European Journal of Law and Economics, 3:83-95 (1996)

© 1996 Kluwer Academic Publishers

Multiproduct Activity and Competition Policy:

The Tetra Pak Case
AURORA GARCIA GALLEGO
Departamento de Economica, Campus de PenyetaRoja, UniversitatJaume L Castell6n, Spain
NIKOLAOSGEORGANTZIS
Departamento de Economica, Campus de PenyetaRoja, UniversitatJaume L Castelldn Spain

Abstract

We suggestthe Tetra Pak case as a real-worldexampleto study the implicationsof multiproductactivity for Euro-
pean CompetitionPolicy.TetraPak, a monopolistin aseptic carton packagingof liquid food, competeswith Elopak
in the nonaseptic sector. The EC Commissionused the effect of Tetra Pak's dominance in the aseptic sector on
its rival's performance as an evidence of the former's anticompetitive behavior. With linear demand and cost
functions and interdependent demands, the Commission'sposition can be supported. However, a more general
model suggests that the Commission's conclusions cannot be supported as the unique outcome of the analysis
of the information available.

Keywords: strategic, production, demand complementarity, substitutability

1. Introduction

In this study we attempt a review of the characteristics of multiproduct 1 activity, which

have to be taken into account by policy makers and case handlers when assessing the behavior
of finns with a broad range of activities in terms of competition law.
Even in the United States, where antitrust legislation has a much longer tradition in Europe,
the concern of competition authorities about the behavior of multiproduct firms is still ex-
pressed in a somewhat naive antidumping law according to which (international) dumping
is either defined as (international) price discrimination, or as marginal-revenue-below-
marginal-costpricing. It is a common place that to treat price discrimination as an illegal
practice may have a negative effect on social welfare, whereas prices that yield a nonnegative
marginal profit are not sufficient to prove the legal, competitive behavior of a multiproduct
finn. The obvious conclusions would be that we need a point between price discrimination
and below-cost pricing to define the limits between legal and illegal behavior of multiprod-
uct finns.
Bulow, Willig, and Panzar (1985) and Kim, R61ler, and Tombak (1992) use similar setups
to study the strategic aspects of multiproduct oligipolies. Strategic interaction appears to
be of great importance and cross-market effects depend on joint economies or diseconomies
and on whether rivals consider their products as strategic substitutes or complements. We
believe that these general results should be used to produce easy-to-apply tests for finn's
competitive or anticompetitive behavior, in order for concrete measures of competition
policy to be suggested to the authorities.
84 GARC[AGALLEGOAND GEORGANTZIS

We propose the Tetra Pak case as an example of competition policy dealing with
multiproduct activity. Our main objective is to show that--under a broad range of demand
and cost conditions--the evidence presented by the Commission to prove Tetra Pak's an-
ticompetitive activity might be the result of the firm's procompetitive behavior (although
it is not necessarily true that any multiproduct firm's behavior is procompedtive as long
as it can be explained with the model). Then, we employ a less general framework--which
is a special case of the general model presented first--assuming linear demand and cost
functions and demand relations alone between the aseptic and the nonaseptic sector. The
resulting model supports the Commission's claim concerning the anti-competitive nature
of any cross-market effect--from the aseptic to the nonaseptic sector--experienced by Tetra
Pak's rival, Elopak.
We, therefore, develop two alternative scenaria under which the Commission's position
may or may not be correct. We suggest that further information--especially on any pro-
duction relation between aseptic and nonaseptic products--could have contributed to assess
with more confidence the accused firm's behavior in terms of competition policy.

