Lecture 9: Antitrust Policy

1

• Lecture 9 Antitrust Policy Merger Analysis
  with Differentiated Products
• An Overview
• The Antitrust Framework.
• Simulation analysis in mergers involving
  differentiated products
• I. An Overview
• In 1998, completed mergers and acquisitions
  totaled 1.273 trillion, a new record. This
  amount was 50 above 1997 levels.
• Indeed, the merger wave of the 1980s peaked with
  completed mergers and acquisitions equivalent to
  10 of GDP, before falling to 3 in 1992. Since
  then, the merger/GDP ratio has risen steadily
  each year, reaching 22 of GDP in 1998.
• Hence mergers are a significant part of economic
  activity.

2

• The analysis of mergers using non-cooperative
  game theory seems very compelling. Yet up until
  the early 1980s, the dominant approach was based
  on the mergers likely effect on collusion in the
  market.
• Now, the Antitrust division typically focuses on
  unilateral effects rather than the potential
  collusion effects.
• Early work using game theory treated firms as
  Cournot competitors, and a merger between two
  firms was equivalent to one of the firms exiting
  the market. This approach has been criticized in
  recent years.
• New merger analysis assumes differentiated
  products and models the firms as independently
  setting the prices of each of their brands. The
  typical assumption is that the merging firms
  continue to sell all products.

3

• II. Mergers The Antitrust Framework
• Mergers fall under section 7 of the Clayton Act
  and the general criterion for illegality is
  whether the effect of the merger would be to
  substantially lessen competition.
• For at least 50 years, the U.S. Courts have
  favored a structural test to see whether such an
  outcome was likely, i.e., a test based on
  concentration and market shares.
• By the 1960s mergers involving firms with small
  market shares were routinely being blocked by the
  courts.
• Because of inconsistent (and overly
  interventionist) decisions, the U.S. Dept. of
  Justice tried to codify the practice of merger
  enforcement with the issuance of a series of
  merger guidelines beginning in 1968.
• The first part of the guidelines involved
  concentration and safe harbors, based on measures
  of market concentration.

4

• For example, the 1982 guidelines created safe
  harbors for mergers in markets with postmerger
  HHIs below 1000 and for mergers that increased
  the HHI by less than 50. Problem is that it was
  assumed that pre and post merger market shares
  would remain the same.
• The 1982 guidelines introduced a major
  innovation in the methodology of market
  definition. A market is defined as a product or
  a group of products and a geographic area in
  which it is sold such that a hypothetical,
  profit-maximizing firm, not subject to price
  regulation, that was the only present and future
  seller of those products in that area would
  impose a small but significant and non-transitory
  increase in price above prevailing or likely
  future levels.
• The purpose is to find a gap in the chain of
  substitute products.

5

• The market that results from this test, is termed
  the relevant antitrust market and critical
  battles have been waged over this issue.
• One problem with the guidelines is that the focus
  was typically on a static oligopoly, that is, the
  guidelines did not place too much emphasis on the
  possible impact of technological change.
• Now, a requirement of the government as part of a
  consent order to a merger is that steps should be
  taken to maintain rivalry in innovation. This
  can be ensured via divestiture or licensing of a
  key proprietary technology.
• The 1992 revision to the Merger Guidelines added
  a new focus on the role of entry. The guidelines
  require that entry be timely, likely and
  sufficient enough to counteract the competitive
  effects of concern.

6

• There is a growing emphasis on efficiencies.
  Without efficiency gains from a merger, prices
  will likely increase. Synergies are often
  critical if a merger is to increase total
  surplus.
• A 1997 revision to the guidelines states that the
  primary benefit of mergers to the economy is
  their potential to generate efficiencies. Only
  the efficiencies that cannot be achieved without
  mergers will be considered in the merger
  analysis.

7

• So how does all of this work in practice?
• Case Study Staples-Office Depot
• In September 1996, Staples and Office Depot, the
  two largest office superstar chains announced an
  agreement to merge.
• The FTC opposed the merger and presented one of
  the first cases to use modern economic and
  econometric analysis.
• This case established the use of
  unilateral-effects analysis (as opposed to
  coordinated-effects analysis).
• The case also confirmed that U.S. courts will
  primarily apply a price standard, that is, a
  contested merger will only be approved if the
  defendants can show that prices will not rise as
  a result of the merger.
• There is a minor role for efficiency gains
  arising from a merger.

8

• The Case in Detail
• By the mid 1990s, there were only three effective
  competitors Staples, Office Depot, and Office
  Max.
• Since these stores were not always in the same
  geographical market, larger cities consisted of
  monopolies, duopolies, and triopolies. (This
  variation was critical to enable the use of
  econometric analysis.)
• The FTC experts first established that Office
  Superstores (OSSs) were the relevant market.
  This was a key step, because OSSs only accounted
  for 6 of total office supply sales.
• Data obtained from the three firms indicated that
  they feared competition from the other two, but
  not traditional stores.
• By comparing prices in monopoly, duopoly, and
  triopoly markets, FTC economists obtained a crude
  measure of the likely effect of both a
  hypothetical monopoly and the proposed merger
  itself.

