1. Top of page
  2. Abstract
  5. EXHIBIT 1. Analyst Rating Classification Systems

The financial well-being of retail investors is impacted by the quality of their investment decisions. Inaccurate or misleading financial information that is misconstrued by investors to be reliable can compromise decision making. This research reports on the results of three studies that show despite the fact that equities with “buy” ratings significantly underperform equities with “hold” ratings, retail investors rely on them when making investment decisions. It also shows analysts' guidance remains inaccurate in the aggregate despite the passage of Sarbanes-Oxley and related legislation/regulation. This article begins a conversation on the implications of this dilemma, specifically the value of affirmative disclosure as a remedy.

Sell-side security analysts play a significant marketing role in brokerage firms. Analysts influence the reputation and status of brokerage firms which, in turn, aids in the recruitment and retention of clients (Brennan and Hughes 1991; Chung 2000; Ribstein 2005). As with any firm, satisfied clients are important to the long-term success of brokerage firms. Client belief in the efficacy of their brokerage firm's analysts is an important factor in their satisfaction (Teoh and Wong 2002). The result of “expert” analysis, analyst reports, and recommendations are widely disseminated as a part of the research services offered by brokerage firms and the financial media. Given their availability, analyst reports also play a critical role in individual investors' process of collecting, interpreting, and analyzing the performance prospects of firms (Cornell 2001). Besides providing raw information, analyst buy–sell–hold recommendations directly facilitate investor decisions (Teoh and Wong 2002).

Securities analysts' exact enormous influence over both retail (unsophisticated) and institutional (sophisticated) investors. Retail investors lacking the requisite financial literacy and investment experience are all too often relegated to an investment strategy of “following the smart money,” which relies heavily on imitating institutional trades and acting on analysts' advice (Damodaran 2003). As a larger segment of retail investors engage in self-directed investments, regulators have become increasingly concerned about the conflict of interest (biases) underlying the issuance of suboptimal investment advice/opinion to unsophisticated retail investors (Mikhail, Walther, and Willis 2004).

Sophisticated and unsophisticated investors, alike, can retrieve investment information through various channels including annual reports, magazines, newspapers, newsletters, blogs, websites, and TV. While the public's desire to earn returns greater than that of the broader market indices proves the catalyst in “making a market” for investment advice, the continued efficacy of the securities analyst market relies on the perception that the discovery and disclosure implicit in sell-side analyst reports is useful in selecting a portfolio capable of generating positive returns (Brody and Rees 1996). Analysts are suggested to assist in the price discovery process by assimilating complicated information including macro-economic externalities, industry-level competitive dynamics as well as quarterly and annual firm-level financial guidance. Since analysts cover specific sectors/industries and a limited number of firms therein, the breadth of industry-specific knowledge and depth of firm-specific expertise affords analysts the privilege of sharing investment advice/opinion through reports including earnings forecasts, price targets and buy–sell–hold recommendations (Wang 2009).

The value-relevance, objectivity, and accuracy of analysts' recommendations have long been a subject of debate (Cowles 1933). More recent research suggests that consensus analyst recommendations do not consistently lead to superior net returns as compared to market averages (Barber et al. 2001, 2003). More disturbingly, there is evidence that analysts are subject to incentive and heuristic-based biases, which can lead them to recommend equities for reasons other than benefiting investors who rely on their objectivity (Michaely and Womack 1999). When the percentage of analysts' “buy” vs. “sell” recommendations reached 74% and 2%, respectively by mid-2000, investors began to voice concerns that analysts' reports were neither a reflection of the analysts' true sentiment nor a material representation of a firms intrinsic value (Barber et al. 2006). In response, regulators began to take a more proactive role in the oversight of financial market research.

Consistent with its stated mission “to protect investors and maintain the integrity of securities markets,” the SEC adopted Reg. FD in October 2000, NASD Rule 2711 and NYSE Rule 472 in May 2002 (Barber et al. 2006) prior to settling with 10 of the largest investment banks for $875 million in April 2003 over allegations that in-house analysts compromised the integrity of their research in an effort to attract investment banking business (Barber, Lehavy, and Trueman 2007). This research further addresses the issue of whether the SEC, alone, or in conjunction with the FTC, per the 1983 FTC Policy Statement on Deception (FTC 1983), has a case in requiring brokerage firms to disclose the performance record of their analysts' recommendations to investors. If investors are influenced by analyst recommendations and if the relationship of analyst buy–sell–hold recommendations with investment returns is tenuous, then affirmative disclosure is arguably reasonable.

The dynamics of the relationship between brokers-analysts and investors is well defined and documented within the political science and economics literatures. The principal-agent theory addresses the contractual motivations between one person-party (agent) making decisions that influence outcomes for a second person-party (principal) in return for some form of payment. The term “contractual” is to be interpreted very broadly as it may refer to a formal document, an implicit contract, or some reward system that is not formally contractual (Rees 1985). While the principal-agent theory has been widely utilized to analyze the cause of potentially inefficient economic outcomes between two parties, the current research is focused on the effect of said inefficient outcomes. In analyzing the effects and outcomes, we will present evidence to adopt affirmative disclosure as a plausible resolution to the informational asymmetry and selective disclosure policies plaguing the relationship between securities analysts and investors, as well as the outcomes internalized by the latter.

There is a dearth of research that examines the accuracy of analyst recommendations from the point of view of the unsophisticated investor. Extant studies looking at analysts' recommendations tend to employ methodologies that impose trading requirements that are onerous to unsophisticated investors (Barber et al. 2001, 2003); examine a small subset of analyst recommendations (Groth et al. 1979) or a particular type of equity rating (Bidwell and Kolb 1980); and/or focus on the influence of analyst ratings on equity prices over time rather than their predictive accuracy (Barber et al. 2001; Womack 1996). This article addresses a gap in the literature by focusing solely on the difference in market-adjusted returns of equities with buy, hold, and sell recommendations, and by comparing analyst performance with investor expectations of analyst performance. No other research of which we are aware has looked at the disparity between actual and perceived analyst performance.

The fundamental empirical question we address is whether analysts' buy, hold and sell recommendations accurately predict (not influence) the relative performance of equities. We also examine the degree to which the accuracy of analyst recommendations has improved since the passage of regulatory actions, most notably Sarbanes-Oxley, which address the integrity of the information discovery and disclosure processes underlying analyst recommendations. Three empirical studies address the issue of affirmative disclosure. Study 1 examines the relationship between analyst recommendations and equity performance in the Dow Jones Industrial Average (DJIA) from January 1999 to May 2013. Study 2 looks at the relationship between analyst recommendations and equity performance in the S&P 500 technology sector from January 1999 to January 2006. Study 3 investigates the degree to which retail investors rely on analyst advice/opinion relative to other information sources when buying and selling equities.

Prior to presenting the results of these studies, which appear to support a case for affirmative disclosure, we will highlight existing research surrounding sources of analyst biases, evidence of analyst accuracy, and the regulatory and legal issues relevant to analyst recommendations and investor interests. The purpose of this discussion is to create a context to understand the issue at hand and to enumerate potential explanations for poor analyst performance. Our empirical studies, however, do not address cause; they are designed to determine whether a gap exists between analyst performance and investor expectations of analyst performance. The presence of such a gap is sufficient in our view to begin a dialogue about the merit of affirmative disclosure.


  1. Top of page
  2. Abstract
  5. EXHIBIT 1. Analyst Rating Classification Systems

Factors that Influence Analyst Behavior

There are several contributing factors to the likelihood that analysts do not accurately predict equity performance through the issuance of “buy,” “hold,” and “sell” recommendations. First, analysts may either overestimate their own abilities or they may apply faulty logic in their decision making, or “herd” on popular opinion/sentiment. Second, analysts can distort guidance in an effort to either curry favor with managers of covered firms or grow in-house investment banking business and commission-based trading revenues. Third, sell-side analysts can be beholden to the portfolio interests of buy-side clients. Finally, but certainly not unique to the fraternity of stock analysts, conflicts of interest can arise due to compensation-based incentives, professional reputation, and career concerns. Each of these will be examined in more detail.

