Guru Stock Recommendations
An Analysis of the Relation between Recommendations and Stock Market Returns
Author:
Sophie Christine Hélène Hofdijk
Supervisor: L. Dam
Second Supervisor: A.J. Meesters
October 2010
University of Groningen Faculty of Economics and Business Master of Science and Business Administration
Guru Stock Recommendations
An Analysis of the Relation between Recommendations and Stock Market Returns
Sophie Christine Hélène Hofdijk1
Abstract
This study examines the relation between the Buy, Hold, and Sell recommendations given by the top 50 Dutch Gurus and stock market returns, by applying the event study methodology. In the period between April 11th, 2003 and May 19th, 2010, 2347 recommendations were given for 147 different companies. This study shows a significantly positive abnormal return in case of a Buy recommendation, and a significantly negative abnormal return in case of a Sell recommendation. Furthermore, the study reveals that Gurus who apply technical analysis realize a better outcome for investors compared to Gurus that use a fundamental approach. Finally, Gurus that are representatives of a broker firm outperform individual analysts.
Key words: Gurus, Buy / Hold / Sell Recommendation, Abnormal Returns, Event
Study
JEL Codes: G14, G24
1
Preface
This event study was a real challenge, because of the fact that I had never done an event study before and because I underestimated the scope of this particular study. An event study commonly consists of around 200 events, but my event study consists of 2347 events, which led to a long process of getting all the related data into Stata and matching the events with the required data. Notwithstanding the issues related to the large scope of this event study, I feel it made my event study extremely interesting at the same time. I would like to thank Lammertjan Dam in particular, for his supervision and guidance during this process. Finally, I also want to thank my parents for supporting me.
Contents
1. INTRODUCTION ...5
2. LITERATURE REVIEW...9
2.1VALUE OF STOCK ANALYST RECOMMENDATIONS...9
2.2REPUTATION OF A STOCK ANALYST...13
2.3FUNDAMENTAL ANALYSIS VERSUS TECHNICAL ANALYSIS...15
2.4INDIVIDUAL STOCK ANALYST OR ANALYST REPRESENTING A BROKERAGE FIRM...16
2.5OTHER EFFECTS...17
2.6MAIN AND SUB QUESTIONS...19
3. DATA AND METHODOLOGY ...20
3.1DATA...20
3.1.1 Data of the recommendation, stock return and characteristics of the Gurus ...20
3.1.2 Descriptive data...21
3.2METHODOLOGY...23
3.2.1 Event study methodology ...23
3.2.2 Variables...25
3.2.3 Tests ...25
4. RESULTS...27
4.1RESULTS OF THE LINEAR REGRESSION...27
1. Introduction
Nowadays, investors are overloaded with information on which stocks to buy, hold or sell. In newspapers, on television, in business magazines and on the internet considerable attention is paid to stocks and stock markets. What will a stock do today and tomorrow? What do the stock analysts predict for the near future? For the ignorant customer, it becomes increasingly challenging to process all this information. With the coming into existence of the internet, this information overload has even grown. For example, a search on Google for “investment advice” on a specific stock will easily yield millions of hits. On a daily basis, thousands of stock analysts and broker firms provide a market analysis and issue stock recommendations. This sometimes ‘free’ advice is taken into consideration by a lot of investors who believe they do not have the necessary knowledge about the stock market themselves. In this study, I will try to shed some light on the value of such recommendations and whether particular analysts are better in providing valuable information than others.
Much is written about the value of recommendations of analysts on stock. Stickel (1995), Womack (1996), Barber et al. (2001), Jegadeesh et al. (2004), Jegadeesh et al. (2006), Barber et al. (2006) and Moshirian et al. (2009) look at the effect of recommendations given by stock analysts. In these studies they mainly focus on the stock analysts working at a brokerage firm in the United States, except for Jegadeesh et al. (2006) who look at analysts active within the G7 countries. But none of these authors have studied the Dutch stock analysts. In my study I analyze the effect of free of charge recommendations of stock analysts for the Dutch market. The aim of my research is to determine whether these Dutch Gurus indeed give sound recommendations to investors and if so, whether particular characteristics of a Guru can explain the quality of his recommendation. Prior international studies are rather old as they focus on recommendations given in the period between 1978 till 2005. The dataset used in this study is more recent namely, from 2003 – 2010. This can give a more up to date result of the situation on the stock market nowadays. In addition, the dataset contains a substantial number of individual analysts, instead of only analysts working at brokerage firms.
to fundamentals may influence the stock price. On the one hand, recommendations given by a stock analyst based on no specific information could operate like a self-fulfilling prophecy. As the recommendation was given by a well-known analyst, the investor follows the advice by investing in or selling the stock, leading to positive or negative reaction on the stock return. On the other hand, the stock analyst could be a well informed predictor. In this case, the advice of this analyst is based on information which the investor does not have access to (yet), or does not understand. This advice also lead to an effect on investors buying or selling the stock, but the positive or negative stock return is in line with the fundamental value of the firm. It is difficult to analyze which of these two types of recommendations leads to a better return, as the basis for the recommendation is never published. The relation between the recommendations given and the stock return is the focus of this study, taking several measurable characteristics of the Guru into account. Realizing that there are many caveats, perhaps the characteristics of the Gurus may give us some idea of whether we are dealing with self-fulfilling prophecies or information provision.
