Seemingly inconsistent analyst revisions*

• 成果类型: Article
• 署名作者: Iselin, Michael; Park, Min; Van Buskirk, Andrew
• 署名单位: University of Minnesota System; University of Minnesota Twin Cities; University of Kansas; University System of Ohio; Ohio State University
• 刊物名称: JOURNAL OF ACCOUNTING & ECONOMICS
• ISSN/ISSBN: 0165-4101
• DOI: 10.1016/j.jacceco.2020.101339
• 发表日期: 2021
• 关键词: EARNINGS FORECASTS

摘要

Sell-side equity analysts produce and disseminate a variety of outputs, including stock recommendations, 12-month target prices, and earnings estimates of various horizons. While these different outputs are generally positively correlated with one another (Bandyopadhyay et al., 1995; Bradshaw, 2004; Ertimur et al., 2007; Da et al., 2016), prior research documents inconsistenciesdcases where one measure conflicts in some way with another measure issued by the same analyst at the same timedanywhere from 26% to 42% of the time (Brown and Huang, 2013; Huang et al., 2014; Kecsk?es et al., 2017). Moreover, many papers characterize these inconsistent observations as low quality, not credible, and/or evidence of analysts? strategic behavior (Barniv et al., 2009; Chen and Chen, 2009; Brown and Huang, 2013; Huang et al., 2014; Malmendier and Shanthikumar 2014; Kecsk?es et al., 2017). Taken together, these conclusions suggest that a sizeable proportion of analyst research is of questionable quality. In about 20%e30% of cases where an analyst revises two outputs (namely, earnings estimates, target prices, or stock recommendations) simultaneously, the two estimates are revised in opposite directions. Existing literature notes that these inconsistent outputs are widespread, and concludes that they are lower-quality, driven by strategic bias, and are viewed as less valid by investors. We find that these characterizations are generally inaccurate. Apparent inconsistency is largely driven by accounting and economic factors, with only limited evidence that investment banking-related conflicts play a role. Moreover, inconsistent outputs are neither less accurate than consistent outputs nor do they resolve less investor uncertainty upon their release. Overall, our results suggest that researchers should be cautious in interpreting the correlation between analyst outputs as a measure of bias or quality, and in using a single analyst output as a proxy for an analyst's overall views. 0 2020 Elsevier B.V. All rights reserved.