Do Analysts Strategically Employ Cash Flow Forecast Revisions to Offset Negative Earnings Forecast Revisions?

Abstract

We investigate whether analysts use cash flow forecasts to reduce the impact of earnings forecast revisions (EFRs) on market participants. In particular, we focus on conflict between an analyst's concurrent cash flow and earnings forecast revisions. We hypothesize and find that analysts are more likely to issue a positive cash flow forecast revision when they issue a negative earnings forecast revision concurrently, but not the opposite, particularly for Fortune 500 firms. Furthermore, our supplementary analyses suggest that (1) some analysts optimistically bias cash flow forecasts when they issue negative earnings forecast revisions; (2) the market pays less attention to the historical accuracy of analyst cash flow forecasts, so analysts have some latitude to present their cash flow forecasts in an optimistic way; and (3) the market reacts mainly to the direction, not the magnitude, of cash flow forecast revisions. Overall, these findings suggest that analysts may strategically use cash flow forecasts in conjunction with earnings forecasts to maintain good management relationships.

JEL Classification:

• G14
• G14
• M41

Acknowledgements

This paper is based on Choong-Yuel Yoo's dissertation, completed at Queen's University in Canada. Choong-Yuel is indebted to his thesis advisor, Steven Salterio, for his advice and encouragement and is grateful for generous research support provided by the ILJU Foundation. We thank A. Rashad Abdel-Khalik, Dan Thornton, Mike Welker, and research workshop participants at KAIST and the 2012 AAA Annual Meeting in Washington DC for comments on previous drafts of this paper.

Notes

¹The US SEC defines sell-side analysts as follows: ‘Sell-side analysts typically work for full-service broker-dealers and make recommendations on the securities they cover. Many of the more popular sell-side analysts work for prominent brokerage firms that also provide investment banking services for corporate clients – including companies whose securities the analysts cover’ (U.S. SEC Analyzing Analyst Recommendations, 2010).

²See Kadan, Madureira, Wang, and Zach (2009) and Barber et al. (2006) for evidence on the great increase in incidence of sell recommendations after Regulation FD and the Global Settlement.

³CEOs have strong preferences to highlight good news and postpone announcing bad news. Thus, they often strategically promote favorable investor perceptions by managing earnings and voluntarily disclosing pro forma earnings and ‘good news’ cash flow forecasts (Daniel, Hirshleifer, & Teoh, 2002; Hirshleifer & Teoh, 2003; Hong, Lim, & Stein, 2000; Schrand & Walther, 2000; Wasley & Wu, 2006).

⁴Favors from and private access to CEOs include disseminating critical information about recent industry developments; facilitating contact between analysts and key personnel; meeting with an analyst's clients; recommending an analyst for jobs; and helping analysts obtain prestigious club memberships (see Clement & Westphal, 2008).

⁵We do not suggest that analysts issue a positive CFFR whenever it suits their interests, though analysts can time their disclosures strategically. Moreover, given the inconsistency in releasing cash flow forecasts across time, analysts, and covered firms, it is possible that analysts might use this discretion to their advantage. We seek to investigate whether analysts indeed exercise this discretion in their best interests.

⁶We also use the ‘actual’ last forecast revision of the fiscal year, which immediately precedes the fiscal year-end date. All test results (untabulated) are qualitatively the same as those reported in the paper.

⁷To ensure that our choice of the last earnings forecast does not drive our results, we re-ran all tests with all analyst forecasts in a given year, increasing our sample size from 11,778 to 64,604. No differences in reported results were found.

⁸We also ran tests using the full I/B/E/S sample without eliminating potential outliers. The results were qualitatively the same whether we included or excluded potential outliers.

⁹Odds indicate the relative likelihood that an event of interest will happen. For instance, when the probability of an event is p (therefore, the probability of no event is 1 – p), the odds of the event occurring are the quotient of the two, or .

¹⁰According to Giampaolo Trasi, vice chairman of the European Federation of Financial Analysts’ Societies, analysts are very likely to drop their coverage of a firm after three consecutive quarters of disappointing earnings, all other things being equal (Rana, 2008).

