Can We Really Trust Expert Reviews? Illustrated By the Case of Recommendations on the Financial Information Sharing Website

Transcription

1 Can We Really Trust Expert Reviews? Illustrated By the Case of Recommendations on the Financial Information Sharing Website Ya-Ling Chiu, Department of Business Administration, Tungnan University, Taiwan. Chiao-Ming Cheng, College of Management, Yuan Ze University, Taiwan. Yuan-Teng Hsu, College of Management, Yuan Ze University, Taiwan. Abstract Recommendation plays a very important part in the decision-making process of consumer behavior, and it is regarded as the efficient communication tool. This study mainly investigates recommendations of security firms research recommended by analyst from stock research sharing website whether it can be the key reference for making decision. As internet is going popular, it affects the way that investors search information. That is, many more people rely on internet to do information research that causes investors receive much more information. Nevertheless, what information, from this information pool, is valid and important to investors? This study takes famous Chinese stock research sharing website (ie., Dubaogao.com) as the investigated scenario. As this website updates recommendation of security firms analyst report, investors may check through the evaluation of each stock s recommendation freely by the website. This study employs event study method to do analysis, and there are 115,267 recommendations in the sample which includes 99,819 buy recommendations, 14,855 hold recommendations and 593 sell recommendations. The study results show two major findings. First, the analyst recommendations do make investors have the abnormal return; that is, the analyst recommendation sheds on the information has value to investors. Second, after categorizing data by analyst characteristics, we find that analyst with few companies analysis (ie., portfolio complexity is low) has better performance than the analyst with more companies (ie., portfolio complexity is high) in the event window (-10,10) which has cumulative average abnormal return (CAAR); furthermore, the finding implies that the star analyst outperforms the non-star analyst. However, the cumulative average abnormal 1

2 return of analyst having higher past performance is lower than the analyst having lower past performance. According to the above knowledge, the findings of the study can offer more useful information for investors to refer before making investment decision. Additionally, we expect the study result may provide suggestions to manager in designing the website system. Keywords: Analyst Stock Recommendations, Abnormal Return, Analyst Characteristics, Information Value, Event Study. JEL classification: G14, M19, M31 2

3 1. Introduction Recommendation is a key reference for consumer purchase decision making, which is considered as an effective communication tool and is applied in various industries (Chen, Liu, and Zhang, 2011; Moe and Trusov, 2011). In marketing, many scholars have done some empirical studies on film entertainment industry, which discussed the influence on the box office generated from online word of mouth (Liu, 2006; Chen and Xie, 2008). In information management, some scholars indicated that the online shop recommender system can effectively reduce the searching costs and the uncertainty when a consumer is looking for some unfamiliar products, which also helps increase the sales (Fleder and Hosanagar, 2009; Pathak, Garfinkel, Gopal, Venkatesan, and Yin, 2010). In financial domain, before making any investment decisions, many investors tend to put analysts recommendations into consideration as one of the sources of investment reference (Raghubir and Das, 2010). Engelberg, Sasseville, and Williams (2012) took the famous American finance television program Mad Money as their research scenario and found out that analyst recommendations do have influence on investors behavior and bring investors short-term excess return. It is revealed in the previous literature that when a consumer is under conditions of uncertainty (i.e. having limited information for the product quality or the service quality), he will then seek useful information sources to back up his behavior decision (Duhan, Johnson, Wilcox, and Harrells, 1997). With the rapid growth of internet and technology development, information is passed on faster and faster, which also increases the information reaching out to any consumers. Thus, figuring out the way to fast find the valuable information for reference is crucial to the consumers. From the business point of view, the way to effectively spread the information of the product and lower the costs of marketing communication has come to the point of concern. This study conducts an empirical research on some well-known stock research shoring website in China, which collects research reports made by the stock analysts from the different investment banks in China. Every investor can read the stock analyst s reviews on this website 1. This study conducts an empirical research on some well-known stock research 1 The website is called Dubaogao ( including some free of charge information like reports category, name of the analysts for a specific report, his/her affiliated institute, date of publication, title of the report, stock symbol for the recommended stock and its rating of recommendation (i.e. Strong Buy, Buy, hold, Underperform and Strong Sell). If an investor wants to attain the complete report, additional fee will then be charged (each IP address is allowed to download three reports free of charge, additional fee will be charged starting from the fourth report.) In addition, it is also provided 3

