Anticompetitive vs Competitive Explanations of Unilateral Practices: the Indentification Problem

1
Anticompetitive vs Competitive Explanations of
Unilateral Practices the Indentification Problem

• Michele Polo
• (Bocconi University, IGIER and LECG)

2
Antitrust enforcement of Article 82 the debate

• More room for economic analysis
• A practice can be a tool for foreclosure or just
  part of a competitive strategy
• The difficult task in enforcement is to
  distinguish anticompetitive from efficient
  practices.
• We have today a widely recognized understanding
  that the nature of the practice dominant
  position is not enough to conclude for an
  anticompetitive behavior.
• The discussion on article 82 guidelines how to
  introduce in a manageable and predictable way the
  suggestions of modern IO in the analysis of cases

3
Antitrust enforcement of Article 82 a proposal
(2)

• We can set up the enforcement activity in the
  form of solving an identification problem
• Drawing from IO literature, select a precise
  anticompetitive explanation that fits the casegtgt
  empirical predictions
• Select a competitive explanationgtgt empirical
  predictions
• Compare the empirical predictions
• Some factual elements are admissible under both
  stories
• Other factual elements are implied just by one of
  the stories these latter allow identifying the
  explanation consistent with the data.

4
Selective price cuts the RDB case

• We apply our identification approach to a recent
  Italian case
• RDB medium size company active in the
  construction materials industry.
• The product market Aerated autoclaved concrete
  (AAC). Very ad hoc market definition. Small
  market with promising perspectives of growth.
  RDB 70 of the market.
• The practice ITGB, a small competitor active in
  centre Italy, complained that RDB adopted in 2005
  a selective price cuts strategy to foreclose the
  market

5
The anticompetitive explanation

• Selective price cuts targeted rebated to
  subtract sales.
• The decision does not specify which model of
  predation is considered (reputation, financial
  predation, etc.)
• It develops mainly an empirical analysis applying
  the AKZO test to over RDBs 40.000 invoices
  during 2005
• It distinguish three categories of clients for
  2005
• Long term historical RDB clients
• Common clients that purchase from RDB and ITGB
• Subtracted clients that abandoned ITGB
  purchasing only from RDB
• It computes the frequency of below cost invoices
  by type of customer

6
The anticompetitive explanation (2)

• Predictions
• Target clients should receive more aggressive
  offers than long term clients
• The frequency of below cost orders is higher for
  common and subtracted clients compared to long
  term clients
• The overall RDB pricing strategy to target
  clients is not replicable by an equally efficient
  competitor
• Total revenues on common and subtracted clients
  are lower than total costs

7
The competitive explanation

• The market works through a network of agents
  decentralized negotiation process
• Two types of clients
• single-sourcing (contact a single provider) ss
• double-sourcing (contact both providers)ds
• For any conceivable process of price formation
  (bargaining, procurement, etc.) we expect that ds
  obtain better terms than ss (coeteris paribus)

8
The competitive explanation (2)

• How can we map the ss-ds classification into the
  long term-common-subtracted classification of the
  data?
• Common clients include only ds clients
• Long term and subtracted clients ex-post purchase
  form a single provider, but they may be either ss
  (from the beginning) or ds (opting for the better
  deal). Hence, long term and subtracted clients
  include both ss and ds types.
• The frequency of ds is higher (100) for common
  than for long term or subtracted clients

9
The competitive explanation (3)

• Predictions
• ds clients should receive more aggressive offers
  than ss clients
• The frequency of below cost orders is higher for
  common than for subtracted or long term clients
• The overall RDB pricing strategy to ITGB clients
  is replicable by an equally efficient competitor
• Total revenues on common and subtracted clients
  are higher than total costs.

10
The empirical evidence

• Although some of the computations of the
  Authority are questionable, we adopt just the
  evidence provided in the decision.
• We consider first a comparison of the
  predictions on the frequency of below cost orders
• and then of those on the replicability of the
  pricing strategy

11
The empirical evidence (2)
12
The empirical evidence (3)

• The frequency of below (ATC) cost orders is not
  significantly different (at 1) between long term
  and subtracted clients
• Common clients have a significantly higher (at
  1) frequency of below cost orders
• This result is confirmed in the econometric
  analysis on price-cost margins
• the dummy of common clients is negative and
  significant,
• that of subtracted client is not significantly
  different from zero (long term clients as the
  reference)

13
The empirical evidence (4)
14
Empirical evidence (5)

• Replicating RDB pricing strategy towards ITGB
  clients is profitable.

15
Conclusions

• If we set up a proper identification test, we can
  find out the empirical predictions that allow to
  discriminate between a competitive and an
  anticompetitive explanation
• The evidence provided in the RDB decision rejects
  all the predictions of the anticompetitive story
  and is consistent with those of the competitive
  one
• The Authority, failing to identify the crucial
  empirical predictions, considers as partial
  evidence of foreclosure what indeed is complete
  evidence of lawful competition.