2. The Tetra Pak case

One of the recent cases subject to the EEC Competition Law concerns Tetra Pak, one of
the world leaders in the field of cartons for liquid food and the equipment and technology
for filling these cartons. Economic analysis has been applied to defend both the Tetra Pak
and the EC Commission positions.
The decision of the EC Commission was based on information that seems to prove Tetra
Pak's anticompetitive behavior and, in particular, that Tetra Pak has a dominant position
in the nonaseptic sector due to its abusive practices with respect to its monopoly in the
aseptic sector.
The Tetra Pak group makes cartons for packing both fresh or nonaseptic and aseptic
liquids. It has also its own technology for the manufacturing of machines for both fresh
and aseptic liquids. While Tetra is a monopolist is aseptic packaging across Europe, it has
one main competitor in the production of cartons for nonaseptic liquids: Elopak. Therefore,
Tetra Pak is a multiproduct firm and Elopak a single product one according to the types
of packaging offered by each one of them. At the same time, both firms are involved in
multimarket activity, given their participation in a large number of national markets.
Elopak brought the case to the EC Commission. More specifically, it accused Tetra Pak
for having infringed Articles 85 and 86 of the EC Treaty (EC Commission, 1988, 1991).
In particular, Elopak argued that its own market shares in the nonasepfic sector were decreas-
ing due to the dominance of Tetra Pak in the aseptic sector.
In Tetra Pak II, the second stage of the case, Elopak Italia (Milan) asked the Commis-
sion to investigate whether or not Tetra Pak Italia and its associated companies were in-
fringing article 86 of the Treaty. In its decision Tetra Pak II (1991), the EC Commission
charged Tetra Pak for having taken advantage of its dominant position in the aseptic sector
in machines and cartons to commit abuses on the related sector of nonaseptics.
In July 1991, the European Commission fined the company a record Ecu 75 million
for abusing its dominant market position in Western Europe claiming that Tetra Pak had
MULTIPRODUCTACTIVITYAND COMPETITIONPOLICY 85

pursued a deliberate policy aiming to eliminate actual or potential competitors. Tetra Pak
appealed to the European Court against the fine, but the Court--after a three-year debate
(Court of Justice, June 10, 1994)--rejected the petition presented by the firm.
The activity of Tetra Pak includes two main sectors, the aseptic (free from infection)
and the nonaseptic (used for packaging of fresh products). The nonaseptically packaged
products cannot be stored for a long time so that they need a very rapid and regular distribu-
tion system and there is always the risk of waste. The aseptic package gives to the product
a shelf life of several months.
Under this definition, there is one variety of an aseptic product and two varieties of a
nonaseptic product. The two products (sectors/markets) are demand-related via weak
substitutability.2 Furthermore, different varieties in the same sector are close substitutes.
The market definition suggested here differs from the one suggested by the Commission
in that, here, the machinery and the final product do not constitute different markets, while
in the study undertaken by the Commission, machines and final products constitute dif-
ferent, vertically related markets. In both our study and that undertaken by the Commis-
sion the multiproduct character of Tetra's activity is merely due to its participation in the
aseptic and the nonaseptic sectors. The participation of the two firms in many national
markets is used as a source of information on the same multiproduct oligopoly pattern,
replicated as many times as the number of national markets in which the two firms com-
pete with each other.
We would, also, expect that a production relation exists between aseptics and nonasep-
tics. The Commission does not explicitly refer to any effect of the coexistence of the two
types of products in the production line of the same firm on production costs. Nevertheless,
it is plausible to assume that a producer of both aseptic and nonaseptic products experiences
some savings in the fixed costs of production or that the production of one type of products
affects unit production costs of the other type (for example, in the case of economies of
scale in the use of one input that is common to the production of both products) (see Baumol,
Willig, and Panzar, 1982).
The only technically acceptable aseptic carton packaging machines commercially available
in the EEC for long shelf life treated liquid food are the Tetra machines. Tetra uses a method
of sterilization that is considered by Elopak as adequate for cartons supplied in continuous
rolls (such as the Tetra brik), but it is less suited to the gable-top cartons such as those
in which Elopak has experience) Then, the nonentry of Elopak in the production of asep-
tics may result from Elopak's inflexible technology.4
Due to technological reasons, entry into the aseptic packaging sector is difficult for a
nonaseptic packaging producer. However, it is relatively easy for a producer in the aseptic
sector to enter into the nonaseptic packaging for fresh liquid food.

3. Quantity-setting multiproduct firms under demand and/or production relations

As a specific application to the Tetra Pak case, we suggest a quantity-setting game with
perfect and complete information. Firm A--Tetra Pak--is the only finn that has the proper
know-how for the production of product 1--aseptics--and, therefore, it is the only firm
involved in the sector of product 1. On the other hand, the same firm produces product 2--
86 GARCIA GALLEGO AND GEORGANTZIS

nonaseptics--which is also produced by finn B--Elopak, the only competitor of firm A

in the sector for product 2. From a technical point of view, the two modes of product 2
are perfect substitutes.