4 shoring website in China, which collects research reports made by the stock analysts from the different investment banks in China. Every investor can read the stock analyst s reviews on this website. In the stock market, the proportion of individual investors (retail investors) is growing bigger and bigger, thus many investors tend to through various channels (i.e. company information, stock analysts, magazines or other media etc.) to collect related information on specific stocks before making an investment decision (Raghubir and Das, 2010). It is shown in the previous research that stock analysts recommendation is an important reference source for investors to evaluate any investment decisions (Barber, Lehavy, McNichols, and Trueman, 2001; 2003; Jegadeesh, Kim, Krische, and Lee, 2004; Womack, 1996). It is especially obvious in the stock market in China, which the proportion of retail investors tops the global stock market. In other words, analysts recommendation is more influential in the China market. However, does the open and easy to access information of recommendation really matter as for reference? This is the topic of concern which all the investors are keen to understand. All the recommendation data in this research are made by various analysts from different Chinese investment banks and the research scenario is more generalized than the scenario focusing on the data of a single analyst. Moreover, the research data are collected from the most complete data bank of stocks and securities research reports sharing website, which is updated every day and any investor can always access the recommended analyst reviews free of charge. It is different from the live TV programs, because the data access is less limited by time and space and the collecting method is also simple and convenient. In addition, most of the previous literature on analyst reviews is mostly based on the stock market in the USA and the data are collected from the pay per view website I/B/E/S 2 (Clement, 1999; Jegadeesh et al., 2004; Womack, 1996), thus it is more difficult for the general investors to access any analysts previous reviews. Therefore this research aims to know if the free of charge and open information provided by such a website is also valuable for investment reference. on the website about the total number of research reports of each institute, the total number of research report of each stock analyst, and the ranking of the best analysts. Further details can be attained on the website. 2 I/B/E/S stands for Institutional Brokers Estimate System, which contains financial analysts earnings forecasts for a wide range of U.S. and international firms. It gathers earnings estimates from over 60,000 firms in about 67 countries. Please see the link below for more information. 4

5 Accordingly, this research intends to answer three important questions and provide suggestions to the investors and the website operators for reference. Firstly, if the information recommended on this website can be taken as reference for investment decision making and generate excess stock return for the investors? With event study method and the published day of the analyst review as the base date, this research intend to verify if analysts reviews can actually bring the investors any excess return. Secondly, if the information of an analyst s recommendation can be considered as an important reference for investor s decision making process, how can we find the real valuable information among those plenty information recommended by the analysts? It is indicated in the previous studies that analyst s competence, portfolio complexity (the number of industry or company one covers) and resources (for example, the business scale of the investment bank behind) are the key factors for the accuracy of an analyst s estimation (Bolliger, 2004; Clement, 1999; Clement and Tse, 2003; 2005; Jacob, Lys, and Neale, 1999). Therefore, we categorize first in this research with analyst s attributes to verify the data and intend to provide more valuable clues to intensify the information value, which can be a reference for investors before they make any decisions. Thirdly, we expect to provide suggestions for the website operators on designing their website attributes. If a website intends to run a pay per view website as its major profit model, then the content value turns out to be the most influential factor of user s willingness to pay (Keeney, 1999). If a user can through a fast and simple way access the feasible information, his loyalty to the website will thus increase, which helps bring up the browsing number of the website and increase the website value with higher market exposure (DeLone and McLean, 2003; 2004). 2. Theoretical Background and Hypotheses According to the study subject and purpose of this research, we furthermore review the related literatures about analyst behavior and the signal theory to propose the model (shown as the figure 1 below), and construct the hypotheses. < Insert Figure 1 about here > 2.1 Literature review about analyst behavior Numbers of previous studies have demonstrated that the recommended stock from analyst is the key factor affecting investor s decision. The explanation is following below: the security analyst possess professional knowledge, market information and the time advantage for integrating and analysis, which conduct the forecast report of recommended stocks (Kross, 5

