Optimal deterrence of collusion in the presence of agency problems within firms. Cédric Argenton Eric van Damme TILEC & CentER - Tilburg University

Transcription

1 Optimal deterrence of collusion in the presence of agency problems within firms Cédric Argenton Eric van Damme TILEC & CentER - Tilburg University

2 Introduction A lot of discussions about the desirability of high or low corporate and/or individual sanctions for violations of competition law Little economic analysis of this issue This study is first attempt at investigating optimal design of liability system when firms/boards lack full control of managers Blank slate approach: considers pure corporate liability regimes, pure individual liability regimes, mixed regimes 2

3 Main idea Moral hazard within firms: managers can engage in illegal activities that boost profit without anyone noticing in real time If firm gives incentives for high profit, incentive for illegal conduct Under corporate liability, collusion can be deterred but at the cost of decreasing the incentives of managers productively to increase profit altogether Addition of individual liability helps firms solve their agency problem because it directly tilts the choice of managers between legal and illegal profit increases 3

4 Model Public authorities, firm, manager Manager maximizes wage minus cost of efforts minus possible legal sanctions Manager can choose one of three (unobserved) possible actions Do nothing (N) Exercise efforts to increase profits with some probability through socially productive action (say, cut costs) (R) Exercise efforts to increase profits for sure through a socially harmful action (say, collusion) (C) C has higher return than R 4

5 Model (2) Firm observes only bottom line (low profit, high profit) and detection Offers an employment contract to manager based on observable/contractible events: Either low profit/high profit only Or low profit/high profit/ detection Public authorities unable or unwilling directly to regulate this contract Only antitrust liability: penalty for firm and/or manager in case of probabilistic detection ; no subsidies 5

6 Model (3) Society prefers R to N to C Manager must get at least outside option Firm must make positive profit Two key assumptions Firm is not more informed than public authorities not obvious that corporate liability is desirable Individual sanctions are limited (for economic or deontological reasons) whereas corporate sanctions can be very high 6

7 Benchmark: no enforcement Firm decides about wage when profit is low and wage when profit is high If bonus for profit, the manager will choose C over R Society gets worse outcome: C Justification for existence of enforcement system 7

8 First case: firm and managers can contract only on low/high profit Pure corporate liability Firm cannot get manager to choose R Society can always deter C with high sanctions N is always final outcome under optimal system Pure individual liability If sanction is high enough, firm can now get manager to choose R But firm can also (indirectly) insure manager against fine through compensation package If sanctions are high enough, get R If not, get C Never N 8

9 First case: firm and managers can contract only on low/high profit (2) Optimal design If individual sanctions can be high, pure individual liability is fine If not, mixture of corporate and individual liability Individual fine allows firms to have a chance to get manager to choose R Corporate fine deters choice of C Get R or N depending on the strengths of individual sanctions One can always find an optimal rule that punishes manager as much as possible for collusion and then add corporate sanctions 9

10 Second case: firm and managers can also contract on detection Pure corporate liability Now firm can incentivize manager to choose R (withdraw compensation in case of detection) but rents might be left in case the detection system is not efficacious (low probability) If those rents are inexistent or low, get R under best system If those rents are high, get N under best system Remember: before, always get N being able to contract on detection (wage schedule, penalties, damages) is a key feature of corporate liability systems Probability of detection as well 10

11 Second case: firm and managers can also contract on detection (2) Pure individual liability Easier for firm to implement R: rents can be diminished one for one with the individual sanction Get R if penalty is high enough to make it profitable Get C otherwise Does not improve on situation where detection is not contractible (C is also easier to implement) but get R more often than corporate liability Optimal system Mixed system Get R whenever more profitable than N Get N otherwise 11

12 Policy implications Corporate liability has one advantage: higher sanctions deter collusion Individual liability has one advantage: directly discourage manager to engage in harmful conduct, which is of help to a firm that wants to comply Unless individual sanctions can be very high, optimal system typically mixed Easy to get to the second-best (N); harder to get the first-best (R) Key feature: probability of detection 12

13 Experimental test Economic agents do not always behave in accordance with economic theory Especially in contracting situations (good faith, reciprocity) Hard to conduct statistical analysis in this area (little variation in enforcement regimes) Experimental methods: recruit real persons, place them in various conditions, subject them to real monetary incentives, compare behavior 13