Numerous studies contend that analysts become overconfident and overweight their ability to identify superior investment opportunities (Massey and Thaler 2005; Odean 1998; Stotz and von Nitzch 2005). Analysts tend to overestimate the likelihood that positive past performance predicts positive future performance (DeBondt and Thaler 1985; Shefrin and Statman 1995; Solt and Statman 1989) while also proving far more likely to follow popular stocks that are highly rated by Standard and Poor's (Chung 2000; Jegadeesh and Kim 2006). Some analysts have also been accused of ignoring their private opinion in favor of “herding” or mimicking the sentiment of other analysts (Clement and Tse 2005; Cote and Goodstein 1999; Lin, Chen, and Chen 2010) while also issuing favorable/negative stock recommendations when investor sentiment is bullish/bearish (Franzzini and Lamont 2008). Paradoxically, existing literature also suggests that securities analysts are traditionally likely to rely on long-term growth (LTG) and price-to-earnings growth (PEG) heuristics to support stock guidance (Jegadeesh et al. 2004) despite evidence suggesting that the residual income model (RIM) is a better predictor of future performance (Bradshaw 2004).

While analysts exhibit a differential ability in providing accurate guidance with accuracy proving persistent over time (Li 2005; Mikhail, Walther, and Willis 2004), many analysts are believed to ignore their own research/opinion in favor of compliance with popular sentiment. Falling in line with the consensus of other analysts, herding, has been shown to provide investors no informational value in the future pricing of securities (Clement and Tse 2005; Cote and Goodstein 1999). Meanwhile, analysts whose stock guidance is positively correlated with investor sentiment also tend to issue less profitable guidance than their peers (Baker and Wurgler 2006; Frazzini and Lamont 2008).

Securities analysts can be pressured by in-house management to provide overly optimistic guidance on stocks covered by the firm. Buy recommendations have historically generated more trading business (Hayes 1998; Irvine 2001, 2004; Jackson 2005; Reuter 2006) as well as underwriting business (Dugar and Nathan 1995; Hunton and McEwen 1997; Ljungqvist, Marston, and Wilhelm 2006; Michaely and Womack 1999; Rajan and Servaes 1997) than sell recommendations. In-house managers are known to apply pressure on staff analysts to artificially inflate stock guidance, on which unsophisticated retail investors often rely (Malmendier and Shanthikumar 2007a, 2007b), for covered firms. Conversely, analysts have also been accused of supporting a strategic downward bias on earnings guidance on which sophisticated institutional investors often rely, in an effort to ensure covered firms beat quarterly expectations while also ensuring analysts retain access to covered firm managerial information (Libby et al. 2008; Lim 2001).

Managers of covered firms are known to exert influence by threatening to “freeze out” analysts promoting stock guidance deemed “too low” (Francis, Hanna, and Philbrick 1997). Similarly, buy-side clients (mutual funds, pensions, etc.) exert influence over sell-side analysts covering those stocks for which their portfolios are heavily weighted (Womack 1996). Conceding that sophisticated institutional investors are better equipped to decode the systematic biases (Jackson 2005) underlying both optimistic stock recommendations and pessimistic earnings forecasts than their unsophisticated retail counterparts, it has been argued that securities analysts communicate to different types of investors in “different tongues” (Malmendier and Shanthikumar 2007a).

Affiliated analysts succumb to professional incentives to issue overly optimistic (vs. pessimistic) stock guidance as future promotion (vs. demotion) is more dependent on “falling in line” with forecasts than being accurate (Hong and Kubik 2003; Lim 2001). According to Ljungqvist, Marston, and Wilhelm (2006), analysts are less likely to succumb to guidance pressure on covered stocks that are highly visible to sophisticated institutional investors for fear of jeopardizing their reputation (Mola and Guidolin 2009) and career prospects (Hong and Kubik 2003; Hong, Kubik, and Solomon 2000).

Analyst Accuracy

Barber et al. (2001) analyzed 360,000 recommendations from 269 brokerage firms and 4,340 analysts over an 11-year period, starting in 1985. They reported greater accuracy in the recommendations associated with small- to medium-sized firms than with larger firms and attributed the performance gap to the lower availability of public information for small- and medium-sized firms. However, after accounting for the transaction costs, investors following analyst guidance for these firms failed to generate abnormal superior returns (see also Jegadeesh et al. 2004; Womack 1996). In a follow-up study tracking performance from 1996 to 2001, Barber et al. (2003) highlight the correlation between the desire to retain-attract investment banking clients and analysts' tendency to cover growth stocks vs. value stocks.

Although industry experience may be a contributor to accuracy (Clement and Tse 2005), empirical evidence suggests that stock guidance issued by “affiliated” analysts is more favorable but proves less accurate over both the short- and long-term horizons (Lin and McNichols 1998; Michaely and Womack 1999). Additionally, affiliated analysts are both slower to downgrade stocks from buy to hold and faster to upgrade from hold to buy (O'Brien, McNichols, and Hsiou-Wei 2005).

Analysts may be truly optimistic/pessimistic about covered firms or they may distort guidance strategically due to institutional pressure from in-house management and/or buy-side clients. If over-optimism underlies analysts' bias, we should expect analysts to express optimism in both recommendations and earnings forecasts. This may not be the case. There is evidence to suggest that analysts target unsophisticated retail investors with bullish recommendations while also targeting sophisticated institutional investors with bearish earnings forecasts. This evidence has led pundits to allege that securities analysts strategically distort in accordance with the degree of financial literacy (Malmendier and Shanthikumar 2007b).

Financial literacy is suggested to affect both direct portfolio choices as well as one's choice of financial information/advisor (van Rooij, Lusardi, and Alessie 2007). Naïve retail investors tend to follow analyst signals (buy, sell, or hold) quite literally while institutional investors are sophisticated enough to scrutinize biased analyst guidance (Malmendier and Shanthikumar 2007b). In addition to supporting the notion that individual investors are less sophisticated processors of complex information, Mikhail, Walther, and Willis (2004) find evidence that retail investors are both more likely to blindly follow analyst guidance and more likely to incur negative returns as a result. Recognizing that the “average” retail investor lacks the requisite skill to integrate and process all of the complex financial information available to him/her and, as a result, at least partially relies on the financial advice published within the “expert” analyst community, we aim to provide evidence supporting the merit of affirmative disclosure as a regulatory safeguard for investors, retail investors in particular.

Regulatory Actions and Analyst Behavior

Concern over the independence and objectivity of sell-side analysts led to a series of regulatory initiatives designed to restore analyst credibility. In October 2000, the SEC enacted Regulation Fair Disclosure (Reg. FD) to eliminate selective disclosure of material investment information among preferred analysts and institutional clients. By eliminating this practice of selective disclosure, Reg. FD sought to ensure unsophisticated retail investors timely access to the same material information that had previously been the protected intellectual property of sophisticated institutional investors (Eleswarapu, Thompson, and Venkataraman 2004).

The Sarbanes-Oxley Act was approved in July 2002 as yet another legislative attempt to strengthen public confidence in securities research and stock guidance. In the name of independence and objectivity, Sarbanes-Oxley sought to segregate the interests of securities analysts' from in-house investment-banking units tasked with procuring underwriting and trade-related revenues. In July 2003, Sarbanes-Oxley was revised to include Section 501, which required the SEC: (1) limit the ability of investment bankers to approve analysts' reports, (2) limit the ability of a firm's investment banking division to influence the compensation of securities analysts, and (3) eradicate internal retaliation by investment bankers against analysts whose reports could harm investment banking business. Clearly, these three measures aimed to moderate the otherwise incestuous business relationship between the analysts and investment banking community (Ebinger 2008).