6 Ro, and Schroeder, 1990). According to the analyst s stock recommendation, investors may evaluate the investment portfolio; furthermore, the progress affects stock price (Fama, 1970). The empirical results from Barber et al. (2001; 2003) and Barber et al. (2001; 2003) show that the recommend stocks from analysts affect investor behaviors, and bring excess return to investors. In other worlds, investors may have better return that market performance by using analyst s stock recommendation to manage portfolio. Based on the above arguments, this study investigates the hypothesis below: H1: The investment decisions based on the analyst recommendation reference have abnormal return. 2.2 The factor about the accuracy of analyst recommendation The findings from previous studies show inconsistent results about the relationship between analyst recommendation and excess return. For example, the studies (Barber et al., 2001; 2003; Womack, 1996) indicate that the analyst stock recommendation always bring excess return to investor; however, Woolridge (2004) shows that the stock from analyst recommendation cannot always have excess return. According to the signal theory, this study discusses the factor affecting the analyst forecast accuracy; therefore, we use analyst character as the category to offer valuable recommendation to investors Portfolio Complexity Previous study (Clement 1999) finds that the analyst forecast accuracy is related to the complexity of portfolio analysis progress. Clement (1999) uses industry number and company number that studding by analyst as the proxy of portfolio complexity: under the same circumstance, the proxy reveals that analyst is much difficult to study more different companies and the industries. That is, if the complexity of portfolio analysis progress is higher, the analyst spends few times on each industry or company; therefore, the stock performance of analyst recommendation is worse. In other words, if analyst just focuses on the less diversity in industry analysis, the investment performance will be better and have higher return. Boni and Womack (2006) indicates the analysts pay more attention on the industries they familiar with, that help to advance the value of recommendation information and have better investment performance. Based on the above arguments, this study investigates the hypothesis below: H2: The lower complexity of portfolio analysis (compared with the higher complexity of portfolio analysis) can bring higher abnormal return to investors. 6

7 2.2.2 Prior Performance Greene and Smart (1999) show that the analyst's prior performance can be the signal that indicates future performance. That is, if the analyst's prior performance is better, then the analyst s stock recommendation performs better than the poor analyst s prior performance dose. In other words, investors can have excess return by buying the recommend stock from the analyst who has better prior performance. Based on the above arguments, this study investigates the hypothesis below: H3: The analyst s stock recommendation with better prior performance (compared with the poor analyst s prior performance) can bring abnormal return to the investors Star Analysts Numbers of studies show that most analysts want to be star analyst (all-star ranking) by the magazine. Once the analyst is selected as the star analyst on the ranking, the analyst reputation increases and the analyst gets higher promotion opportunity, also dramatically raise salary levels. According the above arguments, the analyst has strong incentive to do some particular behavior for being the stat analyst. However, there is argument that whether the star analyst outperforms than others. Stickel (1992) finds that star analysts selected from American security are superior in terms of the accuracy than others. That is, the forecast accuracy from star analyst is higher than other analyst, and the star analyst also bring higher abnormal return to investors. Based on the above arguments, this study investigates the hypothesis below: H4: The star analyst (compared with the non-star analyst) can bring abnormal return to investors Brokerage Size The brokerage size may be the key factor that affects the accuracy of analyst recommendation (Clement, 1999). The large brokerage has a complete rule system and policy; also it offers better education and training that analysts can have more supports and acquire sufficient research information, which improve the analyst s forecast accuracy (Clement, 1999; Jacob, Lys, and Neale, 1999). Based on the above arguments, this study investigates the hypothesis below: H5: The analyst who belongs to larger brokerage (compared with the small brokerage) can bring abnormal return to investors. 7

8 3. Data and Analysis 3.1. Recommendation Data All the data of analyst stock recommendations in this research are collected from the website Dubaogao. Since 2000, the website started to collect all the research reports published by all the stock analysts in China and it is now the biggest stocks and securities research reports sharing website with the most complete databank. Until the end of 2011, there are total 750,000 reports collected in the website databank. The data collected from the website includes research institutes (names of the investment banks), stock names, stock symbols, names of the analysts, published dates of the reports, stock rating etc. We collected data for this study using the analyst stock recommendations available through Dubaogao. Table 1 reports summary statistics for the recommendation data set. The collected data comes from 1 st January, 2007 to 31 st December, All the recommended data in this research are categorized into buying, neutral, and selling, whereas amount to 99,819 (86.60%) cases, 14,855 (12.89%) cases, and 593 (0.51%) cases respectively. The total number of report is 115,267 (see Panel A of Table 1). Panel B of Table 1 reports sample statistics for the analysts. From 2007 to 2011, the number of Brokers increased from 47 to 86. The number of analysts has dramatically increased from 2007 and peaked in Its recommendation average rating of 3.88, all analysts recommended with approximately 1,196 companies each year, almost 56.2% of all. The average of the analysts per covered firms and the covered firms per analyst are 9.2 and 7.6, respectively. a. Data of Stock Prices < Insert Table 1 about here > The stock prices in the research are collected from Yahoo Finance Website ( and all the analyst recommended data are sorted out with stock symbols and the historical stock prices are rearranged on the website. The analyst s recommendation day is taken as the base date and we collect the stock prices of the day and the ones covering 14 days before/after the date, including all the information of the stocks required for the event window. b. Data of Analyst Attributes The attributes of an analyst in this research consists of the level of an analyst s mission complexity, previous performances, award records, and business scale of the analyst s affiliated investment bank. The detailed methods for measuring variables are illustrated as follows: 8

