Political Connections and Insider Trading

Transcription

1 Thomas Bourveau 1 Renaud Coulomb 2 Marc Sangnier 3 1 Hong Kong University of Science and Technology 2 University of Melbourne 3 Aix-Marseille University May 13, 2017 Corruption, Tax Evasion and Institutions Conference, Riga

2 Introduction Motivation Motivation Insider trading based on private information is recognized as a crime and regulated in most countries. Expected costs to society: Reducing public trust in the stock market; Discouraging outsiders from investing in equities; Creating conflicts of interest for board members. High risk insiders: directors (board members). 2 / 23

3 Introduction Motivation Becker (1968) s model of criminality as rational behavior: 3 / 23

4 Introduction Motivation Becker (1968) s model of criminality as rational behavior: Trade-off between the expected returns and costs. 3 / 23

5 Introduction Motivation Becker (1968) s model of criminality as rational behavior: Trade-off between the expected returns and costs. The net expected returns are a function: Punishment; Probability to get caught. 3 / 23

6 Introduction Motivation Becker (1968) s model of criminality as rational behavior: Trade-off between the expected returns and costs. The net expected returns are a function: Punishment; Probability to get caught. Do political connections affects individuals propensity to engage in illegal activities? 3 / 23

7 Introduction Motivation Becker (1968) s model of criminality as rational behavior: Trade-off between the expected returns and costs. The net expected returns are a function: Punishment; Probability to get caught. Do political connections affects individuals propensity to engage in illegal activities? Application to white-collar crime (insider trading). 3 / 23

8 Introduction This paper This paper Use the French 2007 Presidential election as a difference-indifferences setting. Compare non-connected directors to directors connected directors to Sarkozy. Election as President: shift in Sarkozy s power. Trades of directors connected to the newly elected President trigger larger abnormal returns. Longer delays in trades disclosure. More likely to trade closer to firms announcements of results. 4 / 23

9 Introduction This paper Trades disclosure delay 5 / 23

10 Introduction This paper Contributions to the literature Determinants of criminal activity. Political connections favor white-collar crime. Importance of social ties in criminal activities. Political connections. Large (and still growing) literature on political connections. Within the political or administrative systems (targeted subsidies, pork-barrel economics, etc.); Between firms and politicians (government contracts, access to finance, quicker bailouts, lower costs from enforcement actions, etc.); Less knowledge on how political connections affect individuals. Our question: Do political connections affect directors behavior in financial markets? We provide suggestive evidence that political connections favor white-collar crimes. 6 / 23

11 Introduction Framework and context Insider Trading Regulation in France Restriction to trade on private information: First law passed in 1970; Monitoring by the Autorités des Marchés Financiers (AMF); Severe penalties in the 2005 Monetary and Financial Code: Two years of imprisonment; e1,500,000 of fine. Legal disclosure requirements: Derived from the Market Abuse Directive (2003/6/EC). European harmonization that mandates disclosure within five business days. Implemented in France in April 2006.Trades are disclosed to the AMF. Informal rules: Board members code of conduct, etc. No trades before official release of results. 7 / 23

12 Introduction Framework and context Why this (French) election? 8 / 23

13 Introduction Framework and context Why this (French) election? French elites are highly concentrated and politically-connected. 8 / 23

14 Introduction Framework and context Why this (French) election? French elites are highly concentrated and politically-connected. Nicolas Sarkozy was elected President on May 6th / 23

15 Introduction Framework and context Why this (French) election? French elites are highly concentrated and politically-connected. Nicolas Sarkozy was elected President on May 6th France is a Republic where the President has a large power. He appoints the director of the AMF. Limited power before 2007 due to his animosity with President Chirac. Anecdotal evidence on the feeling of impunity among friends of Sarkozy. For firms, Coulomb and Sangnier (2014) provide evidence on the value of friendship connections to Sarkozy. 8 / 23

16 Introduction Framework and context Why this (French) election? French elites are highly concentrated and politically-connected. Nicolas Sarkozy was elected President on May 6th France is a Republic where the President has a large power. He appoints the director of the AMF. Limited power before 2007 due to his animosity with President Chirac. Anecdotal evidence on the feeling of impunity among friends of Sarkozy. For firms, Coulomb and Sangnier (2014) provide evidence on the value of friendship connections to Sarkozy. Relevant event to assess the consequences of a change in the perceived probability of prosecution. 8 / 23