In yet another attempt to more closely regulate analyst research, the National Association of Securities Dealers (NASD) proposed Rule 2711 in 2002. NASD Rule 2711 explicitly prohibits sell-side analysts from being under the supervision or control of investment banking while also prohibiting investment banking from “approving” sell-side analyst reports due to the potential for conflict of interest.

Targeting a similar enhanced regulatory standard to that of NASD Rule 2711, the New York Stock Exchange (NYSE) introduced a modification of Rule 472 in an effort to restrict communication between investment banking and research functions. Rule 472 requires analysts to disclose financial interests in recommended securities, provide ratings distributions assigned to a given stock over the prior 12 months, and disclose the number of buy–hold–sell ratings assigned to all covered stocks. Additionally, in April 2003, the SEC adopted the Analyst Certification Rule requiring all research reports disclose any compensation received by the analyst while certifying that the enclosed guidance reflects the analyst's personal views (Mola and Guidolin 2009).

In April 2003, the SEC disclosed a historic and punitive agreement with ten of the largest investment banks, dubbed the Global Analyst Research Settlement. In light of evidence highlighting the proclivity of affiliated analysts to compromise the integrity of their research to secure investment-banking business, this agreement ultimately confiscated $875 million in penalties and profits from these 10 targeted financial firms. Additionally, this agreement requires investment banks to both separate their investment banking and research departments and to provide independent securities research to clients in order to ensure access to objective advice (Barber et al. 2007).

In light of this recent regulatory intervention, it is interesting to note that the number of buy/hold/sell recommendations as a percentage of total recommendation decreased from 74/24/2 percent in 2000 to 42/41/17 percent by the end of 2003 (Barber et al. 2006). Despite this evidence suggesting that analyst guidance became less optimistic, the literature is mixed with respect to the impact of SEC and NYSE regulatory enhancements. Barber et al. (2001, 2006) suggest that the implementation of NASD Rule 2711 significantly reduces the prevalence of buy recommendations in line with the intended regulatory effect. Conversely, Howe, Unlu, and Xeumin (2009) assert that aggregate analyst recommendations are not attributable to regulatory changes.

Existing literature supports the idea that Reg. FD has inspired timely access to investment information amongst institutional and retail investors alike (Brown, Hilligeist, and Lo 2004; Eleswarapu, Thompson, and Venkataraman 2004). However, the literature is once again mixed with respect to the impact of Reg. FD on analysts' accuracy. While Kwag and Small (2007) suggest that analysts have issued less accurate earnings estimates post Reg. FD, Irani and Karamanou (2004) find that analysts were more accurate post Reg. FD.

As this review discusses, there are serious questions surrounding the issue of analyst accuracy. The extent to which unsophisticated investors place their trust in these recommendations is also important to understand from a public policy perspective. These are issues we now address empirically.


  1. Top of page
  2. Abstract
  5. EXHIBIT 1. Analyst Rating Classification Systems

Studies 1 and 2 (below) assess the accuracy of individual brokerage recommendations over the life of said recommendations. The studies are not looking at the consensus or average analyst recommendation for equities as other studies, e.g., Barber et al. 2001. The unit of analysis is thus not the performance of individual equities partitioned into groups based on their consensus analyst recommendations, but the performance of individual analyst recommendations over the life of the recommendation.

The following assumptions were made when designing Studies 1 and 2. First, it is presumed that as long as an analyst maintains a rating on an equity, he/she is affirming that rating. Second, it is presumed that an analyst takes into account a given equity's relative risk when both making a recommendation and deciding when to change that recommendation. The purpose of a recommendation is to assert the future prospects of an equity relative to other equities. Certainly, the risk profile of an equity is a factor, or should be a factor, when making a recommendation. For this reason, our methodology presumes risk is factored into the rating.

The dependent variable in the studies is the price of each stock, tracked over the time period of the study. Data on stock prices are obtained from the historical stock prices calculated and reported in the Yahoo Finance website platform. These prices are adjusted for dividends and stock splits. The primary independent variable in the study is the analyst rating of each stock included in the study (i.e., buy, sell, or hold). These data are reported by, a financial news and analysis firm, and can be found in the analyst opinion section of company reports in the Yahoo Finance website. For each brokerage firm covering an equity, reports the analyst upgrade and downgrade history including the date that a rating was initiated and the date that the rating terminated.

Exhibit 1 portrays the ratings schemes of eight of the brokerage firms with the most rating assignments reported in As depicted, there is wide variance in how brokerage firms assign ratings. Given the variance in the rating classifications, a simplified three-level classification system was created. The top level, “Positive,” included strong buy, buy, accumulate, overweight, outperform, and any other designation that indicated a relative preference for the equity. The second level, “Neutral,” included hold, maintain, neutral, market perform, in-line, and any other designation that indicated no relative preference. The third level, “Negative,” included sell, reduce, unattractive, underweight, underperform, and any other designation that indicated a relative negative sentiment.

While firms typically had only one rating classification to indicate both neutral and negative sentiment, many firms utilized two “buy” designations to differentiate between preference and “strong” preference. The proposed classification scheme effectively collapsed both levels of preference into a single relative preference category. This aggregation was necessary because there was no way to decide whether a “buy” rating from a brokerage firm using a three-level classification system (buy, sell, hold) should be coded as a “strong buy” or a “buy.”

EXHIBIT 1. Analyst Rating Classification Systems

  1. Top of page
  2. Abstract
  5. EXHIBIT 1. Analyst Rating Classification Systems
1. Strong Buy1. Buy1. Strong Buy1. Buy
2. Buy2. Accumulate2. Buy2. Accumulate
3. Hold3. Maintain Position3. Neutral3. Neutral
4. Sell4. Reduce4. Sell4. Sell
1. Strong Buy1. Strong Buy1. Strong Buy1. Overweight
2. Buy2. Buy2. Buy2. Neutral
3. Neutral3. Market Perform3. Hold3. Underweight
4. Sell4. Market Underperform4. Sell 
1. Strong Buy1. Strong Buy1. Buy1. Outperform
2. Buy2. Buy2. Hold2. In-line
3. Hold3. Market Perform3. Sell3. Underperform
4. Sell4. Market Underperform  
1. Buy1. Overweight1. Buy1. Strong Buy
2. Long Term Buy2. Neutral2. Neutral2. Buy
3. Market Perform3. Underweight3. Reduce3. Hold
   4. Reduce

Study 1


Study 1 examines analyst recommendations for the 30 firms that comprise the Dow Jones Industrial Average (DJIA). The DJIA is designed to be representative of the US large-cap market less transportation and utilities, with the firms included in the index periodically updated in order to maintain representativeness. The firms comprising the index were tracked from January 1999 to May 2013; within this timeframe, three specific periods were tracked. The period prior to the passage of Sarbanes-Oxley (January 1999 through July 2002); the period from initial passage of Sarbanes-Oxley to the lows of the 2008 market collapse (August 2002 to March 2008); and the period from the market lows to May 2013 (April 2008 to May 2013). The first timeframe after the passage of Sarbanes-Oxley represents a volatile period including bull and bear markets and, of course, the market collapse of 2008. Its unusual characteristics arguably made market predictions difficult and could be used to explain poor analyst performance. The second timeframe after the passage of Sarbanes-Oxley, however, reflects a long-term bull market recovery punctuated by a few brief down turns. This market is not characterized by unusual characteristics and, thus, should not have presented analysts with unique circumstances.

The dependent variable in the analysis is a measure of the change in the equity's stock price, created by combining two price changes. The first is the percentage change in the value of the stock price from the time a rating was initiated to the time it was closed. The second is the percentage change in the S&P 500 over the exact same time period. The S&P 500 is an index comprised of the 500 firms listed on the NYSE and NASDAQ with the largest market capitalization. The index is market-capitalization-weighted whereby firms with larger market capitalizations have a greater influence on the index.