9 (1) Level of portfolio complexity As the data of stock recommendations are collected from the website, the number of recommended stocks in the past one season from each analyst will then be calculated, which is taken as the proxy for the level of an analyst s portfolio complexity. (2) Previous performances of an analyst s previous recommendations Through Yahoo Finance Website we collect all the stock prices of the recommended stocks of an analyst and calculate the abnormal returns of each recommended stock. The average of the abnormal returns of an analyst s recommended stocks in the past on year is taken as the proxy for this item. (3) Star analysts This part we refer to the well-known Chinese finance magazine XCF 3, which selects annually the best analysts of the year. (4) Brokerage size of the analyst s affiliated company The number of the employed analysts in each investment bank is taken as the proxy here. c. Event Study Event study method has been applied widely to different domains like finance (Barber and Odean, 2008, Neumann and Kenny, 2007), marketing (Balasubramanian, Mathur, and Thakur, 2005; Wiles and Danielova, 2009), and product innovation (Kalaignanam, Shankar, and Varadarajan, 2007; Sharma and Lacey, 2004; Sood and Tellis, 2009), which assess the impact of a specific event on the stock price of a firm. A large number of researches reviewed the influence on the stock price made from issuing any corporate information (i.e. new product launch, strategic alliance, stock dividend distribution etc.). According to the efficient market hypothesis, in the efficient capital market, the stock prices respond to the new announcements in the market (Fama, 1970). In other words, when a specific event is announced, the investors will evaluate its content for their investment strategy reference. When the corporate announcement shows positive effects, the stock price of the firm will from the day on increase. Therefore, in this research it is intended to review if analysts stock recommendations will generate abnormal returns and cumulative abnormal returns and this method of event announcement is called event study in statistics. 3 Since 2003, the XCF magazine selects annually the best analysts of every industry in order to improve the accuracy of analysts estimations and the quality of their research reports, it has been come to the 9 th yeah in The results are now gradually acknowledged by the public, the corporate investors, and the investment bank research institutes. Thus becoming the best analysts will increase the influence of an analyst s recommendations. Please find the link for more details: 9

10 The reasonable return on investment under no interference of any events is called normal return. The market model is applied to estimate expected rate of return (Brown and Warner, 1980; 1985). R it R.. (1) i i mt it Where R it = the return of stock i at time t, R mt = the return on the market value weighted index for day t, i = a measure of stock i s sensitivity to market changes, and 2 = the error term. 0, it it ~ N i Where i and i are OLS values from the estimation period. This approach uses the returns of the market portfolio as a benchmark for normal returns and then detects any deviations from it. We assume that the event (or recommendation) takes place at t = 0 (see Figure 2). We use a 250-day estimation window (between days -260 and -11) to estimate the normal or expected return and use the (-10, 10) event window when we analyze the pattern of abnormal returns associated with recommendations (MacKinlay, 1997). < Insert Figure 2 about here > Abnormal return is defined as the difference between actual return and expected return in specific securities at specific time. The event period designated in (-10 through +10). The formula is as follow: AR it R it R it Eˆ i R it ˆ ˆ R i mt.. (2) Where AR, R, and R are the abnormal, actual, and normal returns, respectively. it it mt AR it = the abnormal return of stock i at time t, R it = the actual return of stock i at time t, R mt = the return on the market at time t 10