Regarding the process and characteristics of stock recommendations, two economic mechanisms can also be distinguished. One mechanism is the production of the recommendation, and the second mechanism is the decision process of the investor to invest. The production of the recommendation is determined by the process the analyst applies to come to a recommendation to buy, hold or sell a certain stock. What can be measured of this process, is the method used; technical or fundamental analysis, and / or on the profile of the Guru, being either an individual analyst or representative of a broker firm. The second mechanism is the process the investor uses to select a certain Guru and / or his recommendation to base his investor’s decision on. This study can only investigate the mechanism of the recommendations as no information is available on the selection process of the investors. We can however work with the assumption that the investor selects a Guru based on his reputation, expressed in the study by his “ranking”.
In the first step of this study, I create a cross-section of abnormal returns by applying the event study methodology. With this methodology I am able to measure what the effect of the recommendations given by the Gurus is on the stock returns. I then create three sub-samples corresponding to Buy, Hold and Sell recommendations. In the second step multiple linear regressions are performed. Firstly, I regress the abnormal returns on Guru characteristics to identify whether a certain “type” of Guru is more capable in terms of giving positive recommendations. The characteristics taken into account are first, the method the Guru used. This can either be a technical analysis or a fundamental analysis. Secondly, the profile of the Guru is determined. The Guru can be a representative of a broker firm or an individual stock analyst. And finally, the rank of the Guru is used, where number 1 is the highest rank and number 50 is the lowest. Secondly, a linear regression is applied, where I regress the rank of the Guru on the Guru’s characteristics method and profile, to investigate whether these two characteristics have any effect on the ranking.
The results of this study show a significant positive effect on the abnormal return when a Buy recommendation is given and a significant negative effect (which leads also to a positive result for the investor) on the abnormal return when a Sell recommendation is given. Furthermore, when examining the effects of the different characteristics of the Guru, I find that when the Guru performs technical analysis to base his recommendation on, the outcome for the investors is better, compared to those who use fundamental analysis techniques. Finally, when looking at the profile of the Guru, a representative of a broker firm outperforms an individual stock analyst.
results in this respect, it therefore might be an advice to select the Guru on the method (technical) and profile (representative of a broker firm) and not on the ranking.
2. Literature review
In today’s modern financial markets many stock analysts are active, providing market analysis and publishing stock recommendations on a daily basis. This has led to an increased interest in research whether these recommendations lead to positive results and why. In this section a survey of the literature is presented.
2.1 Value of stock analyst recommendations
Much research has been done on the effect of recommendations given by stock analysts on the value of the stock. The effect has been investigated in several ways namely, for different markets, different periods, using different techniques or other relevant parameters.
Stickel (1995) investigates what the average price reactions were on stocks when changes in individual analysts’ recommendations occurred. Looking at 16.957 recommendations given in the period between 1988 and 1991 he performs an event study. With this study he tries to identify the factors, which contribute to the stock price performance of the buy and sell recommendation given. He shows that the reputation of the analyst, the size of the broker and the importance of the change of the recommendation each had a short-term effect on the price of the stock. After a buy recommendation the price increased with 1.16% and a sell recommendation resulted in a decline of price of -1.28%. A positive change of a recommendation of the individual analyst, thus leads to a positive price response at the time of the announcement of the recommendation.
on the stock return, immediately after the recommendation was given but also in later months. Examining the post-recommendation drift, he finds that the sell recommendations have an impact over a longer period on the stock market than buy recommendations do.
After the investigation done by Stickel (1995) and Womack (1996), of publicly available recommendations of security analysts and whether they could lead to profitable investment strategies, Barber et al. (2001) perform a follow-up study. Stickel (1995) and Womack (1996) focus on an analyst and event time perspective. Their approach focuses purely at the changes in the average price, around the time the recommendation is given, from an analyst perspective. Barber et al. (2001) however, take a more investor oriented, calendar time perspective. This approach takes the transaction costs as well as the average stock price reaction into account, but from an investor’s perspective. They extend their investigation to consensus recommendations, examining whether it is possible to earn higher returns by buying the most highly recommended stocks and short selling the least favorable stocks. Their results show that a portfolio consisting of the highly recommended stocks leads to an average annual abnormal gross return of 4.13% and a portfolio consisting of the least favorable recommended stocks leads to an average annual abnormal gross return of -4.91%. This gross return is corrected for market risk, size, book-to-market, and price momentum effects. So if investors use a strategy where they purchase stocks that are most highly recommended and sell short the ones that are least recommended, it yields the best result. But if investors do not rebalance their portfolio on a daily basis or they do not react on changes in analyst’s recommendation this return will be lower.
which is consistent with Stickel (1995), Womack (1996) and Barber et al. (2001). Looking at the sell-side analysts they find that they recommend to sell mostly high growth stocks with “glamour” characteristics. Glamour characteristics are positive momentum, high growth, high volume and relatively expensive stocks.