P~ = PI(qA1, q.a2, qB2) = 3~(qal) + X(qA2 + qB2) (1)

P2 = P2(qA1, qA2, qB2) = f2(qA2, qB2) q- ~kqal (2)

in the regions in which prices and quantities are positive. Parameter )x represents the rela-
tion of demand substitutability--if ~ < 0--0r complementarity--if 9~ > 0--between prod-
ucts 1 and 2.
These functions are continuous for all qA~ > 0, qA2 > 0, qB2 > 0 and such that de-
mands are downward sloping6 and that qa2 and qB2 are quantities of a homogeneous good
produced by different firms:

tgP2 _ OP2 A oP1 - igP1

(3)
CgqA2 t g q B 2 tgqA2 OqB2 "

We also assume that direct demand effects are larger than cross-demand effects. 7
In the production side, the cost structure of firms A and B is defined respectively by
the following total cost functions:

CA = Cl(qA1) -F C2(qA2) + 3qA1 qA2 (4)

ca = C2(qB2) (5)

which are defined and continuous for all output levels qA1 >- O, qA2 > 0 and qB2 >- O.
Ca and CB have continuous first and second partial derivatives for all qAl, qa2, qB -> 0.
Furthermore, the first partial derivatives of the cost functions, CA and Ca, are positive
for all qA1, qA2, qB2 >- O.
Note that the cost functions for the two firms may be identical in the absence of produc-
tion relations between the two sectors. In any other case (production complementarity/
substitutability), we allow for differences in production costs of qA2 and qB2 resulting from
the fact that qA2 is produced jointly with qA1 and/or differences in marginal costs due to
differences between the two firms' scales in the second market. The parameter/3 is a measure
of production substitutability--if/3 > 0--or complementarity--if B < 0--between the two
products.
Each firm maximizes profits, taking the other firm's quantity strategies as given. This
implies that Cournot-Nash equilibrium quantities satisfy the system: OIIa/OqA1 = O,
OIIa/OqA 2 = 0 and OIIB/OqB 2 = 0, which reduces to

f(lqA1 "}-fl + )k(qA2 d- qB2) + )kqA2 = CA1 + ~qA2 (6)

f2A2qA2 + f2 + 2)~qA1 = CA2 + ~qA1 (7)

MULTIPRODUCT ACTIVITY AND COMPETITION POLICY 87

f~ZqB2 + f2 + ~qA1 : c B 2 , (8)

where f~ represents the first derivative of function ~ with respect to the variable qr.
Using the fact that, due to homogeneity of the product in the second market, f2n2 = f~2
and equations (7) and (8), we get

Proposition 1. Equilibrium quantities and prices of the quantity setting firms A and B in
markets 1 and 2 satisfy

CA - /3)aal + (C~ 2 - C~ 2) = f2a2(q82 - qA2) (9)

One of the consequences of Proposition 1 is that demand and/or production relations

between the two markets are directly related to the equilibrum market shares of the two
firms in market 2. That is, the advantage (disadvantage) of the multiproduct firm, as ex-
pressed by the sum of production (and/or demand) complementarity (substitutability) be-
tween the two jointly produced goods, increased (or decreased) by the cost advantage in
the second products' market, as expressed in the difference between the two firms' marginal
costs of producing product 2, is expressed in the multiproduct firm's larger (smaller) market
share in the second market as compared to that of its single-product rival in the same market.
Since f2a2 < 0 and if C2 is a linear function, which implies that the term (Cg 2 - C A2)
in the left-hand side of equation (9) becomes zero due to the fact that the derivative of
C2 is constant for any quantity of the second product produced by any of the two firms,
it is of interest to observe that

()k -- ~) > 0 =~ qA2 > qB2 (10)

()k -- ~) < 0 =~ qB2 > qA2" (11)

Under this assumption, let us consider the case in which t3 = 0 so that there is no pro-
duction relation between the products 1 and 2. In such a case, from Proposition 1 and
considering that aseptics and nonaseptics are substitutes in demand 0, < 0), firm A will
have lower sales of product 2 than firm B, for all positive quantities of product 1--that
is, qB2 > qA2.
Similarly, if products 1 and 2 are substitutes in production (/3 > 0) as well as in de-
mand, firm B will produce more of product 2 than firm A--that is, qB2 > qAE-
If, on the contrary, products are substitutes in demand and complements in production,
which is the most likely--among the three scenaria--to be the case here, so that k < 0
and/3 < 0, three results are possible:

1. Ixl > 1/31, If the degree of demand substitutability between products 1 and 2 is larger
than the degree of production complementarity between them, firm B is expected to
produce more than firm A of product 2--that is, qs2 > qA2.
2. Ixl = 1/51. If the degrees of demand and production relations between the two products
are equal, both firms A and B will produce exactly the same quantity of product 2--that
is, qB2 = qA2.
88 GARC[AGALLEGOAND GEORGANTZIS

. Ix] < 1/31. When the degree of production complementarity between the two goods
is bigger than the existing demand relation of substitutability, then firm A will produce
more than finn B of product 2--that is, qa2 > qB2.