17 Data & methodology Data Data Data we need: 1 Group of politically connected directors (treatment group). 2 Counter-factual. 3 Measures of behaviors on financial markets. 9 / 23

18 Data & methodology Data Political connections Directors connected to Nicolas Sarkozy Sarkozy affiliates using two sources: 1 Large contributors to Sarkozy s electoral campaign: List of 584 names leaked by Mediapart.fr in September 2012; Accurate and a priori comprehensive. 2 Direct friends: List of 27 businessmen constructed by Coulomb and Sangnier (2014) from books written by journalists and political pundits. Visible connections. 10 / 23

19 Data & methodology Data 11 / 23

20 Data & methodology Data 12 / 23

21 Data & methodology Data Non-treated directors & trades data Exhaustive insider trading data from Directors Deal: Trader s identity, transaction dates, announcement date, trades size, directors position; 7, 385 trades by 1, 643 unique traders between mid-2006 and mid % of the transactions operated by 43 Sarkozy affiliates. Sarkozy affiliates belong to larger firms and make larger trades. 13 / 23

22 Data & methodology Data Counterfactuals 1 Full: All 1, 600 other traders. 2 Affiliates firms: 351 non-affiliated traders who belong to the same board as at least one Sarkozy affiliate. 3 Matched board members: 63 non-affiliated traders who do not belong to a board that includes a Sarkozy affiliate, matched on firm size, industry, executive status, etc. 14 / 23

23 Data & methodology Data Measures of directors behavior on financial markets Abnormal return. Change in a stock s return that is not explained by the market evolution. 7, 385 calibrations and predictions. Proxy of the trade s information content. 15 / 23

24 Data & methodology Data Measures of directors behavior on financial markets Abnormal return. Change in a stock s return that is not explained by the market evolution. 7, 385 calibrations and predictions. Proxy of the trade s information content. Disclosure within 5 business days limit. Compliance with legal requirements. 15 / 23

25 Data & methodology Data Measures of directors behavior on financial markets Abnormal return. Change in a stock s return that is not explained by the market evolution. 7, 385 calibrations and predictions. Proxy of the trade s information content. Disclosure within 5 business days limit. Compliance with legal requirements. Time to the firm s next results announcement. Compliance with informal rules (and legal requirements). Detailed information about French listed firms public announcements (scrapped from TradingSat.com). Implicit assumption: privately available information is at least weakly increasing as the time to the firm s next public announcement decreases. 15 / 23

26 Data & methodology Methodology Identification method Difference-in-differences design: y it = βsarkozy affiliate i Post-election t + γsarkozy affiliate i + δpost-election t + α + ε it, where β captures the relative change in behavior of Sarkozy affiliates compared to other directors after the election. 1 year before and after the election. Not a discontinuity-style approach. 16 / 23

27 Results Abnormal returns Abnormal returns Dependent variable: 2-day abnormal return on purchases at announcement date Control group: Full (N = 2, 581) Sarkozy affiliate Post-election 0.007*** 0.007*** (0.002) (0.003) Trade/trader covariates & firm fixed effects Yes Control group: Affiliates firms (N = 888) Sarkozy affiliate Post-election 0.006** 0.005* (0.003) (0.003) Trade/trader covariates & firm fixed effects Yes Control group: Matched board members (N = 488) Sarkozy affiliate Post-election 0.015*** 0.015*** (0.004) (0.004) Trade/trader covariates & firm covariates Yes *** p<0.01, ** p<0.05, * p<0.1. White heteroskedastic standard errors in parentheses. OLS regressions. 17 / 23