In contrast to the literature mentioned so far, Jegadeesh et al. (2006) look at the value of recommendations in specific markets namely, the stock market for the G7 countries. After examining the distribution of the analyst recommendation levels in each country of the G7, they test the value of the analyst recommendation in each country. They use two approaches, the first similar to the methodology of Womack (1996) but for the period of November 1993 till July 2002. In the second approach, the performance of calendar-time trading strategies is analyzed. In these strategies stocks with recommendations upgrades are bought and stocks with recommendation downgrades are sold. They find that in all the countries, except for Italy, the stock prices respond significantly to the recommendations given at the day itself and at the day after. Large price reactions were identified around the recommendations revisions and a large post-revision price drift effect was present. This is especially found in the largest market, the USA. According to Jegadeesh et al. (2006) this good performance of the US analysts can best be explained by the fact that the US analyst might possess special skills in identifying undervalued priced stocks.
of Jegadeesh et al. (2004), they also find that the emerging markets respond more positively to firms, which have higher growth opportunities.
Concluding, all these articles mentioned above, show that the recommendations identify positive abnormal returns so investors can benefit from these recommendation by taken them into account in their investment decisions. Table 1.1 shows an overview of the literature on the value of stock recommendations mentioned in this subsection.
Table 1.1 Overview of literature on the value of stock recommendations
This table shows an overview of the existing literature on the value of stock analyst recommendations. For every paper the authors, journal, aim of the research, the data used, the methodology and the results are presented.
Paper Aim research Data Methodology Result
Stickel (1995) Financial Analysts Journal Average price reactions on stocks when changes in recommendations occurred Period of 1988 – 1991 Source: Zacks Investment Research
Event study +1.16 % price increase
Buy recommendations. -1.28% decrease Sell recommendations. Womack (1996) The Journal of Finance Average price and volume reactions when changes in recommendations occurred and if analyst recommendations have investment value. Period of 1989 – 1991
Source: First Call Boston
Event study +2.4% increase for Buy
recommendations. -9.1% decrease for Sell recommendations. Barber, Lehavy, McNichols and Trueman (2001) The Journal of Finance If investors can profit from the publicly available recommendations of the security analysis from an investors oriented perspective. Period of 1985 – 1996 Source: Zacks Investment Research
Calender time Portfolio with highly recommended stocks 4.13% annual abnormal gross return, portfolio least recommended stocks -4.91% annual abnormal gross return.
Jegadeesh , Kim, Krische and Lee
(2004) The Journal of Finance Investigate the source of the investment value that is provided by analyst stock recommendations and the changes in recommendations Period of 1985 – 1998 Source: Zacks Investment Research CON (consensus recommendation level) and CHGCON (consensus recommendation change) is calculated
Jegadeesh and Kim (2006)
The Journal of Financial
Markets
Look at the value of stock recommendations in the G7 countries Period of 1993 – 2002 Source: IBES Event time performance and calender-time
Stock prices responded significantly to the recommendations that where given at the day itself and the day after, except for Italy.
Moshirian, Ng and Wu (2009) International Review of Financial Analysis
Look at the value of the stock analyst in an emerging market Period of 1996 – 2005 Source: Recommendations from IBES, stock returns from MSCI and accounting data from S&P emerging market database Event-time performance with Buy-and-Hold abnormal returns (BHARs)
Stock prices react significantly to recommendations on the event and on the following days.
2.2 Reputation of a stock analyst
With so many different advices of stock analysts, the investor has to make a choice which recommendation he is going to follow. So how to select the best advice is the mechanism the investor has to deal with. Barber and Odean (2008) look at the effect of news and attention on the buying behavior of institutional investors and individuals. They find that individual investors buy more stocks, which are in the news, stocks, which experience high abnormal trading volume and stocks with extreme one-day returns, so-called ‘grabbing stocks’. These findings indicate the existence of a kind of reputation of the stock. Like Barber and Odean (2008) stated ‘When there are many alternatives, options that attract attention are more likely to be considered, hence more likely to be chosen, while options that do not attract attention are often ignored’. Based on this analysis you might state that people tend to be attention-driven. Stock analysts with a good reputation could have a similar power of drawing the attention of the investor in their investment strategy.
All-Americans have more impact on the stock returns than other analysts; they have on average a 0.21% greater impact on security prices.
Consistent with these two results, also Stickel (1995) shows that the reputation of the analysts has a short-term price effect on the stock.
In both articles, Stickel shows that the stock analysts with a higher reputation will be followed more active than the ones with a low reputation and they perform better. Are stock analysts with good reputation better performers than analyst with a lower reputation? For an investor it is important how this reputation is determined. It could be the result of a one time hit or on a continuous series of recommendations. The investors should take this determination of the reputation into account before following the advice blindly.
Table 1.2 shows an overview of the literature on the reputation of the stock analyst described in this subsection.
Table 1.2. Overview of literature on the reputation of the stock analyst
This table shows an overview of the existing literature on the reputation of the stock analyst. For every paper the authors, journal, aim of the research, the data used, the methodology and the results are presented.