¹¹In the I/B/E/S sample, we have multiple observations (analyst forecasts) per firm year. The likelihood of an OPDFR by analyst i for firm j may be associated with firm j’s specific characteristics. Also, researchers argue that critical events in the early 2000s (e.g. the crash of the Internet bubble, the decimalization of the U.S. stock exchanges, the demise of Enron, WorldCom's acknowledgement of accounting errors and bankruptcy filing, and the U.S. economic recession) have changed the information playing field for analysts and managers (e.g. Agrawal, Chadha, & Chen, 2006; Ahmed & Schneible, 2007; Bailey, Li, Mao, & Zhong, 2003; Heflin, Subramanyam, & Zhang, 2003; Mohanram & Sunder, 2006). Thus, it is possible that our pooled logit regression results are affected by omitted explanatory variables at the cluster level of the firm-year pair. To control for firm and year fixed effects, we use the conditional maximum likelihood estimation for Models 2 and 4 (Allison, 1999; Chamberlain, 1980) that includes year and firm effects.

¹²Unlike Models 1 and 3, Models 2 and 4 do not have intercepts, because α_k parameters that represent each firm-year k (i.e. α₀ of Equation 1) are cancelled out in the conditional maximum likelihood estimation (Allison, 1999, pp. 188–192).

¹³To interpret the logit regression results in terms of adjusted odds rather than the coefficient estimate per se, we compute the adjusted odds ratio by 100 x (e^β-1), where β is a coefficient estimate (see Defond & Hung, 2003).

¹⁴It is interesting that CFFR for S2 is informative (i.e. its coefficient estimate is marginally significant at the 10% level). It may be interpreted that the market reacts only to a positive CFFR (i.e. as positive reinforcement) in the presence of a positive EFR, but not in the presence of a negative EFR.

¹⁵Analyst interview studies tend to include small numbers, given the challenges involved in obtaining access to analysts. Most published studies were done before the financial crisis of 2008. For example, Roberts et al. (2006) used a group setting to study analysts at 13 UK-listed companies in the aftermath of the 2002 accounting scandals. Other studies (e.g. Barker, 1998, 2000) are even older, with a common set of 32 analysts and focusing on analysts’ beliefs about the efficient-market hypothesis and their understanding of accounting information more generally, albeit with better access to analysts.

¹⁶The interviewees have a mean 12 years of equity analyst experience. The descriptive statistics of our I/B/E/S sample indicates that the sample analysts have a mean forecasting experience of 8.5 years, suggesting that the interviewees are senior analysts. The job titles of the interviewees include the head of an equity research group and four managing directors. Most of the interviewees have been ranked first several times in their earnings forecast performance and stock picking skills by Thomson Reuters's business units or other similar investor services. Nine of the interviewees have MBA degrees, and two have graduate degrees in engineering. Ten are Chartered Financial Analysts, and two are members of the Canadian Institute of Chartered Accountants. Only one of the interviewees has no formal business or economics education or professional designations. Overall, the interviewees are highly experienced and educated informants who can offer insights into the forecast revision process.

¹⁷Multiple readings of the interviews with various types of organizations and coding were carried out, as suggested in Yin (1994).

¹⁸To protect the anonymity of the interviewees, all names used in the paper are pseudonyms and are not associated with the real names of the interviewees.

¹⁹Of the four interviewees who did not release cash flow forecasts, one asserted that he did not provide cash flow forecasts for a particular company because that was the norm among analysts who followed the same company. Consistent with his argument, we found in the I/B/E/S database that only one analyst among more than 20 analysts who followed that company during 2007 issued concurrent EFRs and CFFRs. Five years later, a majority of analysts who follow that company issue concurrent EFRs and CFFRs, including our interviewee, though some analysts still issue none and some issue infrequently. This pattern allows analysts to determine whether to strategically release concurrent CFFRs and EFRs.

²⁰The demand for positive cash flow forecasts to balance negative EFRs is different from the demand for cash flow forecasts when the usefulness of earnings information is low, as suggested by Defond and Hung (2003). Whereas the latter suggests analysts responding to investors’ needs for accurate valuation inputs, the former describes analysts presenting cash flow forecasts strategically, as they understand investors’ and managers’ preferences toward favorable news. That is, Defond and Hung's (2003) demand hypothesis cannot explain our field study finding (the interviewees’ perceptions) that investors do not demand cash flow forecasts when analysts issue positive EFRs.