1.4. Sistemas usados durante el desarrollo de esta tesis
1.4.3. SION (Scalable I/O Network)
There are two major reasons why analysts’ career concerns have a determinant impact on analyst accuracy. First, financial analysts’ accuracy will be influenced by their incentives to forecast accurately. These incentives are highly dependent on the relation between analysts’ forecasting performance and career concerns such as reputation, compensation, and career termination. Different incentives to produce accurate forecasts across international labor markets for financial analysts may also cause differences in financial analyst forecast accuracy across countries. Second, career concerns such as reputation, have been shown by academic theory (see Scharfstein and Stein, 1990; Trueman, 1994) to influence imitation or herding among financial agents. This means, herding among analysts may contribute to the optimistic bias documented by past research and to some differences in forecast accuracy among analysts. In this section, I review the literature that examines the relationship between accuracy and career concerns as well as the literature that investigates the link between analysts’ career concerns (reputation) and herd behavior.
2.4.1 Analysts’ career concerns and forecast accuracy
An important number of past research investigates whether relative analysts’ accuracy improves as analysts age. Examples of such research include Mikhail et al. (1997), Jacob et al. (1999), and Clement (1999) for the US market as well as Clement et al. (2000) for a few non-US markets. The evidence appears mixed and sensitive to the variable used to measure financial analyst experience. However, a majority of the investigations conducted on the US market show that relative analysts’ accuracy improves with age; that is, analysts have incentives to improve their accuracy during their careers. For non-US markets, a weak or insignificant relationship between analysts’ experience and accuracy has been documented. Unfortunately, these studies do not consider the nature of the implicit incentives (career concerns) faced by security analysts. In other words, they do not investigate why analysts may (or may not) have interests to improve their forecast accuracy during their careers.
Stickel (1992) is the first to document a positive relationship between analysts’ performance and career concerns. He finds that Institutional Investor All-American analysts, who are typically better compensated than other analysts (see Michaely and Womack, 1999 among others), provide more accurate earnings forecasts and tend to revise their recommendations more frequently than other analysts. Mikhail et al. (1999) document that poor relative performance among analysts leads to job turnover. Both studies suggest a positive relationship between forecast accuracy and positive career concerns. More recently, Hong and Kubik (2003) find that relatively accurate analysts have a greater probability of being hired by a prestigious brokerage house (i.e. to get higher compensation) than other analysts. Conversely, relatively inaccurate analysts are more likely to move down the brokerage house hierarchy. Furthermore, controlling for forecast accuracy, they find that analysts who issue relatively optimistic forecasts (forecasts greater than the consensus) are more likely to move up the brokerage house hierarchy. They conclude that brokerage houses do not only reward accurate forecasts but also relatively optimistic forecasts. This suggests that accurate analysts experience favourable job separations outcomes and that, among the group of accurate analysts, the more optimistic analysts experience even more favourable career-development outcomes.
2.4.2 Analysts’ career concerns and herding
Analysts may be influenced by the recommendations of other analysts. The degree to which they are influenced may be related to their career concerns. Scharfstein and Stein (1990) suggest that analysts herd to protect their reputation. Specifically, analysts may herd because, at some stages of their careers, it is less costly for their reputation to be wrong when other analysts make the same mistake than to be wrong alone. Welch (2000) tests for herding in analysts’ recommendations. He detects behaviour consistent with mutual imitation among analysts. More precisely, he reports a significant influence of the prevailing consensus on subsequent analyst forecasts. This influence is not significantly stronger when the consensus turns out to be correct in its prediction of future stock price movements suggesting that analysts herd based on little or no information. Even if he does not identify the reasons why analysts herd, his results are consistent with reputation-based theories. Hong et al. (2001) are the first to investigate whether analysts herd in order to manage their reputation. They find younger analysts to herd more (they forecast closer to the consensus) than their older counterparts. Older analysts issue more timely forecasts and revise less than younger analysts.
They conclude that the young analysts’ need for preserving their reputation leads them to herd more than older analysts whose reputation is generally established. Phillips and Zuckermann (2001) also provide evidence consistent with theories relating herd behaviour and reputation. They find that Institutional Investor All-American analysts issue significantly more “sell” recommendations than other analysts. Cooper et al. (2001) report significant differences in the accuracy of lead analysts (timely analysts) and follower analysts (who are suspected to herd). They find that lead analysts provide less accurate forecasts than follower analysts. They conclude that lead analysts trade timeliness against accuracy. However, the stock price reaction following analyst forecast revisions is significantly higher for lead analysts than for followers ones. Their results suggest that herding influences the relative accuracy of financial analysts as well as the relative informativeness of their revisions.