What makes a stock risky? Evidence from sell-side analysts ...

[Pages:55]What makes a stock risky? Evidence from sell-side analysts' risk ratings

Daphne Lui Lancaster University Management School

Lancaster LA1 4YX 44 (01524) 593 638 d.lui@lancaster.ac.uk

Stanimir Markov Goizueta Business School

Emory University Atlanta, GA 30322

(404) 727-5329 Stanimir_Markov@bus.emory.edu

Ane Tamayo London Business School

Regent's Park London NW1 4SA 44 (020) 7262 5050 atamayo@london.edu

April 26, 2006

Abstract

We examine the determinants and the informativeness of financial analysts' risk ratings using a large sample of research reports issued by Salomon Smith Barney, now Citigroup, over the period 1997-2003. We find that the cross-sectional variation in risk ratings is largely explained by variables commonly viewed as risk proxies such as idiosyncratic risk, size, leverage, and accounting losses. We also find that the risk ratings can be used to predict future return volatility, after controlling for other predictors of future volatility. Both findings establish the important role of financial analysts as providers of information about investment risk.

We thank Teresa Dau, Arantza Urra and, especially, Inma Urra for excellent research assistance. We also thank Sudipta Basu, Larry Brown, Marty Butler, Francesca Cornelli, Miles Gietzmann, Connie Kertz, Michael Kimbrough, Grace Pownall, Henri Servaes, Greg Waymire, the seminar participants at Emory University, The Hong Kong University of Science and Technology, London Business School, Tulane University, Singapore Management University, the 16th Annual Conference on Financial Economics and Accounting at the University of North Carolina, and the 29th Annual Congress of the European Accounting Association in Dublin for their helpful comments.

1. Introduction

Information about investment risk is critical for making investment decisions. In a world

of uncertainty, the desirability of an investment depends not only on the expected payoff, but

also on the risk of the future payoffs. For that reason, in addition to forecasting the levels of

future cash flows, earnings, or stock prices, and providing stock recommendations, financial analysts often provide information about investment risk.1 Despite Zmijewski's (1993, p.337)

call for more research into how analysts make risk assessments, no prior research has

systematically studied the determinants of financial analysts' risk assessments. This paper takes

the first step by providing evidence on the cross-sectional determinants and the usefulness of

analysts' risk ratings.

Our main sample includes 6,098 research reports issued in the period 1997-2003 by

Salomon Smith Barney, now Citigroup, (or SSB henceforth), and available through Investext.

This brokerage house rates stocks as Low, Medium, High, and Speculative, based on price

volatility and predictability of financial results. To ensure that our results are not unique to SSB

and can be generalized to other information providers, we also analyze risk ratings issued by

Merrill Lynch and Value Line Investment Service.

We find that stock characteristics suggested in prior research as measures of risk are

important determinants of the risk ratings. Analysts rate stocks with high leverage and low

1 Currently, there are regulatory and litigation-related reasons for providing information about investment risks. Following the bursting of the internet bubble in 2000, NYSE's Rule 472 and NASD's rule 2210 were amended to require that research reports disclose "the valuation methods used, and any price objectives must have a reasonable basis and include a discussion of risks" (Exchange Act Release # 48252 (July 29, 2003)). Since 2002, more than 60 class action suits alleging that analysts committed federal securities fraud in their research reports have been filed. In dismissing the class action suit against Merrill Lynch and Henry Blodget, the court used the provided risk ratings and qualitative discussions about specific sources of risk to conclude that the plaintiffs did not overcome "the Bespeaks Caution Doctrine" (In Merrill Lynch and Co. Research Reports Securities Litigation, 272 F. Supp. 2d 351 (2003)). Under this doctrine, Merrill Lynch is protected from liability because it warned investors about the risks of investing in the two concerned stocks (7/24 Real Media and Interliant), and there was no allegation of misrepresentation that made Merrill Lynch's cautionary statements fraudulent.

1

market capitalization as riskier, which supports the interpretation of these variables as measures of risk rather than mispricing (e.g., Fama and French, 1992, 1993). Analysts also view firms reporting losses as being riskier. Losses signal poor earnings prospects and could be capturing distress risk as in Fama and French (1992). Finally, we find no evidence that either SSB or Merrill Lynch analysts view high book-to-market stocks as being riskier but document that Value Line analysts do.

Our evidence on beta, just like prior evidence on beta from the analysis of stock returns (e.g., Kothari et al., 1995; Fama and French, 1992; Daniel and Titman, 1997; Easley et al., 2002), is mixed. Univariate regressions and regressions that control for book-to-market, leverage and size, suggest that high beta stocks are considered riskier by analysts. However, once we control for idiosyncratic risk, defined as the standard deviation of stock returns unexplained by the market, we find that only Value Line analysts view high beta stocks as being riskier. We conclude that idiosyncratic risk plays a much bigger role as a determinant of the risk ratings than beta.

