See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/240114299

Mergers with Differentiated Products

Article · January 1995

CITATIONS READS
114 729

1 author:

Carl Shapiro
University of California, Berkeley
84 PUBLICATIONS   22,407 CITATIONS   

SEE PROFILE

All content following this page was uploaded by Carl Shapiro on 09 October 2014.

The user has requested enhancement of the downloaded file.

Mergers with Differentiated Products

T
he antitrust treatment of hori- centration, they can significantly supple- cern ourselves with the prospect that the
zontal mergers by the Justice ment those steps. prices for one or more brands sold by the
Department and the Federal To place the topic in context, it is merging parties will rise after the merger,
Trade Commission is one of instructive to compare mergers with even if prices do not uniformly rise
the most well developed and closely scru- homogeneous products to those involving throughout the relevant market. Such
tinized areas of antitrust law. The differentiated products. For homoge- concerns are not present, or are far less
enforcement agencies have extraordinary neous products, the traditional structural pronounced, in markets for homogeneous
experience reviewing mergers and the approach of defining markets and mea- products.
merger bar is no less sophisticated. From suring market shares and market concen- The danger that a merger of Brands A
my perspective as an antitrust economist, tration has deep roots, along with a rich and B will cause anticompetitive price
this sophistication permits merger empirical tradition linking market struc- increases for one or both of these brands
enforcement to be at the cutting edge ture to performance. In these markets, it is greatest if the merging brands are
when it comes to incorporating econom- is both natural and appropriate to count “close,” in the sense that many customers
ic learning into competition policy. up each firm’s unit or dollar sales, or using one brand consider the other brand
The 1992 DOJ and F TC Merger capacity, to measure market shares, and their second choice. Ultimately, unilateral
Guidelines have placed important atten- to make inferences about a merger’s anticompetitive effects are based on the
tion on the “unilateral effects” of a merg- effects based on market structure, includ- following logic: As the price of Brand A
er, i.e., the tendency of a horizontal merg- ing the HHI of market concentration. The rises, some customers will shift from
er to lead to higher prices simply by danger of collusion is surely related to Brand A to Brand B. Prior to the merger,
virtue of the fact that the merger will market concentration (although quanti- these customers would be lost to the firm
eliminate the direct competition between fying this relationship is difficult), and owning Brand A. After the merger, this
the two merging firms, even if all other economists’ primary model of non-coop- same firm owns Brand B and thus does
firms in the market continue to compete erative oligopolistic competition among not lose these customers. As a result, the
independently. Unilateral effects are con- manufacturers of homogeneous goods price increase is more profitable to the
trasted to “coordinated effects,” i.e., the relates market structure to performance.1 merged entity.
danger that the merger will lead to collu- Although economists continue to debate To explicate this logic, consider a
sion between the merged entity and its the empirical relationship between mar- merger between Brands A and B. The
remaining rivals. ket structure and performance, there likely post-merger price increase for
This article discusses some of the exists a solid foundation for using market Brand A is driven by two variables, each
methods used by the Antitrust Division to structure prominently in evaluating hori- of which we have a good chance of
analyze unilateral effects in mergers zontal mergers involving homogeneous observing with a proper investigation.
involving differentiated products. While goods.2 The first I call the Diversion Ratio from
the methods outlined here do not replace This traditional structural approach Brand A to B. This Diversion Ratio is the
the standard steps of defining markets towards merger policy, which dates back answer to the following question: If the
and measuring market shares and con- to the 1960s but has been refined as just price of Brand A were to rise, what frac-
described, is less attractive for differenti- tion of the customers leaving Brand A
ated products. When products are highly would switch to Brand B? The second
Carl Shapiro is the Deputy Assistant differentiated, concerns about coordinat- variable is the Gross Margin — the per-
Attorney General in charge of economics ed effects may be secondary to concerns centage gap between price and incre-
at the Antitrust Division, U.S. Department about unilateral effects. And, to assess mental cost — for Brand B.3 Roughly
of Justice. He is currently on leave from his unilateral effects most accurately, it is speaking, a valuable index of the poten-
position as the Transamerica Professor of highly desirable to go beyond industry tial anticompetitive unilateral effects is
Business Strategy at the Haas School of concentration measures to look directly obtained by multiplying the Diversion
Business and Professor of Economics in at the extent of competition between the Ratio by the Gross Margin. Any danger
the Economics Department at the merging brands. This is especially true if of a unilateral price increase may be alle-
University of California at Berkeley. Dr.
competition is “localized,” i.e., if some viated by product repositioning, entry, or
Shapiro thanks Joseph Kattan, George
Rozanski, Steve Salop, Greg Werden, and
brands are especially close substitutes for efficiencies. Nonetheless, the Diversion
Robert Willig for valuable comments on other brands due to their product charac- Ratio and the Gross Margin are the key
this paper. teristics or image. To put this differently, variables in the demand-side portion of
at the Antitrust Division we must con- the analysis.

