Please rotate your device to landscape mode to view the charts.

Selecting Directors Using Machine Learning

Journal: Review of Financial Studies

Date: 20210701

Author: Erel, Isil; Stern, Lea H.; Tan, Chenhao; Weisbach, Michael S.

Abstract:
Can algorithms assist firms in their decisions on nominating corporate directors? Directors predicted by algorithms to perform poorly indeed do perform poorly compared to a realistic pool of candidates in out-of-sample tests. Predictably bad directors are more likely to be male, accumulate more directorships, and have larger networks than the directors the algorithm would recommend in their place. Companies with weaker governance structures are more likely to nominate them. Our results suggest that machine learning holds promise for understanding the process by which governance structures are chosen and has potential to help real-world firms improve their governance.

Link: Google Scholar

Key Findings

Machine Learning Predicts Director Performance

Algorithms successfully predict which directors will receive low/high shareholder support, with directors predicted to perform poorly receiving -3.1% excess votes compared to +1.2% for those predicted to perform well.

Market Reacts to Predicted Performance

Directors predicted to perform poorly by the algorithm have negative announcement returns (-1.94%) while those predicted to perform well have positive returns (+0.75%).

Governance Quality Affects Director Selection

Firms with weaker governance (higher E-index scores) are more likely to select predictably poor-performing directors, suggesting agency conflicts influence nomination decisions.

Director Performance Predictions

  • XGBoost algorithm predictions show clear relationship between predicted and actual performance
  • Directors in bottom decile of predicted performance receive -3.1% excess votes
  • Directors in top decile receive +1.2% excess votes

Market Response to Director Appointments

  • Clear market differentiation between predicted high and low performing directors
  • Negative announcement returns for predicted poor performers (-1.94%)
  • Positive announcement returns for predicted strong performers (+0.75%)

Governance Quality and Director Selection

  • Higher E-index scores indicate weaker shareholder rights
  • Positive relationship between E-index and selection of predictably poor directors
  • Suggests agency conflicts influence director nomination process

Contribution and Implications

  • First study to demonstrate machine learning algorithms can effectively predict director performance
  • Provides evidence that agency conflicts affect director selection process
  • Suggests potential for algorithmic tools to improve board selection decisions
  • Highlights importance of governance quality in director nomination process

Data Sources

  • Excess votes performance chart based on Table 3 showing predictive accuracy of machine learning models
  • Announcement returns visualization based on Table 5 showing CARs around director appointments
  • Governance quality chart based on Table 7 showing probit regression results for predictably bad director selection