Regulatory Sanctions and Reputational Damage in Financial Markets

Transcription

1 Regulatory Sanctions and Reputational Damage in Financial Markets John Armour (Oxford) Colin Mayer (Oxford) Andrea Polo (Pompeu Fabra) BFI Lecture April 10, 2013

2 Overview 1. Motivation 2. Theory and prior literature 3. Institutional structure and hypotheses 4. Data and methodology 5. Results 6. Conclusion and implications

3 Motivation

4 On 3 June 2010, JP Morgan was fined 33m ($51m) by the UK s Financial Services Authority ( FSA ) for failing to protect client money by segregating it appropriately. On 9 September 2010, Goldman Sachs was fined 17.5m ($27m) for weaknesses in controls resulting in failure to provide FSA with appropriate information

5 Research questions What is market reaction to announcements of penalties imposed by regulators? Do financial regulators public announcements of sanctions for misconduct by publicly-traded companies impose reputational losses on those companies, over and above the sanctions themselves? Does naming lead to shaming? If so, how big are these losses?

6 Calibrating Enforcement To deter, regulation requires effective enforcement How do we measure effectiveness? formal penalties (fine dollars)? what about informal (reputational) sanctions? Regulators need to understand reputational sanctions to avoid over- or under- deterrence

7 Regulatory policy Coffee, 2007 Traditionally, UK regulators relied far less on formal sanctions than did their US counterparts. But the FSA has recently signalled a significant increase in formal penalties (FSA, 2010). Where do reputational sanctions fit in?

8 The press view The threat of fines from the FSA are seen as a footling expense, just another cost of doing business, no different from paying the quarterly phone bill. The embarrassment factor no longer counts for much, alas. There is not much shame in being on the receiving end of a fine. Only the size of the fine has come to matter. In some areas, this has proved laughably inadequate in producing better behaviour (The Times, July 7, 2009)

9 Theory and Prior Literature

10 What is a reputational loss? Reputation: expectations of trading partners as to quality of performance: affects terms of trade Reputational loss: misconduct causes trading partners (customers, suppliers, investors) to become less willing to trade with firm in the future: terms of trade become more expensive.

11 Measuring reputational loss Abnormal Returns Company X around the announcement of a penalty Financial payment Reputational loss Total market loss Reputational loss = total market loss financial payment

12 Prior literature on reputational losses General literature on reputation Reputation as an asset: Klein and Laffler, 1981; Shapiro,1983. Reputational losses on revelation of poor performance: e.g. Peltzman,1981 (false advertising); Jarrel and Peltzman, 1985 (product recalls); Mitchell and Maloney, 1989 (air safety disasters). Early literature on reputation and enforcement of regulation Evidence of reputational losses being additive to legal penalties (Karpoff and Lott, 1993; Alexander, 1999) Recent literature on reputation and enforcement of regulation Study Misconduct Financial payments? Loss in market value? Karpoff, Lott & Wehrly (2005) Environmental damage by US firms Yes: fines and compensation Yes: equal to financial payments Karpoff, Lee & Martin (2008) Financial misstatements by US firms Yes: fines and compensation Yes: bigger than financial payments

13 The US enforcement environment gives noisy measures E.g.: Karpoff et al. (2008) a) Multiple events b) SEC enforcement actions as trigger event for private litigation Need to assume investors make good estimates of sizes of subsequent fines and class action settlements when news of misconduct becomes public

14 Institutional Framework and Hypotheses

15 Why study the UK? Existing literature on reputational sanctions (focusing on US) gives only a noisy estimation of reputational losses. This is due to the structure of the SEC enforcement process and the role of class action lawsuits. Enforcement actions in the UK (by the FSA and LSE) provide a uniquely clean dataset to calculate pure reputational losses

16 UK institutional structure FSA was UK s integrated financial regulator, Conduct of business and prudential rules, also insider dealing and listing rules (~ banking and securities regulation) Listing Rules for AIM administered by LSE Powers to sanction: Public censure: FSA, LSE Compensation/restitution orders : FSA, LSE Civil penalties (fines): FSA, LSE Criminal penalties: FSA Announcement of sanction usually a unique event No prior announcement of investigation (FSA, 2010) Announcement reveals financial payment No private enforcement of securities law (Armour et al, 2009)

17 Hypotheses on existence of reputational sanctions H1 Market Reaction announcement has impact on share prices H2 Reputational sanction impact on mkt cap > size of any financial payments H3 Second party vs third party wrongs reaction larger where wrongdoing harms customers/investors than unconnected third party H4 Regulatory penalties reputational loss correlated with size of financial penalty

18 Data and Methodology

19 The Universe of FSA fines

20 The Universe of FSA Fines (2)

21 Database construction FSA/LSE: 83 cases of press statements announcing sanctions imposed on listed companies, or their subsidiaries, in period 2001 Jan 2011 For each event, search Factiva and LexisNexis: for leakage of information about the misconduct in the previous 2 years (28 dropped) For confounding events in the announcement window (5 dropped) Drop cases with change of ownership in the investigation period (3 dropped) Drop cases with uncertainty about the amount of financial payment (e.g. compensation to be determined -7 dropped) 40 cases remain

