Compensation, status, and press coverage of managers in the U.S. follow a highly skewed distribution: a small number of ‘superstars’ enjoy the bulk of the rewards. However, the “tournament” for CEO status and public attention is not designed by shareholders as an incentive device, but is largely conducted by the media. As a result, the value consequences of superstar status are unclear. While increased media exposure may boost profitability, it could also shift power towards the CEO and induce perquisite consumption. In our paper, Superstar CEOs, which was recently accepted for publication in the Quarterly Journal of Economics, we analyze the ex-post value consequences of the managerial superstar system.
We use several empirical methods to identify a credible counterfactual for the winning CEOs. As our main identification strategy, we construct a nearest neighbor matching estimator. We estimate a logit regression to identify observable firm and CEO characteristics that predict CEO awards. We then match each award winner to the non-winning CEO who, at the time of the award, had the closest predicted probability of winning. Lastly, we verify that award winners and the control sample are indistinguishable along most observable dimensions, including firm and CEO characteristics not explicitly included in the match procedure. We exploit shifts in CEO status due to CEO awards conferred by major national media organizations. We link award-induced changes in status to corporate performance and CEO decision-making, using matched non-winning CEOs as a benchmark.
We find that firms with award-winning CEOs subsequently underperform, both in terms of stock and operating performance. At the same time, CEO compensation increases, CEOs spend more time on activities outside the company like writing books and sitting on outside boards, and they are more likely to engage in earnings management. The ex-post effects are strongest in firms with poor corporate governance, compared to a matched sample of non-winners with no ex ante differences in governance. Our findings suggest that the superstar system has negative ex-post value consequences for shareholders. The net effect of the superstar system, after accounting for ex-ante incentives created by the tournament for status, is hard to assess. However, the ex post value destruction we measure is large and it appears to be avoidable. Firms with strong shareholder rights do not experience a decline in performance when their CEOs win awards, suggesting that it is optimal to increase monitoring after CEOs win awards.
The full paper is available for download here.