Measuring and Evaluating the Performance of Competition Agencies

1
Measuring and Evaluating the Performance of
Competition Agencies

• Julian L. Clarke Simon J. Evenett,
  www.evenett.com

2
Contents of this presentation

• Five possible contributions of empirical analyses
  of competition regimes.
• Empirical analyses of cartel cases and
  enforcement regimes.
• Empirical analyses of merger cases and
  enforcement regimes.
• Concluding remarks including
• Notes of caution on quantification.

3
Possible contributions of empirical analyses of
competition regimes

• Evaluations of the effects of the competition
  regime on the national economy (and constituents
  thereof).
• Evaluations of the value for money spent on
  competition enforcement.
• Evaluations of the competition regime against
  nationally-determined targets.
• Evaluations of the performance of a national
  competition agency against international peers.
• Identifying steps which could enhance the
  performance of the competition regime (perhaps
  along the metrics implicitly identified in the
  statements above.)

4
Empirical evaluations of cartel cases (1)

• Price impacts
• Distinguish between the price increase caused by
  a cartel and the price reduction that may follow
  a cartels demise after a successful enforcement.
• Observed price changes (e.g. football shirt
  investigation by OFT Levenstein and Suslow (ALJ
  2003) in the international context).
• But for analyses Sproul (JPE 1993).
• Uses prices of related goods to forecast
  counterfactual prices.
• Examines effects of penalties on price impact of
  enforcement.

5
Empirical evaluations of cartel cases (2)

• Estimates of overcharges
• Clarke and Evenett (AB 2003) on vitamins cartel.
• Method used underestimates consumer welfare harm.
• Estimated annual overcharges in UK 14.7m.
• Comparing across countries, we could also
  estimate the additional overcharge if the UK did
  not have an active cartel enforcement programme.
• Estimated additional annual overcharge in the UK
  29.65m.
• Latter as a percent of the budget of the
  competition authorities 64 (computed at time of
  study).

6
Empirical evaluation of merger cases (1)

• Analyses of type I and type II errors made during
  decisions on merger reviews
• requires an independent, and ideally, objective,
  benchmark against which to compare decisions.
• see, e.g., Duso, Neven, and Röller (2003) on EC
  merger.
• Identify type I and II errors in decision making
  and estimate the factors that correlate with
  errors.
• Found type I errors were made in 28 of
  prohibitions.
• Found type II errors were made in 23 of allowed
  mergers.
• Found no evidence that talking to competitors
  induced errors four other factors were important
  including country-, industry-, market definition,
  and procedural aspects.

7
Empirical evaluation of merger cases (2)

• Ex-post analyses of remedies.
• Long history of studies going back to Elzinga
  (1969) (and maybe before!) See Duso, Gugler, and
  Yurtoglu (2005) for a survey.
• Studies by enforcement agencies as well (see
  recent commentary by Davidson, AAI 2006).
• Duso, Gugler, and Yurtoglu (2005) also examine
  type I and II errors in EU remedies decisions.
• EC made type I errors in 46 of pro-competitive
  mergers and type II errors in 57 of
  anti-competitive mergers.

8
Towards estimates of the deterrent effects of
merger review?

• Could cross-country data on the distribution of
  proposed mergers reveal anything about the
  deterrent effects of national merger review?
• Do nations with robust merger review regimes see
  fewer larger mergers proposed in sectors with
  high levels of concentration? (NB Concerns about
  market definition.)
• Evenett (IBA 2003) examined effects of different
  merger review regimes on the amount and average
  size of cross-border MA by US firms
• Found the mandatory pre-notification regimes had
  important effects on these two variables.
• More disaggregated data would allow for plenty of
  room for refinement.
• Could benchmark UK merger review regime against
  other jurisdictions whose regimes are thought to
  be tough.

9
Concluding remarks

• Statistical estimates should complement, rather
  than substitute for, other estimates of agency
  performance.
• Dont put cost-benefit analysis on a pedestal.
• The case for competition law should be made on
  other grounds.
• Seize opportunities in data-rich and important
  cases.
• The search for robust off-the-shelf techniques is
  doomed for two reasons.
• Novelty of the subject matter.
• Concerns about robustness.