In an exhaustive empirical analysis of the situation, this relation has to be taken into
account. We suggest that the demand system expressed in equations (1) and (2) and the cost
function of firm A, as expressed in (4), should be estimated. Demand and production sub-
stitutability or complementarity should be tested for, by looking at the estimated values of
the parameters X and 13. For example, it would be very important to verify whether X < 0
and/] < 0 or not--that is, whether or not the two products are demand substitutes and
production complements. While the Commission supports the former of the two hypoth-
eses with satisfactory arguments, production complementarity between aseptic and nonasep-
tic products is not tested for. Finally, the estimated values for X and/3 should be used to
compare sign{X - / 3 } to sign{qa2 - qB2}.
Imagine, as an example, that it happens that X < 0 and/~ < 0, so that demand substi-
tutability and production complementarity are estimated between the two products and that
those values are such that IX[ < I 1, Then, the observation of quantities of product 2 pro-
duced by firm A higher than quantities of product 2 produced by firm B (qa2 > q82) may
result from firm A's competitive behavior. In that case, one cannot use the relation between
the two firms' outputs in the market for nonaseptics as an evidence of abuse of Tetra's domi-
nant position in the market for aseptics.
In fact, using the data available to the EC Commission, s it is easy to estimate equation
(9). At the same time, we test the validity of Elopak's claim that its own market share in
the nonaseptic sector has decreased because of Tetra Pak's dominance in the aseptic sector.
The results imply that the negative effect of Tetra Pak's sales in the aseptic sector on
Elopak's sales in the nonaseptic sector is not significant. This can be considered to con-
tradict Elopak's complain.
Nevertheless, and even if the cross-market effect on Elopak's performance were signifi-
cantly different from zero and negative, 1/31 > Ixl would be sufficient to argue that such
negative effect might result from Tetra Pak's procompetitive behavior.

4. A linear quantity-setting model with demand relations

We present here a variation of the model assuming linear demand and cost functions and
demand relations alone. An interesting implication is that, in absolute terms, the equilibrium
profits and the output of the single-product firm, as well as the price in the second market
are not different from what they would be in the standard Cournot-duopoly case. In other
words, the participation of the multiproduct firm in the first market affects only its own
profits and strategies. This could justify the worries inspired to the Commission by Elopak's
complain concerning a loss in the nonaseptic sector due to Tetra Pak's abuse of its power
in the aseptic sector.
Consider that demands for the two goods are interrelated through substitutability or com-
plementarity, depending on the sign of the parameter X in the system of the following in-
verse demand functions:
MULTIPRODUCTACTIVITYAND COMPETITIONPOLICY 89

Pl = a - blQ1 - RQ2 (12)

P2 = h - b2Q 2 - }~Q1 (13)

in the regions in which both prices Pl and P2 as well as quantities are positive, and where
Q1 = qal and Q2 = qa2 + qB2 are, respectively, the total quantities of products 1 and 2.
The parameters a, h, bl, and b2 are nonnegative. The parameter X is a measure of
substitutability--if X > 0--or complementarity--if X < 0--between the two products. Fur-
thermore, we assume that Ixl < m i n { b b b2}, which in RiSller and Tombak (1990) is in-
terpreted as a higher own price effect as compared to the cross-price effect.
Production technologies are described by the linear cost functions

Ca = c l q a l + c2qa2 (14)

CB = czqt~z, (15)

where CA a n d CB are, respectively, the total costs of production for firms A and B and
Cl, c2 are the marginal costs of production for products 1 and 2.
Under this specification, profit functions are given by

I I A = (a - blqAa - X(aa2 + qB2))qal + (h - b 2 ( q A 2 -}- q B 2 ) -- ~qA1)qA2 --

(clqal + c2qa2) (16)

I I B = (h - b2(qA 2 + qB2) -- ~kqA1)qB2 -- c 2 q s 2 . (17)