11 ˆ, ˆ is the best estimate value of stock i in the market model equation during the research i i period. The cumulative abnormal return is defined by the following equation: b CAR... (3) it AR it t a CAR it (as shown in equation 3) indicates the cumulative return of stock i that analyst announced from time a to the time b, and t=0 is the recommend day. In other words, it s the cumulative abnormal return of recommended stock within the period that before and after announcement by analyst. 4. Results 4.1 Results of Analyst Recommendation and Abnormal Return. In Figure 3 and Table 2, the average abnormal return is positive prior to 10 days of event day. The average abnormal returns in analysts recommended day (ie, event day t = 0) is 0.32% and statistically significant at the 1% level. After the event day (t = 1) and (t = 2), the average abnormal returns were 0.16% and 0.05%, respectively. Both of them were also statistically significant at the 1% level. But there are negative average abnormal returns from t=3 to t=10. Figure 4 show the results. From 10 days before the event day (t = -10), cumulative average abnormal returns increased gradually from 0.01 percent, and rapid increase to 1.15% from three days before the event day(t = -3) to the two day after the event day (t = 2), and then it appears to be slowly decreasing from (t = 3) to (t=10). These results are consistent with our hypothesis1. < Insert Table 2 about here > < Insert Figure 3 about here > < Insert Figure 4 about here > 4.2 Buy Recommendations across Different Segments Results of Portfolio Complexity Table 3 shows cumulative average abnormal returns in the event window (-10, 10), we find that analyst with few companies analysis (i.e. portfolio complexity is low) has better performance than the analyst with more companies (i.e., portfolio complexity is high), the 11

12 cumulative average abnormal returns were 1.15% and 0.64% (0.51% difference, p-value < 0.05), respectively. This results support our hypothesis H2. < Insert Table 3 about here > Results of Prior Performances We consider high and low past performance in Table 4, we find cumulative average abnormal returns in the event window (-10, 10) that analyst having higher past performance (0.61%, p-value < 0.05) is lower than the analyst having lower past performance (1.47%, p-value < 0.001). This results contrary to H3. In addition, we also find evidence of a significant difference (p < 0.01) across the two subsamples in the event window (2, 10), the cumulative average abnormal returns were -0.39% and 0.06%, respectively. < Insert Table 4 about here > Results of Star Analysts Table 5 presents cumulative average abnormal returns in the event window (-10,10), we find that star analyst has better performance than the non-star analyst, the average cumulative abnormal returns were 1.72% and 0.73% (0.99% difference, p-value < 0.01), respectively. This results support our hypothesis H4. < Insert Table 5 about here > Results of Brokerage Size The results of the analysis across the two subsamples appear in Table 6. There is some support for H5 because cumulative average abnormal returns in the event window (-1, 1), we find that large brokerage size has better performance than small (0.24% difference, p-value < 0.001). However, in the event window (-10, 10) the two subsamples has non-significant difference. < Insert Table 6 about here > 5. Conclusions This study aims to find the valuable information, as the information collector, for investment decision from numbers of recommendation, and offers design suggestions for the investment information web. The contributions are following below; (1) This study shows that the analyst recommendation from financial website brings abnormal return to investors, which is the actual valuable information. (2) We use analyst charter as the category indicator, and the finding shows that the abnormal return with fewer recommended companies (as the lower portfolio complexity) in event window (-10, 10) outperforms the abnormal return with more recommended companies. 12

13 (3) This study also finds that the forecast accuracy is higher from the star analyst s stock recommendation (Stickel, 1992). In other words, the analyst award can be an effective signal to investor for making investment decision. (4) The opposite finding from our hypothesis is that the analyst s abnormal return with better prior performance is worse than those analysts with poor prior performance. According to the above all study results; we provide suggestion to website mangers for resource management. For example, website may add company number that studying by analyst, the tag of star analyst, and the prior performance to offer more valuable information that user can evaluate. Reference Balasubramanian, Siva K., Ike Mathur, and Ramendra.Thakur (2005), The Impact of High-Quality Firm Achievements on Shareholder: Focus on Malcolm Baldrige and J. D. Power and Associates Awards, Journal of the Academy of Marketing Science, 33(4), Barber, Brad, Reuven Lehavy, Maureen McNichols, and Brett Trueman (2001), Can Investors Profit From The Prophets? Security Analyst Recommendations and Stock Returns, Journal of Finance, 56 (2), Barber, Brad, Reuven Lehavy, Maureen McNichols, and Brett Trueman (2003), Reassessing the Returns to Analysts Stock Recommendations, Financial Analysts Journal, 59 (2), Barber, Brad M. and Terrance Odean (2008), All That Glitters: The Effect of Attention and News on the Buying Behavior of Individual and Institutional Investors, The Review of Financial Studies, 21 (2), Bolliger, Guido (2004), The Characteristics of Individual Analysts Forecasts in Europe, Journal of Banking and Finance, 28 (9), Boni, Leslie and Kent L. Womack (2006), Analysts, Industries, and Price Momentum, Journal of Financial and Quantitative Analysis, 41 (1), Brown, Stephen J. and Jerold B. Warner (1980), Measuring Security Price Performance, Journal of Financial Economics, 8 (3), Brown, Stephen J. and Jerold B. Warner (1985), Using Daily Stock Returns: The Case of Event Studies, Journal of Financial Economics, 14 (1), Chen, Yubo, Yong Liu, and Jurui Zhang (2011), When Do Third-Party Product Reviews Affect Firm Value and What Can Firms Do? The Case of Media Critics and Professional Movie Reviews, Journal of Marketing, 75 (2),

