1 / 24

Finding Bernie Madoff: Detecting Fraud by Investment Managers

Finding Bernie Madoff: Detecting Fraud by Investment Managers. Stephen G. Dimmock and William C. Gerken. Fraud. On December 11, 2008 Bernie Madoff was charged with a $65 billion investment fraud.

sonya-berg
Download Presentation

Finding Bernie Madoff: Detecting Fraud by Investment Managers

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript


  1. Finding Bernie Madoff: Detecting Fraud by Investment Managers Stephen G. Dimmock and William C. Gerken

  2. Fraud • On December 11, 2008 Bernie Madoff was charged with a $65 billion investment fraud. Introduction Data Prediction Alpha and Fees Hidden Info Reaction Conclusion

  3. The Questions • We test if fraud is ex ante predictable. If so, what predicts fraud? • Is it possible to improve the disclosure requirements mandated by the SEC? • Are there real economic consequences to fraud? Introduction Data Prediction Alpha and Fees Hidden Info Reaction Conclusion

  4. Our Study • The main disclosure requirement in U.S. securities law is that investment advisors with more than $25 million in assets must file Form ADV with the SEC. • We use a panel of all ADV filings from 2000-2006. • 13,579 distinct investment managers • Over 20 million investors • More than $32 trillion in assets under management Introduction Data Prediction Alpha and Fees Hidden Info Reaction Conclusion

  5. Data • Panel of all Form ADV filings from 2000 through 2006. We also have disclosure reporting pages (DRP) which list all criminal violations in detail through 2007. • Current forms are available at: http://www.adviserinfo.sec.gov • Firms must file annually or in the event of a material change. • Measure variables as of August 1st and DRP filings over subsequent year Introduction Data Prediction Alpha and Fees Hidden Info Reaction Conclusion

  6. Data: Filings and Removal Introduction Data Prediction Alpha and Fees Hidden Info Reaction Conclusion

  7. Internal Policies and Fraud Introduction Data Prediction Alpha and Fees Hidden Info Reaction Conclusion

  8. Predicting Fraud: Table 4 Part 1 Introduction Data Prediction Alpha and Fees Hidden Info Reaction Conclusion

  9. Predicting Fraud: Table 4 Part 2 Introduction Data Prediction Alpha and Fees Hidden Info Reaction Conclusion

  10. Robustness • Fraud and number of employees is highly correlated. We control for this but want to be sure this does not inadvertently drive our results. • We estimate a placebo model, where the dependent variable equals one if the firm reports a non-investment crime such as drunk driving. • Also, we split the sample into small and large firms. Introduction Data Prediction Alpha and Fees Hidden Info Reaction Conclusion

  11. Predicting Fraud: Tradeoff 73.3% identified at 5% false positive rate 59.3% identified at 1% false positive rate 33.2% identified at 0.2% false positive rate Introduction Data Prediction Alpha and Fees Hidden Info Reaction Conclusion

  12. Alpha and Fees • Using data from PSN (institutional funds) and CRSP MF (mutual funds), we determine if fraud risk is compensated. • No relation between fraud risk and alpha • No relation between fraud risk and fees Introduction Data Prediction Alpha and Fees Hidden Info Reaction Conclusion

  13. Hidden Information • Firms are required to disclose crimes and regulator violations for 10 years, unless the offender leaves the firm. • If the offender leaves, the violation disappears. • Many violations disappear without explanation. Introduction Data Prediction Alpha and Fees Hidden Info Reaction Conclusion

  14. Predicting Fraud with Hidden Information • Can removed information predict fraud? • Can information that is difficult to observe due to the format of Form ADV predict fraud? • Include the same controls as in Table 4, but do not show them in the interest of brevity. Introduction Data Prediction Alpha and Fees Hidden Info Reaction Conclusion

  15. Predicting Fraud with Hidden Information Introduction Data Prediction Alpha and Fees Hidden Info Reaction Conclusion

  16. Predicting Fraud with Hidden Information Introduction Data Prediction Alpha and Fees Hidden Info Reaction Conclusion

  17. Consequences: Firm Death • Does fraud kill firms? • Estimate a survival hazard model of firms dying in the next year • Report hazard ratios – show the relative probability of firm death compared to other firms • Include controls used in previous regressions Introduction Data Prediction Alpha and Fees Hidden Info Reaction Conclusion

  18. Consequences: Firm Death Introduction Data Prediction Alpha and Fees Hidden Info Reaction Conclusion

  19. Consequences: Flows • Do investors withdraw their money following the disclosure of fraud? • Estimate panel regressions with firm fixed-effects and controls for: returns, portfolio value, assets under management, firm age, # of employees, and time fixed-effects Introduction Data Prediction Alpha and Fees Hidden Info Reaction Conclusion

  20. Consequences: Contemporaneous Flows Introduction Data Prediction Alpha and Fees Hidden Info Reaction Conclusion

  21. Conclusion • Fraud is predictable • Predict 73.3% of frauds with public information • Conflicts of interest and history of violations • Improve predictions using hidden information for high fraud risk firms • Investors react to fraud • Transparent disclosure: 549% increase in firm death, 32% outflows • Non-transparent disclosure: No Effect

  22. Conclusion • Four simple changes would improve investor welfare: • Report the number of past violations • Disclose investment and non-investment crimes separately • Force disclosure of violations in the past year even if removed • Require firms to disclose the number of violations removed before 10 years has passed Introduction Data Prediction Alpha and Fees Hidden Info Reaction Conclusion

  23. Conclusion • Since the SEC has this hidden information on record, and firms are required to report it, the marginal cost of disclosing this information to investors is essentially zero. • Disclosing this information would allow investors to avoid frauds and likely increase the market penalty for fraud. Introduction Data Prediction Alpha and Fees Hidden Info Reaction Conclusion

  24. What happens if investors use our results? Introduction Data Prediction Alpha and Fees Hidden Info Reaction Conclusion

More Related