The Momentum Effect And The Winner Finance Essay
Bachelier first indicates the existence of the EMH in his paper in which he proposes that there is no predictable relationship between publicly available information and stock price movements. It was not until 1965 that the term EMH became common and the theory was widely accepted by academics. This was due to Fama (1965), who proposes that in an efficient market, in general, new information in the stock market is immediately taken into account in stock prices. Therefore, one cannot seek to take advantage of any publicly available information, as it has already been included in the stock price.
Fama (1970) expands his theory further by categorising three different forms of the EMH, which allow for a more detailed analysis of the effects of information on stock price movement. These forms are referred to as "weak form", "semi-strong form" and "strong form". Weak-form efficiency dictates that all information derived from past prices is reflected in the current price of an asset. Semi-strong form goes a step further to include the effects of all publicly available information in the current price. Finally, strong-form efficiency asserts that the current price reflects not only publicly available information (including past price information), but also insider information. Fama (1970) comments in his conclusion that, in contrast to many economic theories, the evidence for the EMH is overwhelming while the evidence against it is unusually sparse.
The Overreaction Hypothesis and the Winner-Loser Effect
Struggling with your essay?
While the winner-loser effect had been documented in previous studies, DeBondt and Thaler's (1985) landmark paper on the effect drew much attention from academics to the anomaly. This was due to the connection the authors discovered between the overreaction hypothesis and the winner-loser effect. The overreaction hypothesis dictates that investors tend to overreact to especially negative (or positive) news on stocks, causing stocks to fall (or rise) dramatically, temporarily deviating from their equilibrium price.
To test this theory, DeBondt and Thaler (1985) examine the monthly returns of New York Stock Exchange (NYSE) common stocks. They create portfolio's consisting of the 35 most extreme winners and losers by ranking each company's performance over a three year period. At the end of this period, they then hold these portfolios for three years and test their performance against the average return of the NYSE as a whole. They find that, on average over a 57 year period, investors using this strategy could generate abnormal returns of 24.6% in excess of the NYSE average return.
Bremer and Sweeney (1991), Shiller (1981) and Jegadeesh (1990) find further evidence for the overreaction hypothesis, along with DeBondt and Thaler (1987) who find further evidence to support their initial paper.
The winner-loser effect is not confined to US markets. Extensive studies have been conducted across a wide range of markets, and the effect has been found to be present often.
One example is a study conducted by Baytas and Cakici (1999), which analyses several European equity markets. They take a more detailed approach than DeBondt and Thaler (1985) by creating winner and loser portfolios using one, two and three year rank and holding periods. They find that their portfolios generate abnormal returns of between 21.6% and 62.9%.
Schiereck et al (1999) provide further evidence of the effect in European markets in the study they conduct on companies listed on the Frankfort Stock Exchange. Using a thirty year period, data from 357 listed companies and portfolios that ranged in size from ten to forty companies, they find that profits of between 16.7% and 26.8% can be generated. Perhaps the most interesting finding in this study is not the confirmation of the winner-loser effect in the German market, but rather the conclusion that having accounted for factors such as company size and risk, they conclude that their results are still significant.
Other examples of the prevalence of the winner-loser effect in stock markets include Yulong et al's. (2005) evidence in the Nasdaq, Chen et al's. (2012) study of the Shenzen and Shanghai markets and Galoriotis' (2004) proof in the Athens Stock Exchange.
However, there is also ample evidence of the absence of the effect. Kyrzanowski and Zhang (1992), Karen and Kapusuzoglu (2010) and Yulong et al. (2005) find the contrarian strategies to be ineffective in the Toronto Stock Exchange, the Istanbul Stock Exchange and the NYSE, respectively.
Academics have proposed several explanations as to what is causing the effect. Offerman and Sonnemans (2004) examine two theories commonly believed to be causes of the overreaction effect. The recency hypothesis, which suggests that market traders place too much focus on recent information leading them to become overly pessimistic of loser stocks and overconfident in the performance of winner stocks, could not explain their results from the data. However, the hot-hand hypothesis, which states that market traders who find trends in firms' past performance and overestimate their accuracy thereby fuelling the overreaction effect, could possibly explain their results.
Since the initial documentation of this market anomaly, many authors have published findings that suggest DeBondt and Thaler (1985) may have overlooked certain factors that could explain the winner-loser effect. Chan (1988) argues that they neglected to account for time-varying risk and that contrarian profits can be attributed to adverse risk. Zarowin (1990) finds that much of the abnormal returns from contrarian portfolios can be attributed to the small-firm effect.
