Rating Reports: Do Investors Get The Text Message?
University of Ulm – Department of Mathematics and Economics
Brazilian School of Public and Business Administration (EBAPE), Getulio Vargas Foundation (FGV)
June 30, 2016
We investigate whether the linguistic tone of Moody’s rating reports predicts future rating changes, earnings and stock returns of firms in the United States. We are the first to consider reports on rating changes as well as on events with no rating change. Our main findings are:
(1) Irrespective of report type, a negative linguistic tone predicts future downgrades and lower earnings;
(2) stock returns overreact to the tone of negative change reports, but they do not or not immediately react to the tone of positive changes and no-change reports; and
(3) higher investor attention leads to a faster incorporation of rating news.
The evidence suggests that investors’ response to the text message is biased: they overreact to salient news and they under react to less salient news
Rating Reports: Do Investors Get The Text Message? – Introduction
The volume of information in financial markets is huge. However, extensive evidence in the psychology literature suggests that the cognitive resources available to market participants to process this information is limited. While traditional economic models ignore the attention constraints of market participants, there is a growing literature that explores the implications of their limited attention for financial markets. The existing evidence primarily relates to limited investor attention in explaining asset pricing anomalies.
One potentially important manifestation of limited attention in financial markets is the tendency to rely on coarse (“macro”) information pertaining to a firm’s broad category, i.e. category-level information, as opposed to fine (“micro”) firm-level information. For example, Mullainathan (2002) argues that using categories results in an information processing bias when investors with limited attention use coarse categorizations, and overly rely on an asset’s category to infer about the asset.2 In Peng and Xiong (2006), investors with attention constraints process category-level information to the exclusion of asset-level information, more so when the investor exhibits more attention constraints. Such category-based oversimplifications during information processing could be widespread in financial markets, since the categorization of a large universe of assets into coarse groups is ubiquitous in the process of organizing and disseminating information. In the stock market, these groups or categories are usually based on perceived shared attributes of the stocks, such as industry (e.g. Oil and Gas) or style (e.g. Growth). An example of category-level information is past returns of an industry (e.g. Oil and Gas), whereas the past returns of a stock (e.g. Exxon Mobil) is an example of firm-specific information.
Despite the theoretical studies mentioned above, there is still a considerable gap in our understanding of how limited attention in general, and category-based information processing in particular, may affect price discovery and market efficiency. Moreover, few studies have focused specifically on sophisticated market participants, such as managers, market makers, and institutional investors.3 This study explores whether limited attention, as reflected in the reliance on category-level information relative to firm-level information, affects security analysts. Although analysts are typically viewed as sophisticated information intermediaries and play a crucial role in the price discovery process, there is significant evidence that analysts are not immune to biases.4 Following the theoretical predictions of Peng and Xiong (2006), when an analyst faces attention constraints, we expect the analyst to process more category-level information and less firm-specific information. The degree to which analysts vary in their reliance on category-level information in issuing firm-level forecasts could contain information about their susceptibility to limited attention, their forecasting performance and, consequently, their contribution to the price discovery process for the stocks they follow.
The first research question we address is whether the analysts’ reliance on category-level information relative to firm-level information is associated with their forecasting performance. We relate analysts’ reliance on categories to other analyst attributes that previous studies have linked to their forecasting ability. We also examine the relation between the analysts’ reliance on categories and the quality and quantity of their information production, as measured by forecast accuracy and forecast frequency, respectively. The second research question is whether the stock price impact of analysts’ forecast revisions is a function of their tendency to rely on category-level information. Lastly, we explore whether analysts’ tendency to rely on category-level information has real consequences for the analysts’ reputation and career outcomes.
See full PDF below.