Paper Aim research Data Methodology Result
Barber and Odean (2008) The review of Financial Studies
Look at the effect of news and attention on the buying behavior of institutional investors and individuals Period of 1991 – 1996 Source: Large discount brokerage, Small discount brokerage, Large-full service brokerage and the Plexus Group
Time series Individual investors are
Stickel (1995) Financial Analyst Journal Average price reactions on stocks when changes in recommendations occurred Period of 1988 – 1991 Source: Zacks Investment Research
Event study Reputation of the
analysts has a short-term price effect on the stock
2.3 Fundamental analysis versus technical analysis
A stock analyst can give a recommendation based on either fundamental analysis or technical analysis. Where a technical analysis is more based on historical prices combined with information from the market, the fundamental analysis is more concerned with financial information and economic developments of the company within their the field of operation. This characteristic (method used) can be an important determinant in the choice of the investor to select a recommendation of a stock analyst. Furthermore, it can also be an important determinant for the analyst to base his recommendation on. So it influences both the economic mechanisms.
Because of the fact that the use of technical analysis is frequently not well understood, Menkhoff (2010) does a survey among 692 fund managers in five different countries to study the use of the technical analysis by fund managers. He finds that mostly small asset management firms tend to use the technical analysis technique. Overall, the survey shows that 68% rated the fundamental analysis more important, 22% the technical analysis and 10% opted for a combination of the two. But when Menkhoff (2010) solely looks at the forecasting of horizons, the technical analysis is more actively used in the long term forecasting. Furthermore, his survey shows that psychological factors have a big influence on the prices. The fund managers using a technical analysis have a tendency to rely on trend-following behavior. Finally, his findings support the fact that fund managers tend to use technical analysis because of the high costs of the information needed by fundamental analysis.
Table 1.3. Overview of literature on the technical and fundamental analysis
This table shows an overview of the existing literature on technical and fundamental analysis. For every paper the authors, journal, aim of the research, the data used, the methodology and the results are presented.
Paper Aim research Data Methodology Result
Lo, Mamaysky and Wang (2000) The Journal of Finance Investigate whether technical analysis provides usefull information to predict the development of stock prices Period of 1962 – 1996 Source: US stocks Nonparametic Kernel regression Technical analysis applied over a longer horizon do provide incremental information. Menkhoff (2010) Journal of Banking & Finance Look at the effect of the technical analysis on fund managers Period of 2003 – 2004 Source: questionnaire in Switserland, Germany, United States, Italy and Thailand
Multivariate correlation
Techinical analysis became more important in the forecasting of oncoming weeks.
2.4 Individual stock analyst or analyst representing a brokerage firm
analyst or he/she can be a representative of a brokerage firm, which publishes the recommendation. A representative of a brokerage firm, who gives the recommendation, has to follow a different process before the recommendation can be presented to the public than the individual analyst. The individual analyst can publish the recommendation without having to agree with anyone else, while the representative of a brokerage firm has to find agreement within the firm. On the other hand, the brokerage firms have more employees who prepare recommendations, so are able to publish recommendations on different stocks on a regular basis, where an individual analyst has a lower production being just one person.
Furthermore, the target groups of the two types of analysts can differ, where the brokerage firms will focus more on the big clients, such as for example insurance companies, individual analysts are more focused on recommending stocks of interest to individual investors. This difference in type of clients can perhaps lead to a bigger impact on the volume of the stock involved in the case the broker firms give a recommendation.
Because of the effect of this difference in operation, this difference in profile is considered as an important parameter in this study.
2.5 Other effects
The production of the recommendation can be influenced by the method used and / or by the profile of the stock analyst. For the selection of the recommendation by the investor also the herding effect, advance knowledge, or the economic context can play a role. These possible factors are mentioned but will not be examined further in this study.
analyst turns out to be a more precise forecaster of the stock returns, the effect is even stronger.
An article more focused on the information advantage that stock analyst might have, is the article of Green (2006). He investigates whether the short-lived information advantage associated with the early access of brokerage clients to the recommendation changes, gives them more investment value. Often, the same information will be reported on newswire services after they are shared with clients. So he investigates if these clients receive any additional investment value because this information is shared with them first. This article, in contrast to the other articles, looks at the value of the recommendation given by the stock analyst from a client’s perspective. He eventually shows that in the sample period of 1999-2002 clients with early access to these kind of recommendations, realize an average of two-day returns of 1.02% by purchasing directly when an upgrade was given and returns of 1.50% by selling short when downgrades were given. So an early access to stock recommendations based on advance knowledge gives clients of brokerage firms a higher return on investment.
of the worsening economy, the brokers are less inclined to issue buy recommendations.
2.6 Main and sub questions
Based on these research papers mentioned in this section, the main research question of the study is:
‘What is the relation between the advice given by Dutch Guru analysts and the return of the stock and can this relation be explained by the characteristics of the
Gurus’
Many possible variables are mentioned in the literature review which can have an effect on the stock return after a recommendation is given by a stock analyst. To be more specific the main question is supported by the following sub question, which will be tested in this study.