14 Chen, Yubo and Jinhong Xie (2008), Online Consumer Review: Word-of-mouth as A New Element of Marketing Communication Mix, Management Science, 54 (3), Clement, Michael B. (1999), Analyst Forecast Accuracy: Do Ability, Resources, and Portfolio Complexity Matter? Journal of Accounting and Economics, 27 (3), Clement, Michael B. and SenyoY. Tse (2003), Do Investors Respond to Analysts Forecast Revisions as if Forecast Accuracy is All That Matters? The Accounting Review, 78 (1), Clement, Michael B. and SenyoY. Tse (2005), Financial Analyst Characteristics and Herding Behavior in Forecasting, Journal of Finance, 60 (1), DeLone, William H. and Ephraim R. McLean (2003), The DeLone and McLean Model of Information Systems Success: A Ten-Year Update, Journal of Management Information Systems, 19 (4), DeLone, William H. and Ephraim R. McLean (2004), Measuring E-Commerce Success: Applying the DeLone & McLean Information Systems Success Model, International Journal of Electronic Commerce, 9 (1), Duhan, Dale F., Scott D. Johnson, James B. Wilcox, and Gilbert D. Harrell (1997), Influences on Consumer Use of Word-of-Mouth Recommendation Sources, Journal of the Academy of Marketing Science, 25 (4), Engelberg, Joseph, Caroline Sasseville, and Jared Williams (2012), Market Madness? The Case of Mad Money, Management Science, 58 (2), Fama, Eugene F. (1970), Efficient Capital Markets: A Review of Theory and Empirical Work, Journal of Finance, 25 (2), Fleder, Daniel and Kartik Hosanagar (2009), Blockbuster Culture s Next Rise or Fall: The Impact of Recommender Systems on Sales Diversity, Management Science, 55 (5), Greene, Jason and Scott Smart (1999), Liquidity Provision and Noise Trading Evidence from the Investment Dartboard Column, Journal of Finance, 54 (5), Hong, Harrison, Jeffrey D. Kubik, and Amit Solomon (2000), Security Analysts Career Concerns and Herding of Earnings Forecasts, RAND Journal of Economics, 31 (1), Jacob, John, Thomas Z. Lys, Margaret A. Neale (1999), Expertise in Forecasting Performance of Security Analysts, Journal of Accounting and Economics, 28 (1), Jegadeesh, Narasimhan, Joonghyuk Kim, Susan D. Krische, and Charles M. C. Lee (2004), Analyzing the Analysts: When Do Recommendations Add Value? Journal of Finance, 59 (3),