The dependent variable was calculated by subtracting the return of the S&P 500 from the return of the equity. This approach controlled for the overall performance of the stock market during the rating time period. For example, if a stock had a return of 9.0% during a rating time period and the S&P 500 also had a 9.0% return during the rating time period, the dependent variable was coded as 0%. Inherent in this approach is the presumption that the skill of an analyst is determined by his/her ability to steer investors towards stocks that will have greater than average returns and away from stocks that will have below average returns.

The primary independent variable in the study is the analyst rating of the stock (i.e., buy, sell, or hold). The second independent variable is the timeframe of the study relative to the passage of Sarbanes-Oxley in July of 2002. Time periods were coded as either pre-Sarbanes (rating initiated and closed between January 1999 and July 2002), post-Sarbanes I (rating initiated and closed between August 2002 and February 2008) or post-Sarbanes II (rating initiated and closed between March 2008 and May 2013). The purpose of this coding is to assess whether the efficacy of analyst stock recommendations changed after the passage of Sarbanes-Oxley. A final control is whether each rating represented an upgrade, downgrade, or no change.

Usable observations were those rating assignments that had identifiable start dates and stop dates, and for which historical stock prices could be clearly tracked. Further, to be included in the analysis, the start and stop dates of a recommendation had to fall wholly within one of the three specific time periods noted above.

The DJIA analysis produced 1,381 usable observations. Analyst ratings from 69 brokerage firms were included in the analysis. Ratings from the 10 brokerage firms with the greatest number of rating assignments comprised 58.9% of the observations. The average length of time any equity remained at a given rating was 342.9 days, with a standard deviation of 329.1 days.


A GLM regression analysis using the independent and control variables as classification variables was conducted rather than an ANOVA because the design was nonorthogonal. In accordance with the trends in ratings noted in early discussion, only 6.3% of analyst ratings in the sample were “negative”; 51.6% and 42.1% were “positive” and “neutral,” respectively. Due to the presence of extreme outliers in the data, three separate analyses are reported to assess the performance of equities with positive, neutral and negative ratings. The secondary analyses were conducted to determine if results could be attributed to outliers. The full sample analysis is reported in Table 1. The second and third analyses eliminated observations beyond 2.5 and 2.0 standard deviations from the dependent variable mean, respectively. This eliminated 18 and 70 extreme observations, respectively.

Table 1. GLM Regression Results Full Sample (Dependent Variable: Equity Performance)
 DJIA Analysis (N = 1,381)S&P 500 Technology Sector Analysis (N = 3,092)
 dfF ValuePr > FdfF ValuePr > F
Full model103.76.00171.92.062
Independent variables      
Analyst rating (AR)24.16.01623.80.022
Sarbanes (S)21.93.14611.70.192
AR × S41.65.15820.75.472
Change in rating20.13.88123.62.027

As reported in Table 1, the full sample analysis produced a significant model, F(10, 1370) = 3.76, p < .016. There was a significant difference in the performance of equities across ratings, F(2,1368) = 4.16, p < .016, but there was no significant difference in performance across the three time periods. There was also no interaction between rating performance ratings and the three time periods, which indicates no change in analyst performance patterns before and after the passage of Sarbanes-Oxley.

As Table 2 reports, there was no evidence that analyst recommendations reliably predicted equity performance. In all three analyses reported in Table 2, the performance of equities with positive ratings lagged equities with neutral and, surprisingly, negative ratings. In each analysis, the difference in performance between equities with positive and neutral ratings was statistically significant (p < .05). Equities with neutral ratings outperformed equities with positive ratings. Controlling for outliers made no difference in the results. As can be seen, these findings run contrary to any expectation that equities with buy ratings outperform equities with hold or sell ratings.

Table 2. DJIA Analysis Main Effects Analysis (Dependent Variable: Equity Performance)
 Descriptive StatisticsStock Returns by Analyst Rating
 Sample SizeMean ReturnSDLowHighVariable F-TestPositive RatingNeutral RatingNegative Rating
  1. Notes: Mean percentages with different superscripts are significantly different from one another (p < .05). Sample sizes for positive ratings were 713, 703, 672, respectively; for neutral ratings were 582, 575 and 557, respectively, and for negative ratings were 86, 85, 82, respectively.

Full sample1,3810.9%22.7%−151.0%127.7%4.16p < .016  −.3%12.5%21.1%1,2
Sample within 2.5 SD1,3631.0%19.9% −61.7% 78.8%3.25p < .039−.2%12.0%22.2%,1,2
Sample Within 2 SD1,3111.1%16.6% −42.9% 48.3%3.14p < .044  .2%12.1%21.6%1,2

As Table 3 reports, the pattern of results did not change across the three time periods. In each time period, equities with neutral ratings outperformed equities with positive ratings. Although the rating by time period interaction, as reported above, was not significant, the gap between equities with buy and hold ratings trended larger after passage of Sarbanes-Oxley, which is contrary to the purpose of the regulatory actions. This suggests that the passage of Sarbanes-Oxley has had no discernable aggregate impact on the quality of analyst stock recommendations. Equities with “neutral” ratings continued to perform at least as well, in the aggregate, as equities with “positive” ratings.

Table 3. Pre-Post Sarbanes Interaction Analysis
 DJIA*S&P 500*
 “Positive” Rating“Neutral” Rating“P”−“N” Difference“Positive” Rating“Neutral” Rating“P”−“N” Difference
  1. Notes: Negative recommendations were removed due to small sample sizes. Since the performance of equities with negative recommendations was less extreme than those with hold recommendations their removal would not impact significance tests. Sample size for DJIA analysis was 1,381; sample size for S&P 500 analysis was 3,244.

  2. *Mean percentages with different superscripts are significantly different from one another (p < .05). #Mean difference that is marginally significant (p < .10).

Post-Sarbanes I−3.4%1−.9%2−2.5%5.6%112.1%2−6.6%
Post-Sarbanes II−5.6%15.0%2,#−10.6%   

Study 21


Methods of Study 2 closely mirror Study 1, except that it tracks the performance of analyst recommendations for companies within a single sector of the economy, and the time frame is truncated to two-and-a-half years on either side of the passage of Sarbanes-Oxley (January 1999 to January 2006). S&P 500 firms are classified into one of 10 sectors: financials, industrials, technology, energy, consumer discretionary, consumer staple, health care, materials, utilities, and telecommunications; we chose to utilize the firms that comprise the S&P technology sector in our analysis as it contains some of the most dynamic, visible, and monitored equities (e.g., Google, Apple, Microsoft, Cisco Systems, Oracle).

As in Study 1, the primary independent variable was the analyst rating of the stock (i.e., positive, neutral, or negative) and the same criteria were used to assign rating codes. Also, as in Study 1, the change of status in the rating was employed as a covariate. A similar dependent variable was created using two measures. The first was the percentage change in the value of the stock price from the time a rating was initiated to the time it ceased. The second was the percentage change in the S&P 500 technology sector over the exact same time period. The dependent variable was calculated by subtracting the return of the S&P 500 technology sector from the return of the rated stock. This approach controlled for the overall performance of the technology sector over the proposed time period.

Usable observations were those rating assignments that had clearly identifiable start dates and stop dates, and for which historical stock prices could be clearly tracked. Only the analyst rating cycles that were fully contained within one of the two time periods (pre-Sarbanes and post-Sarbanes) are included in the analysis. The sample consists of 3,092 usable observations. The average length of time that any stock remained at any given rating was 231.5, days with a standard deviation of 212.5 days and range of 1 to 1,898 days.