Firms are risk neutral and set quantities to maximize profits. From equation (17) we
get O 2 I I J ( O q , & l B 2 ) = -X, which implies that the cross-effect of firm A's strategies with
respect to product 1 on firm B's strategies with respect to product 2 on firm B's profit
is positive (negative) if products are demand complements (substitutes). Derivatives
0 2 I I A / ( a q , aqa2) and a2IIB/(aqat&/82) measure the d i r e c t e f f e c t that the demand relation
between the two products has on firm's profits.
Each firm takes its rival's output as given. Equilibrium is reached simultaneously with
respect to the two products. The assumptions on Ixl, bl and bE guarantee that equilibrium
is locally strictly stable.
Firms decide the output strategies that maximize their profit. The equilibrium quantities
are such that

qa~ = b2(a - ct) - X(h - c2) (18)

2(bib2 - X 2)

qA2 = (2bib2 + x z ) ( h - c2) - 3 b z X ( a - ca) (19)

6 b 2 ( b l b 2 - ~k2)
90 GARCIA GALLEGO AND GEORGANTZIS

for firm A and

h - c2
qB2 -- (20)
3b2

for firm B.
Equilibrium outputs are assumed to be positive in order for the industry configuration
to be nontrivial. We assume that (a - cl) > 0 and (h - c2) > 0.
Equilibrium prices for the two goods are given by

3b2(a + el) -- )~(h - c2)

Pl = (21)
6b2

h + 2c2
P2 -- - - (22)
3

Equilibrium profits for firm A are given by

Ha = 9b2(a - cl)2 - 18b2X(h - c2)(a - Cl) -]- (5)~ 2 "[- 4blb2)(h - c2) 2 (23)
36b2(blb2 - )~2)

ii a _ (h - c2) 2 (24)
9b 2

It is worth observing that

Proposition 2. Under linear demand and cost functions, when products are demand
substitutes, firm A will produce less of at least one of them as compared to what it would
produce if products were unrelated.

Proof. See Appendix 1.

Proposition 3. In the linear case firm B' s output and profits are the same in the presence
of demand relations between products 1 and 2 as it would be if the two products were
unrelated. Moreover, the price of product 2 is the same as if it were sold in a duopoly
in the absence of demand relations with product 1.

Proof Straightforward from equations (20), (22), and (24), observing that they coincide
with the formulas corresponding to the standard Cournot duopoly. []

According to this proposition, the strategic (within the nonaseptic sector) and the direct
demand (cross-market) effects of the multiproduct firm's strategies in the aseptic sector
on its single-product rival's performance and equilibrium price in the nonaseptic sector
cancel out entirely.
MULTIPRODUCTACTIVITYAND COMPETITIONPOLICY 91

Therefore, in the Tetra Pak case, decreases in Elopak's sales and profits in the nonaseptic
sector, as a result of Tetra Pak's participation in the aseptic sector, could contradict Tetra
Pak's procompetitive behavior. Then, the approach adopted by the Commission could have
been strongly supported by showing that the linear model presented offers a close enough
description of the debated situation and by using empirical evidence to confirm that the
cross-market effect claimed by Elopak really exists.

5. Conclusions

Using the Tetra Pak case as an example, we have studied the coexistence of demand and
production relations in a multiproduct oligopoly.
When the two products are demand substitutes or complements, there are two different
effects of the equilibrium with respect to the first product on the equilibrium in the second
product's market: the direct effect and the strategic effect. In the linear version of the model,
the two effects cancel out entirely as far as the single product firm's sales and profits as
well as the second product's equilibrium price are concerned. In general, the direct effect
of a firm's actions with respect to one of its products on the strategies of its rival in the
supply of another product is not trivial.
We have produced testable hypotheses that, instead of establishing a criterion that could be
used to detect any anticompetitive elements in the behavior of multiproduct firms, could give
some idea on whether the observed situation could result from the firm's procompetitive be-
havior. A possible application of this study in the Tetra Pak case would require the estimation
of production and demand relation coefficients between the aseptic and the nonaseptic sector.
The theoretical model is constructed, considering a setup similar to one of the Tetra
Pak case. The model suggests that the sum of production and demand relation coefficients
determines the size and the sign of the effect of multiproduct activity on market share dif-
ferences in the sector in which the single-product firm competes with the multiproduct one.
However, in the Tetra Pak case, the negative effect of Tetra Pak's sales claimed by Elopak
on its own sales does not appear to be significant. Moreover, and even if such effect were
significant, dominance of the production complementarity coefficient over the coefficient
of demand substitutability (1 1 > Ixl) would be sufficient to argue that such a negative
effect could result from Tetra Pak's procompetitive behavior. Then, the Commission may
have ignored an important--procompetitive in an antitrust sense--feature of Tetra Pak's
behavior that justifies its dominance in the nonaseptic sector as a result of its participation
in the aseptic sector.
In the absence of production relations and if the two firms involved act in the way im-
plied by the linear version of the model, the Commission would be right to worry about
the effect of Tetra Pak's participation in the aseptic sector on its rival's performance in
the nonaseptic sector. However, it would be necessary to prove first that, in the case under
investigation, such an effect really exists.
Summarizing our conclusions, we would like to point out that the advantage of participating
in more than one market (or producing more than one product) does not necessarily imply
losses to those fh'mst that do not have the possibility of multiproduct activity. However,
the existence of some negative cross-market effect, even when rigorously tested, should
not be automatically interpreted as an evidence of anticompetitive behavior.
92 GARCL~k GALLEGO A N D GEORGANTZIS

