Transcript The Role of Econometric Analysis in Antitrust

Vertical Restraints & Effects of
Upstream Horizontal Mergers:
Does the Retail Sector Matter?
Luke Froeb
April 12, 2002
University of Florida
Vanderbilt University
1
Related work

Pass Through rates and the Price
Effects of Mergers


mba.vanderbilt.edu/luke.froeb/papers/
Coauthors, Steve Tschantz & Greg
Werden

Views are my own, do not purport to
represent those of my co-authors or Justice
or FTC.
Vanderbilt University
2
Talk Outline

Policy & Theory Background




I will go over fast unless I get questions
3 Games of retailer-manufacturer
behavior
Empirical Example of a Chicago Bread
Merger in each Game
Conclusions & Unanswered questions
Vanderbilt University
3
Background: Retail Sector is
Consolidating in US

In US, Wal-Mart, Kmart, Target, Costco, and
Sears—account for 60 percent of generalmerchandise sales


General-merchandise is 15 % of all retail sales
Productivity advantage over smaller retailers



Economies of scale
Economies of purchasing
Economies of distribution
Vanderbilt University
4
Productivity Gains Associated
with Industry Consolidation
Vanderbilt University
5
Retail consolidation also in
Europe

In EU, top 10 grocery stores forecast to
increase share to 50-60%


Currently at 38%
Wal-Mart entering Europe

Also entering South America
Vanderbilt University
6
Policy Reaction to Retail
Consolidation

FTC challenging some retail mergers



Blocked Kroger + Winn-Dixie
Blocked Staples + Office Depot
Competitive analysis based on increase
in local (within-city) horizontal market
power

“standard” horizontal analysis
Vanderbilt University
7
Standard Horizontal Analysis:
Benefit-Cost of Merger

Goal: quantitative estimate of merger
effect.


Necessary to weigh efficiencies against loss
of competition
Two methodologies


Model-based simulations
“Natural” experiments, e.g. Staples-Office
Depot
Vanderbilt University
8
Model-based simulation



Model current competition
Estimate model parameters
Simulate loss of competition using
estimated parameters

Merger effects modeled as difference
between pre- and post-merger Nash
equilibria
Vanderbilt University
9
Model-based Simulation:
Kroger + Winn Dixie

Estimate
“Gravity” choice
model


Survey density in
Charlotte, NC
Dots represent
grocery stores
Vanderbilt University
10
Pre-merger Equilibrium:
Share of Kroger + Winn-Dixie
Vanderbilt University
11
Post-merger Equilibrium:
Share of Kroger+Winn-Dixie
Vanderbilt University
12
Empirical Comparisons
e.g., Staples-Office Depot




Prices 6% higher in 1superstore cities
15% pass through
40%=6%/.15
compensating MC
reduction
big pass-through 
big merger effect,


Both depend on demand
curvature
So estimate merger
effects and pass-through
rates together.
Vanderbilt University
13
Theory & Policy Towards
Horizontal vs. Vertical restraints

Horizontal

Widespread consensus on how to model horizontal
restraints.



Collusion or “unilateral” effects
Policy debate is empirical
Vertical


No consensus on how to model vertical restraints.
Policy debate is theoretical

Or on “necessary conditions,” e.g., market-share screens
Vanderbilt University
14
Questioning the Consensus on
Horizontal Restraints

How do vertical restraints change
standard horizontal merger analysis
which ignores retail sector?


Focus of paper
How do vertical restraints affect our
understanding of retail consolidation?

Does standard horizontal analysis suffice?

Not going to answer this one.
Vanderbilt University
15
Methodology: put Monopoly
Retail sector on top of Bertrand
Manufacturing Oligopoly

Strategic form bargaining game (n+1 players)


Retailer chooses the best set of offers


Pre-merger Nash equilibrium
Then, two upstream manufacturers merge


Upstream Bertrand oligopolists (n) make take-itor-leave-it offers to retail monopolist (1)
Merger effect is difference between pre- and postmerger equilibria
What happens to retail prices and quantities?
Vanderbilt University
16
Results: the retail sector
matters--a lot.

