Sell-side analysts are important information intermediaries in capital markets and as a result their research has been under scrutiny. While a large number of studies document that analyst coverage and analyst forecasts have economic consequences (Bailey et al., 2003; Jackson, 2005) an equally large number of studies document that analyst forecasts are influenced by conflicts of interest (Beyer and Guttman, 2011; Cowen et al., 2006; Hong and Kubik, 2003; Jackson, 2005; Lim, 2001; Richardson et al., 2004; Schipper, 1991).
In this paper we concentrate on the banking industry and investigate whether analyst forecasts are biased because of career concerns. Past studies have documented that analyst forecasts can be biased because of underwriting activities in the investment banking business, pressure to generate trading commissions, and career concerns (Hunton and McEwen, 1997; Lin and McNichols, 1998; Michaely and Womack, 1999; Dugar and Nathan, 1995; Dechow et al., 2000; O’Brien et al., 2005; Hong and Kubik, 2003). In terms of career concerns, past studies have documented that more optimistic analysts tend to experience favorable job separations (Hong and Kubik, 2003) and younger analysts tend to herd more (Hong et al., 2000). In these studies, the underlying source of career concerns is pressure from investment banking and/or brokerage business to please companies or buy-side portfolio managers respectively. In this paper, we concentrate on a different source of conflicts of interest. Banking analysts issue forecasts for companies that constitute a large part of their outside opportunities in terms of employment. These analysts view the banks that they issue forecasts for as potential sources of employment, thereby increasing their incentives to satisfy those clients. This is independent of incentives to generate investment banking business or trading commissions, which exist for all companies they cover.
In order to examine whether this pressure to satisfy future potential employers is influencing analyst forecasts we examine the pattern in the bias of their forecasts. In our research design we hold the analyst constant by requiring that the same analyst is forecasting earnings for companies with sell-side equity departments (‘employers’) and for companies with no sell-side equity departments (‘non-employers’). We then show that banking analysts issue forecasts that are relatively more optimistic for employers in the beginning of the year. At the end of the year the opposite is true; banking analysts issue forecasts that are relatively more pessimistic for employers. Therefore, our research design is similar to a differences-indifference specification where we observe the forecasting pattern early and late in the year and we compare this pattern for employers and non-employers for the same set of analysts. We limit our sample to those analysts who are not employed by the top investment banks and therefore could have relatively greater career concerns. Analysts that are already working for bulge investment banks have greater career opportunities and less incentives to move as they already work at the most reputable banks. Therefore, we treat analysts working at the top banks as a control group that allow us to scale our dependent variable for forecast bias. We report results using both this relative bias variable and an absolute bias variable relative to the earnings of the firm and document similar results.
This paper investigates how career concerns of analysts that forecast the performance of potential future employers influence their forecasts. We find evidence of a walk-down to 25 beatable earnings when forecasting earnings of future employers but not of companies that are unlikely to be future employers. Moreover, this pattern is more pronounced after the Global Settlement which likely exacerbated career concerns of analysts by limiting their outside opportunities. Consistent with career concerns about future employment biasing forecasts we find that bias in potential future employers’ forecasts lead to favorable career outcomes. No such effect is found for bias in non-employer forecasts.
Our paper documents a source of conflict of interest for research analysts that is widely discussed in other settings, such as auditing. The generalizability of the phenomenon to the analyst setting is important as it suggests that other information intermediaries might be affected by such conflicts. Our findings open up opportunities for future research. Do investors recognize this source of bias when incorporate analyst earnings forecasts in market prices? Do bias incentives from revolving doors generalize to investment recommendations?
full report below