‘What is the influence of the method used by the Guru, the profile of the stock analyst and the ranking of the Guru on the return of the stock in the case of either a
3. Data and Methodology
3.1 Data
In this section I discuss the data sources and descriptive statistics of the analysis of the recommendations and of the characteristics of the Gurus.
3.1.1 Data of the recommendation, stock return and characteristics of the Gurus
Buy, Hold and Sell recommendations of the top 50 Dutch Gurus at May 19th, 2010 were collected from www.guruwatch.nl. This website presents recommendations given by many professional analysts on stocks and markets developments. Furthermore, the average performance of each of the Gurus is being calculated based on the yield of their recommendations over time. The yield of a recommendation given is being calculated as follows, for a Buy and Sell recommendation the stock value at present (or at the moment of closing) is divided by the stock value on the moment of the recommendation given minus 1, this gives the score in percentage. For a Hold recommendation it is the same calculation, but the result is divided by 2. Furthermore, to determine the yield of the Guru the following calculation is done: when a Guru gives four Buy recommendations in a period of three months with the following results +15%, -15%, -10%, +20%. His / her yield for this period will be (15-15-10+20) / 4 = 2.5%. Based on this method the Guru site presents the average performance by the Guru over several years. As a result the scores identify the best / worst Gurus of the Netherlands at a certain moment. In order to rank the different Gurus, three different types of recommendations are distinguished. Namely, a Buy recommendation, Hold recommendation or a Sell recommendation. In appendix A, table A.9, the top 50 of the Dutch Gurus on May 19th, 2010 is presented with the number of Buy, Hold, or Sell recommendations per Guru, and the yield for all his recommendations.
consists of 2347 recommendations, either a Buy, Hold or Sell, given by the Dutch Gurus in the period between April 11th, 2003 and May, 19th 2010 and the relevant stock return data.
In addition to the returns of the Buy, Hold and Sell recommendations given and the rank of the Gurus, also the different characteristics of the Gurus are gathered. From www.guruwatch.nl the method the Guru uses to base his recommendation on is collected. This can either be a fundamental or a technical analysis. And from the top 50 of Gurus, it is determined whether the Guru is an individual stock analyst or someone representing a broker firm.
3.1.2 Descriptive data
Table 1.4 Descriptive statistics, frequency of the variables recommendation, method and profile of the Guru Frequency Percentage Recommendation Buy 1554 66.21% Hold 157 6.69% Sell 636 27.10% Total 2347 100.00% Method Technical Analysis 1301 55.43% Fundamental Analysis 1046 44.57% Total 2347 100.00%
Profile of the Guru
Company 1735 73.96%
Individual 612 26.04%
Total 2347 100.00%
Table 1.5 shows the distribution of the Guru recommendations for both technical analysis and fundamental analysis. Most of the Buy and Hold recommendations are based on fundamental analysis, while the most Sell recommendations are based on technical analysis.
In appendix B, table A.10, the method that the Guru uses is being compared with the profile of the Guru. It shows that broker firms base their recommendations in 65% of the cases on technical analysis, where an individual stock analyst relies more (around 70%) on fundamental analysis performed.
Table 1.5 Descriptive Statistic, frequency of the variables recommendation and the method
Method
Technical Analyst Fundamental Analyst Total
Recommendation Frequency Percentage Frequency Percentage Frequency Percentage
Buy 694 44,66% 860 55,34% 1554 100%
Hold 29 18,47% 128 81,53% 157 100%
Sell 578 90,88% 58 9,12% 636 100%
Total 1301 1046 2347
individual stock analyst. As the table demonstrates, in more than 90% of the Sell recommendations given, the Guru is working for a broker firm. For Buy or Hold recommendations, the Guru works in most of the cases also for a broker company. But the individual stock analyst and the broker firms both have in common that the majority of their recommendations are advice to buy stocks.
Table 1.6 Descriptive Statistic, frequency of the variables recommendation and the profile of the Guru
Profile
Company Individual Total
Recommendation Frequency Percentage Frequency Percentage Frequency Percentage
Buy 1038 66,80% 516 33,20% 1554 100%
Hold 114 72,61% 43 27,39% 157 100%
Sell 583 91,67% 53 8,33% 636 100%
Total 1735 612 2347
3.2 Methodology
In this section the methodology, which is used throughout this study is described. First, the event study methodology is explained. Second, the dependent and independent variables are introduced. Finally, the tests that are performed are discussed.
3.2.1 Event study methodology
In order to measure whether the advice given by a Guru has an effect on the stock prices, an event study is performed following the theory of Brown and Warner (1980 and 1985) and MacKinlay (1997). MacKinlay (1997) states ‘Using financial data, an event study measures the impact of a specific event on the value of a firm’. The aim of this event study is first to determine how the value of the stock changes in an estimation period from t = -90 till t = -30, in order to define the normal value of the stock changes. In the model this is called the estimation window. According to MacKinlay (1997) ‘The normal return is defined as the expected return without conditioning on the event taking place’. This event study uses an event window of 1 day, which is the day the recommendation is published (t=0) online on
given because it is not expected that investors have advance knowledge, so no inside information is available. As a result, only the day the recommendation is given will be used to determine if an effect will take place yes or no.