Upstream horizontal mergers can have
a variety of effects when “filtered”
through retail sector



Transparent retail sector
Opaque retail sector
Double marginalization


Can amplify merger effects, or
Attenuate them
Vanderbilt University
17
Three Different Games

Game 1: retailer must carry all profitable
products


Game 2: retailer has option of exclusive
dealing


Transparent retail sector
Opaque retail sector
Game 3: manufacturers limited to offering
wholesale unit prices independent of quantity

Double marginalization

Can amplify or attenuate merger effects
Vanderbilt University
18
Game 1: retailer must carry all
profitable products

Retailer


Manufacturer


internalizes price
effects between
products
Set w’ to maximize
profitability on own
product
Wholesale price below
marginal cost to
induce Bertrand prices
Vanderbilt University
19
Game 1: continued



Manufacturers set wholesale prices
below mc to induce Bertrand (noncooperative) pricing at retail level
Collect all profit (rev.) with fixed fees
Pricing & Merger effects are same as
would occur if Manufacturers sold
directly to consumers.
Vanderbilt University
20
Game 2: Retailer has Option
of Exclusive Dealing

Retailer has four options





Exclusive dealing with Mfg. 1, total profit=T1
Exclusive dealing with Mfg. 2, total profit=T2
Joint dealing with Mfg. 1 and 2, total profit=TJ
Neither, total profit =0
Mfg.’s make contingent offers to retailer using
two-part prices


O’Brien & Shaffer (1997) and Bernheim &
Whinston (1998).
Equilibrium is “efficient” in that Mfg’s transfer at
mc, and retailer maximizes joint profitability
Vanderbilt University
21
Game 2: Continued


Profit split: winner must outbid next best alternative
1 is most valuable alternative: T1>max(T2,TJ)




1 and 2 are substitutes: T1+T2 > TJ > max(T1,T2)




1’s profit=T1 - T2;
2’s profit=0;
retailer’s profit=T2;
1’s profit=TJ - T2;
2’s profit=TJ - T1;
retailer’s profit=T1 + T2 - TJ
1 and 2 are complements: TJ > T1+T2 > max(T1,T2)

retailer’s profit=0
Vanderbilt University
22
Game 2: continued


Manufacturers set wholesale prices at
marginal cost
Retailer maximizes total profit


Retailers paid their marginal contribution to total
profit
Mergers


do not change retail prices
But do transfer profit from retailer to merged
manufacturers
Vanderbilt University
23
Game 3: Wholesale prices are
independent of quantity

Retailer profit (same)


internalizes price effects
between products
Manufacturers face
derived demand, q*



Bertrand equilibrium
with derived demand.
Wholesale margins are
functions of elasticity of
derived demand
Depends on passthrough rates from
wholesale to retail
Vanderbilt University
24
Game 3: continued


Prices “too high”—double
marginalization.
Big wholesale margins

Derived demand is usually less elastic than
retail demand because



Wholesale prices are lower (which makes
percentage price changes bigger)
Pass-through rates can be less than one
Merger effects can be higher or lower
Vanderbilt University
25
3 Retail Games Illustrated:
White Pan Bread in Chicago

All calibrated to same prices, quantities,
pre-merger elasticities (logit demand)
Vanderbilt University
26
Model Calibration
Vanderbilt University
27
Merger of Brands 1+2
Vanderbilt University
28
Merger of 1+2 w/AIDS Demand
Vanderbilt University
29
Logit vs. AIDS Demand

Pass through rates for logit (95%),
linear (near 50%) demand are less than
one.


Relatively inelastic derived demand
Pass through rates for AIDS (180%),
Constant Elas (near 200%) demand are
higher

Relatively elastic derived demand
Vanderbilt University
30
Conclusions

Retail sector matters a lot for standard
horizontal merger analysis



Constant mark-up or percentage mark-up
usually assumed which is transparent case.
Not correct if “opaque” or “double
marginalization”.
Empirical Identification of retail game

Games have negative, zero, and positive
wholesale margins, respectively.
Vanderbilt University
31
Unanswered Questions

How do retailer’s behave?



Sales agents compensated on revenue commission
Positive wholesale margins
Complex nonlinear contracts with promotional
allowances, quantity discounts


is two-part pricing a good metaphor?
What about the n X k case (n manufacturers,
k retailers)?


Retailers compete on selection, price,
convenience.
Does opaque equilibrium hold for n X k case?
Vanderbilt University
32