Does EU competition policy support inclusive growth?

Does EU competition policy support inclusive growth? Adriaan Dierx and Anna Thum-Thysen European Commission* 2 nd Annual Conference on Structural Reforms "Taking Structural Reforms Forward Why and How?" GPEARI Ministry of Finance of Portugal Lisbon, 6 July 2017 *This presentation has been prepared under the sole responsibility of its authors. It does not represent the views of the European Commission 1

This presentation is based on a study made by the Competition Policy Directorate General of the European Commission in collaboration with the Directorate General for Economic and Financial Affairs and the Joint Research Centre. See: Adriaan Dierx, Fabienne Ilzkovitz, Beatrice Pataracchia, Marco Ratto, Anna Thum-Thysen and Janos Varga, "Does EU competition policy support inclusive growth?", paper accepted for publication in the Journal of Competition Law and Economics. 2

Outline Presentation Does EU competition policy support inclusive growth? I. Motivation II. III. IV. Analytical Framework Main assumptions The model V. Main results VI. Further research 3

I. Motivation Advocacy and legitimacy I. Motivation Increased skepticism about benefits of competition policy Greater need to assess the effects of competition policy not only on GDP growth but also on job creation and inequality. 4

I. Motivation Income Inequality has increased in most OECD countries since the mid-1980s. 5

II. Framework II. Analytical framework 6

II. Framework Novelties of this approach 1. Measure the strength of EU competition policy by the number and importance of key merger and cartel decisions taken by the European Commission 2. Reinforce the micro-foundations of a macroeconomic assessment of EU competition policy by making a link between the strength of competition policy and mark-up shocks 3. Propose a method to integrate the deterrent effects of competition policy 4. Investigate not only the macroeconomic but also the distributional effects of competition policy 7

III. Main assumptions III. Main assumptions used in the modelling exercise i. Assumptions regarding the direct effects of competition policy decisions ii. Assumptions regarding the deterrent effects of competition policy decisions iii. Use standard calibration of DSGE model QUEST and apply mark-up shock 8

III. Main assumptions i. Direct effects of competition policy decisions Competition policy intervention Cartel prohibition Merger decision Affected turnover Avoided price increase* Turnover of cartel members Case-specific information Total turnover in the relevant market 10-15-20% Empirical literature: Median price overcharge for cartels = 16-25% Case-specific information 1-3 - 5% Empirical literature: Mean price change due to remedied mergers 3% Expected duration of harm avoided 1/3/6 Case-specific information: Summary indicator of the stability of the cartel 2/3/5 Case specific information: Assessment of barriers to entry and expansion * Baseline scenario in bold 9

III. Main assumptions ii. Deterrent effects of competition policy decisions Deterrent effects extend from affected market to subsector defined at the NACE 4-digit level: - Mergers are more likely to be abandoned or modified following a recent inquiry in the sector - Cartel detection in a market decreases the rate of cartel formation in related markets To avoid implausible multiplication effects, thresholds* for the ratio of deterrent effects over direct effects are defined: - 5-10-15 for merger decisions (literature: 6-17) - 10-20-30 for cartel decisions (literature: 10-30) * Baseline scenario in bold 10

III. Main assumptions iii. Mark-up shocks applied to the model Database of EU merger and cartel decisions provides information on overcharges, their duration and the size of the affected market in 2012, 2013, 2014, 2015 and 2016 allows calculation of direct mark-up shock. Permanent mark-up shock is applied. Information on size of direct and deterrent effects under the baseline scenario is used to calculate a total mark-up decline of 0.9 pp, i.e. a reduction of the mark-up from 12.7% to 11.8%. 11

IV. The model IV. The model 2 region open economy DSGE model 2 types of households low skilled, liquidity constrained high-skilled, non-constrained Product market: monopolistic competition with firms charging a mark-up over marginal costs Monetary and fiscal authority Forward-looking utility and profit maximization 12

QUEST3 models: micro founded, Dynamic Stochastic GE model Monetary authority Interest rate (Taylor)-rule Interest rates Representative Household: Maximise life-time utility Interest rates Interest rates Taxes Consumption, investment Labour and capital income Transfers, benefits Firms: Monopolistic competition Maximise profits Taxes Subsidies Fiscal authority: Budgetary rules 13