S P R I N G 1 9 9 6 · 2 3
 A R T I C L E S • A N D • F E A T U R E S

Monopolistic Competition the analysis of the dangers of collusion. a price increase unprofitable, and unless
To a greater or lesser degree, virtually all While we are fairly confident in listing synergies imply that the price increase
markets involve some element of product and analyzing factors that facilitate or will not in fact raise profits, the merger
differentiation.4 Even in a classic homo- hinder collusion, including market struc- will injure consumers and be anticom-
geneous-goods market — such as the ture, there is no single accepted method petitive.
market for an agricultural commodity or of quantifying the increased likelihood There are many valid ways to carry
for a specific chemical compound — pro- of collusion attendant to a merger. out this analysis. As much as anything,
ducers often attempt to differentiate Our ability to predict unilateral effects the method chosen depends upon the
themselves based on product quality, based on demand patterns is only as good available data. Generally, however, I find
reliability, or customer service. as the available data. Very often we must it useful to structure my thinking about
My emphasis here is on markets make do with qualitative evidence or unilateral effects in differentiated product
where the brands are distinct in important rather rough estimates. However, even markets in terms of the following four
and long-lasting ways. It is helpful to when the data are limited, the theory of steps.7 Consider a merger between brands
keep two categories in mind: (1) branded monopolistic competition can provide A and B, and focus attention on the dan-
consumer products, where each brand of some very helpful predictions based on ger that the price of Brand A will be ele-
pens, bread, facial tissue, computer soft- premerger Gross Margins and market vated after the merger:8
ware, or cereal, is distinct; and (2) phys- shares, which often can be measured (1) Consider a price increase for
ical facilities that distribute or deliver using data routinely collected during the Brand A of, say, 10 percent. Try to mea-
goods or services, such as supermarkets, second request process. These predictions sure what fraction of the sales lost by
department stores, branch banks, or hos- will need to be checked against the views Brand A due to this price increase would
pitals, where the differentiation is based of industry participants, company docu- be captured by Brand B. I call this frac-
on location. These two broad categories ments, and other information sources, but tion the Diversion Ratio (from Brand A to
correspond to the familiar tasks of prod- they do constitute a vital part of merger Brand B).
uct and geographic market definition. analysis. (2) Based on premerger Gross Mar-
A final caveat: the analysis below gins and the estimated Diversion Ratio,
Unilateral Competitive Effects applies when the firms independently set calculate the post-merger price increase,
The predominant approach taken by uniform prices for their branded prod- assuming no synergies or rival supply
economists studying markets with differ- ucts. To the extent that firms engage in responses.
entiated products is to model the firms price negotiations on a customer-by-cus- (3) Try to account for any likely and
as independently setting the prices of tomer basis or engage in other forms of timely changes in prices or product offer-
each of their brands. As usual in eco- price discrimination, the analysis must ings by nonmerging parties, including
nomics, we assume that each firm seeks be modified or replaced with another product repositioning and entry.
to maximize its own profits. This method approach. (4) If there are credible and docu-
of analysis fits perfectly with the “uni- mented synergies that lower marginal
lateral effects” portion of the Merger Estimating Post-Merger Prices: costs, reduce the predicted post-merger
Guidelines (§ 2.2). Four Steps prices accordingly. This step can lead to
It is fair to say that economic analysis Economists in the agencies and those a predicted price decrease.
of differentiated-products mergers at the working for private parties largely agree The first two steps capture the
Division typically focuses on unilateral on what to look for to estimate unilateral demand-side analysis; the third step is
effects, unless there are structural factors competitive effects, although they may the supply-side analysis; the final step
facilitating collusion following the merg- differ in specific implementation steps. accounts for possible efficiencies. If these
er or there is a history of collusion in the We seek to estimate the post-merger steps indicate that the merged entity
industry. This emphasis represents a sig- Bertrand equilibrium in prices, account- would find it optimal to impose a signif-
nificant shift in a fairly short period of ing for the new market structure in which icant price increase following the merger,
time. some brands are jointly owned that had then the merger is very likely to be anti-
Economic analysis regarding unilat- previously been independent and for the competitive.
eral effects is more amenable to quantifi- new cost structure of the merged entity.6
cation than is economic analysis of the The analytical steps in this exercise The Diversion Ratio
dangers of collusion. Ultimately, we are can be described succinctly in a manner The concept underlying the Diversion
trying to measure the added incentive to consistent with the methodology of the Ratio is easily comprehensible: “If you
raise price caused by the merger. Merger Guidelines: If a significant pro- raise your price, what fraction of your
Employing a combination of game-theo- portion of consumers considers the merg- lost customers will turn to your rival
retic and econometric methods, we now ing firms’ products to be their first and (now merger partner)?”
have the capability to estimate consumer second choices (at premerger prices), The Diversion Ratio is a close cousin
demand using industry data and, based then the merged entity will have an incen- of the cross-elasticity of demand between
on these demand estimates, to derive spe- tive to impose a nontrivial price increase the merging brands.9 If the unit sales of
cific predictions regarding post-merger following the merger. Unless product the merging brands are equal prior to the
prices.5 This contrasts somewhat with repositioning or entry would render such merger, the Diversion Ratio from Brand