22 Descriptive statistics Mis-selling of products Misleading advertisements Timing and content of announcements to the market Other Compliance money laundering rules Market capitalization ( m) Financial payment: fine + compensation (as a % of market cap) Wrongs vs Customers/ Investors Wrongs vs.third Parties Market misconduct Transaction reporting failures Other Total Total Post-crisis: 17 FSA: 37 Financial companies: 29 Max Min Median Mean

23 Results

24 Results: ΔVt Announcement window Announcement return (%) t-statistic (0) -1.26*** (0,1) -1.16** (-1,1) -1.68** -1.97

25 Who suffers? Customers/Investors vs. Third parties 0.50% 0.00% -0.50% Total Customers/ Investors Third party -1.00% -1.50% Market reaction -2.00% -2.50% -3.00% Time window: (-1,1) Total Customers/ Investors Third Party Market reaction -1.68** -2.62*** 0.24 t statistic

26 Who suffers? Customers/Investors vs. Third parties 1.50% 1.00% 0.50% 0.00% -0.50% -1.00% -1.50% Related Party Third Party -2.00% -2.50% -3.00% -3.50%

27 How much of this is fines? 1.00% 0.50% 0.00% -0.50% Total Customers/ Investors Third party -1.00% Fine Market reaction - fine -1.50% -2.00% -2.50% -3.00% Total Customers/ Third party Investors Fine (%)

28 How much of this is compensation? 1.00% 0.50% 0.00% -0.50% Total Customers/ Investors Third party Compensation -1.00% Fine Reputational sanction -1.50% -2.00% -2.50% -3.00% Total Customers/ Third party Investors Compensation

29 Is the reputational sanction significant? Market reaction (%) Financial payment (%) Reputational effect (%) Rep effect t statistic Total Customers/ Investors Third party -1.68** -2.62*** ** -2.31**

30 Robustness checks Non parametric test statistics for significance of market reactions yield the same inferences. Outliers: winsorized abnormal returns (5 th and 95 th percentiles) but also robust to dropping outliers Choice of the model of expected returns: same results with market model (a=0,b=1) Results robust to exclusion of LSE cases and also to the exclusion of 5 cases for which it may be questioned that they were totally unexpected (not reported) Test of the effect of the filter: we obtain the same results in terms of share price reactions if we reintroduce the filtered cases. Similar results in the sample of 80 cases.

31 Determinants of reputation effect Customers/Investors 0.028*** (0.001) 0.015** (0.034) Financial penalty (0.240) Market size * (0.075) Post-crisis 0.021*** (0.009) Industry fixed effects (financial vs. non financial) No Intercept (0.363) No. obs Yes 0.083* (0.060) F Prob.>F (0.001) (0.000)

32 Robustness checks Very similar results if we: run Tobit regressions instead of OLS; winsorize or drop outliers for the variable financial penalty, both in the OLS and in the Tobit regressions (except Market size which is no longer significant); bootstrap the standard errors; Control for potential endogeneity of Financial penalty: 2SLS regressions where Clearly-profit enhancing is used as an instrument for Financial penalty gives same results

33 Do reputational losses reflect forgone profits? Clearly profit enhancing (0.979) (0.858) Market size ** (0.014) ** (0.044) Post-crisis 0.028*** (0.009) 0.028** (0.016) Financial penalty (0.574) Industry fixed effects (financial vs. non financial) Yes Yes Intercept 0.109*** 0.124** (0.006) (0.023) No. obs F Prob.>F (0.001) (o.001)

34 Long- term impact: failure rate of unlisted companies penalised for harming customers Year Number of sanctioned companies Companies not active in December 2011 Matching sample A (n=85) Matching sample B (n=170) Tot Failure rate 33.0% 10.6% 12.9%

35 Conclusions and Implications

36 Conclusions FSA and LSE enforcement sanctions provide a unique dataset to calculate pure reputational sanctions Key findings: Reputational sanctions are substantial for wrongdoing against customers/investors: stock price impact is on average almost nine times larger than the financial payments The market ignores (if not rewards) wrongs against third parties Reputational losses are: Not correlated with size of penalties levied Higher reputational following financial crisis (Weak significance) higher for smaller firms Enduring adverse effect on the solvency of private companies penalised for harming customers

37 Policy suggestions Regulators / other public enforcers should take into account private market reaction to enforcement announcements Market provides powerful sanction where corporate reputation with customers and investors is at stake Such reputational losses are hard for regulators to calibrate To avoid Type I error, announce investigation only if firm found guilty Where wrongs are to third parties then regulatory penalties have to do all the work in deterring wrong-doing Penalties should be set larger than for harms to customers / investors No need for secrecy regarding investigations