CLS Blue Sky Blog

Algorithmic Trading and How it Affects What Directors Learn from Stock Prices

Algorithmic trading (AT) is one of the most notable financial innovations in several decades and constitutes a substantial portion of recent trading in stock markets. However, evidence on the economic consequences of AT is mixed. On one hand, prior research finds that AT expedites the incorporation of public information into stock prices, especially around earnings announcements. On the other hand, studies document that AT reduces price informativeness, thus adversely influencing the extent to which managers learn from stock prices when making investment decisions. In our paper, we examine how AT affects the extent to which directors on corporate boards learn information from stock prices when making CEO turnover decisions.

Learning from Stock Prices

Stock prices aggregate information that is otherwise dispersed among market participants, and corporate decision-makers use it to guide their decisions. Evidence supporting this proposition is mainly based on how managers – but not directors or other decision-makers – use the information to make investment and earnings forecasting decisions.

Directors, particularly outside directors, require high-quality information in conducting a monitoring role. Their monitoring role includes making CEO hiring and firing decisions that significantly affect shareholder value. However, managers do not always give directors information that may adversely affect the managers’ jobs. At the same time, outside directors are unlikely to become as informed as the managers, so stock prices provide useful information to directors in making CEO turnover decisions. Investors make costly efforts to acquire and trade on information that affects firm value, and that information should include CEO performance and any conflicts between the CEO and the company. If such information is unknown to directors but accounted for in stock prices, directors will have incentives to use it in deciding whether to keep or replace underperforming CEOs.

However, prior research suggests that AT discourages investors’ costly information gathering efforts by free-riding on order flows or front-running informed trades. This is because, facing diminished expected returns on information acquisition efforts, investors reduce the production of information, and as a result, stock prices incorporate less information that could otherwise inform directors about CEO performance and CEO-firm match. Hence, directors would reduce the extent to which they rely on stock prices in CEO turnover decisions when AT increases.

Main Findings

We find that the adverse effect of poor stock-price performance on the likelihood of forced CEO turnover (i.e., turnover-return sensitivity) decreases with AT, consistent with AT’s impeding directors’ ability to learn investor information from stock prices in CEO turnover decisions. Using the 2016 Tick Size Pilot (TSP) program as an exogenous shock that decreases AT, we find that turnover-return sensitivity becomes more negative for treatment firms during the TSP period compared with control firms, mitigating endogeneity concerns associated with the effects of AT on turnover-return sensitivity.

To provide further support for the idea that stock prices offer important information, we conduct two sets of cross-sectional tests and investigate whether the effect of AT on turnover-return sensitivity is greater in firms where price-based director learning is predicted to be stronger.

First, learning models commonly assume that investors collectively have information advantages in assessing a company’s growth opportunities and how macroeconomic factors affect the company. Research also suggests that investors, unlike directors, have an information advantage because they are located around the world. Consistent with these arguments, we find that the negative effect of AT is more pronounced for growth firms, firms with greater exposure to macroeconomic factors, and firms with geographically dispersed investors, where the information that AT crowds out is more likely to be about new to directors.

Second, gleaning decision-relevant information from stock prices can be a challenge because, while directors need to distinguish between price movements due to decision-relevant information and those due to noise, they lack expertise or time. This implies that directors with greater expertise are more likely to learn information from stock prices. Learning models also posit that decision-makers rely more on stock prices when they have relatively little information. Consistent with these arguments, we find that AT moderates the sensitivity of CEO turnover to stock returns to a greater extent when directors have industry expertise as CEOs and when directors are less informed.

Conclusion

Our findings suggest that stock prices aggregate information about CEO performance and CEO-firm match that is otherwise unavailable to directors and that directors incorporate this information into their CEO turnover decisions. Our evidence highlights that the effect of AT goes beyond financial market participants and extends to directors in determining whether to replace poorly performing CEOs. The findings are particularly relevant for policymakers because any conclusions based solely on financial market consequences may be incomplete in assessing the overall effect of algorithmic trading on the economy.

This post comes to us from Jaewoo Kim, an associate professor at the University of Oregon, Jun Oh, a PhD candidate at Cornell University’s Samuel Curtis Johnson Graduate School of Management, Hojun Seo, an assistant professor at Purdue University’s Krannert School of Management, and Luo Zuo, a professor at Cornell University’s Samuel Curtis Johnson Graduate School of Management. It is based on their recent paper, “Algorithmic Trading and Directors’ Learning from Stock Prices: Evidence from CEO Turnover Decisions,” available here.

Exit mobile version