2 4 · A N T I T R U S T
 A R T I C L E S • A N D • F E A T U R E S

A to Brand B is equal to the cross elas- of white pan bread in five regional model at this point do not account for
ticity of demand from Brand A to Brand markets. The proposed Final Judgment product repositioning, entry, or synergies.
B, divided by the own-price elasticity of orders Interstate and Continental to divest In all cases, however, this demand-side
demand for Brand A. 10 Suppose, for certain white bread brands in each analysis did take account of competition
example, that Brand A has an own-price geographic market. Our competitive con- from private-label products as well as
elasticity of demand of 2.0 (i.e., a 1 per- cerns, as well as the ultimate relief, were branded goods. In the bread merger, the
cent increase in the price of Brand A greatly informed by our economic analy- computer model predicted price increas-
causes unit sales to decline by 2 percent), sis of the merger’s anticompetitive uni- es in the 5 to 15 percent range for
and that the cross-price elasticity of lateral effects. Continental’s and Interstate’s premium
demand from Brand A to Brand B is 0.5. The Kimberly-Clark acquisition of white pan breads in the Los Angeles and
If the two brands’ premerger unit sales Scott Paper Company posed a threat to Chicago areas. In baby wipes, we esti-
are equal, then the Diversion Ratio from competition in the markets for baby mated that there would have been sub-
Brand A to Brand B is (0.5)/(2.0) or 25 wipes and facial tissues.14 Again, our stantial price increases following the
percent. Stated differently, one-quarter of analysis was significantly informed by merger. Price increases were also pre-
the unit sales lost by Brand A if its price econometric analysis of the demand for dicted in facial tissue, especially for the
rises are captured by Brand B. the various brands in these markets, from smaller Scotties brand.
In some cases, the Diversion Ratio which we were able to estimate likely While econometricians dream about
from Brand A to Brand B will be closely price increases following the acquisition. this type of “high-tech” analysis, in real-
linked to Brand B’s market share. In par- In both of these merger investigations, ity the data are rarely available to do this
ticular, if all sales lost by Brand A are we had access to excellent data on prices type of full-blown simulation analysis
captured by other brands in the market, and unit sales derived from checkout with assurance. If econometric estima-
and if all brands are “equally close” to scanners at retail locations. These data tion of elasticities based on transactions
each other, then the Diversion Ratio from provided information regarding competi- data is not possible, relevant consumer
Brand A to Brand B may be stated as tion both from other brands and from survey data may still be available to
sB/(1-sA), where sA and sB are the brands’ private label products. In the bakeries directly estimate the Diversion Ratio.
respective market shares.11 In the more merger, we focused on the brands of pre- Survey data are not as good as actual
realistic situation where some customers mium white bread sold by Continental transactions data, but can still be reliable
substituting away from Brand A switch to and Interstate. In the baby wipes market, if valid sampling procedures were used
products outside the market entirely, this we focused on competition between and if the results are not overly sensitive
Diversion Ratio will be proportionately Kimberly-Clark’s Huggies brand of baby to the framing of the questions. If reliable
lower. For example, if 20 percent of the wipes and Scott’s two brands of baby survey data are also unavailable, it still
customers lost by Brand A leave the mar- wipes, Baby Fresh and Wash-a-Bye may be possible to use company docu-
ket entirely, the Diversion Ratio from Baby; together the merging parties con- ments and other qualitative information
Brand A to Brand B will instead be trolled over 50 percent of the sales in this regarding consumers’ brand preferences
(0.8)sB/(1-sA). $500 million market. In the facial tissue to estimate Diversion Ratios.
Since the Diversion Ratio plays a cru- market, we estimated the cross-elasticity The firm’s market shares can be very
cial role in this analysis, in differentiated- between Kimberly-Clark’s Kleenex helpful in estimating Diversion Ratios if
product mergers the Antitrust Division brand and Scott’s Scotties brand; togeth- none of the brands in the market are espe-
will invariably want to know the best er, the two parties controlled nearly 60 cially “close” to or “distant” from each
estimate of the Diversion Ratio based on percent of the sales in this $1.34 billion other. As noted above, Diversion Ratios
the available evidence. market. will be proportional to market shares in
In these two mergers we were able, this case. For example, consider such a
Estimating Unilateral with considerable work and making var- market in which Brand A has a 25 per-
Competitive Effects in Practice ious assumptions about the structure of cent share and Brand B has a 15 percent
Steps 1 and 2: Demand-Side Analysis. In demand, to estimate a complete model of share. Suppose also that very few cus-
practice, the demand-side analysis demand for the various brands. These tomers of Brand A would reduce their
depends heavily on the availability of methods subsume the Diversion Ratio overall purchases in the market if Brand
data.12 Two recent merger cases at the concept I stressed above. The calibrated A were to raise its price; instead these
Antitrust Division that led to consent model of consumer demand derived from customers would by and large pick
decrees illustrate what can be done with the supermarket scanner data was then among the other brands. In this case, the
detailed data. used, in a high-tech version of Steps 1 Diversion Ratio from Brand A to Brand
The merger of Interstate Bakeries and 2 above, to predict the likely post- B is 20 percent.16 The Diversion Ratio
Corporation and Continental Baking merger price increases for the various will be lower, to the extent that some of
Company involved the first and third merging brands.15 Brand A’s customers reduce their total
largest bakers of fresh bread in the In this prediction exercise, we purchases in the market when the price of
United States.13 The Division concluded assumed that all firms in the industry set Brand A rises. If half of the customers
that the deal would substantially lessen prices independently after the merger to dropping Brand A leave the market alto-
competition in the production and sale maximize profits. The predictions of the gether, or if customers switching to other

S P R I N G 1 9 9 6 · 2 5
 A R T I C L E S • A N D • F E A T U R E S

brands in the market reduce their pur- ity demand functions. With constant-elas- a Diversion Ratio of 0.2 (i.e., 20 percent
chases by half when switching away from ticity demand, and assuming that the two of the sales lost when the price of
their favorite Brand, the Diversion Ratio merging brands are symmetric prior to Brand A goes up are captured by
from Brand A to Brand B will only be 10 the merger, the merged entity’s profit- Brand B), then the optimal post-merger
percent. maximizing percentage price increase is price increase in percentage terms is
Under this analysis, if the merging mD/(1–m–D).17 Here m is the premerger (0.4)*(0.2)/(1–0.4–0.2) = 0.2: a 20 per-
brands are similar in characteristics, or if Gross Margin and D is the Diversion cent price increase would maximize
the merging brands have large shares Ratio between the two merging brands. profits.
within a broader product category, the For example, suppose that the pre- This formula must be used with great
Diversion Ratio is likely to be high. Note merger price is $100, and the cost per caution because it relies on several strong
in particular that the Diversion Ratio is unit is $60, so the premerger markup, assumptions, as I have noted. To the
likely to be high for a brand that is merg- m, is 40 percent, not uncommon at all extent that the elasticity of demand for a
ing with a dominant brand: the large mar- for differentiated products. If we assume brand rises as the price of that brand rises,
ket share of the dominant brand makes it
likely that customers switching away
from the smaller brand will divert to the
dominant brand rather than elsewhere.
On the other hand, if the merging brands Diversion Ratio Example
are usually sold to different types of con-
sumers, or through different channels, or