As reported in Table 4, the full sample analysis produced a marginally significant model, F(7, 3084) = 1.92, p < .062. There was a significant difference in the performance of equities across ratings, F(2, 3084) = 3.80, p < .022, but there was no significant difference in performance across the three time periods. As with the study of the DJIA, a second and third analysis was conducted to assess analyst performance across ratings after extreme outliers were removed. In these analyses, observations beyond 2.5 and 2.0 standard deviations from the mean were removed. This eliminated 41 and 142 extreme observations, respectively. As in Study 1 (Table 4), in both outlier analyses equities with neutral ratings outperformed equities with positive ratings (p < .05).

Table 4. S&P 500 Analysis Results
 Descriptive StatisticsStock Returns by Analyst Rating*
 Sample sizeMean ReturnSDLowHighModel F-TestPositive RatingNeutral RatingNegative Rating
  1. Notes: Sample sizes for positive ratings were 1,914, 1883, 1804, respectively; for neutral were 982, 974 and 959, respectively; and for negative were 196, 194, 187, respectively.

  2. *Mean percentages with different superscripts are significantly different from one another (p < .05).

Full sample3,09210.9%54.5%−108.4%1358.5%3.80p < .02210.3%112.4%210.1%1, 2
Sample within 2.5 SD3,0528.8%39.6%−68.8%248.9%4.57p < .017.9%110.4%29.2%1,2
Sample within 2 SD2,9517.0%31.8%−50.0%136.1%8.77p < .0015.6%19.2%28.4%1,2

Also as in Study 1 (Table 3), there was no interaction between rating performance and the two time periods, which indicates no change in analyst performance patterns before and after the passage of Sarbanes-Oxley. As was the case in Study 1, equities with negative ratings only accounted for a small minority of the study observations (6.3%); equities with positive and neutral ratings comprised 61.9% and 31.8% of the sample, respectively. As Table 3 reports, although the interaction was not significant, the performance gap between equities with positive and neutral ratings in the post-Sarbanes time period widened from .9% (11.9% vs. 12.8%) in the pre-Sarbanes period to 6.5% (5.6% vs. 12.1%) in the post-Sarbanes period. The gap in the post-Sarbanes period is statistically significant (p < .05). This corroborates the finding from Study 1 that the passage of Sarbanes-Oxley has had no discernable positive impact on the quality of analyst stock recommendations, at least during the time period in question.


  1. Top of page
  2. Abstract
  5. EXHIBIT 1. Analyst Rating Classification Systems

Studies 1 and 2 examined whether analysts' buy, sell, and hold recommendations are accurate predictors of equity performance. Findings suggest that they are not only inaccurate, but trend in the opposite direction of expectations. Study 3 examines whether a significant proportion of retail investors use analyst recommendations as a factor in their investment decisions. It also explores whether this is based on a belief that analyst recommendations are positively correlated to stock performance.

A survey of individual investors was conducted to answer the two questions raised above with three objectives in mind. The first objective was to assess the relative influence of four different sources of financial information on investor decisions. These sources were analyst recommendations, financial newsletters, fundamental analysis (revenue growth, income growth, gross margins, PE and PEG ratios, etc.), and technical analysis (i.e., performance chart patterns). The second objective was to determine the degree to which individual investors believe that stocks with “buy” ratings outperform stocks with “hold” ratings. The third objective was to assess the extent to which investor reliance on each of the eight information sources was correlated with the individual's investment returns.


A nationally representative consumer panel managed by Market Tools, Inc., was employed to reach active retail investors. Among the demographic variables used to profile panel members was investing behavior. This permitted the creation of a targeted sample. The survey was conducted in two waves with the first wave conducted in July 2006 and the second wave conducted in May 2012. The second wave was conducted to assess whether the financial turmoil of 2008 and 2009 shifted investors' use and trust of analyst recommendations in their investment decisions. For each wave, an e-mail invitation to participate in the survey was sent to 1,700 and 1,250 targeted retail investors, respectively, with 782 and 540 investors participating (46.3% and 43.2% response rates, respectively). In an effort to qualify each respondent, all were asked a series of questions to ensure that each had: (1) actively managed personal assets through an online brokerage account within the past two years, (2) possessed individual stocks in their investment portfolio, and (3) been personally involved in the buy-sell decisions in their portfolio.

The decision to limit participation to individuals with online brokerage accounts was based on the presumption that these individuals are most likely to make self-directed investment decisions without a reliance on third-party influence. Three hundred twenty and three hundred individuals qualified to complete the first and second survey, respectively; 40.9% and 55.6% of the full pool of respondents, respectively. All qualifiers completed the survey.


To assess investor reliance on each of four types of information guidance (fundamental analysis, technical analysis, analyst recommendations, and financial newsletters), a single item measure was employed for each. This approach is supported by recent research indicating that there is no difference in the predictive validity of multiple- and single-item measures for marketing constructs representing concrete singular objects or behaviors (Bergvist and Rossiter 2007). Utilizing a 7-point scale anchored by “never use” and “always use,” respondents were asked to “rate the extent to which you review each of the information types listed below before making individual stock buy-sell decisions.” Using a similar Likert scale, investor perceptions of the efficacy of analyst recommendations were assessed by asking to agree or disagree with the statement, “on average, stocks with ‘buy’ ratings significantly outperform stocks with ‘hold’ ratings.” Two individual difference measures, age and sex, were collected as control variables. The measure of investor portfolio performance was aligned with the measure of performance used in Studies 1 and 2 to measure the degree to which equity performance may be controlled by overall market performance. Respondents were asked to assess the performance of their personal portfolio relative to market performance averages. On a 5-point Likert scale anchored by “significantly below the market average” and “significantly above the market average,” respondents were asked, “Which of the following statements best describes the financial performance of your online brokerage account(s) over the past three years?”


Table 5 reports the inter-variable correlations of the six primary variables in the study. Table 6 reports the results of three analyses designed to assess the reliance of retail investors on analyst recommendations, the relationship between analyst recommendations on portfolio performance, and the belief that equities with “buy” recommendations outperform equities with “hold” recommendations.

Table 5. Variable Correlations
1. Fundamental analysis5.101.41     
2. Technical analysis4.981.49.58    
3. Analyst recommendations4.531.58.40.44   
4. Financial newsletter4.  
5. “Buy” outperforms “Hold”4.301. 
6. Portfolio performance3.
Table 6. Retail Investor Survey Results
 Analysis One Source Influence on Investors
 Fundamental AnalysisTechnical recommendationsAnalyst newslettersFinancial analysis
 Analysis Three Relationship of Source Influence to Portfolio PerformanceAnalysis Two Relationship of Individual Factors to Investor Reliance on Analysts
 Performance Analyst Influence
 t-ValueSig. t-ValueSig.
  1. *Means with different superscript numbers differ significantly (p < .05).

Analyst Recommendation (AR)−.23p < .822“Buys” Beats “Hold” (BH)2.81p < .005
Time period (T)1.95p < .052Time (T)2.08p < .038
AR × T−.38p < .706BH × T−1.15p < .249
Fundamental analysis2.38p < .018Fundamental analysis4.05p < .001
Financial newsletter1.97p < .050Technical analysis4.63p < .001
Technical analysis.73p < .466Financial newsletter8.33p < .001
Gender−1.08p < .281Gender2.20p < .028
Age3.36p < .001Age−1.76p < .079

Analysis One in Table 6 reports the mean level of influence of each of the four financial information source constructs. A mixed model ANOVA with information type as a within-subjects variable and time (wave 1 in 2006 or wave 2 in 2012) as a between-subjects variable was conducted. Information: F(3, 616) = 84.2, p < .001, time; F(1, 616) = 10.7, p < .001 and the information–time interaction: F(1, 616) = 32.2, p < .001 were all statistically significant. The relative influence of each information type was fundamental analysis (5.10), technical analysis (4.98), analyst recommendations (4.53), and financial newsletters (4.17); each mean was significantly different from every other mean (p < .05). The overall aggregate influence of information was greater in the 2012 wave (4.86) than the 2006 wave (4.55). The interaction of information by time was a function of the significant increase in influence of analyst recommendations (4.19 vs. 4.90; t = 5.74, p < .001) and financial newsletters (3.79 vs. 4.59; t = 5.92, p < .001). There was no significant change in the influence of fundamental analysis (5.18 vs. 5.02; t = 1.45; p < .148) or technical analysis (5.03 vs. 4.92; t = .87, p < .383). The evidence suggests that retail investors attempted to diversify the type of information on which they relied after the financial traumas of 2008–2009. Analyst recommendations and financial newsletters benefited from this diversification.

