Investment Banking Relationships and Analyst Affiliation Bias

In our study Investment Banking Relationships and Analyst Affiliation Bias: The Impact of the Global Settlement on Sanctioned and Non-Sanctioned Banks, we examine the impact of the 2003 Global Analyst Research Settlement on affiliation bias in sell-side analyst recommendations. Affiliation bias refers to the well-known finding that analysts are overly optimistic when their employers have underwriting relationships with covered firms. Using a broad measure of investment bank-firm relationships, we find a substantial reduction in analyst affiliation bias following the settlement for the 12 sanctioned banks. However, we find strong evidence of bias both before and after the settlement for affiliated analysts at non-sanctioned banks. Our results suggest that the settlement led to an increase in the expected costs of issuing biased coverage at sanctioned banks, while concurrent self-regulatory organization (SRO) rule changes were largely ineffective at reducing the influence of investment banking on analyst research at large non-sanctioned banks.

Conflicts of interest within financial institutions have received significant attention from regulators, academics, and the popular media. We focus on the conflict of interest that arises when financial institutions provide both analyst research and investment banking services. Underlying this conflict is the idea that analysts provide optimistic research coverage in an attempt to curry favor with their firm’s existing clients or to win future investment banking business from covered firms. Attention to this type of conflict reached a crescendo in 2001 when New York Attorney General, Elliot Spitzer, began investigating Merrill Lynch and the actions of the firm’s well-known internet analyst, Henry Blodget, in the wake of the dot-com bubble.

The increased scrutiny of analyst research ultimately led to the 2003 Global Settlement, which sanctioned 12 major investment banks. A primary goal of both the settlement and concurrent SRO rule changes was to reduce conflicts of interest by separating the investment banking and research roles within banks. While previous studies suggest that analysts changed their behavior following the settlement, survey evidence and continuing enforcement actions suggest that the conflicts may not have been completely eliminated. Further, prior studies provide little evidence on the differential impact of the settlement on sanctioned and non-sanctioned banks or on the relative effectiveness of the settlement vs. industry-wide SRO rule changes. We argue that the investigation and punishment of the 12 sanctioned banks led to a substantial increase in the expected costs of issuing biased recommendations for this subset of banks. We therefore expect a more pronounced decrease in affiliation bias at sanctioned banks than at non-sanctioned banks.

To test this hypothesis, we investigate analyst affiliation bias at sanctioned and non-sanctioned banks between 1998 and 2009. Our main variable of interest is the analyst’s relative recommendation, defined as the difference between the analyst’s recommendation and the median recommendation across all analysts covering the stock. We test for differences in relative recommendations between affiliated and non-affiliated analysts, allowing for differences before and after the settlement and between sanctioned and non-sanctioned banks. Following prior research, we define an affiliated analyst as one whose employer has an investment banking relationship with the covered firm. However, unlike prior studies which focus primarily on equity underwriting, we measure affiliation through equity underwriting, debt underwriting, and M&A advising relationships.

Consistent with prior studies, we find strong evidence of affiliation bias prior to the settlement for both sanctioned and non-sanctioned banks. However, results from the post-settlement period point to stark differences across banks. While we find some evidence of lingering affiliation bias at sanctioned banks after the settlement, the bias is dramatically reduced—dropping by as much as 81% relative to the pre-settlement period. In contrast, affiliated analysts at non-sanctioned banks continue to exhibit strong bias after the settlement. These results are not driven by the shift of many investment banks from five-tier to three-tier recommendation schemes following the settlement. In addition, the continued post-settlement affiliation bias at non-sanctioned banks is evident in both more frequent positive recommendations and less frequent negative recommendations.

A more detailed analysis of the post-settlement period reveals that affiliation bias at sanctioned banks continues to dissipate in the years following the settlement and is eliminated by the end of our sample period. Moreover, the lingering bias at sanctioned banks immediately following the settlement appears to reflect the recommendations of analysts who were employed prior to the settlement. Over time, these analysts are replaced with new analysts who exhibit no affiliation bias. This distinction between old and new analysts suggests that the long-term reduction in affiliation bias at sanctioned banks reflects a shift in culture, hiring, or training practices following the settlement. We find no such distinction at non-sanctioned banks, where both old and new analysts exhibit affiliation bias throughout the post-settlement period.

Our research contributes to the literature on conflicts of interest within financial institutions and particularly to studies that examine the effects of the Global Settlement on analyst behavior. These prior studies show that Buy (Sell) recommendations became less (more) frequent after the settlement, with the reduction in optimism being most pronounced for investment bank, and particularly sanctioned bank, analysts. We add to this literature by examining the differential impact of the settlement and contemporaneous regulatory changes on analyst affiliation bias at sanctioned and non-sanctioned banks. We also examine the link between affiliation bias and a more robust measure of investment banking relationships that includes equity, debt, and M&A components.

In summary, we identify a sharp reduction in analyst affiliation bias at the 12 banks that were sanctioned in the Global Settlement. This result is consistent with the settlement leading to a significant increase for sanctioned banks in the expected costs of producing biased coverage. At the same time, the limited impact on non-sanctioned banks suggests that industry-wide SRO rule changes were largely ineffective at reducing the influence of investment banking on analyst research.

The complete article is available for download here.