T
if consumers’ preferences are such that he importance of the Diversion Ratio and the Gross Margin can be seen with a few
they can easily substitute to a broad range numerical examples. Consider a situation in which Brands A and B each have pre-
of products (e.g., gifts instead of premium merger prices of $100 and premerger unit sales of 1000. Suppose that each brand has
pens, or breakfast foods instead of ready- a marginal cost of $60, so the premerger Gross Margin is ($100–$60)/$100 or 40%. Prior
to-eat cereals), the Diversion Ratio is to the merger, Brand A makes profits (often, contribution to fixed costs) of $40 per unit times
likely to be lower, other things equal. 1000 units, or $40,000. Consistent with premerger optimal pricing by Brand A, suppose that
Merely asserting that there are numer- a 10% price increase by Brand A would lead to a 25% reduction in sales, to 750 units. At this
ous products to which consumers could higher price, Brand A could earn profits of $50 per unit times 750 units, or $37,500. Since
substitute, and thus the Diversion Ratio this is less than the $40,000 figure, prior to the merger Firm A did not find it profitable to set
must be low, ignores the importance of the higher price.
diverse consumer preferences and does How would a merger between Brands A and B change this calculation? Suppose that the
not replace this step of the analysis. Nor Diversion Ratio from Brand A to Brand B is 30%. In other words, of the 250 units lost by Brand
is a merger immunized merely because A due to the price increase, 30%, or 75 units, are diverted to Brand B. The merged entity would
the merging brands are not next-closest take into account the additional profits earned by Brand B from these customers when con-
substitutes, as some parties claim, any sidering raising prices from $100 to $110. Assuming the price of Brand B also rises to $110,
more than a merger is immunized mere- these 75 diverted sales generate profits of $50 per unit, or $3750 in total. Adding these to the
ly because the merged entity still faces $37,500 from the premerger calculation, the post-merger profits per brand at the elevated
some post-merger competition. prices are $41,250. The merger has made it profitable to raise prices 10% above premerger
To complete the demand-side analy- levels. In this example, a 10% price increase will be profitable after the merger if and only if
sis, the estimated Diversion Ratio can be the Diversion Ratio is at least 20%.*
used, along with premerger Gross Compare this example to one in which the premerger Gross Margin is smaller. Specifically,
Margins and perhaps other industry mea- suppose now that the marginal cost is $80, so the premerger Gross Margin is only
sures, to give a rough prediction of the ($100–$80)/$100 or 20%. The premerger profits on Brand A are now only $20 per unit or
post-merger price increases for the merg- $20,000. Now, consistent with optimal premerger pricing, suppose that a 10% price increase
ing brands. The principle here is that would result in a 50% loss of sales, to 500 units. Prior to the merger, a 10% price increase
high Gross Margins and high Diversion would reduce profits to $30 per unit times 500 units or $15,000, some $5000 less than could
Ratios suggest large post-merger price be earned at a price of $100.
increases. The tricky part is that the actu- What about after the merger? Now a 30% Diversion Ratio would mean 30% of the 500
al calculation of the post-merger price lost units, or 150 units, would be captured by Brand B; at a $30 profit margin per unit, this
increase depends upon the specific shape adds $4500 in profits. But this $4500 of extra profit is still not enough to make up for the loss
assumed for the demand curve. of $5000 in profits on Brand A. With the smaller Gross Margin in this example, a larger
A simple formula can be derived for Diversion Ratio would be necessary to make a 10% price increase profitable.
the post-merger price increase if one is
willing to assume that consumer demand *In these examples, I ask only whether a 10% price increase would generate more profits than premerger prices.
Below, I discuss how to calculate the optimal post-merger price increase. Fortunately, with linear demand, the
functions exhibit constant elasticity over
maximal profitable price increase is exactly twice the optimal price increase. So, if a 10% price increase is prof-
the relevant range of prices. Very often itable, it will be optimal to raise prices at least 5%. Throughout this article, I use 10% price increases purely for
when economists estimate demand using illustrative purposes.
data, they employ such constant-elastic-