9

• The FTC also constructed an econometric model of
  the industry, including both large and small
  non-OSS. The model predicted that a merger to
  monopoly would raise prices by 8.49, more than
  needed for an antitrust market.
• Once the relevant market had been established,
  the FTC used the econometric model to predict the
  effect of the proposed Staples-Office Depot
  merger.
• The model predicted that the merger would
  increase prices by an average of 7.3 in in the
  two and three firm market cities where both firms
  were present.
• The FCC also used an event study of stock market
  prices immediately preceding and following the
  proposed merger announcement to show that their
  econometric estimates were consistent with
  (stock) market expectations of the possible
  merger.

10

• The FCC also demonstrated that barriers to entry
  were very high in the OSS market.
• The defense claimed that efficiency gains
  (primarily from economies of scale) would cause
  prices to fall by 3.0 following a merger. (The
  defense also claimed that the gross price effect
  of a merger ignoring efficiencies was less than
  1. The FTC argued that economies of scale were
  nearly exhausted and that they would amount to
  less than 1.
• The judge agreed with the FTC in almost every
  respect and the proposed merger was not approved.

11
III. The use of simulation analysis in mergers
involving differentiated products Economic
Analysis regarding the unilateral effects is more
amenable to quantification than is economic
analysis of the dangers of collusion. Ultimately,
we are trying to measure the added incentive to
raise price caused by the merger. Employing a
combination of game-theoretic and econometric
methods, we now have the capability to estimate
consumer demand using industry data, and based on
these demand estimates to derive specific
predictions regarding post merger prices. This
contrasts sharply with with the analysis of the
dangers of collusion. While we can easily list
the factors that facilitate collusion, there is
no accepted method of quantifying the increased
likelihood of collusion as a result of a merger.
12

• In the mid 1990s, the Justice Department showed
  an interest in a completely new approach to
  conducting merger analysis.
• Pioneered by Froeb and Werden, the simulation
  methodology using discrete choice models of
  product differentiation removes the need for
  defining a relevant antitrust market.
• In the same way, market shares have no particular
  meaning.
• While well focus on the methodology, it is
  important to keep the pitfalls in mind
• One problem with this approach is that it may
  take decades before judges and antitrust lawyers
  become comfortable with the methodology.
• It remains to be seen that this approach leads to
  better decisions that an approach based on market
  definition and market shares.

13

• Methodology
• Econometrically simulating the effects of
  differentiated product mergers involves four
  steps.
• Make an assumption about the type of conduct
  (competition) in the industry. In most cases,
  Bertrand, or price competition is employed.
• Econometrically estimate the relevant demand
  parameters. Here a specification needs to be
  chosen. (We will not do this now.)
• Use the estimated demand parameters and
  information on prices and shares to get marginal
  cost estimates from first order conditions
• Use demand and cost information to simulate the
  effect of a merger.

14
Estimating the Effects of a merger Suppose that
two firms merge in a four firm industry where
firms sell differentiated products (Step 1
Assume Bertrand competition.) Before merger ?i
(pi mci) qi , i1,4. After merger ?12 (p1
mc1) q1 (p2 mc2) q2 ?3(p3 mc3) q3 ?4
(p4 mc4) q4 Step 3 If we know the demand
function and have information on prices and
quantities, we can we can back out the marginal
costs using the before merger industry
structure. Step 4 Once we know marginal costs,
we can simulate the effect of a merger, by
solving for prices (and quantities) in after
merger industry structure. Hence, we have
estimated the effect of a merger.
15
Step 2 Estimating Demand Functions Following
Berry (1994) we employ a random utility model of
the form

• xj vector of observable characteristics,
• pj is the observed price of automobile j,
• and ? mean valuations of the observable
  characteristics,
• The last two terms of the above equation are
  error terms
• ?j is the average value of product j's unobserved
  characteristics
• ?ij represents the distribution of consumer
  preferences around this mean. (?ij introduces
  heterogeneity and its distribution determines the
  substitution patterns among products.)

16
Well assume that the are i.i.d extreme value
(Weibull) distribution function. This
(multinomial logit) model is popular because of a
closed form solution. The probabaility of
choosing product j is
(2)
17
Demand If we normalize the mean utility of the
outside good (k0) to be 0, and take logs of
equation (2), we obtain the demand function for
each product (3) Where s0 is the share
of the outside good. Equation (3) will be
estimated. This will yield estimates for ? and
?.
18
Oligopoly (Bertrand) price competition Let
pjprice of product j, qjquantity of product i,
The profits of a single product firm are ?j
(pj - mcj) qj FOC for profit maximization
imply qj (pj - mcj)? qj/? pj 0. For the
logit model, the FOCs are pj mcj 1/?(1-sj)?
pj 0 (4) Given pj,sj, and ?, marginal costs
can be estimated from (4).