To be able to measure the effect of the event, which in this case is the recommendation to Buy, Hold or Sell given by the Guru, on the stock return, the so-called abnormal return has to be calculated. The abnormal return is the return in the event window corrected for the estimated (normal) return. According to MacKinlay (1997) the abnormal return is defined as:
ARiτ = Ri - E (Riτ | Xτ)
Where ARiτ is the abnormal return for stock i on date τ, Riτ is the actual return and E (Riτ | Xτ) is the normal return expected for time period τ. The Xτ is the conditioning information for the normal return model.
But according to Brown and Warner (1980) you can measure the abnormal return in three different ways, namely mean adjusted return, the market adjusted returns and the market and risk adjusted return. For this study the abnormal return is measured on the basis of the market and risk adjusted return model. The market and risk adjusted return model takes into account the securities systematic risk and the market return. Thus, to be able to measure the abnormal return, the alpha and the beta of the stock return have to be obtained. These values are calculated by using the normal return measured in the estimation period for the market index. In this study the market index, which is used is the S&P 500. This index is a worldwide known index, which is a good measure of the world economy. According to Brown and Warner (1980) the abnormal return is measured by the use of the market and risk adjusted model as follows:
ARiτ = Riτ - αi - βiRmτ
stock i on date τ and the return on the market portfolio. The actual return of the stock is measured by:
Riτ = ( Sτ – Sτ-1 ) / Sτ-1
Sτ is the stock price on date τ and Sτ-1 is the stock price on the day before Sτ.
The market return is calculated at the same way but with the use of the market S&P 500 index.
3.2.2 Variables
In this event study the influence of different variables on the stock return is analyzed. The dependent variable in this study is the abnormal return of the event. The independent variables taken into account are first of all, the method which is used by the Guru. This can be either fundamental analysis or technical analysis. Second, the profile of the Guru is taken into account. The profile of a Guru can either be a representative of a broker firm or an individual stock analyst. The last independent variable is the rank of the Guru. The rank of the Guru is a proxy of the reputation of the Guru. Stickel (1992) and Stickel (1995) both show that the better the reputation the better the performance. In this study, it is assumed that the higher the rank of the Guru the higher his / her reputation, where rank number 1 is the highest and rank number 50 is the lowest.
3.2.3 Tests
In this last section, the tests used in this study will be explained. A two steps method is used throughout this study. The first step is to create a cross section of AR (abnormal return) by performing an event study. Next, corresponding to each Buy,
influence on the change of the abnormal return, are shown in the following regressions:
ARi = β0 + β1 METHOD + β2 PROFILE + β3 RANK + ε
The variables method and profile are dummy variables. When the dummy variable method results in a 1 than the Guru uses a technical analysis and when it is a 0 the Guru uses a fundamental analysis technique. For the dummy variable profile the 1 stands for an individual person who gives the recommendation and the 0 for a representative of a broker firm. The variable rank is a number from 1 till 50, since the different Gurus are ranked based on their performance from 1 (high) till 50 (low). This regression is performed for each sub-sample. For all the regressions in this study, the robustness for the standard error is used.
The second linear regression is done taking the variable rank as dependent and the variables method and the profile as independent variables to assess the impact of the method and the profile of the Guru on the rank of the Guru. This leads to the following regression:
Rank = β0 + β1 METHOD + β2 PROFILE + ε
In the Appendix C the results of additional linear regression is being presented. In table A.11 the following linear regression is shown:
ARi = β0 + β1 METHOD + β2 PROFILE + ε
4. Results
In this section the results of the event study are presented and analyzed.
4.1 Results of the linear regression
In this part the results of the linear regression are presented. First of all, the results are shown where the regression is determined by recommendations. In this analysis the abnormal return is used as the dependent variable and the method, profile of the Guru and the rank as independent variables. After this analysis, another regression analysis is done by taking the variable rank as the dependent and the method used and the profile of the Guru as independent variables. In this regression all the recommendation given by the Guru are combined. It purely looks at the relation between the rank and the independent variables.
Table 1.7 shows the results of the first regression analysis, which demonstrates that a Buy recommendation leads to a 2.4% significant positive abnormal return and a significant negative abnormal return of -2.0% in case of a Sell recommendation. When the Guru gives a Buy recommendation the technical analysis done by the Guru performs 2.6% better than those based on fundamental analysis. Furthermore, when an individual stock analyst gives a Buy recommendation, their performance is 1% lower than from a representative of a broker firm. Finally, for the Buy recommendations you can see that a better ranking (lower number) leads to a higher abnormal return, in line with what one would expect. So, in case of a Buy recommendation, all the variables are significant at a 1% significance level. For the
To test the stability of the results, the regression analysis is repeated excluding the insignificant variables. The results of the significant level and the R-squared did not improve.
Table 1.7 The relation between the abnormal return and the Guru characteristics.