2 6 · A N T I T R U S T
 A R T I C L E S • A N D • F E A T U R E S

the constant-elasticity-of-demand calcu- post-merger price increase confidently entiated-product markets, brands enter
lations will overestimate the post-merger based on these two variables alone. and exit with some regularity, and exist-
price increases. This overestimation can In some cases, we can measure Gross ing products may be repositioned either
be significant, especially if the formula as Margins quite well, but may be uncertain through design changes or revised mar-
stated generates a large percentage price about the Diversion Ratios. In this situa- keting strategies. As a general rule, the
increase.18 Certainly it is desirable to find tion we can use either of the two formu- “farther” a brand must be moved to com-
support for the constant-elasticity las above to ask how large the Diversion pete more effectively with the merging
assumption in documents, in the testimo- Ratio must be in order for the optimal brands, the less likely it is that such a
ny of industry participants, or in the data post-merger price to be at least 10 percent move would in fact occur in response to
before using this formula or variants on it. (say) above premerger levels. Using the a post-merger price increase. As noted
If demand instead takes a linear form, linear demand formula just given, the generally in the Merger Guidelines, the
the elasticity rises as the price rises, mak- post-merger price increase will be at greater the sunk costs associated with
ing the optimal post-merger price least 10 percent if and only if the product entry or repositioning, and the
increase smaller. In this case, an alterna- Diversion Ratio is at least 0.2/(m+0.2). If longer such supply responses take, the
tive formula can be derived: the optimal the premerger Gross Margin is 40 per- less likely they are to deter or defeat an
post-merger percentage price increase cent (m=0.4), prices will go up at least anticompetitive price increase.
with linear demand (again with premerg- 10 percent if the Diversion Ratio is at Notwithstanding the foregoing, rivals’
er symmetry between the two brands) is least one-third. Again, the larger the responses do not necessarily reduce the
given by mD/ 2(1–D). This formula is Gross Margins, the lower the Diversion profitability of a post-merger price
quite different from the earlier one. Using Ratio necessary to raise anticompetitive increase. Game-theoretic analyses of
the same numerical example as above, concerns. pricing competition with differentiated
with a premerger Gross Margin of 40 per- Steps 1 and 2 invariably will lead to products indicate that rivals will typical-
cent and a Diversion Ratio of 20 percent, the interim prediction that prices will rise ly find it optimal to raise their prices in
the post-merger price increase would be after the merger, if indeed Brands A and response to higher prices set by the merg-
“only” 5 percent.19 B compete with each other. After all, a ing firms. Accounting for these accom-
I am keenly aware that the two for- merger between rival brands does elimi- modating responses tends to increase, not
mulas presented above give quite differ- nate competition between those brands, decrease, the predicted post-merger price
ent predictions. This observation does not which in and of itself leads to higher increase.
take away from my main point — that prices.22 But we do not condemn all hor- Merging parties in consumer-goods
Diversion Ratios and Gross Margins are izontal mergers, of course. To begin with, industries may be tempted to argue that
key variables to explore in a merger we recognize that this incentive to raise brand name is unimportant, but they
investigation involving differentiated price is very slight for some mergers, should be cautious in doing so. Such
products — but should serve as a warn- even horizontal ones. Steps 1 and 2 claims are not credible if the parties
ing. The fact is, the profit-maximizing should detect this, in the form of a very themselves have made substantial invest-
post-merger price increase is sensitive to small predicted price increase. But there ments in brand equity, or if the deal price
consumers’ ability to substitute away are two more important reasons why hor- itself reflects substantial brand equity.
from the merging brands as prices rise.20 izontal mergers often are not anticom- Attempts to downplay the importance of
It is simply not possible, and one should petitive. First, the true price increase may brand names are particularly problemat-
not expect, to fully predict price changes be far smaller than predicted by Steps 1 ic if new brands historically have found it
on the basis of two numb ers, the and 2, or negligible, because rivals may difficult to gain a secure foothold in the
Diversion Ratio and the Gross Margin, respond to defeat any price increase. market.
alone. One way or another, however, the Second, the merger may reduce costs. In the recent bread merger the
Diversion Ratio (or its close cousin) and Steps 3 and 4 are where these crucial con- Division concluded that brand names
the Gross Margin will be an integral part siderations come into play. were very important. The evidence also
of the analysis. Step 3: Product Repositioning and showed that a brand that achieved success
Finally, even if data are limited and Entry. If the merging brands are “close” in one region might not meet with accep-
precise predictions of post-merger price in attributes, the Diversion Ratio is like- tance in another area. Furthermore, new
increases are difficult to make with con- ly to be high, and Step 2 will suggest a entry required significant sunk invest-
fidence, these formulas still are a useful significant price increase. In precisely ments in brand promotion, as well as
starting point in gaining a sense of the this situation, however, it may well pay investments in a route delivery network.
likely magnitude of any post-merger for a rival firm to reposition its brand, or For these reasons and more our concerns
price increase, if full-scale demand esti- introduce a new brand, closer to the were not alleviated by the prospect of
mation is not possible.21 And the numer- merging brands. And this threat could product repositioning or entry.
ator in both of these expressions, mD, i.e., well deter the price increase in the first In the facial tissue and baby wipes
the product of the Gross Margin and the place. Alternatively, a de novo entrant markets, we also found that entry takes
Diversion Ratio, is a very useful quick could locate its brand near Brands A and significant time and expense. The baby
guide to the likely anticompetitive dan- B if prices of these brands were above wipes market had unique aspects that
gers, even if we cannot predict the competitive levels. Very often in differ- warranted an especially close look at the