28 Results Non-compliance with legal time limit Non-compliance with legal time limit Dependent variable: Non-compliance with legal disclosure delay {0, 1} Control group: Full (N = 7, 385) Sarkozy affiliate Post-election 0.174*** 0.204*** (0.035) (0.033) Trade/trader covariates & firm fixed effects Yes Control group: Affiliates firms (N = 2, 324) Sarkozy affiliate Post-election 0.207*** 0.251*** (0.039) (0.039) Trade/trader covariates & firm fixed effects Yes Control group: Matched board members (N = 987) Sarkozy affiliate Post-election 0.145** 0.143** (0.057) (0.058) Trade/trader covariates & firm covariates Yes *** p<0.01, ** p<0.05, * p<0.1. White heteroskedastic standard errors in parentheses. OLS regressions. 18 / 23

29 Results Time to firm s next results announcement Time to firm s next results announcement Dependent variable: Time to firm s next results announcement Control group: Full (N = 6, 390) Sarkozy affiliate Post-election *** *** (8.955) (9.453) Trade/trader covariates & firm fixed effects Yes Control group: Affiliates firms (N = 2, 181) Sarkozy affiliate Post-election *** *** (10.013) (10.636) Trade/trader covariates & firm fixed effects Yes Control group: Matched board members (N = 884) Sarkozy affiliate Post-election *** *** (15.020) (15.680) Trade/trader covariates & firm covariates Yes *** p<0.01, ** p<0.05, * p<0.1. White heteroskedastic standard errors in parentheses. OLS regressions. 19 / 23

30 Results Robustness checks Robustness checks Abnormal returns: Compliance with legal time limit: Time to firm s next results announcement All: 20 / 23

31 Results Robustness checks Robustness checks Abnormal returns:! 7-, 60-, and 120-day market models.! Changes in length of compound abnormal returns.! Sales at announcement date: no effect.! Purchases at transaction date: no effect. Compliance with legal time limit: Time to firm s next results announcement All: 20 / 23

32 Results Robustness checks Robustness checks Abnormal returns:! 7-, 60-, and 120-day market models.! Changes in length of compound abnormal returns.! Sales at announcement date: no effect.! Purchases at transaction date: no effect. Compliance with legal time limit:! Probit model.! Absolute number of days.! Log of number of days. Time to firm s next results announcement All: 20 / 23

33 Results Robustness checks Robustness checks Abnormal returns:! 7-, 60-, and 120-day market models.! Changes in length of compound abnormal returns.! Sales at announcement date: no effect.! Purchases at transaction date: no effect. Compliance with legal time limit:! Probit model.! Absolute number of days.! Log of number of days. Time to firm s next results announcement! Log of time.! Less than one month from event {0, 1}.! Yearly results announcement only.! Time to next event of other type: no effect.! Time since firm s last announcement: no effect. All: 20 / 23

34 Results Robustness checks Robustness checks Abnormal returns:! 7-, 60-, and 120-day market models.! Changes in length of compound abnormal returns.! Sales at announcement date: no effect.! Purchases at transaction date: no effect. Compliance with legal time limit:! Probit model.! Absolute number of days.! Log of number of days. Time to firm s next results announcement! Log of time.! Less than one month from event {0, 1}.! Yearly results announcement only.! Time to next event of other type: no effect.! Time since firm s last announcement: no effect. All:! Standard errors clustered at the firm and at the trader level.! Exclusion of influential observations.! 4 fictitious and 4 non-presidential elections dates ( ): no effect. 20 / 23

35 Results Robustness checks Sarkozy affiliates payoff How much money did Sarkozy affiliates make? 21 / 23

36 Results Robustness checks Sarkozy affiliates payoff How much money did Sarkozy affiliates make? Not possible to observe complete evolution of portfolios (no pre information). 21 / 23

37 Results Robustness checks Sarkozy affiliates payoff How much money did Sarkozy affiliates make? Not possible to observe complete evolution of portfolios (no pre information). Latent benefits: Difference between a stock s evolution and the market index over the 30-day period that follows a purchase. Median (mean) difference-in-differences estimates: (0.039) (0.039) total value of stock purchases by Sarkozy affiliates during the year following the election 30 million euro. 21 / 23

38 Conclusion Conclusion This paper offers suggestive evidence that political connection leads directors to engage (more) in fraudulent activities in financial markets following a shift in power of the politician they are connected to. Political connections may promote white-collar crime on financial markets. 22 / 23

39 Conclusion Thank you for your attention. 23 / 23