1.3: The Momentum Effect
Jegadeesh and Titman (1993) find that by creating momentum portfolios, one of which buys stocks that are performing well and another that sells stocks that are underperforming, using holding periods that varied in range between three and twelve months, they could generate abnormal returns. They claim that these strategies do not expose the trader to adverse risk.
Lewellan (2002) studies the stock market anomaly further, taking into account elements that had been neglected by Jegadeesh and Titman, including the small-firm effect and the industry of the companies in their dataset. He finds that the momentum effect is still significant even while accounting for the above factors, supporting Jegadeesh and Titmans initial work.
This essay is an example of a student's work
Rouwenhorst (1998) applies the same approach as Jegadeesh and Titman and finds his results from his European dataset are consistent with their results in the U.S. More importantly, Rouwenhorst accounted for firm size so as to determine whether his results could be attributed to the size effect and found the opposite was the case. The majority of his portfolio consisted of large cap firms, indicating that momentum profits in European markets are not due to adverse risk.
Miffre and Rallis (2005) identify thirteen portfolios that use the momentum effect in the commodity futures market that return over 9% on average per annum. Compared to the average return of all the commodity futures utilised in their study, which resulted in a loss of over two and a half percent, these finding provide further evidence of the robustness of momentum strategies across a wide range of markets. While it should be noted that this study neglected to account for trading costs, the authors conclude that these costs would not have any significant effect on their results. (Rephrase)
Griffin et al (2003) conduct a broad study across a range of markets in varying economic climates, finding momentum strategies to be profitable in all macroeconomic states.
An oddity of momentum is that, unlike other stock market anomalies, returns from strategies based on this anomaly continue persist well after information of it has been extensively distributed. In their paper, Chordia and Shivakumar (2002) "show that profits to momentum
strategies are explained by a parsimonious set of macroeconomic variables that are related to the business cycle. The evidence in this paper is consistent with time-varying expected returns being a plausible explanation for stock momentum". (need to edit and reword)(2 more papers to be added)
Literature Map of Key Studies
Efficient Market Hypothesis
Quantitative -Time Series
30 stocks on the Dow Jones Industrial Average, Daily price changes, 5 years.
Stocks behave in a 'random walk' pattern.
Efficient Market Hypothesis
Many authors findings regarding market efficiency
Three forms of market efficiency; weak, semi-strong and strong.
De Bondt and Thaler (1985), U.S.
Quantitative - Time Series
57 years of NYSE monthly price data, portfolios of 35 companies, 3 year rank and holding periods
Abnormal returns using winner-loser portfolios are significant.
Baytas and Cakici (1999), Europe
Quantitative - Time Series
1, 2 and 3 year rank and holding periods, may European markets
Contrarian Profits ranging between 21.6% and 62.9%
Schiereck et al (1999), Frankfurt.
Quantitative - Time Series
357 companies, varying size of portfolios, monthly data
Contrarian Profits significant when risk and company size accounted for
Offerman and Sonnemans (2004) Europe
Recency Hypothesis and Hot-Hand Hypothesis
N = 35
Hot Hand hypothesis may explain contrarian profits from overreaction hypothesis
Jegadeesh and Titman (1993), U.S.
Quantitative - Time Series
54 years, CRSP monthly returns, 1-12 month lagged returns
0.93% to 1.99% returns, accounting for time-varying risk and the January effect
Quantitative - Time Series
15 years, 12 countries, weekly data.
Winner Portfolios outperform Losers, returns negatively related with firm size
Miffre and Rallis (2005)
Quantitative -Time Series
31 Commodity Futures, 5 year period
Momentum Strategies that trade backwardated and contagoed contracts are the most profitable
Literature Review Conclusion
The review of the literature on both the winner-loser effect and the momentum effect has raised several issues that need to be addressed. Firstly, the effects of both anomalies are not witnessed in all markets. It is not clear from the literature whether market size or age has any bearing on this issue. Secondly, the extent of the effect varies across different markets. Lastly, in the case of the winner-loser effect, several factors must be taken into account when assessing the authenticity of the results from the studies examined, such as firm size and time varying risk.
Freelance Essay Writing Vacancies
Looking to become a writer?
Due to high demand for our services we're always looking for new writers. If you're interested in becoming a freelance writer then why not read more about our freelance writer jobs.
It is from these issues that the objectives of this paper are drawn.
Research Problem, Literature Gap and Objectives.