15 Kalaignanam, Kartik, Venkatesh Shankar, and Rajan Varadarajan (2007), Asymmetric New Product Development Alliances: Win-Win or Win-Lose Partnerships? Management Science, 53 (3), Karniouchina, Ekaterina V., William L. Moore, and Kevin J. Cooney (2009), Impact of Mad Money Stock Recommendations: Merging Financial and Marketing Perspectives, Journal of Marketing, 73 (Nov), Keeney, Ralph L. (1999), The Value of Internet Commerce to the Customer, Management Science, 45 (4), Kross, William, Byung Ro and Douglas Schroeder (1990), Earnings Expectations: The Analysts Information Advantage, The Accounting Review, 65 (2), Liu, Yong (2006), Word of Mouth for Movies: Its Dynamics and Impact on Box Office Revenue, Journal of Marketing, 70 (3), MacKinlay, A. Craig (1997), Event Studies in Economics and Finance, Journal of Economic Literature, 35 (1), Mikhail, Michael B., Beverly R. Walther, and Richard H. Willis (2004), Do Security Analysts Exhibit Persistent Differences in Stock Picking Ability? Journal of Financial Economics, 74 (1), Moe, Wendy W. and Michael Trusov (2011), The Value of Social Dynamics in Online Product Ratings Forums, Journal of Marketing Research, 48 (3), Neumann, John J. and Peppi M. Kenny (2007) Does Mad Money Make the Market Go Mad? The Quarterly Review of Economics and Finance, 47 (5), Pathak, Bhavik, Robert Garfinkel, Ram D. Gopal, Rajkumar Venkatesan, and Fang Yin (2010), Empirical Analysis of the Impact of Recommender Systems on Sales, Journal of Management Information Systems, 27 (2), Raghubir, Priya and Sanjiv R. Das (2010), The Long and Short of It: Why Are Stocks with Shorter Runs Preferred? Journal of Consumer Research, 36 (6), Stickel, Scott E. (1992), Reputation and Performance among Security Analysts, Journal of Finance, 47 (5), Sharma, Anurag and Nelson Lacey (2004), Linking Product Development Outcomes to Market Valuation of the Firm: The Case of the U.S. Pharmaceutical Industry, Journal of Product Innovation Management, 21 (5), Sood, Ashish and Gerard J. Tellis (2009), Do Innovations Really Payoff? Total Stock Market Returns to Innovation, Marketing Science, 28 (3),

16 Wiles, Michael A. and Anna Danielova (2009), The Worth of Product Placement in Successful Films: An Event Study Analysis, Journal of Marketing, 73 (4), Womack, Kent (1996), Do Brokerage Analysts Recommendations Have Investment Value? Journal of Finance, 51 (1), Woolridge, Randall J. (2004), Performance of Stocks Recommended by Brokerages, Journal of Investing, 13 (1),

17 Social Sciences (EAR15Italy Conference) ISBN: Figure 1: Conceptual Framework Recommendations Analyst Attributes 1. Portfolio Complexity 2. Prior Performance 3. Star Analysts 4. Brokerage Size H2 H3 H4 H5 H1 Abnormal returns Figure 2: Event Study Timeline Estimation Window Event Window t -260 t -11 t -10 t 0 t 10 Recommendation 17

18 Average Abnormal Returns Proceedings of the First European Academic Research Conference on Global Business, Economics, Finance and Figure3: Average Abnormal Returns for Buy Recommendations 0.35% 0.32% 0.30% 0.25% 0.20% 0.20% 0.22% 0.16% 0.15% 0.10% 0.05% 0.00% -0.05% -0.10% 0.08% 0.05% 0.01% 0.02% 0.02% 0.03% 0.04% 0.02% 0.00% 0.00% % % % -0.03% -0.04% -0.06% -0.06% Days 18

19 Cumulative Average Abnormal Returns Proceedings of the First European Academic Research Conference on Global Business, Economics, Finance and 1.40% Figure 4: Cumulative Average Abnormal Returns for Buy Recommendations 1.20% 1.00% 0.94% 1.15% 1.13% 1.10% 1.08% 1.04% 1.03% 1.00% 1.00% 0.95% 0.89% 0.80% 0.63% 0.60% 0.40% 0.20% 0.00% 0.08% 0.10% 0.13% 0.01% 0.03% 0.03% 0.05% 0.22% 0.41% Days 19

20 Panel A. Recommendations Table 1: Summary Statistics Buy (Strong Buy and Buy) Hold Sell (Strong Sell and Underperform) Total Year N % N % N % Total Panel B. Analysts per Covered Firms per Listed Covered % of Covered Firm Analyst firms firms Total Year Mean Median Mean Median Brokers Analysts Average All Years Average Rating 20

