Transcript Greg.Day1.Intro

Antitrust Analysis Using
Calibrated Economic Models
Gregory J. Werden
Senior Economic Counsel
Antitrust Division
U.S. Department of Justice
The views expressed herein are not purported
to reflect those of the U.S. Department of Justice
Calibrated Economic Models Are:
 Standard
economic tools
• Monopoly and oligopoly models
• Consumer demand models
 Applied
in a straightforward way
• Using known facts (e.g., prices and shares)
• Making reasonable assumptions
Applications of Calibrated Economic
Models In Merger Cases
 Market
delineation
Modeling the hypothetical monopolist
with the standard model of monopoly
 Merger
simulation
Modeling oligopoly interaction
with a standard model of oligopoly
Applications of Calibrated Economic
Models In Non-Merger Cases
 Market
delineation
Non-merger cases may present prospective
market power issues just as mergers do
 Analyzing
competitive effects
Modeling can shed light on the impact of
many exclusionary practices
Selection and Calibration of Models
 Select
a model that generally describes
the actual competitive interaction
 Make
necessary assumptions that are
(1) supported by data, (2) unimportant,
or (3) biased against the modeler
 Calibrate
the model by fitting it
precisely to a “pre-merger equilibrium”
Calibrated Economic Models
Sharpen Focus
 Models
are built on assumptions
 Assumptions
can and should be tested
against the established facts
 Modeling
indicates:
• why price effects are large or small
• how experts reached different conclusions
• where resources should be concentrated
Calibrated Economic Models
Increase Accuracy
 Calculation
replaces intuition
• Expert intuition on quantitative matters is
a highly unreliable black box
• Modeling replaces the black box with
transparent calculations

Measurement replaces assumption
• Real-world data is the basis for calculations
Calibrated Economic Models
Enhance Persuasiveness
 The
analysis is firmly anchored in facts
• Model assumptions are tested against facts
• Model parameters are based on data
 The
analysis employs economic science
• The models are standard economic theory
• Econometrics may be used
• Economists do what they normally do
Artful Construction of
Calibrated Economic Models
 Highly
realistic models are useless
• Calibration most likely is infeasible
• Predictions most likely are unclear
 Useful
models capture what matters
most over the relatively short run
• The essence of the competitive process
• Critical firm and product attributes
Critical Elasticity of Demand
and Critical Sales Loss
 Indicate
how much substitution is too
much for the hypothetical monopolist
 Qualitative application
Providing a lens for viewing qualitative
evidence of substitutability
 Quantitative
application
Using demand elasticities estimated from
transaction or survey data
Modeling a Hypothetical Monopolist
 Demand
models
• Multiple uses with significantly differing
elasticities of demand
 Cost
models
• Marginal costs that vary with output
• Fixed costs for blocks of capacity
 Multiple
prices
• Disproportionate price increases
Modeling the Unilateral Competitive
Effects of Mergers
 Select
a standard oligopoly model
capturing important industry features
 Calibrate
the model so it perfectly
predicts the pre-merger equilibrium
 Compute
the post-merger equilibrium,
which internalizes the competition
among merging products
Presumptions of Merger Simulation
 Merger
simulation presumes that the
fundamental nature of the competitive
interaction is not changed by a merger.
 Merger
simulation presumes that
everything “outside the model” is
unaffected by the merger.
Oligopoly Models Commonly
Used in Merger Simulation
 Dominant
firm
Monopoly with respect to residual demand
 Cournot
Quantity setting with a homogeneous product
 Bertrand
Price setting with differentiated products
 Auctions
Non-cooperative bidding
Bertrand Merger Simulation:
Model Calibration
 Calibrate
prices and shares
• Normally use those prevailing pre merger
• May project changes “but for” the merger
 Calibrate
demand elasticities
• Commonly estimated from scanner data
• Can be inferred from price-cost margins
Bertrand Merger Simulation:
Modeling Demand
 Every
demand model is restrictive
Each has inherent “curvature” properties
 Flexibility
is a blessing and a curse
Trading off bias and variance
 The
logit model
Substitution proportionate to shares
Bertrand Merger Simulation:
Checking the Bertrand Assumption
 Compare
Bertrand price-cost margins
with “actual” price-cost margins
 Examine
substantial differences for the
merging products and major rivals
 Intensity
of competition may not be as
implied by the Bertrand assumption
Bertrand Merger Simulation:
Continuing Controversies
 Multifaceted
 Product
 The
marketing strategies
repositioning
retail sector
 Demand
estimation problems
Merger Simulation in Perspective I

Merger simulation is simply a useful device
for clarifying the implications of established
facts and plausible market conditions.

Merger simulation combines what is known
with reasonable assumptions about what is
not known, and evaluates their significance in
a precise, objective manner by substituting
calculation for intuition.
Merger Simulation in Perspective II
 Models
simplify the world
• Models are tractable because they simplify
• Modeling is as much art as science
 Calibration
is inexact
• Assumptions fill gaps in measurement
• Measurement is inherently imperfect
 Predictions
are useful rough estimates