Stock market anomalies have attracted a plethora of studies over the years due to their predictive power and their inconsistency with the EMH. However, the results produced by these papers are often inconsistent with other authors' findings, resulting in a need for further, more detailed, analysis. This has led to confusion as to whether the EMH is definitively applicable to all markets. While the momentum effect and the winner-loser effect have been thoroughly documented in the majority of developed and developing markets, certain markets lack the same level of scrutiny. Two such markets are those in Thailand and Singapore. Singapore has been included in several multi-market studies by authors such as Bolvers et al (2000), Brown et al (2008) and Huang (2006), however these studies lack a detailed company level analysis of the anomalies. Thailand has been the focus of one detailed study by Du et al (2009), but for the most part is similar to Singapore in that the vast majority of research conducted that includes this country is multi-market studies. Examples include Naranjo and Porter (2007), Shen et al (2005) and Barros and Haas (2008).
In many cases, studies neglect to account for certain factors that could have a significant impact on the profitability of momentum and contrarian strategies. Zarowin (1990) asserts that many contrarian portfolios do not account for the effects of firm size on their results. Chan (1988) indicates that studies have neglected to account for time-varying risk. It is the authors' intention to address issues such as these in a detailed analysis of Singaporean and Thai markets.
This lack of research, in comparison to U.S. and European markets, provides a research gap for the author of this paper and justifies the purpose of this paper.
The research will seek to answer several questions:
Is the momentum effect present in Thailand or Singapore? (This also serves as a test of weak-form market efficiency in these markets)
Is the winner-loser effect present in Thailand or Singapore?
Do other market anomalies exacerbate the returns from either the momentum portfolio or the winner-loser portfolio? (Such as the January effect and the small-firm effect).
Did the recent financial crisis have any effect on the returns from either portfolio?
Did the 1997 financial crisis in Asia have any effect on portfolio returns?
It is the authors' belief that determining the answers to the previous questions will provide sufficient evidence to settle the debate over the predictive power of momentum and contrarian strategies, at least for this time period.
(Literature for the effect of market instability on momentum and contrarian returns is required)
4: Research Methodology
4.1 Data and Portfolios
As Thailand and Singapore are the target markets for this research, price data will be obtained from companies that are traded on their main indices, the SET and SGX, respectively.
A list of 20 companies will be compiled, with price data going back 21 years for each company. This time period analysed in this study will be from 1st of January 1991 to 31st of December 2011. Ideally the dataset would include more companies and analyse a longer period of time, however due to the difficulty of obtaining stock price information this will not be possible. Analysing data up to December 2011 has the added benefit of ascertaining whether either the momentum effect or the winner-loser effect have disappeared in these markets in more recent times. The Thompson One database will be used to acquire weekly stock price data for both these stocks and their respective index.
Once the data has been collected it will be sorted into seven ranking periods, which are as follows:
January 1st 1991 - December 31st 1993
January 1st 1994 - December 31st 1996
January 1st 1997 - December 31st 1999
January 1st 2000 - December 31st 2002
January 1st 2003 - December 31st 2005
January 1st 2006 - December 31st 2008
January 1st 2009 - December 31st 2011
4.2 The Adjusted Market Model
A stocks performance in the ranking period will be used to determine whether it should be part of the winner or loser portfolio. The adjusted market model used by DeBondt and Thaler (1987) will be used to calculate abnormal returns, which is written as follows:
Rit = Î±i + Î² (Rmt) + Îµit
In their adjusted market model, DeBondt and Thaler (1987) note that Î± equals zero and Î² equals one. This allows for the calculation of abnormal returns using the following formulas:
Rit - Rmt = Îµit
Or Rit - Rmt = Î¼it
Or ARit = Rit - Rmt
Rit denotes the rate of return on security i at time t, Rmt denotes the rate of return on the market at time t and Îµit is the error term.
Using the above formulas it will then be possible to calculate the abnormal returns for each stock allowing for them to be sorted into winner and loser portfolios. Using the three year ranking periods, the top ten stocks, based on abnormal returns, will be sorted into the winner portfolio and the bottom ten stocks will be sorted into the loser portfolio. These stocks will then either be bought (winner portfolio) or short sold (loser portfolio) and held for a three year period, known as the holding period. This process will then be repeated for each ranking period, resulting in seven holding periods.