This table shows the linear regression with abnormal return as dependent variable and the characteristics of the Guru as independent variables. These characteristics are the method the Guru uses, the profile of the Guru and the rank of the Guru. In this regression a distinction is made between a Buy, Hold and Sell recommendation. The abnormal return is calculated by performing an event study to be able to perform this linear regression. The characteristics method and profile of the Guru are dummy variables. The method is either a fundamental analysis versus a technical analyses and the profile of the Guru is either a representative of a broker firm versus an individual stock analyst. The rank of the Guru is a number of 1 till 50. * = significant at 10% level, ** = significant at 5% level and *** = significant at 1% level.
Coefficient (Std. Error) BUY Constant 0.024 *** (0.003) Method 0.026 *** (0.002) Profile -0.009 *** (0.002) Rank -0.001 *** (0.000) R-Squared 0.162 Number of Observations 1549 HOLD Constant -0.009 (0.001) Method 0.014 ** (0.006) Profile -0.009 (0.009) Rank 0.000 (0.000) R-Squared 0.040 Number of Observations 156 SELL Constant -0.020 ** (0.008) Method -0.022 *** (0.004) Profile 0.018*** (0.004) Rank 0.000 (0.000) R-Squared 0.065 Number of Observations 632
these two independent variables on the rank of the Guru. The table shows that only the variable method is significant at a 5% significance level. This result indicates that when the Guru uses a technical analysis, he / she perform better on the ranking list, then when a fundamental analysis technique is used. The fact that the result is a negative number (-1.282) is due to the fact that a high ranking has a low rank number, so number 1 is the best performing Guru. The profile of the Guru is not a significant parameter, so this characteristic does not have any influence on the ranking.
Table 1.8 The relation between the rank and the Guru characteristics method and profile.
This table shows the linear regression with the variable ‘rank’ as dependent variable and the characteristics as the independent variables. These characteristics in this table are the method the Guru uses and the profile of the Guru. The dependent variable is a number of 1 till 50, where number 1 is the highest and 50 the lowest. The independent variables are both dummy variables. The method is either a fundamental analysis versus a technical analyses and the profile of the Guru is either a representative of a broker firm versus an individual stock analyst.* = significant at 10%, ** = significant at 5% and *** = significant at 1%
Coefficient (Std. Error) Method -1.282 ** (0.518) Profile 0.444 (0.641) R-Squared 0.005 Number of Observations 2350 4.2 Discussion
the opportunity instead of sell which decision requires more consideration. This effect is strengthened as the study has an event window of just one day.
The better results of the technical analysis for all Buy, Hold and Sell recommendations might be explained by the fact that in this study the event window is compared with the estimation window, which is closer to the technical analysis method then the fundamental analysis. The data used are based on the past performance of the company stock, while the fundamental analysis is based on the economical context of the market the company operates in.
The sample used in this study shows that 44% of the recommendations given by a Guru, which are based on a technical analysis, lead to a Sell recommendation. In the case of a fundamental analysis, only 5% are Sell recommendations. This can be clarified by the fact that fundamental analysis is based on less information grounding their analysis. Therefore, they might be reluctant to give a Sell recommendation because this can lead to a loss of money for the investors. While in the case of a Buy recommendation, this has a potential for investors to gain money. The technical analysis is based on more objective criteria as it follows the information from the past while the fundamentalists use more subjective interpretations, looking at current economic situation and development of the firm within the industry.
The results also indicate that a representative of a broker firm are better able to give sound Buy or Sell recommendations than the individual stock analyst. However in this study more Gurus are working at a broker firm than as an individual stock analyst namely, 59%. Besides, the broker firms publish in total more recommendations, almost 73.9%, than the individual stock analyst. A possible explanation that the broker firms perform better, is due to the fact that the broker firms have more access to information and tools, which the individuals do not have. Looking at table A.10 in Appendix B it shows that the broker firms also apply more technical analysis which reinforces this effect.
5. Conclusion
This study focuses on the relation between the advice given by Dutch Guru analysts and stock returns, and whether this relation can be explained by the characteristics of the Gurus. To answer the question, this study examined recommendations given by Dutch Gurus which were published on
www.guruwatch.nl. A two steps method is used throughout this study. The first step is a cross-section of abnormal returns created by performing an event study. The event study consists of a sample of 2347 recommendations given in a period of April 11th, 2003 till May 19th, 2010. Using an estimation window of 60 days and an event window of one day, the abnormal return of each recommendation is determined. The recommendations to Buy, Hold and Sell are put into three sub-samples. In the second step, the abnormal returns are regressed on Guru characteristics to identify whether a certain “type” of Guru is more capable in giving a “good” recommendation. The characteristics taken into account are; the method the Guru uses, either fundamental or technical analysis; the profile of the Guru, either a representative of a broker firm or an individual stock analyst and the rank of the Guru, where 1 is the highest rank and 50 the lowest. After performing several linear regressions, this study finds significant positive abnormal returns of 2.4% in the case of a Buy recommendation and significant negative abnormal returns of -2.0% when a Sell recommendation is given. Prior studies have similarly concluded that the recommendations lead to positive abnormal returns, so investors can benefit by taking recommendations into account in their investment decisions.