S P R I N G 1 9 9 6 · 2 7
 A R T I C L E S • A N D • F E A T U R E S

dynamics of entry: the most likely ing products, their competitive signifi- brands are close together, the Diversion
entrants appeared to be firms with well- cance may well be understated by their Ratio is likely to be high, and any given
established brand names in related con- exclusion. level of market concentration is more
sumer products, and entry is surely made Since we need to define markets to troubling. The reverse is true if the brands
easier by the fact that there is relatively identify the lines of commerce affected are distant.
rapid turnover in demand in this market, by the merger, it may be tempting for If all the brands in the market are
since individual babies, and even parents counsel to argue for a very broad market roughly equidistant from each other, then
with two or more children, soon outgrow if there is no clear break in the chain of the market shares of various brands will
the need for wipes. Nonetheless, entry substitutes. This may be inconsistent, be proportional to their Diversion Ratios,
into this market has proven difficult, and however, with a careful application of the making emphasis on the two brands’
the prospect of entry did not alleviate our Merger Guidelines, which can lead to a market shares very appropriate. The
concerns. market boundary between very similar Guidelines look to both the sum of the
In both cases, as in all mergers, the products. In the pens case, testifying on market shares of the merging brands (to
Division analyzed whether a price behalf of Gillette, I agreed with the see if this sum exceeds 35 percent in eval-
increase, or a reduction in quality, would Division that one could define a market uating unilateral effects) and to the prod-
remain profitable, after accounting for boundary based on price, even though uct of the market shares (which will
rivals’ supply-side responses. The histo- pens were sold at a continuum of prices. reflect the number of consumers who
ry of brand entry, exit, and positioning, Judge Lamberth accepted this principle in regard the merging brands as their first
and the associated costs, is relevant for his opinion.24 In the case of geographic and second choices, and determines the
this portion of the analysis. markets the mere fact that there are gas increase in the HHI).
Step 4: Synergies. If a post-merger stations or supermarkets all over Los As illustrated above, if the brands are
price increase is profitable, even after Angeles does not necessarily imply that equidistant from one another, informa-
accounting for rivals’ responses as well the geographic markets for these prod- tion about market shares can be com-
as consumer substitution, the merger is ucts are as broad as the entire L.A. area. bined with a measure of the overall mar-
likely to be anticompetitive. Consumers Nor is there a single market for all food, ket elasticity of demand to estimate the
likely will be harmed by the combina- despite the fact that it is difficult to draw Diversion Ratio between the merging
tion, unless it truly offers substantial effi- boundaries between food groups. brands. This estimated Diversion Ratio
ciencies that lower incremental costs. On the other hand, an anticompetitive can then be combined with information
Reductions in incremental costs can off- merger cannot be disguised by arguing about premerger margins to give at least
set the incentive to raise price, since the for a very narrow market so as to quar- a rough estimate of the profit-maximizing
merged entity, like any firm, will have an antine the merging parties in separate post-merger price increase.
incentive to set a lower price, the lower “markets.” Suppose that Brands A and B In arguing for a broad market, a com-
are its incremental costs. propose to merge, but Brand X is situat- mon tactic is to calculate a “critical elas-
To be relevant, the cost savings must ed between them. Suppose further that a ticity” of demand for a group of products
truly stem from synergies specific to the merger between Brands A and X would being considered as a market,25 and then
merger. If one firm alone can achieve lead to at least a 5 percent price increase, argue that the true elasticity is above this
lower costs by expanding its scale of and likewise for a merger between critical level, making a 5 percent price
operations, that should occur through Brands B and X. Defense counsel might increase unprofitable. This method must
competition, not merger. Furthermore, if be tempted to argue that A and X form a be used with great caution in the context
they are to benefit consumers, the syner- market, and that B and X form a market, of differentiated products, to avoid at
gies typically must lower incremental but that A and B are not in the same mar- least two pitfalls. First, there is no reason
costs.23 ket. Still, if the Diversion Ratio between to restrict attention to a uniform price
A and B is significant (albeit smaller than increase of 5 percent for the purposes of
Market Definition and between A and X or between B and X), market definition if a single firm control-
Market Shares the merger of A and B could well harm ling the entire product category would
In my experience the main battlefield in consumers. In this case, it is appropriate find it optimal to increase the prices of
litigated merger cases is market defini- for the Division to use a price increase different brands by different amounts.
tion. However, any attempt to make a of more than 5 percent in defining the Second, care must be taken to ensure that
sharp distinction between products “in” market, as suggested in the Merger the claimed “market” elasticity is consis-
and “out” of the market can be mislead- Guidelines. tent with information about each brand’s
ing if there is no clear break in the chain Once markets are defined, using the own elasticity of demand and the cross-
of substitutes: if products “in” the market Guidelines approach there remains the elasticities of demand among the prod-
are but distant substitutes for the merging issue of how to use the firms’ market ucts in the category. Remember, the
products, their significance may be over- shares in differentiated-products markets. “market” elasticity will be lower than the
stated by inclusion to the full extent As the Guidelines point out, market share individual brand elasticities of demand,
that their market share would suggest; numbers must be interpreted in conjunc- and significantly so if the Diversion
and if products “out” of the market have tion with evidence about the proximity Ratios are large. If each brand sells at a
significant cross-elasticity with the merg- of the merging brands. If the merging high markup, this is strong evidence of a

2 8 · A N T I T R U S T
 A R T I C L E S • A N D • F E A T U R E S

low price elasticity for each brand, which it was appropriate to focus on unilateral consensus view among economists of
is inconsistent with a high “market” elas- effects, we disagreed about how to inter- how to analyze competition in differen-
ticity of demand. If the premerger pret the evidence of substitution between tiated-product industries, together pro-
markups are large and the Diversion fountain pens and roller ball and ball vide a consistent, valid, and reliable way
Ratios among the brands are large, claims point pens, and on the difficulty of prod- of evaluating proposed horizontal merg-
of a large “market” elasticity of demand uct repositioning. ers involving differentiated products.
are not credible. Similarly, in the cereals merger Central to the analysis are the Diversion
between Kraft and Nabisco,26 economic Ratios between the two merging brands,
Unilateral Effects in the Courts experts for both sides, relying on super- which measure the fraction of customers
In addition to the growing use of the market scanner data and survey evidence, of each brand that consider the merging
approach described here by Antitrust spent considerable time estimating elas- brand their second choice. In cases
Division economists, courts have also ticities of demand for the purposes of where detailed price and quantity data
incorporated this type of analysis into evaluating unilateral effects. Judge are available, Diversion Ratios can be
their reasoning. For example, in the pens Wood’s opinion discussed unilateral calculated based on estimated elastici-
case, Gillette, Parker Pen, and other firms effects at great length, emphasizing the ties of demand. Alternatively, the
such as Cross, Schaeffer, and Mont econometric estimates of cross-price elas- Diversion Ratios can be estimated based
Blanc, offered fountain pens, roller ball ticity between the key merging brands, on whatever pieces of evidence are avail-
pens, ball point pens, and mechanical Grape Nuts and Shredded Wheat, for able, including more qualitative infor-
pencils at a continuum of prices. I em- evaluating possible anticompetitive uni- mation.
ployed the methods sketched out here lateral effects. If a significant proportion of con-
to develop my testimony on behalf of sumers regard the merging brands as their
Gillette, although we lacked data on Conclusion first and second choices, the Diversion
prices and quantities for premium pens to Mergers in markets with differentiated Ratios will be high, and the merger will
econometrically estimate the Diversion products may seem a confusing area, and indeed create an incentive to raise price,
Ratio. This same general methodology the case law provides less guidance than particularly if premerger Gross Margins
was employed by George Rozanski in his one might like regarding how to define are large. This incentive can be undercut
analysis for the Antitrust Division. markets to include “reasonable substi- by rivals’ product repositioning, by entry,
Although Gillette and the Division agreed tutes.” The Merger Guidelines, and the or by credible synergies. ●