IV. The model Modelling distributional effects Low-skilled, liquidity constrained households: Income from wages, transfers and benefits only Consume their income every period can increase consumption due to declining prices and increasing wage income High-skilled, non-constrained households: Additional income from capital ownership and the financial market Their income can decrease as a result of competition policy as lower mark-ups lead to lower profits 14

% % V. Main results Macroeconomic and distributional effects 1 0.8 5 0-5 Consumption Wage Profits 0.6-10 Rich Poor 50 years effects 0.4 0.2 0 1 5 10 20 50 Years GDP Consumption Labour productivity Note: deviation from no-shock scenario More intense competition increases GDP, consumption and labour productivity. Competition policy is more beneficial for the low-skilled, liquidity-constrained households in terms of disposable income and consumption. Main income channel: only the high-skilled, financially unconstrained households bear the consequence of lower profits as prices decline. 15

V. Main results Robustness of GDP impact Scenario Merger overcharge Cartel overcharge Sector spill-over Merger deterrence threshold Cartel deterrence threshold GDP effect after five years Baseline 3% 15% Yes 10x 20x +0.41% Lower bound overcharge 1% 10% Yes 10x 20x +0.23% Upper bound overcharge 5% 20% Yes 10x 20x +0.58% Lower bound deterrence threshold 3% 15% Yes 5x 10x +0.23% Upper bound deterrence 3% 15% Yes 15x 30x +0.54% 16

V. Main results Key policy messages 1. Competition policy seems to have a positive and sizeable impact on GDP growth and employment. 2. Competition policy has strong redistributive effects: poorer households (low skilled and consuming all their resources in each period) increase consumption proportionally more than richer households: four times as much after 5 years. 17

V. Main results Key policy messages 3. GDP effects after 5 years increases from 0.2% to 0.6% if we move from the lower bound to the upper bound overcharge scenario. This would plead in favour of competition interventions in more concentrated markets. 4. It would be misleading for competition authorities to define priorities only on the basis of the direct effects of their interventions as the deterrent effects play such an important role. 18

VI. Further research VI. Further research Improve the scenarios including deterrent effects by assuming that the diffusion of such effects at the sector level follows a logistic function Include an assessment of competition policy interventions in other areas than cartel and merger control (abuse and State aid control, e.g.) Model other transmission channels of competition policy (business dynamism, innovation incentives) Continue the work analysing the differential effects of competition policy decisions affecting different sectors 19

Background slides 20

Mark-up shock applied to the model Definition of mark-up: P = 1 + MUP C Can be written as: ΔP P = Δ(1+MUP) + ΔC (1+MUP) C Assume ΔC C ΔP P = ΔMUP (1+MUP) = 0 and Δ(1 + MUP) ΔMUP : ΔMUP = ΔP P (1 + MUP) 21

Mark-up shock applied to the model ΔP P = ΔMUP (1+MUP) ΔMUP = ΔP P 1 + MUP ΔP P k = ΔP MS P nk + n n M k n C k ΔP P n MS nk ΔMUP N = k K N ΔP P k 1 + MUP k GO k GO 22

IV. The model Modelling labour market effects Labour supply: Trade unions act as an intermediary between households and firms Wage levels for both low-skilled and high-skilled workers are set by trade unions in monopolistically competitive markets Utility maximisation by households given the set wage level determines the labour supply Labour demand: Cost minimisation of firms given the set wage level determines labour demand for both low-skilled and high-skilled workers 23

V. Main results Labour market effects (I) More intense competition has a positive effect both on: Labour demand (due to the increased demand for products associated with lower prices and increased incomes); Labour supply (due to the higher real wages) Employment increases for both skill groups Competition policy is beneficial for employment 24

V. Main results Labour market effects (II) Δ ( pp) after n years Baseline scenario 1 5 10 20 50 Employment 0.11 0.28 0.32 0.37 0.35 Positive employment effect both in the short run and in the long run In practice, however, competition policy decisions may lead to lay-offs in the 'very-short' run Labour market adjustment would mitigate this 'very-short' run effect 25