Geographic Market Definition Presentation to the CBA Competition Section

1
Geographic Market DefinitionPresentation to the
CBA Competition Sections Young Lawyers
CommitteeNeil Campbell, McMillan LLP Lilla
Csorgo, Competition BureauMargaret Sanderson,
Charles River Associates November 19, 2009

2
Disclaimer

• Discussions of particular policies and cases are
  for teaching purposes only.
• These slides are part of a presentation and
  cannot be fully understood separately from that
  presentation. Ideas presented here are
  preliminary and their intent is to promote
  further discussion and analysis.
• These slides do not necessarily reflect the views
  of the authors organizations.

3
Topics for Discussion

• Identifying geographic market definition issues
• Merger analytical framework
• Practical indicators
• Hypothetical monopolist test
• Testing for regional price discrimination
• Critical loss analysis
• Conduct cases and the cellophane trap
• Canada Pipe
• Making use of market definition evidence

4
Identifying Geographic Market Definition Issues

• Where do the merging parties have overlaps?
• Sales offices / production facilities
• Customers
• Does geographic market matter?
• Market share similarities / differences
• Differences in competitors presence / position
• Price differences
• Note need to consider product market definition
  in parallel
• Similar questions

5
Merger Analytical Framework
Product Geographic Market Definition Market Shares /Concentration Competitive Effects(s. 93) Factors

• Market definition is based on substitutability
  and focuses on demand responses to changes in
  relative prices (MEGs, 3.3)
• May not correspond to standard company or
  industry segmentations
• May not follow political boundaries

• 35 unilateral effects safe harbour
• 65 CR4 coordinated effects safe harbour

• Effectiveness of remaining competition
• Removal of vigorous competitor
• Entry
• Change / innovation
• Countervailing power
• Failing firm
• Efficiencies (possible defence)
• Other

6
Practical Indicators for Defining Geographic
Markets

• Closeness of substitution possibilities
• Demand side extent to which buyers would switch
  suppliers
• Supply side scope for expansion or repositioning
• Potentially relevant factors (MEGs, 3.21-3.26)
• Buyers and trade views
• End use and product attributes
• Price relationships and levels
• Switching costs
• Transportation costs
• Shipment patterns
• Foreign competition

7
Hypothetical Monopolist Test

• Product dimension
• Smallest group of products over which it would be
  profitable for a single firm acting as a
  monopolist to implement and sustain a small yet
  significant non-transitory increase in price
  (SSNIP)
• Geographic dimension
• Smallest geographic area over which it would be
  profitable for a single firm acting as a
  monopolist to implement and sustain a SSNIP
• Conceptual, iterative process

8
(No Transcript)
9
(No Transcript)
10
(No Transcript)
11
Hypothetical Monopolist Test (cont)

• What is a SSNIP?
• Usual test is 5 price increases over 1 year
  (MEGs, 3.4)
• May vary depending upon the product
• What base is used?
• Prevailing prices for mergers (even if not
  competitive)
• Demand responses depend upon price faced by
  customer(e.g., transportation plus tipping fees
  in waste disposal)
• Supply responses depend upon margin faced by
  entrant(e.g., FOB mill price in case of
  delivered pricing where supplier pays for freight
  cost)
• Is price discrimination relevant?
• If different buyers pay different prices for the
  same relevant product, subsets of buyers may
  comprise separate markets over which hypothetical
  monopolist profitably can impose a SSNIP

12
Testing for Regional Price Discrimination

• Merger involving two firms based in Western
  Canada
• Merging parties consider the relevant geographic
  market to be North America given their extensive
  exports to U.S. customers
• There are no imports into Canada
• Regression analysis employed to test whether
  geographic price discrimination exists
• Test whether the netback (per unit sales
  revenues less freight costs) from the merging
  firms sales to customers in the U.S. is similar
  to the netback earned from customers in Western
  Canada
• If netbacks (as a measure of margin) are
  materially higher for sales to customers in
  Western Canada than to U.S. customers, the threat
  of entry from the U.S. does not appear to
  discipline prices in Western Canada
• Common netbacks are a necessary but not a
  sufficient condition for finding that Western
  Canada is part of a wider geographic market

13
Merger Case Practical Example (cont)

• Transaction level data is obtained from merging
  firms
• Prices, quantities, discounts, freight costs,
  customer location, product type for all shipments
  to U.S. and Canadian customers
• Regression specification used
• Log(netback per tonne) a ßlog(volume)
  ?(region dummy)
• d(month dummy) ?(product type dummy) error
• Log-linear regression means the coefficient ?
  measures the difference in netback per tonne
  for sales to customers in other regions relative
  to netbacks earned from customers located in B.C.
• If no regional price discrimination exists, the
  netbacks for sales to customers in B.C. should
  not be significantly (either economically or
  statistically) different from the netbacks for
  sales to customers in other regions

14
Merger Case Practical Example (cont)

• Netbacks are 10-11 lower for sales to customers
  in the U.S. compared to netbacks on sales to
  customers in B.C. and Prairies
• Suggest B.C. Prairies are in one geographic
  market, but U.S. Pacific Northwest and U.S.
  Southwest are in a separate geographic market

