Are Cartel Fines Optimal? Theory and Evidence from the European Union

The damages are cumulative: the amount of damages to be paid if the cartel is detected is the sum of the welfare loss in the current period (possibly multiplied by a given factor) plus the damages in the previous periods, discounted to take account of the difficulty to evaluate past welfare losses. [...] The deterrence benchmark is the dynamic deterrent fine DF defined and characterized previously, while the compensation benchmark CF is the total amount of excess profit captured by the firm over the total duration of the cartel, that is, with the 13We keep this cartel for the dynamic fines estimations, but we remove it from the sample for the assessment of the compensatory fine. [...] Interestingly, they find that the explanatory variables of the second group also have significant impacts on the overcharge estimate although the regressors of the first group explain the variability of the overcharge estimates to a greater extent than the regressors of the second group. [...] Our methodology follows that of Combe and Monnier (2011) who compare in each case the actual fine with their definition of both the compensatory fine, which corresponds to the annual excess profit as a percentage of cartel sales times the duration of the cartel, and the deterrent fine, which corresponds, in their approach, to the compensatory fine divided by the annual probability of detection. [...] But in other cases, the EC gives a range of values: we then report the average value of the range, and the error term indicates the precision of the range.