Discussion of “The Role of Managerial Overconfidence in the Design of Debt Covenants” Sudipta Basu Is the paper interesting? No accounting content Behavioral finance – but old theory… Managerial hubris... Roll (1986) Still interesting to understand whether bondholders “pierce the corporate veil” (i.e. look through firm characteristics to managers actually making decisions) What’s the paper about? Are overconfident managers more restricted in their business activities? Do managers who do not rush stock option exercises have their judgment questioned by bondholders? Merger restrictions correlated with overconfidence, “abnormal” accruals, M/B, ROA, financing restrictions… Libby Boxes Independent Conceptual Overconfidence Dependent 1 X (X ) Distrust Y (Y) 2 Operational Option delay X 3 5 Debt covenants Y 4 Control Agency costs, Transparency, Profitability Vs (prior), Zs (current causes) Validity threats related to Libby Boxes From easiest to hardest to think about: • Statistical Conclusion Validity (5) • Internal Validity (4) • Construct Validity (2 and 3) • External Validity (1 generalizes to ??) Are the results persuasive? Basic question is how well are potential validity threats addressed? Is a definitive/sharp test presented? Are multiple test results consistent? What is the totality of the evidence? Thin (regression) versus thick (case study, institutional detail) association Theory/External Validity Independent Conceptual Overconfidence X (X ) Dependent 1 Distrust Y (Y) Do we usually distrust overconfident people? How much is “optimal” confidence? How do we identify overconfidence? Can you be both depending on domain? Is overconfidence the same as optimism? Review PSYCHOLOGY research Construct Validity (1) Independent Conceptual Overconfidence X (X ) 2 Operational Option delay X NegativeDependent Proxy 1 Founder CEO? Distrust Y (Y) Family CEO? Very wealthy CEO? 3 Power-hungry CEO Debt covenants Y Try Positive Proxies like N successive optimistic forecasts, MD&A or conference call tone, press or analyst descriptions… Construct Validity (2) Independent 1 What are distrustful actions? Overconfidence Conceptual Avoidance X (X ) Recommend others 2 Limited trial (credit line?) Boundaries (covenants) Option delay Operational X (interest) Compensation Model other distrustful actions —selection bias issues Dependent Distrust Y (Y) 3 Debt covenants Y Internal Validity Easier to observe proxy for overconfidence? repeated optimistic management forecasts conference call 2behavior or MD& A tone 3 Operational Option delay 5 X Debt covenants Y 4 Control Agency costs, Transparency, Profitability Vs (prior), Zs (current causes) WHO observes and decides? Other minor issues Why don’t shareholders fire these CEOs? Abnormal accruals = accounting ignorance Conservatism reduces control denominators Time-series clustering? (mid-1980s?) Benchmark model with only control variables Medians not means (very skewed variables) Include predicted signs in tables