We also examine whether analysts minimize the risks associated with purchasing stocks of companies whose equities offerings their firm underwrote. In contrast with stock recommendations (e.g., Dugar and Nathan, 1995; Lin and McNichols, 1998; Michaely and Womack, 1999), risk ratings do not appear to be biased when underwriting relations exist. Combined with the evidence that the same risk variables drive the ratings of sell-side analysts and Value Line, this result suggests that even if sell-side analysts have incentives to minimize investment risks, there are competitive forces holding these incentives in check.

Overall, we conclude that analysts' risk ratings mainly incorporate information about various stock characteristics commonly viewed as risk measures in the literature. The financial

2

analysts' notion of risk is multidimensional, and very similar to the notion of risk used in the academic literature.

In a world of costly information search and bounded rationality, gathering information about risk and providing a single summary statistic (as financial analysts do) can have value. The value of the risk ratings, however, would be even greater, if they provided information incremental to the already-available information. Thus, we take the perspective of an investor who is interested in predicting future volatility, and examine whether the risk ratings alone and in the presence of other variables, predict the cross-sectional variation in future price volatility.

Our results suggest that the risk ratings alone explain almost 50% of the cross-sectional variation in future return volatility. For the SSB sample, the spread in future volatility between Low and Speculative risk stocks is 9.83% per month, which is large given a cross-sectional standard deviation of future return volatility of 7.83% per month. The relation between future return volatility and risk rating weakens when other predictors of future volatility are incorporated, but the risk ratings remain incrementally informative about future volatility. For example, controlling for past volatility, the difference in monthly volatility between Low and Speculative risk stocks is 1.73%. Controlling for other stock characteristics reduces the difference to 1.42% per month.

The evidence on the incremental information content of risk ratings is not unique to SSB; both Merrill Lynch's and Value Line's risk ratings help predict the cross-sectional variation of future volatility. We conclude that our findings are consistent with the view that analysts play an important role as providers of information about investment risks.

The evidence provided in this paper is useful for investors. It helps them understand what information is included in (or excluded from) analysts' risk assessments, so they can

3

optimally combine analysts' information with their own information. In addition, our findings on the predictive ability of analysts' risk ratings suggest that investors can improve their forecasts of future volatility by using analysts' risk ratings in addition to other public information.

Understanding how analysts determine their risk ratings is also of interest to researchers. Since investors' notions of risk are not observable, in developing and testing asset pricing models researchers make various auxiliary assumptions whose validity may be hard to ascertain (Brav and Heaton, 2002). In his presidential address to the American Finance Association, Elton (1999) questions the tests' continuous reliance on the assumption that realized returns are a good proxy for expected returns, and calls for alternative ways of testing theories that do not use realized returns. We believe that our analysis can potentially help us interpret existing evidence about the relation between firm characteristics and average returns. For example, if analysts rate small stocks as riskier than large stocks, then it is more likely that size is a risk proxy. This assumes that SSB's risk assessments influence, or are correlated with, marginal investor's notion of risk.2

Our analysis also complements the experimental literature on how individuals and investors view risk (e.g., Alderfer and Bierman, 1970; Cooley, 1977; Mear and Firth, 1987; Olsen, 1997; Bloomfield and Michaely, 2004). Unlike experimental and survey evidence, our evidence comes from a market setting: research reports are supplied by analysts and demanded by investors.3

2 This assumption is consistent with large body of evidence that analyst research reports contain information that influences marginal investor's earnings and cash flow expectations. For example, information provided by analysts in the form of earnings forecasts and stock recommendations (Francis and Soffer 1997; Markov, 2001), cash flow forecasts (Defond and Hung, 2003), price targets (Brav and Lehavy, 2003), and justifications of analyst opinion (Asquith et al., 2005) has an effect on stock prices. 3 A general discussion of the advantages and disadvantages of experimental evidence vis-?-vis market evidence is provided in Libby et al. (2002).

4

The rest of this paper is organized as follows. Section 2 describes the sample. The crosssectional determinants of analysts' risk assessments are examined in section 3. Section 4 examines the informativeness of analysts' risk ratings. Additional analyses are presented in Section 5. Section 6 concludes the paper.

