From Kelly Shue:
I explore how executive social interactions can affect managerial decision making and firm policies using the historical random assignment of MBA students to sections at Harvard Business School. Under the identifying assumption that social bonds are stronger within randomly assigned sections than across sections in the same class year, I test whether executive and firm outcomes are more similar among section peers than among class peers. I find evidence of significant peer effects in firm investment, leverage, interest coverage, and firm size, with the strongest effects in executive compensation and acquisition activity. Section peers are 10% more similar than class peers in terms of compensation and acquisitions. Under the additional structural assumptions of the linear-in-means model, I estimate a substantial lower bound for the elasticity of individual outcomes to mean section peer characteristics of 10%–20%.
I show that peers are also important determinants of executive career outcomes, such as choice of industry, firm, and geographical locale. However, past peer interactions leading to selection into executive roles and into similar types of firms do not drive peer effects in compensation and acquisitions. Rather, the underlying mechanism is due to contemporaneous social interactions: peer similarities in compensation and acquisitions are more than twice as large in the year following staggered alumni reunions, which act as shocks to contemporaneous interactions. Tests of “pay for friend’s luck” further narrow the mechanism driving peer effects in compensation. I find that compensation responds to lucky industry-level shocks to peers in distant industries after controlling for own firm and industry performance. Because these industry shocks alter peer outcomes while leaving peer fundamentals unchanged, peer effects in compensation are not driven only by the sharing of productive managerial skills within peer networks. Rather, relative compensation directly affects individual compensation.
Executive peer effects imply that executives matter for firm policies in a systematic way that can affect aggregate patterns. Multipliers arising from social interactions imply that the aggregate effect of a change in the fundamental determinants of compensation or acquisition activity will be 20% larger than the direct effect because of contagion among connected agents. Positive peer effects also lead to reduced within-group variation and increased across-group variation, that is, clustered financial outcomes. In the context of HBS sections, I find that the ratio of between- to within-section variance is 20%–40% greater than expected under the null hypothesis of no peer effects. A similar peer inter- action mechanism can potentially amplify fundamental differences across other types of executive peer groups operating at the industry or geographic level.
Whereas executive peer effects have clear implications for our understanding of managerial decision making, consequences for social welfare are less obvious. Peer effects in acquisitions can be efficient if private information about how to conduct value-enhancing mergers is transmitted through social networks. However, I find evidence that peer influence in acquisitions does not operate through fully efficient information sharing and may lead to weakly less efficient acquisitions. Similarly, peer influence in compensation can lead to movements in compensation that do not reflect changes in firm or managerial productivity. Empirical investigation of the efficiency implications of peer influence among executives is a promising direction for future research.
HT: Seth Zimmerman