4.3 Cumulated Abnormal Returns
The cumulated average returns (CAR) approach will allow for the calculation of the winner-loser effect and also act as a means to test for the momentum effect. DeBondt and Thaler (1987) adapt this approach in their methodology and calculate CAR by aggregating single weekly returns over each period. The formula used in this study will be one that Conrad and Kaul (1995) suggest, which specifies how CAR is acquired for a three year period and is written as follows:
The first formula allows for the calculation of the cumulative abnormal returns while the second allows for the calculation of the average cumulative abnormal return of the portfolios. At this point the cumulative average residual returns will be calculated. The residuals are calculated by regressing each company's returns against its respective exchange's returns. These residuals can then be used to calculate the cumulative average residual returns of all of the stocks in each portfolio. As previously mention, CAR is then calculated by aggregating all weekly returns. These are then averaged, resulting in ACAR. Finally, the difference between the portfolios is calculated and if the winner portfolios indicate price continuation, the momentum effect is confirmed in the market. Alternatively, if the winner portfolios show price reversals, the author can confirm the presence of the winner-loser effect.
As mentioned in the literature, Chan (1988) suggests that contrarian profits may be due to adverse risk in compiled portfolios. To account for such risk the CAPM risk model will be used. Betas will be calculated for the test period and regression analysis of the Thai and Singapore markets will be conducted as follows:
Ln Rpt - ln Rft = Î±p + Î²p (ln Rmt - ln Rft) + Îµt
Ln R lt -ln Rwt = Î±l-w + Î²l-w (ln Rmt - ln Rft) + Îµt
According to Dissanaike's (1997) description of the model, Î± is the continuously compounded average monthly excess return, ln Rpt denotes the rate of return on the relevant portfolio, ln Rmt signifies the continuously compounded return on the market portfolio in month t and ln Rft is the compounded risk free rate of return in month t.
The model will allow the author to confirm whether the EMH and CAMP hold in the subject markets when Î± equals zero. The second equation is used to determine whether stocks in the loser portfolios carry more risk than those in the winner portfolios, given by the Î²l-w term.
4.5 The small-firm effect and the January effect
Both of these anomalies have been known to impact the profitability of momentum and contrarian strategies, calling into question the validity of such strategies. To counteract this problem, once the initial regression analysis has been carried out, the sample will be adjusted to account for company size by market capitalisation. Additionally, the months of January data will be analysed separately to discover the extent of the impact of this anomaly on the portfolios returns.
Due to the difficulty in acquiring reliable data going back even as far as 21 years, the study is quite limited in the number of companies that can be included in the portfolio. It is for this reason that the sample period is still subject to change, which would allow for a greater number of companies to qualify for the sample. Transaction costs will not be accounted for as these vary over the time period and are difficult to acquire accurately.
1st Meeting: November 22nd
I met with my supervisor, Cormac O'Keeffe, and received feedback for my dissertation proposal. The data and methodology, as well as target countries were of major concern to me up until this meeting. My supervisor confirmed that Thailand and Singapore were acceptable markets for study, and indicated that further research on my target markets was required. A meeting was arranged for mid-January, for which I am to have made significant progress on the literature review section of my dissertation.
2nd Meeting: Mid-January
At this point I aim to have made significant progress on my literature review. I will discuss area's in which I believe my literature is lacking. I will also be submitting my literature review for feedback at this point. Mr O'Keeffe indicated that he would provide feedback on my research proposal at this meeting.
3rd Meeting: Mid February
At this point I hope to receive feedback from my supervisor on my literature review. I will incorporate this feedback into my literature review and aim to finish this section of my dissertation by the end of February. I intend to have a detailed plan for my approach to the methodology section, and will discuss any weak areas in my plan. I will then agree on a date to submit my methodology chapter with my supervisor
4th Meeting: Early May
I intend to spend the large time period between February and May collecting data for my target markets. This will mostly likely be an intensely time-consuming process and thus, beyond creating a plan for my introduction chapter, I expect to have little other work done at this point. Due to the nature of the data collection, I do not anticipate any significant issues regarding the data, however if I do I will raise the issue with my supervisor during this meeting. I will receive feedback on my methodology chapter and adjust it accordingly.
5th Meeting: Early June
For this meeting I will have prepared a template for the presentation of my findings. As with other meetings, I will adjust my work to date according to the advice of my supervisor. Soon after this I will submit my findings for feedback.
6th Meeting: Late June
During this meeting I will be discussing all remaining sections of my dissertation. With feedback for my findings section I aim to have that section finished within a week of this meeting. This will allow more time to be spent on the last few sections. I will have all other sections prepared for the final meeting.
7th Meeting: Late July
By this stage the vast majority of the work will be finished and all that will remain is any area's I am having particular difficulties with regarding the final dissertation write up.
If you are the original writer of this essay and no longer wish to have the essay published on the UK Essays website then please click on the link below to request removal: