Which side are you on? How institutional positions affect financial analysts' incentives.

Recent scandals involving financial analysts shed light on conflicts of interest in the stock market. Financial analysts were accused of releasing dishonest "buy" recommendations for stocks and biased earnings forecasts. In 2002, the New York attorney general released internal e-mails obtained during his investigation, in which analysts disparaged the very stocks they were urging outside fund managers or investors to buy. In the following months, a number of analysts received lifetime suspension from the securities business, and several banks had to pay severe penalties.

Interestingly, the problem has, in most cases, been treated as an individual matter, as if it only concerned deviant behaviors by certain individuals. But it can also be viewed as a problem of rule. In this study, I endeavor to provide evidence that analysts' behaviors are deeply influenced by the organizations to which they belong and, hence, are influenced by the regulations governing these organizations. In other words, using the terminology of Saced Parto (2005), the institutions implicated when an analyst decides to release intentionally biased earnings forecasts are regulative institutions rather than behavioral institutions.

In the remainder of the paper, I first present academic observers' explanations of the bias in analysts' forecasts; I then describe the data I used and the methods leading to my main results; and I conclude with a discussion of an institutional framework for stock market regulators.

Explaining Analysts' Forecasting Behavior

For convenience I will distinguish between two broad explanations of analysts' behavior discussed in the research literature: (1) the first is related to individual career considerations, and the second to those of investment banking and trading.

It is frequently asserted that analysts sometimes give untruthful forecasts due to career considerations. The theoretical source of such statements lies in reputation-based "herding models" (see for example Scharfstein and Stein 1990). Analysts with better reputations are more likely to issue accurate forecasts; hence this provides an incentive for other analysts to "herd." Harrison Hong and Jeffrey Kubik (2003) showed that more accurate analysts are more frequently hired by high-status brokerage houses. An accurate analyst is rewarded by becoming a member of a prestigious firm. Members of low-status firms are more prone to overoptimism in their forecasts. Similarly, Michael Mikhail, Beverly Walther, and Richard Willis (1997) and Hong, Kubik, and Amit Solomon (2000) treated experience as a key variable: on average, younger and less experienced analysts make more biased forecasts than their more experienced counterparts. However, all these researchers used experience or career as proxy for compensation, since the data on compensation were not available. Analysts are paid with both a salary and a variable bonus, depending on their performance, as direct rewards for their work. The assumption is that an analyst producing high-quality forecasts is compensated highly.

But an intriguing fact has led to further inquiries. Controlling for experience, Hong and Kubik (2003) showed that more optimistic analysts were better rewarded, and Andrew Leone and Joanna Shuang Wu (2002) found that, although analysts with better reputations had less bias toward being optimistic, the forecast bias nonetheless …