In addition, the results of the linear regression also show that the profile of the Guru has a significant effect on the Buy and Sell recommendations. In both cases, the representatives of a broker firm tend to outperform the individual stock analysts.
The two economic mechanisms identified in this study are first of all, the production of the recommendation and secondly, the process of the investor making an investment decision. The two characteristics which lead to the best results in the first mechanism, is when the Guru uses a technical analysis and / or when the Guru is a representative of a broker firm. For the second mechanism, only in case of a Buy recommendation it shows that a higher rank leads to a higher abnormal return. Because the Sell and Hold recommendation show insignificant results in this respect, it therefore might be an advice to select the Guru on the method (technical) and profile (representative of a broker firm) and not only look at the ranking.
Looking at the other mechanisms, the self-fulfilling prophecy versus the provision of information, the results of the higher abnormal returns of Buy recommendation in case of a higher ranking support the hypothesis that recommendations by popular analysts (higher ranked) can create self-fulfilling prophecies.
6. References
Barber, B.M., Lehavy, R., McNichols, M., Trueman, B., 2001, Can investors profit from the prophets? Security analyst recommendations and stock returns, The
Journal of Finance, vol.56, Issue 2, pp. 531-563
Barber, B.M., Lehavy, R., McNichols, M., Trueman, B., 2006, Buys, hold and sells: The distribution of investment banks’ stock ratings and the implications for the profitability of analysts’ recommendations, Journal of Accounting and Economics, vol.41, Issue 1 – 2, pp. 87-117
Barber, B.M., Odean, T., 2008, All that glitters: the effect of attention and news on the buying behavior of individual and institutional investors, The Review of
Financial Studies, vol.21, Issue 2, pp. 785-818
Brown, S.J., Warner, J.B., 1980, Measuring security price performance,
Journal of Financial Economics, vol.8, Issue 3, pp. 205-258
Brown, S.J., Warner, J.B., 1985, Using daily stock returns: The case of event studies, Journal of Financial Economics, vol.14, Issue 1, pp. 3-31
Green, T.C., 2006, The value of client access to analyst recommendations,
Journal of Financial and Quantative analysis, vol.41, Issue 1, pp. 1-24
Jegadeesh, N., Kim, J., Krische, S.D., Lee, C.M.C., 2004, Analyzing the Analysts: When do recommendations add value?, The Journal of Finance, vol.59, Issue 3, pp. 1083-1124
Lo, A.W., Mamaysky, H., Wang, J., 2000, Foundations of technical analysis: Computational Algorithms, Statistical Inference, and Empirical Implementation, The
Journal of Finance, vol. 55, Issue4, pp. 1705-1765
MacKinlay, A.C., 1997, Event studies in economics and finance, Journal of
Economic Literature, vol.35, Issue 1, pp. 13-39
Menkhoff, L., 2010, The use of technical analysis by fund managers: International evidence, Journal of Banking & Finance, vol.34, Issue 11, pp. 2573-2586
Moshirian, F., Ng, D., Wu, E., 2009, The value of stock analysts’ recommendations: Evidence from emerging markets, International Review of
Financial Analysis, vol.18, Issue 1 – 2, pp. 74-83
Stickel, S.E., 1992, Reputation and performance among security analysts, The
Journal of Finance, vol.47, Issue 5, pp. 1811-1836
Stickel, S.E., 1995, The anatomy of the performance of buy and sell recommendations, Financial Analysts Journal, vol.51, Issue 5, pp. 25-39
Welch, I., 2000, Herding among security analysts, Journal of Financial
Economics, vol.58, Issue 3, pp. 369-396
7. Appendix
7.1 Appendix A
7.2 Appendix B
Table A.10. Frequency of the variable method combined with variable profile
This table shows the method the Guru uses combined with the variable ‘profile’. Thus whether it is a representative of a broker firm who gives the recommendation or an individual stock analyst. It shows that representatives of a broker firms base their recommendation in most of the cases on a technical analysis where the individual stock analyst relies more on a fundamental analysis.
Profile
Company Individual Total
Method
Fundamental 621 425 1046
Technical 1114 187 1301
7.3 Appendix C
Table A.11 The relation between the abnormal return and the Guru characteristics without the variable rank.
This table shows a linear regression with the abnormal return as dependent variable. The independent variables are method and the profile of the Guru. In this regression a distinction is made between a Buy, Hold and Sell recommendation. The abnormal return is calculated by performing an event study. The characteristics method and profile of the Guru are dummy variables. The method is either a fundamental analysis versus a technical analysis and the profile of the Guru is either a representative of a broker firm versus an individual stock analyst. In this table the influence of the variable rank is left out of the regression. This is done to see if the rank can lead to endogeneity when it is taken together with the abnormal return in the linear regression. The results show that the R-squared is even lower when the rank is left out of the regression. Furthermore, the same variables are still significant. In conclusion, the rank does not have a big influence on the results of the linear regression, so there is no strong endogeneity present. * = significant at 10%, ** = significant at 5% and *** = significant at 1%