1
I am referring here to the Cournot model of quantity competition, which ROBINSON, THE ECONOMICS OF IMPERFECT COMPETITION (1933); EDWARD
dates back to 1838. See Joseph Farrell & Carl Shapiro, Horizontal CHAMBERLIN, THE THEORY OF MONOPOLISTIC COMPETITION (1933). By
Mergers: An Equilibrium Analysis, 80 AM. ECON. REV. 107 (1990) (apply- the 1990s economists have made great progress in understanding
ing this model to mergers); Asset Ownership and Market Structure in competition with differentiated products, offering a sound foundation for
Oligopoly, 21 RAND J. ECON. 275 (1990) (exploring the welfare proper- merger enforcement in such industries.
ties of this model, with applications to mergers and other transfers of assets 5
Let me emphasize that the reliability of this analysis depends very much
among rivals). on the data and other evidence upon which it is based. Ideally, the analyst
2
In assessing mergers among suppliers of homogeneous products, the main should conduct a sensitivity analysis to make sure the results are not over-
amendment to a structural approach urged by economics is that it is impor- ly sensitive to specific simplifying assumptions that must be made.
tant to consider the elasticity of demand facing today’s suppliers in the 6
Bertrand Equilibrium refers to the pattern of prices that prevails if each
aggregate. If entry is easy, or if other distinct products offer good substi- firm sets the price of its brands taking as given the prices of the other
tutes, the elasticity facing today’s suppliers of the homogeneous good will brands. This is also known as a “Nash Equilibrium” in prices. Nash
be high. And, if today’s suppliers indeed face highly elastic demand, Equilibrium is the dominant method by which economists model “non-
anticompetitive effects are less likely to be significant, for any given cooperative,” i.e., non-collusive, behavior among rivals. Bertrand equi-
market structure. In fact, a strong case can be made that the danger of anti- librium is not the only possible equilibrium concept, but, absent clear evi-
competitive effects in these markets can be gauged by the ratio of the HHI dence to the contrary, it is a very useful workhorse.
to the market elasticity of demand. One of the many strong points of the 7
I do not want to leave the impression that the actual analysis precisely
Merger Guidelines is that their “hypothetical monopolist” approach to
tracks the four steps outlined below, rather these four steps form a con-
market definition explicitly incorporates the aggregate elasticity of
ceptual road map.
demand into the market definition exercise.
8
This analysis will then have to be repeated for Brand B to see if its price
3
It is important to measure incremental cost properly in calculating the pre-
will rise after the merger. If the merged entity has an incentive to raise
merger margin. For example, if a firm is facing capacity constraints, one
either price significantly, the merger may be anticompetitive. Furthermore,
must include some capital costs in the measure of incremental cost. Also,
if the firms own multiple brands, the four steps outlined here will need to
the time frame and scale over which incremental costs are measured
be repeated for each individual brand owned by either firm. With multi-
should be commensurate with the unilateral effects being studied.
ple brands, not only must the analysis be repeated, but we must keep track
4
Economists have long realized that firms selling differentiated products of diversion to all brands owned by the merger partner when considering
have some “market power” in a technical economic sense, although typ- the repricing of any given brand.
ically not nearly enough to rise to the level of “monopoly” power. In the 9
In practice, economists often estimate the merging brands’ own- and
1930s Joan Robinson and Edward Chamberlin developed the theory of
cross-price elasticities, from which the Diversion Ratio can be calculat-
“monopolistic competition” to describe markets in which each firm has a
ed. I find the Diversion Ratio more intuitive and easier to work with, so I
distinct product, but competes with several or many other firms. JOAN
frame the discussion in terms of the Diversion Ratio.