Variable Parameter Estimate T-statistic
Intercept 5.970 10.83
Volume -0.005 -1.47
Prairies -0.030 -1.13
U.S. Pacific Northwest -0.104 -2.07
U.S. Southwest -0.113 -2.11
Observations 1,194
R2 0.481
Statistically significant at 1 level
Statistically significant at 5 level
15
Critical Loss Analysis in Market Definition

• No regional price discrimination is a necessary
  but not a sufficient condition for finding a
  single, common geographic market
• Sufficient test is the hypothetical monopolist
  test (critical loss analysis)
• If hypothetical monopolist imposes a SSNIP, are
  the gains in revenue from customers who still
  purchase (area A) greater than the losses from
  customers who are no longer buying (area B)?

P
P1
Demand Curve
A
P0
B
Marginal Cost
Q
Q0
Q1
16
Market Definition Tool Critical Loss Analysis
First Step Contribution Margin
Second Step Critical Loss in Sales
Third Step Lost Customers

• Percentage of customers the hypothetical
  monopolist could lose before the price increase
  becomes unprofitable

Estimate whether the hypothetical monopolist
would lose customers beyond the critical level if
it increased price
Estimate the hypothetical monopolists initial
per unit margin
Contribution margin (initial price marginal
cost) initial price
Critical loss in sales SSNIP ? (SSNIP
contribution margin)
Involves quantifying how many customers would be
lost either to rival firms or by not purchasing
when faced with the SSNIP
17
Critical Loss Analysis - Implications

• As the critical loss formula indicates
• The greater the contribution margin, the smaller
  the critical loss for a given postulated price
  increase
• The greater the postulated price increase, the
  larger the critical loss for a given contribution
  margin

Percentage Contribution Margin Percentage Price Increase Percentage Price Increase Percentage Price Increase
Percentage Contribution Margin 5 10 15
0 100.0 100.0 100.0
10 33.3 50.0 60.0
20 20.0 33.3 42.9
30 14.3 25.0 33.3
40 11.1 20.0 27.3
50 9.1 16.7 23.1
18
Critical Loss Analysis - Adjustments and Caveats

• Important adjustments and considerations
• Alternative production facility use may attenuate
  the effects of lost sales
• Joint product production
• Margins may be high for a reason (e.g., inelastic
  demand) therefore do not assume high margins
  always mean broad geographic markets because the
  critical loss is small
• Assumptions
• No price discrimination
• Marginal cost is constant over the range of
  output relevant to the postulated price increase
• Average variable cost is often used as a proxy
  for marginal cost
• Accounting data is typically used to calculate
  the average variable cost, but this introduces
  measurement problems

19
Critical Loss Analysis Practical Example

• Geographic market analysis for a possible ChemCo
  / TargetCo merger separate East and West
  markets vs. all Canada?
• Available data only for ChemCo plants throughout
  North America
• Assume hypothetical monopolist of all Western
  Canada plants
• Hypothesize 5 price increase in Western Canada
  only
• Price increase is realized by shutting down
  enough Western Canada capacity to increase price
  by 5 given elasticity of demand
• Calculate the contribution margin associated with
  the shut down capacity by assuming production is
  shut down at the highest cost plants find 2
  plants need to be shut down to remove enough
  volume in Western Canada to raise price by 5
• Contribution margin associated with the 2
  shut-down plants is roughly 20
• Forgone margin 20, price increase 5 ?
  critical loss 20

20
Critical Loss Practical Example (cont)

• Will actual loss gt 20 if hypothetical
  monopolist in West raises price by 5?
• Assume prices in East remain the same and assume
  netbacks earned by rival firms are similar to
  ChemCo netbacks in East
• Calculate existing ChemCo netbacks (Transaction
  Price - Freight Cost) for each eastern plant for
  sales to current customers
• Assume prices in Western Canada rise by 5 and
  calculate potential netbacks from each eastern
  plant to ship to customers in the West
• Compare post-price-increase netbacks in West with
  netbacks for current customers in East if
  netback to shift an Eastern sale to a Western
  customer is improved assume East volume is
  diverted to West
• Calculate total volumes that would be diverted to
  West, assuming diversion begins with the highest
  netback opportunity
• Is the total volume diverted from East to West gt
  20 of total Western volume?