While analyst recommendations were less influential than fundamental or technical analysis, their influence was substantial. Across both survey waves, 54.9% of respondents report using analyst recommendations on a consistent basis (i.e., 5 or higher on the 7-point scale) vs. 19.5% that suggested analyst recommendations provided limited value in their investment decision process (3 or lower). In the second wave survey, only fundamental analysis was rated to be a more influential type of information than analyst recommendations.

Analysis Two in Table 6 assesses the degree to which individual investors believe that stocks with “buy” ratings outperform stocks with “hold” or “sell” ratings. Using OLS regression, the dependent variable, investor reliance on analyst information, was regressed against the belief that equities with “buy” ratings outperform those with “hold” ratings, time, and their interaction. Investor age and investor gender were included as covariates as well as investor reliance on fundamental analysis, technical analysis, and financial newsletters. The model proved significant, F(7, 602) = 53.2, p < .001. As reported in Table 6, retail investors that believed that stocks with “buy” ratings outperform stocks with “hold” ratings were significantly more likely to rely on analysts when making investment decisions. On average, twice as many respondents agreed with the statement that equities with “buy” ratings outperform equities with “hold” ratings (40.5% vs. 21.0%).

This tendency to rely on analysts' recommendations did not vary by time, the time × “buy vs. hold” interaction was not significant (t = −1.15, p < .249). As also reflected in Analysis One, the positive impact of time on analyst reliance (t = 2.08, p < .038) reflects greater reliance on analysts in 2012 than in 2006. There was also a positive relationship between reliance on fundamental analysis (t = 4.05, p < .001), technical analysis (t = 4.63, p < .001), and financial newsletters (t = 8.33, p < .001) with the belief that equities with buy ratings outperform equities with hold ratings. This evidence, coupled with the moderate level of inter-correlation among each of the four information types reported in Table 5, reinforces the notion that retail investors both rely on third party information and rely on multiple sources of information, including analyst recommendations.

Analysis Three in Table 6, employing OLS regression, assesses the extent to which: (1) investor reliance on analyst recommendations is correlated to investment returns, and (2) the degree to which reliance varied between the first and second waves of the survey. The same controls employed in Analysis Two were also employed in this analysis. The dependent measure was investors' assessment of their three-year portfolio performance relative to market averages. The overall model was significant, F(8, 601) = 6.59, p < .001, but neither the use of analyst recommendations (t = −.23, p < .822) or the use of analyst recommendations by time interaction (t = −.38, p < .706) approached statistical significance. This perceptual finding is consistent with the objective findings of Studies 1 and 2, which showed that equities with positive recommendations do not outperform equities with neutral or negative ratings.

It is important to note that the simple correlation between the use of analyst recommendations and performance, as reported in Table 5 (r = .06), was not statistically significant. Additionally, there was a significant effect of time (t = 1.95, p < .052) with performance perception improving from the 2006 to the 2012 wave. Recognizing that stock market performance was much stronger from May 2009 to May 2012 (50.71% increase in S&P 500) than from July 2003 to July 2006 (23.6% increase), this time effect may be an artifact of stock market performance. Interestingly, investor usage of fundamental analysis (t = 2.38, p < .018) and financial newsletters (t = 1.97, p < .05) were positively related to investment performance, while technical analysis was not so related (t = .73, p < .466).


  1. Top of page
  2. Abstract
  5. EXHIBIT 1. Analyst Rating Classification Systems

The empirical portion of this article addressed the following questions. Are analyst recommendations sufficiently accurate to serve as a reliable tool for investors to make “buy,” “sell,” or “hold” decisions? Do a significant proportion of retail investors use analyst recommendations to guide their investment decisions? Our results suggest that the answer to the first question is “no” and to the second questions is “yes.” Based on this, we believe that (1) there is evidence that investors, naïve investors in particular, are being misled and (2) there is merit in beginning a dialogue about affirmative disclosure as an additional safeguard for protecting investors.

Analyst Recommendations and Affirmative Disclosure

As professional agents endowed with the “expert” status synonymous with information discovery, equity market efficiency, and resulting media disclosure, the financial analyst community has garnered a unique authority that investors are likely to both consider and rely on when making strategic portfolio decisions. This influence is exemplified by the short-term price movements associated with changes in analyst recommendations. The SEC's adoption of Reg. FD, NASD Rule 2711, NYSE rule 472, and Sarbanes-Oxley as well as the 1983 FTC Policy Statement on Deception (FTC 1983) recognize the potential for analyst behavior to deceive reasonable consumers. Theory on consumer behavior predicts that analysts' influence is likely to be very strong when investors do not have the motivation, ability, or opportunity to rely solely on fundamental stock analysis (Petty and Cacioppo 1981) or when investors are making decisions in what they perceive to be ambiguous situations (Hoch and Ha 1986). While most retail investors are likely to be motivated to make good investment decisions, both their ability and their opportunity (i.e., the time required) is limited relative to third party experts. The nature of the equities markets renders almost every investor decision complex and ambiguous, especially for the unsophisticated retail investor lacking the experience, education, and expertise to decipher analyst-speak. As a result, analyst recommendations are likely to be used to aid in and influence the interpretation of more specific fundamental and technical performance data. This research provides empirical confirmation that unsophisticated retail investors rely on analyst recommendations.

General consensus within the literature does suggest that the average retail investor fails to beat market averages over the long-term (Barber et al. 2001, 2003; Bidwell and Kolb 1980; Groth et al. 1979; Womack 1996). Given the incongruence between this evidence surrounding long-term market returns and investor psychology justifying the pursuit of investment returns that exceed market averages, might not all secondary investment information or tools be subject to our deception standard? No. Within all service-based professions, specifically those characterized by complexity, there exists a small segment of individuals labeled “expert” as a by-product of education, license, experiential learning, or some combination thereof. While there is no shortage of people, product, or platform-based noise within the investment marketplace, the investment analyst community is the lone constituency designated “expert” amongst peers, within financial institutions, across the media and, perhaps most importantly, according to government licensing agencies. Assuming that the “average” or “naïve” investor does rely on analyst guidance, we would argue that there must exist a heightened regulatory burden for the “experts” as compared to unsolicited investment tools and talking points.

Implications and Implementation of Affirmative Disclosure

For some firms, affirmative disclosure may provide a means to differentiate and establish relative credibility in the market. Firms with analysts able to successfully pick equities with better than average performance afford themselves the potential to market their agents and organizational capabilities to establish a competitive advantage over other firms. This is arguably good for the evolution of the industry and the well-being of its customers, particularly retail customers. For the industry as a whole, the requirement of affirmative disclosure is likely to force firms to more closely scrutinize analyst practices and the methodologies on which analysts base their recommendations. As more investors voice concern about the efficacy of financial market agents, there is certainly incentive to self-police through innovate processes. If, over time, analyst performance does not improve at the firm level, it is likely that those firms unable to perform will drop analyst recommendations altogether. This too may lead to positive changes in how firms structure investment products, counsel customers, and disseminate investment-based information.

An important issue in affirmative disclosure is the trade-off between the benefit of the disclosure and its cost. It is beyond the purview of this research and expertise of the authors to recommend the preferred framework and/or format for tracking and disclosing proprietary analyst recommendation results. That said, it seems logical to believe that brokerage houses could track the daily, monthly, quarterly, and yearly performance of in-house analysts' guidance—both open and closed. With daily opening, intraday, and closing prices instantly accessible for all publicly traded equities, the relative accuracy of all analyst recommendations can be calculated to the second. For logistical reasons, affirmative disclosure might otherwise become a quarterly expectation.