21 Social Sciences (EAR15Italy Conference) ISBN: Table 2: Market Reaction to Buy Recommendations Panel A. Market reaction over 20 days around buy recommendations AR CAR Day Mean t-test Sign test Sign rank Mean t-test Sign test Sign rank % *** *** 0.01% *** *** % * *** *** 0.03% *** *** % *** *** 0.03% *** *** % * *** *** 0.05% * *** *** % ** *** *** 0.08% ** *** *** % ** *** *** 0.10% *** *** *** % ** *** *** 0.13% *** *** *** % *** *** *** 0.22% *** *** *** % *** *** *** 0.41% *** *** *** % *** *** *** 0.63% *** *** % *** *** *** 0.94% *** *** *** % *** *** *** 1.10% *** *** *** % *** *** *** 1.15% *** *** *** % *** *** 1.13% *** *** *** % *** *** *** 1.08% *** *** *** % *** *** *** 1.04% *** *** *** % *** *** 1.03% *** *** *** % *** *** *** 1.00% *** ** *** % *** *** 1.00% *** *** % ** *** *** 0.95% *** *** % *** *** *** 0.89% *** ** *** Panel B. Event windows to buy recommendations Days N Mean CAR Positive: Negative t-test Sign test Sign rank [-10,-2] % 32786: *** *** *** [-1,1] % 34149: *** *** *** [2,10] % 29139: ** *** *** [-10,10] % 33228: *** ** *** *p <.05, **p <.01, ***p <

22 Social Sciences (EAR15Italy Conference) ISBN: Days Groups Mean CAR Table 3: Results of Portfolio Complexity by Firm Number Positive: Negative t-test Sign test Sign rank [-10,-2] High 0.29% 16066: *** *** [-10,-2] Low 0.54% 16712: *** *** DIFF (H minus L) -0.25% * (0.0605) [-1,1] High 0.66% 17401: *** *** *** [-1,1] Low 0.73% 16749: *** *** *** DIFF (H minus L) -0.07% (0.2792) [2,10] High -0.31% 14836: ** *** *** [2,10] Low -0.10% 14297: *** *** DIFF (H minus L) -0.21% *** (0.2231) [-10,10] High 0.64% 16923: ** *** [-10,10] Low 1.15% 16284: *** *** DIFF (H minus L) -0.51% * (0.0913) *p <.05, **p <.01, ***p <

23 Social Sciences (EAR15Italy Conference) ISBN: Days Groups Mean CAR Table 4: Results of Prior Performance by Analyst Positive: Negative t-test Sign test Sign rank [-10,-2] High 0.32% 14637: *** *** [-10,-2] Low 0.67% 13292: *** *** DIFF (H minus L) -0.34% ** (0.0187) [-1,1] High 0.68% 15203: *** *** *** [-1,1] Low 0.75% 13738: *** *** *** DIFF (H minus L) -0.07% (0.3018) [2,10] High -0.39% 14836: ** *** *** [2,10] Low 0.06% 14297: *** *** DIFF (H minus L) -0.45% ** (0.0188) [-10,10] High 0.61% 14762: ** *** [-10,10] Low 1.47% 13552: *** *** *** DIFF (H minus L) -0.86% ** (0.0137) *p <.05, **p <.01, ***p <

24 Social Sciences (EAR15Italy Conference) ISBN: Days Groups Mean CAR Table 5: Results of Star Analyst Positive: Negative t-test Sign test Sign rank [-10,-2] Star 0.65% 5492: *** *** [-10,-2] Non-Star 0.37% 27293: *** *** *** DIFF (H minus L) 0.29% (0.1624) [-1,1] Star 1.00% 5835: *** *** *** [-1,1] Non-Star 0.63% 28314: *** *** *** DIFF (H minus L) 0.36% *** (0.0000) [2,10] Star 0.09% 4885: *** *** [2,10] Non-Star -0.26% 24252: *** *** *** DIFF (H minus L) 0.35% (0.1092) [-10,10] Star 1.72% 5697: *** *** *** [-10,10] Non-Star 0.73% 27528: *** *** *** DIFF (H minus L) 0.99% ** (0.0349) *p <.05, **p <.01, ***p <

25 Social Sciences (EAR15Italy Conference) ISBN: Days Groups Mean CAR Table 6: Results of Brokerage Size Positive: Negative t-test Sign test Sign rank [-10,-2] Large 0.45% 16145: *** *** *** [-10,-2] Small 0.39% 16069: *** *** *** DIFF (L minus S) 0.06% (0.6298) [-1,1] Large 0.82% 17053: *** *** *** [-1,1] Small 0.58% 16484: *** *** *** DIFF (L minus S) 0.24% *** (0.0000) [2,10] Large -0.22% 14322: *** *** [2,10] Small -0.19% 14305: *** *** *** DIFF (L minus S) -0.03% (0.8812) [-10,10] Large 1.03% 16429: *** *** [-10,10] Small 0.76% 16185: *** *** *** DIFF (L minus S) 0.27% (0.3893) *p <.05, **p <.01, ***p <