21
Critical Loss Practical Example (cont)

• Findings
• Price increases up to 5.2 in Western Canada
  induce cumulative potential divertible volume
  from ChemCo plants of 41,414 tonnes
• Scaling ChemCo divertible volumes by ratio of
  total plant capacity to ChemCo plant capacity in
  Eastern Canada (5.64) yields potential industry
  divertible volume from Eastern Canada to Western
  Canada of 233,706 tonnes
• This represents 37 of total Western Canada
  sales, which exceeds the critical loss of 20
• Assumes competitors have freight costs, netbacks
  and diversion opportunities similar to those of
  ChemCos Eastern plants
• Assumes no contract constraints would restrict
  diversion of sales from Eastern customers to West

22
Conduct Cases and the Cellophane Trap

• The wrong base price can lead to overly large
  markets (cellophane trap) or overly small ones
  (reverse cellophane trap)
• Cellophane trap If the prevailing price is one
  that already exhibits substantial market power as
  result of anticompetitive conduct, a further
  (hypothetical) price increase in relation to that
  price may cause buyers to switch to products they
  would not normally consider as substitutes
• U.S. v. Dupont the alleged monopolist Dupont had
  priced its cellophane wrap product so high that
  substitution of less desirable wrapping materials
  finally occurred
• Reverse Cellophane Trap Some suggest that the
  competitive price be used as the base price
  instead to avoid the trap, but using the
  competitive price when market power has already
  been exercised can lead to overly narrow markets

23
Reverse Cellophane Trap Example Canada Pipe

• In Canada Pipe, prevailing prices were used
• No switching to imports in face of a price
  increase was found, but the absence of switching
  was possibly due to the anti-competitive act (the
  stocking distribution program) and the related
  penalties associated with switching
• It is expensive to transport cast iron product
• As a consequence, if price is at the competitive
  level (i.e., at marginal cost or even average
  cost), applying the hypothetical monopolist test
  would likely find that imports are not in the
  relevant geographic markets
• The price that would prevail absent the
  anti-competitive act (the but-for price) did
  not appear to be the competitive price
• At the but-for price, imports would be in the
  market

24
Abuse Case Example Canada Pipe

• Six regions of Canada were defined as separate
  markets
• B.C., Alberta, the Prairies, Quebec, Ontario and
  the Maritimes
• Commissioner submitted that Canada Pipe had at
  least 82 market share in each geographic market
• Canada Pipe has production facilities in only one
  of the geographic markets, Quebec
• Finding on geographic market would imply that
  buyers located in B.C. would not find sellers
  located in Quebec to be adequate substitutes
• How can the location of Canada Pipe be reconciled
  with its participation in all six geographic
  markets?
• In considering the potential effects of the
  alleged anti-competitive acts, the Tribunal
  considered the effects on imports from the U.S.
• How can this be reconciled with a geographic
  market that consists of regions of Canada?

25
Conundrum 1 Correlations and Price Discrimination

• Tribunal decided in favour of six markets based
  on evidence that prices across those six regions
  were not correlated (para. 112).
• MEGs When price discrimination is feasible, it
  may be appropriate to define relevant markets
  with reference to the characteristics of the
  classes of buyers or to the particular locations
  of the targeted buyers. (para. 3.9)
• Telecom Abuse Bulletin The Bureau generally
  aggregates locations that have the same
  competitive alternatives (within the product
  market) for the relevant telecommunications
  services into a single geographic market.
  (section 2.6)

26
Canada Pipe - Location of Manufacturers and
Importers

27
Sellers of Relevant Product in North America
Seller Location Area of Distribution
Canada Pipe Ste Croix, QC Canada
Vandem Hamilton, ON Ontario
Fernco Sarnia, ON Canada
New Centurion Nanaimo, BC British Columbia (?)
Sierra Abbotsford, BC British Columbia and Alberta (?)
Mission Rubber California ?
Ideal Indiana and Arkansas ?
28
Conundrum 2 Hypothetical Monopolist Test in
Abuse Cases

• Base Price
• The Tribunal noted that markets are based on
  sufficiently close substitutes. Substitutes are
  considered close if buyers are willing to switch
  from one product to another in response to a
  relative price change (para. 68)
• Nowhere does it consider the relevant base price
• Based its conclusions on the Commissioners
  submissions on price correlations
• Those correlations were based on prices that
  prevailed during the stocking distribution
  program (the alleged anticompetitive conduct)
• The prevailing price is not typically the right
  base price when defining markets in abuse cases

29
Market Definition in Conduct Cases

• Test hypothetical monopolist
• Base price an appropriate benchmark
• Likely the price level that would prevail absent
  the alleged anti-competitive act(s)
• Allows for a determination of the products and
  geographic areas that an allegedly abusive firm
  would have to control, or otherwise adversely
  affect, in order to be able to raise price above
  the price that would prevail absent the
  anticompetitive act

30
Abuse Case Example

• Sections of relevant geographic markets can
  overlap
• The alternative supply locations buyers have
  available to them need not be uniform
• For example, a product is imported into both
  Eastern and Western Canada from Central America
• Within Canada, the product imported into or
  produced in the East is not shipped past
  Manitoba, and product imported into or produced
  in the West is not shipped past Saskatchewan
• Buyers throughout Canada have the option of
  buying product from Central America
• The relevant geographic markets are Eastern
  Canada plus Central America, and Western Canada
  plus Central America

31
Making Use of Market Definition Evidence

• Who is proving (or disproving) what?
• Use visuals (maps, graphs, tables)
• Disclose methodology
• Assumptions
• Data sets
• Reality check
• Is the economic analysis compatible with key
  business documents and the clients market
  behaviour?