Quarterly analyst disclosures might well reference all relevant dates for each analyst's covered position (i.e., initial recommendation, historical changes in guidance, and halted positions) as well as comparative quarterly performance (both past and present quarters) relative to historical guidance and changes therein (buy, hold, sell). It seems plausible that brokerage houses would be expected to disseminate the aggregate efficacy of in-house analyst recommendations using the various traditional (TV, print, radio, etc.) and nontraditional (websites, corporate, and individual analyst blogs, etc.) channels that investors frequent. Given the expectation of regulatory oversight, the SEC might otherwise require brokerage firms submit their analyst performance records on a monthly or quarterly basis to an independent information-clearing house to ensure objectivity.


  1. Top of page
  2. Abstract
  5. EXHIBIT 1. Analyst Rating Classification Systems

Adopting the perspective of retail investors using analysts' buy, hold, and sell ratings as guidance to manage personal portfolios, we believe the evidence in this research suggests that the “expert” advice proffered by investment analysts has no predictive validity. In fact, we would argue that analyst guidance is potentially misleading given our results that highlight the peculiar inconsistency whereby equities with “hold” ratings consistently outperform equities with “buy” ratings. We believe that, taken as a whole, the evidence herein is sufficient to begin a dialogue and/or debate surrounding the merits of affirmative disclosure as a tool capable of differentiating superior vs. inferior analyst guidance that retail investors use to justify portfolio decisions.

Uniquely qualified as a result of education, preparation, and market experience, financial analysts are licensed by federal securities regulators and projected within the media as the investment experts among us. Both the financial models-metrics underlying analyst guidance and the historical accuracy of analyst guidance are objectively measurable. “Buy,” “sell,” and “hold” recommendations are not vague calls to action; they are method-driven guidance girded by complex and ostensibly objective analysis. The impact of this guidance on short-term market movements has been observed and therefore legitimizes its impact.

To be clear, this research does not represent an attempt to present conclusive proof of investor deception within the securities industries. We do, however, assert the potential for affirmative disclosure as a means for investors to better weigh the merits of analyst recommendations when making investment decisions. Recognizing that the FTC, SEC, and other regulatory authorities are tasked with ensuring integrity among all market actors, we believe it is in the best interest of all investors, particularly the average retail investor, to further explore the merit of affirmative disclosure. Ultimately, affirmative disclosure promotes transparency and represents a balancing mechanism through which all investors will better ascertain the relative and historical value of each analyst's “expert” recommendation.


