1), pp. 65-91.

Rolf Banz (March 1981). The Relationship between Return and Market Value of Common Stocks. Journal of Financial Economics, Vol. 9(1), pp. 3-18.

Werner F. M. De Bondt, Richard Thaler (July 1985). Does the Stock Market Overreact?. The Journal of Finance, Vol. 40(3), pp. 793-805.