Penalising on the basis of the severity of the offence: A sophisticated revenue-based cartel penalty Yannis Katsoulacos (Athens University of Economics & Business) Evgenia Motchenkova (Vrije Universiteit Amsterdam) David Ulph (University of St Andrews)
Background In most jurisdictions penalties for infringement of competition law are based on the revenue earned by infringing firm(s). Moreover the baseline penalty rate applied to this is pre-announced and fixed Some discretion for Competition Authority (hereafter CA) to modify penalty rate often as a way of penalising/rewarding other behaviours. In many jurisdictions this baseline penalty is 10% So can characterise most existing penalty regimes as what we call simple revenue-based penalty regimes: fixed penalty rate applied to revenue as base
Attractions Relevant revenue is relatively easy to observe Fixed rate that applies across all cases/industries No need for any other information/calculations Together these imply that simple revenue-based penalty regimes have two attractive features: Legal Certainty at time firms make decision to take anti-competitive action they know the penalty they will suffer if successfully prosecuted Ease of Implementation Relatively easy/cheap for CA to implement - not much scope for appeal.
Issues 1: Penalty Rate In context of cartels, John Connor and co-authors have published an extensive number of papers arguing that existing penalty rates are too low and should be raised by factor of at least 5. Katsoulacos & Ulph (2013) argued that existing evidence supported a penalty rate of around 30% A key factor determining the appropriate penalty rate is the average cartel overcharge.
Issues 2: Penalty Base Katsoulacos & Ulph (2013), Bageri, Katsoulacos & Spagnolo (2013) and others have shown that under any simple revenue-based penalty regime cartels will set price above monopoly price, and, moreover, the higher the penalty rate the higher will be the cartel price. Intuition: Absent CA, cartel would set monopoly price where MR = MC > 0. With CA it faces risk of fine based on revenue so wants to cut revenue. Since MR > 0 it cuts output and so drives up price. As we show in this paper, under simple revenue-based penalty regime deterrence varies across industries and is lower in those industries with higher monopoly overcharge. So these simple revenue-based penalty regimes have poor welfare properties.
Background Literature In Katsoulacos, Motchenkova and Ulph (2015) hereafter KMU (2015) - we showed that an overcharge-based penalty regime whereby the penalty imposed by the CA was calculated as a fixed penalty rate applied to a base that was the (percentage) overcharge actually set by the cartel multiplied by the revenue it would have earned in the counterfactual situation had the desirable welfare properties of: (i) inducing all cartels to price below the monopoly price; (ii) achieving same deterrence as the existing simple revenue-based penalty regime. BUT while CAs often need to calculate the overcharge for calculation of damages (and there are well-developed techniques for doing so), the need to calculate counterfactual revenue is much more problematic and so suggests this penalty regime does badly in terms of both legal certainty and ease of implementation
Contribution of this paper We show that a sophisticated revenue-based penalty regime whereby the penalty base is once again revenue, but the penalty rate applied to that base varies with the cartel overcharge according to a simple formula that is constant across all industries/cases performs well on ALL criteria: Like an overcharge-based regime it has the desirable welfare-property of inducing all cartels to price below the monopoly price. Deterrence is same across all industries whatever the overcharge, and but can achieve equivalent deterrence to that of existing simple revenue-based regimes for the average industry Because it requires no information other than the overcharge set by the cartel and is based on a simple formula that can be pre-announced this generates legal certainty For the same reason it also satisfies ease of implementation.
The Model 1: Industries Range of industries In a given industry there are n firms, each with constant unit costs of production, c > 0, producing a homogeneous product with demand Q( p) pq( p) characterised by elasticity function ( p) 0 Q( p) which is assumed to be strictly increasing. Bertrand competition, so counterfactual price is c p c The percentage overcharge is, so pc(1 ) c Industry Profits (resp. Revenue) are: ( ) c Q c(1 ), resp. R( ) c(1 ) Q c(1
The Model 2: Competition Authority There is a constant probability, 0 1 that in each period a cartel will be detected, successfully prosecuted and penalised. The penalty imposed on cartel with detected overcharge, θ, and revenue R is ( )R As is KMU (2015), Chen and Rey (2013) we assume that following prosecution, a cartel re-forms, so if cartel sets an overcharge, θ, the expected present value of profits per cartel member is: (1 ) ( ) (1 ) (1 ) c Q c c Q c V ( ) where Δ = n(1 δ) is intrinsic difficulty of holding cartel together - KMU(2015) and δ, 0 < δ < 1 is discount rate.
The Model 3: Cartel Stability If cartel member defects, for a single period it can set a price that undercuts cartel and gets entire industry profits. Thereafter cartel implements grim trigger strategy and reverts to competition for ever more. Since cartel might set overcharge above monopoly price, defection profits are: M M d, ( ) M ( ), Cartel stability condition: d V ( ) ( )
Main Results: Cartel Pricing Cartels will choose θ to maximise V(θ) subject to cartel stability condition. Pricing Result Provided (1) then in ANY industry the cartel price will be below the monopoly price. Notes (a) ( ) 1 ( ) (1 ) Because penalty base is revenue, then ceteris paribus there is still a pressure to raise prices as explained in intuition above. To offset this the penalty rate has to rise sufficiently fast with the overcharge, and (1) makes precise what sufficiently fast means. (b) Condition (1) doesn t depend on industry characteristics, so penalty rate schedule ρ(θ) can be chosen to be same across industries and so no information other than overcharge actually set is needed to determine penalty. So legal certainty (c) Simple linear penalty schedule ( ) 1. satisfies (1) so easy to implement (d) Overcharge-based penalty equivalent to sophisticated revenue-based penalty that satisfies (1) but it incorporates industry characteristics and so lacks legal certainty and ease of implementation
Main Results: Deterrence 1 Under a simple revenue-based penalty regime with a fixed penalty rate 0 0 the fraction of cartels that would have formed in the absence of a CA that are deterred from forming in the presence of a M 1 CA is D0 0 M and so varies across industries. Indeed it is a decreasing function of the monopoly overcharge. Under a linear sophisticated revenue-based penalty regime with a penalty rate schedule. 0 1 the fraction of cartels that would have formed in the absence of a CA that are deterred from forming in the presence of a CA is D1 1 and so is the same across all industries.
Main Results: Deterrence 2 If we set where then: M 1 0 1 M M is the average monopoly overcharge across all industries 1. In industries with an average monopoly overcharge a linear sophisticated revenue-based penalty regime will achieve exactly the same deterrence as a simple revenue-based penalty regime. 2. But it will achieve greater deterrence in industries with aboveaverage monopoly overcharges.
Choosing the penalty schedule So if CAs are to use a linear sophisticated penalty regime of the sort proposed here, how do we choose 1? Connor & Bolotova (2006) suggest that the average cartel overcharge is approximately 30% while Boyer & Kotchoni (2011) suggest a figure of around 15%. Since, under existing simple revenue-based penalty regimes, cartel overcharges are above the monopoly overcharge, corresponding figures for average monopoly overcharges could be 0.25 and 0.125 respectively. So using a figure of gives corresponding figures of. 0 0.1 1 0.5 and 0.9 Applied to industries with the average cartel overcharge the penalty rates would be 15% and 13.5% respectively. If a cartel set an overcharge 4 times the average, the penalty would be 50-60% depending on what view had been taken on average overcharge.