2. Sample and Variable Description Our data on analysts' risk assessments were hand-collected from analysts' written reports

available on Investext. Well-known information providers such as IBES, First Call, and Zacks gather and make available in electronic form various types of analysts' provided information, but not analysts' risk assessments. Due to the large number of contributors to Investext we examined research reports by seven major brokerages (Bear Stearns, Credit Suisse First Boston, Deutsche Bank, Merrill Lynch, Morgan Stanley Dean Witter, Salomon Smith Barney, and Warburg Dillon Read). Salomon Smith Barney (SSB) and Merrill Lynch have provided quantitative risk assessments at least since 1997 and 1998. Credit Swiss First Boston and Morgan Stanley have provided such risk ratings at least since 2004. The other three firms do not include risk ratings in their reports, but do provide qualitative information about risk.

Our main sample includes reports issued by SSB over the period 1997-2003, which we supplement with a sample of Merrill Lynch reports issued in 1998. We were not able to expand the Merrill Lynch sample since Merrill Lynch had discontinued the practice of making its reports available to Investext.

We acknowledge that evidence from the analysis of the reports of two investment firms may not be generalizable to other investment research providers as the notion of risk may vary across investment firms. Thus, we view the risk assessments of SSB and Merrill Lynch only as

5

useful proxies for the "consensus" risk assessment. The weight of these two firms in this "consensus" assessment is likely to be significant for several reasons. First, these firms not only employ a very large number of analysts, about 10% of all analysts on IBES in 2003, but are viewed by institutional investors as the premier providers of investment research. In every year in the period 1996-2005 SSB and Merrill Lynch, as well as Morgan Stanley and Credit Suisse First Boston, were ranked by The Institutional Investor Magazine among the top 10 providers of investment research. Second, SSB and Merrill Lynch have significant retail operations; together they employ about 25,000 financial advisors (2003 SSB/Citigroup and Merrill Lynch Annual Reports). Thus, their investment research reaches, and potentially influences, the opinions of large numbers of individual investors.

The question why SSB and Merrill Lynch provide quantitative risk assessments while other firms provide only qualitative information about risk is an important one. As a profit maximizing entity, an investment firm would produce quantitative risk assessments as long as the costs of producing them are lower than the revenues generated. What distinguishes these two firms is that they both have very large private clients groups.4 We suggest that the benefits from the provision of quantitative information about investment risks are perhaps greater for an investment firm that derives a great portion of its revenues from serving individual investors. Individual investors are more likely to find this information useful than large institutions with the resources to generate it internally.5

4 In 1999 Merrill Lynch and SSB had client assets of $1,222 and $852 billion, followed by Charles Schwab and Morgan Stanley Dean Witter with $595 and $529 billion, also providers of risk ratings. V. Kasturi Rangan, and Marie Bell, "Merrill Lynch: Integrated Choice", HBS # 500-090 (Boston: Harvard Business School Publishing, 2001), p. 25. 5 Explaining a firm's choice to provide quantitative risk assessments would require that we sample the reports of all firms, rather than the reports of only seven firms and gather data on firm characteristics potentially related to the costs and benefits of providing such information such as size, the existence and importance of private clients group, etc. This is an interesting venue for future research.

6

2.1 Definition of risk rating The notion of risk as expected price volatility is common to all four brokerages. Since our

main sample consists of SSB's research reports, the rest of our discussion discusses SSB's risk ratings policy.

From 1997 to September 2002 SSB rated stocks using 5 categories: "L (Low risk): predictable earnings and dividends, suitable for conservative investor. M (Medium risk): moderately predictable earnings and dividends, suitable for average equity investor. H (High risk): earnings and dividends are less predictable, suitable for aggressive investor. S (Speculative): very low predictability of fundamentals and a high degree of volatility, suitable for sophisticated investors with diversified portfolios that can withstand material losses. V (Venture): indicates a stock with venture capital characteristics that is suitable for sophisticated investors with high tolerance for risk and broadly diversified investment portfolios." 6

From September 2002 onwards, the firm no longer assigned stocks to the Venture category. After 2002 all stocks are rated as Low [L], Medium [M], High [H], or Speculative [S].7 The risk ratings and the forecasts of total return (price appreciation plus dividends) are the basis for the stock recommendations. Stocks with risk ratings Low, Medium, High, and Speculative are rated Buy when the analyst forecasts total return of at least 10% or more, 15% or more, 20% or more, and 35% or more respectively; Hold when the analyst forecasts total return of 0%-10%, 0%-15%, 0%-20%, and 0%-35% respectively; Sell when the analyst forecasts negative return.

6 Spencer Grimes, Liberty Media Group, Salomon Smith Barney, December 29, 1998, via Thomson Research/Investext, accessed January 30, 2006. 7 Lanny Baker and William Morrison, , Citigroup Smith Barney, July 22 2004, via Thomson Research/Investext, accessed January 30, 2006.

7

 ................
................

In order to avoid copyright disputes, this page is only a partial summary.

Google Online Preview   Download