  1. Top of page
  2. Abstract
  5. EXHIBIT 1. Analyst Rating Classification Systems
  • Baker, Malcolm and Jeffrey Wurgler. 2006. Investor Sentiment and the Cross-Section of Stock Returns. Journal of Finance, 61 (4): 16451680.
  • Barber, Brad, Reuven Lehavy, Maureen McNichols, and Brett Trueman. 2001. Can Investors Profit From the Prophets? Securities Analysts' Recommendations and Stock Returns. Journal of Finance, 56 (2): 531563.
  • Barber, Brad, Reuven Lehavy, Maureen McNichols, and Brett Trueman. 2003. Reassessing the Returns to Analysts' Stock Recommendations. Financial Analysts Journal, 59 (2): 8895.
  • Barber, Brad, Reuven Lehavy, Maureen McNichols, and Brett Trueman. 2006. Buys, Holds and Sells: The Distribution of Investment Banks' Stock Ratings and the Implications for the Profitability of Analysts' Recommendations. Journal of Accounting and Economics, 41 (1–2): 87117.
  • Barber, Brad, Reuven Lehavy, and Brett Trueman. 2007. Comparing the Stock Recommendation Performance of Investment Banks and Independent Research Firms. Journal of Financial Economics, 85 (2): 490517.
  • Bergkvist, Lars and John Rossiter. 2007. The Predictive Validity of Multiple-Item Versus Single Item Measures of the Same Constructs. Journal of Marketing Research, 44 (2): 175184.
  • Bidwell, Clinton M. and Robert Kolb. 1980. The Impact and Value of Broker's Sell Recommendations. The Financial Review, 15 (3): 5868.
  • Bradshaw, Mark T. 2004. How Do Analysts Use Their Earnings Forecast in Generating Stock Recommendations? The Accounting Review, 79 (1): 2550.
  • Brennan, Michael and Patricia Hughes. 1991. Stock Prices and the Supply of Information. Journal of Finance, 46 (5): 27.
  • Brody, Richard G. and Lynn Rees. 1996. The Performance of Popular Investment Magazine Stock Analysis. Journal of Applied Business Research, 12 (1): 4246.
  • Brown, Stephen, Stephen Hilligeist, and Kin Lo. 2004. Conference Calls and Information Asymmetry. Journal of Accounting and Economics, 37 (3): 343366.
  • Chung, Kee H. 2000. Marketing of Stocks by Brokerage Firms: The Role of Financial Analysts. Financial Management, 29 (2): 3554.
  • Clement, Michael and Senyo Tse. 2005. Financial Analyst Characteristics and Herding Behavior in Forecasting. Journal of Finance, 60 (1): 307341.
  • Cornell, Bradford. 2001. Is the Response of Analysts to Information Consistent With Fundamental Valuation? The Case of Intel. Journal of Financial Management, 30 (1): 113136.
  • Cote, Jane and Jerry Goodstein. 1999. A Breed Apart? Security Analysts and Herding Behavior. Journal of Business Ethics, 18 (3): 305314.
  • Cowles, Alfred. 1933. Can Stock Market Forecasters Forecast? Econometrica: Journal of Econometric Society, 1: 309324.
  • Damodaran, Aswath. 2003. Investment Philosophies: Successful Investment Philosophies and the Greatest Investors Who Made Them Work. Hoboken, NJ: John Wiley & Sons, Inc.
  • DeBondt, Warner and Richard Thaler. 1985. Does the Stock Market Overreact?'' Journal of Finance, 40 (3): 793805.
  • Dugar, Amitabh and Siva Nathan. 1995. The Effects of Investment Banking Relationships on Financial Analysts' Earnings Investment Recommendations. Contemporary Accounting Research, 12 (1): 131160.
  • Ebinger, Louis. 2008. Sarbanes-Oxley Section 501(a): No Implied Private Right of Action, and a Call to Congress for an Express Private Right of Action to Enhance Analyst Disclosure. Iowa Law Review, 93: 19211929.
  • Eleswarapu, Venkat R., Rex Thompson, and Kumar Venkataraman. 2004. The Impact of Regulation Fair Disclosure: Trading Costs and Information Asymmetry. Journal of Financial and Quantitative Analysis, 39 (2): 209225.
  • Francis, Jennifer, J. Douglas Hanna, and Donna R. Philbrick. 1997. Management Communications with Securities Analysts. Journal of Accounting and Economics, 24 (3): 363394.
  • Franzzini, Andrea and Owen A. Lamont. 2008. Dumb Money: Mutual Fund Flows and the Cross-Section of Stock Returns. Journal of Financial Economics, 88 (2): 299322.
  • Grossman, Sanford J. 1995. Dynamic Asset Allocation and the Informational Efficiency of Markets. Journal of Finance, 50 (3): 773778.
  • Groth, John C., Wilbur G. Lewellen, Gary G. Schlarbaum, and Ronald C. Lease. 1979. An Analysis of Brokerage House Securities Recommendations. Financial Analysts Journal, 35 (1): 3240.
  • Hayes, Rachel. 1998. The Impact of Trading Commission Incentives on Analysts' Stock Coverage Decisions and Earnings Forecasts. Journal of Accounting Research, 36 (2): 299320.
  • Hoch, Stephen J. and Young-Won Ha. 1986. Consumer Learning: Advertising and the Ambiguity of Product Experience. Journal of Consumer Research, 13 (2): 221233.
  • Hong, Harrison and Jeffrey D. Kubik. 2003. Analyzing the Analysts: Career Concerns and Biased Earnings Forecasts. Journal of Finance, 58 (1): 313351.
  • Hong, Harrison, Jeffrey D. Kubik, and Amit Solomon. 2000. Securities Analysts' Career Concerns and Herding of Earnings Forecasts. Rand Journal of Economics, 31 (1): 121144.
  • Howe, John S., Emre Unlu, and Xeumin Sterling. 2009. The Predictive Content of Aggregate Analyst Recommendations. Journal of Accounting Research, 47 (3): 799821.
  • Hung, Kee H. 2000. Marketing Stocks by Brokerage Firms: The Role of Financial Analysts. Financial Management, 29 (2): 3554.
  • Hunton, James E. and Ruth Ann McEwen. 1997. An Assessment of the Relation Between Analysts' Earnings Forecast Accuracy, Motivational Incentives and Cognitive Information Search Strategy. The Accounting Review, 72 (4): 497515.
  • Irani, Afshad J. and Irene Karamanou. 2004. A Study of the Economic Consequences of Regulation FD (Fair Disclosure). Research in Accounting Regulation, 17: 195212.
  • Irvine, Paul. 2001. Do Analysts Generate Trade for Their Firms? Evidence From the Toronto Stock Exchange. Journal of Accounting and Economics, 30 (2): 209226.
  • Irvine, Paul. 2004. Analysts' Forecasts and Brokerage Firm Trading. The Accounting Review, 79 (1): 125149.
  • Jackson, Andrew. 2005. Trade Generation, Reputation and Sell-Side Analysts. Journal of Finance, 60 (2): 673717.
  • Jegadeesh, Narasimhan and Woojin Kim. 2006. Value of Analyst Recommendations: International Evidence. Journal of Financial Markets, 9 (3): 274309.
  • Jegadeesh, Narasimhan, Joonghyuk Kim, Susan D. Krische, and Charles Lee. 2004. Analyzing the Analysts: When Do Recommendations Add Value? Journal of Finance, 59 (3): 10831124.
  • Kwag, Seung-Woog and Kenneth Small. 2007. The Impact of Regulation Fair Disclosure on Earnings Management and Analyst Forecast Bias. Journal of Economics and Finance, 3 (1): 8798.
  • Li, Xi. 2005. The Persistence of Relative Performance in Stock Recommendations of Sell-Side Financial Analysts. Journal of Accounting and Economics, 40 (1-3): 129152.
  • Libby, Robert, James E. Hunton, Hun-Tong Tan, and Nicholas Seybert. 2008. Relationship Incentives and the Optimistic/Pessimistic Pattern in Analysts' Forecasts. Journal of Accounting Research, 46 (1): 173198.
  • Lim, Terence. 2001. Rationality and Analysts' Forecast Bias. Journal of Finance, 56 (1): 369385.
  • Lin, Hsiou-Yi and Maureen McNichols. 1998. Underwriting Relationships, Analysts' Earnings Forecasts and Investment Recommendations. Journal of Accounting and Economics, 25 (1): 101127.
  • Lin, Wen-Yi, Po-Jung Chen, and Sheng-Syan Chen. 2010. Stock Characteristics and Herding in Financial Analyst Recommendations. Applied Financial Economics, 21 (5): 317331.
  • Ljungqvist, Alexander, Felicia Marston, and William J. Wilhelm. 2006. Competing for Securities Underwriting Mandates: Banking Relationships and Analysts' Recommendations. Journal of Finance, 61 (1): 301340.
  • Malmendier, Ulrike, and Devin Shanthikumar. 2007a. Do Securities Analysts Speak in Two Tongues? NBER Working Paper No. 13124.
  • Malmendier, Ulrike, and Devin Shanthikumar. 2007b. Are Small Investors Naïve About Incentives? Journal of Financial Economics, 85 (2): 457489.
  • Massey, Cade, and Richard H. Thaler. 2005. The Loser's Curse: Overconfidence vs. Market Efficiency in the National Football League Draft. Duke University.
  • Michaely, Roni and Kent L. Womack. 1999. Conflict of Interest and the Credibility of Underwriter Analyst Recommendations. Review of Financial Studies, 122 (4): 653686.
  • 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): 6791.
  • Mola, Simona and Massimo Guidolin. 2009. Affiliated Mutual Funds and Analyst Optimism. Journal of Financial Economics, 93 (1): 108137.
  • O'Brien, Patricia C., Maureen F. McNichols, and Lin Hsiou-Wei. 2005. Analyst Impartiality and Investment Banking Relationships. Journal of Accounting Research, 43 (4): 623650.
  • Odean, Terrance. 1998. Volume, Volatility, Price, and Profit When All Traders Are Above Average. Journal of Finance, 53 (6): 18871934.
  • Petty, Richard E. and John T. Cacioppo. 1981. Attitudes and Persuasion: Classic and Contemporary Approaches. Dubuque, IA: Wm. C. Brown Co.
  • Rajan, Raghuram and Henri Servaes. 1997. Analyst Following of Initial Public Offerings. Journal of Finance, 52 (2): 507529.
  • Rees, Ray. 1985. The Theory of Principal and Agent—Part I. Bulletin of Economic Research, 37 (1): 326.
  • Reuter, Jonathan. 2006. Are IPO Allocations For Sale? Evidence From Mutual Funds. Journal of Finance, 61 (5): 22892324.
  • Ribstein, Larry E. 2005. Sarbanes-Oxley After Three Years. Working Paper, College of Law, University of Illinois.
  • Richards, Jeff I. and Ivan L. Preston. 1992. Proving and Disproving Materiality of Deceptive Advertising Claims. Journal of Public Policy and Marketing, 11 (2): 4556.
  • Shefrin, Hersh and Meir Statman. 1995. Making Sense of Beta, Size and Book-to-Market. Journal of Portfolio Management, 21 (2): 2634.
  • Solt, Michael E. and Meir Statman. 1989. Good Companies, Bad Stocks. Journal of Portfolio Management, 15 (4): 3944.
  • Stickel, Scott E. 1995. The Anatomy of the Performance of Buy and Sell Recommendations. Financial Analysts Journal, 51 (5): 2539.
  • Stotz, Olaf and Rudiger von Nitzch. 2005. The Perception of Control and the Level of Overconfidence: Evidence From Analyst Earnings Estimates and Price Targets. Journal of Behavioral Finance, 6 (3): 121128.
  • Teoh, Siew Hong and T.J. Wong. 2002. Why New Issues and High Accrual Firms Underperform: the Role of Analysts Credulity. The Review of Financial Studies, 15 (3): 869900.
  • Van Rooij Maarten, AnnaMarie Lusardi, Rob Alessie. 2007. Financial Literacy and Stock Market Participation. NBER Working Paper No. 13565.
  • Wang, Sean. 2009. When Informed Agents Disagree. Doctoral dissertation. Retrieved from ProQuest Dissertation and Theses Database (AAT 3376692).
  • Womack, Kent L. 1996. Do Brokerage Analysts' Recommendations Have Investment Value? Journal of Finance, 51 (1): 137167.
  1. 1

    To test whether the use of only “closed” analyst recommendations (those with clear start and stop dates) would bias the results herein, an analysis of all DJIA recommendations from March 2008 to March 2013 was conducted. In this analysis, 256 of the recommendations were closed and 363 were open. A GLM regression analysis mirroring the other analyses, but including a dichotomous recommendation status independent variable (closed or open) and its interaction with rating type (buy, hold, or sell) was conducted. There was no significant interaction between recommendation status and rating type, meaning that the pattern of results by rating type did not depend on whether the recommendation was open or closed.