S P R I N G 1 9 9 6 · 2 9
 A R T I C L E S • A N D • F E A T U R E S

10
More generally, the formula is DAB = (EAB/EA)(xB/xA), where DAB is the 15
The elasticities obtained from the econometric estimation of demand
Diversion Ratio from Brand A to Brand B, EA is the own-price elasticity were introduced into an industry model of the pricing of all brands to esti-
of Brand A, and xA and xB are the unit sales of Brands A and B respec- mate the unilateral effects of the merger. Some of the modeling methods
tively. Robert Willig, Merger Analysis, Industrial Organization Theory, employed are described further in Gregory Werden & Luke Froeb,
and Merger Guidelines, BROOKINGS PAPERS ON ECONOMIC ACTIVITY: Simulation as an Alternative to Structural Merger Policy in Differentiated
MICROECONOMICS 281 (1991), also notes that the ratio of the cross-price Products Industries (Economic Analysis Group Discussion Paper EAG
to the own-price elasticity measures the share of the marginal sales of one 95-2 Sept. 1995).
brand that will divert to another in response to a price increase. 16
This figure is obtained by dividing Brand B’s market share of 15% into
11
In other words, among a group of brands that are all equally “close” or the combined share of all the brands to which customers of Brand A turn,
“distant” substitutes, Diversion Ratios are proportional to market shares. which is 75%. This is an application of the sB/(1-sA) formula provided
This is the essence of the “logit” model of demand. See Willig, supra note above.
10, for a discussion of the role of market shares in differentiated product 17
Symmetry means that the two brands have equal unit sales and Gross
markets using the logit model. For further detail on using the logit model Margins prior to the merger and that the Diversion Ratio from A to B is
to analyze mergers, see Gregory Werden & Luke Froeb, The Effects of the same as that from B to A. This formula also assumes that each merg-
Mergers in Differentiated Products Industries: Logit Demand and Merger ing firm sells a single brand prior to the merger. The analysis is more
Policy, 10 J.L., ECON. & ORG. 407 (1994). involved and the formulas much more complex if the brands are not sym-
12
Of course the effects of a merger cannot be predicted based on data analy- metric or if the merging firms sell multiple brands prior to the merger.
sis alone. To be reliable, any data analysis must pass a reality check based 18
In fact, the constant elasticity assumption becomes logically untenable as
on business documents and the testimony of industry participants. the sum of the Gross Margin and the Diversion Ratio approaches or
13
United States v. Interstate Bakeries Corp., Civil Action No. 95C-4194 exceeds unity. The dangers associated with the constant elasticity assump-
(N.D. Ill. filed July 20, 1995) tion are greater, the larger are m and D.
14
United States v. Kimberly-Clark Corp., Civil Action No. 3:95 CV 3055-P 19
I believe that the percentage price increases predicted by the logit model
(N.D. Tex. filed Dec. 12, 1995). tend to be comparable to those of the linear model, although I am unaware
of any comparably simple formula in the two-firm symmetric logit model.
20
The magnitude of the optimal price increase depends not only on the pre-
http://www.abanet.org merger Gross Margin and Diversion Ratio, but also upon how quickly
elasticity itself rises as price rises above the premerger level. In principle,
http://www.abanet.org this information can be obtained from the data or from other evidence of
What’s New on the how consumers would respond to price changes. Incorporating such
http://www.abanet.org
http://www.abanet.org
Internet from the information is an essential “reality check” if simple formulas like the
ones displayed here are to be used.

http://www.abanet.org
Antitrust Section 21
Although it is difficult to generalize, and the Antitrust Division certainly
is not wedded to any single approach, there are some sound reasons to use
http://www.abanet.org formulas based on linear or logit demand systems for these purposes in
http://www.abanet.org preference over constant-elasticity systems. This is what the Division has
http://www.abanet.org done in recent investigations. As I noted above, however, econometricians

H ave you visited the Antitrust often estimate constant-elasticity of demand systems empirically. All I can
http://www.abanet.org say with confidence is that the demand system used to predict post-merg-
Section on the Internet lately? er price increases should, as much as possible, conform to the qualitative
http://www.abanet.org The Section’s site currently contains: as well as quantitative evidence that is available.
http://www.abanet.org ☛ a list of Section committee descrip- 22
Oligopoly theory generally predicts that horizontal mergers will lead to at
least marginally higher prices if they generate no efficiencies, although the
http://www.abanet.org tions with a committee application tendency towards higher prices can be thwarted by product repositioning
and membership form; or entry. This tendency is perhaps clearest in the case of differentiated
http://www.abanet.org
products and pricing (Bertrand) competition, where rivals typically will
http://www.abanet.org ☛ a list of Section publications and choose to raise prices if the merging parties do so. See Raymond
descriptions with an order form; Deneckere & Carl Davidson, Incentives to Form Coalitions with Bertrand
http://www.abanet.org Competition, 16 RAND J. ECON. 473 (1985). However, the same result
☛ a calendar of upcoming Section applies with quantity (Cournot) competition, even though rivals
http://www.abanet.org programs; and typically increase output as the merging firms restrict output. Even
http://www.abanet.org with these responses, horizontal mergers under quantity competition
☛ a form for comments and questions lead to higher prices unless they generate synergies. This is the “No
http://www.abanet.org that can be sent to the Section’s Synergies Theorem” proven in Farrell & Shapiro, Horizontal Mergers,
http://www.abanet.org e-mail address: supra note 1.
antitrust@attmail.com. 23
The issue of how to measure incremental costs remains, however. This
http://www.abanet.org involves questions of time and questions of scale, at least. The longer the
Visit us at the 44th Annual Antitrust time frame over which we are looking, the more costs tend to be variable,
http://www.abanet.org Spring Meeting! Stop by the Internet or incremental, rather than fixed. And some categories of costs may be
http://www.abanet.org display at the Antitrust Expo and see incremental with respect to large customers, but not small ones. In the
what the Section’s home page has to extreme case of one customer, such as the Department of Defense for some
http://www.abanet.org weapons systems, cost savings in virtually any category may be passed
offer. Speakers’ papers from the 1996 along to the customer, at least to some degree.
http://www.abanet.org Spring Meeting will be posted and 24
United States v. Gillette Co., 828 F. Supp. 78 (D.D.C. 1993).
http://www.abanet.org future developments are planned for the 25
The “critical elasticity” is the minimum elasticity for which a 5% price
site. We welcome your suggestions. ◆ increase would be unprofitable.
http://www.abanet.org
26
New York v. Kraft General Foods, Inc., 1995-1 Trade Cas. (CCH)
http://www.abanet.org ¶ 70,911 (S.D.N.Y. 1995).

3 0 · A N T I T R U S T

View publication stats