Three Essays on Corporate Governance and CEO Dismissal

This thesis comprises three separate empirical studies which investigate the key issues in corporate governance, CEO incentives, and firm value. The first study examines the impact of Institutional Shareholder Services (ISS) policy changes on share pledging, focusing on financially constrained (FC) firms. It finds that the 2012 prohibition on share pledging leads to a greater decline in shareholder value for FC firms compared to non-financially constrained (NFC) firms. In response, FC firms adjust CEO compensation by reducing equity and increasing cash pay to mitigate managerial risk exposure and liquidity concerns. The second study applies machine learning (ML) techniques to predict performance-induced CEO dismissals, demonstrating that ML models significantly outperform traditional statistical models in out-of-sample predictive accuracy across multiple performance dimensions. Using SHapley Additive exPlanations (SHAP) values, the study decomposes model predictions to reveal the relative importance of firm-specific factors, where market-based performance metrics contribute materially to dismissal likelihood. Long-short portfolio analyses show that firms managed by CEOs with high ML-predicted dismissal probabilities underperform, yielding positive and significant monthly abnormal returns. These firms also exhibit deteriorating future accounting performance, measured by industry-adjusted ROA, suggesting that ML models effectively identify CEO quality and are effective in predicting firm performance. The third study extends the second study by examining the timing of CEO dismissals. The study classifies dismissal events into delayed, timely, and unexpected dismissals based on ML-predicted probabilities and finds that governance structures play a crucial role in dismissal timing. Specifically, firms with entrenched governance—high E-index scores and CEO duality—are more likely to delay dismissals. Event study analyses further reveal that the market reacts negatively to unexpected dismissals, while expected dismissals (delayed + timely dismissals) elicit no significant response, suggesting the ML model’s capacity to align with market expectations regarding CEO ability.

Name thesis:degree_name

PhD

Level thesis:degree_level

Doctoral

Discipline thesis:degree_discipline

Finance

Grantor dc:publisher

ResearchSpace@Auckland

Year dc:date.issued

2025

Author dc:creator

• Zeng, Juebin

Advisors dc:contributor.advisor

• Berkman, Henk
• Helen, Lu

• Finance
• Corporate Finance
• CEO turnover
• Share pledging

Statement dc:rights

• Items in ResearchSpace are protected by copyright, with all rights reserved, unless otherwise indicated.

Licence dc:rights.uri

• https://researchspace.auckland.ac.nz/docs/uoa-docs/rights.htm

Handle dc:identifier.uri

https://hdl.handle.net/2292/73576

OAI identifier oai:identifier

oai:researchspace.auckland.ac.nz:2292/73576