Appendix 1: Proof of Proposition 2

Consider the case of demand substitutability, that is ~ > 0. Remember that

~, < min{b 1, b2}

is a necessary condition for a strictly stable equilibrium. Firm A's output of product 1
is higher than in the absence of demand relations between the two products if

)~ > b l h - c2 _ b~
a - C1

Finn A's output of product 2 is higher than in the absence of demand relations between
the two products if

a - C1 p
h > b2h c2 - b2.

We want to show that there is no value of X that makes the three last inequalities hold
simultaneously.
Let a - cl = tl and h - c 2 = t2. For each {i, j} = {1, 2}, it is true that bi' < bi
= bj'>_ bj.
In that case, it must hold that X > bj >_ bj, which violates ~ < rain {bl, b2}. []

Appendix 2: The data

We present here the data available on the two finns' outputs in the two sectors in the EC.

Table 1. Data.

Variable B DK F D GR EI I NL P SP UK LU

STas 502 237 2916 3262 66 15 2756 269 539 1408 1108 502
STna 21 343 253 793 10 337 1288 379 20 127 821 21
SE 50 650 200 1100 50 150 400 800 0 100 1200 50

Source: Annexes of the EC Commission Decision, Tetra Pak II, L 72/1, July 24, 1991.

The columns correspond to the twelve countries that belong to the European Community.
Sales are measured in 106's of product units. The variables are

STas Tetra Pak's sales in the aseptic sector for cartons

STna Tetra Pak's sales in the nonaseptic sector for cartons
SE Elopak's sales in the aseptic sector for cartons
MULTIPRODUCT ACTIVITY AND COMPETITION POLICY 93

Acknowledgments

T h e r e s e a r c h presented h e r e was financed by the Instituto Valenciano de Investigaciones

E c o n 6 m i c a s 0 V I E ) , w h o s e industrial e c o n o m i c s area is s p o n s o r e d by the IMPIVA. Part
o f the r e s e a r c h was written w h e n the authors w e r e visiting the F a c u l d a d e de E c o n o m i a
o f the U n i v e r s i d a d e N o v a de Lisboa. We are grateful to the faculty m e m b e r s o f that univer-
sity and especially to Luis Cabral and Pedro Pita Barros for their hospitality and stimulating
discussion. We are grateful to F r a n c i s c o Caballero, A m p a r o Urbano, G o n z a l o Olcina,
Vicente Orts, and R o b e r t Waldmann for e n c o u r a g i n g c o m m e n t s . A n earlier version ap-
p e a r e d as an I V I E w o r k i n g p a p e r ( W P - E C 95-04) and was p r e s e n t e d at the 1993 A S S E T
meeting in Barcelona and the 1994 E c o n o m i c s S e m i n a r at the University of Valencia, Spain.
T h e p a p e r benefitted substantially f r o m c o m m e n t s and discussion by the participants in
both events, as well as f r o m c o m m e n t s by two a n o n y m o u s referees of this journal. All re-
maining errors are the authors' responsibility alone.

Notes

1. We use the term multiproduct as a synonymous of multiservice, with more than one activity, mulfiplant, multi-
location, or multimarket. In that sense, multiproduct firms may be the ones that operate in more than one
market (including geographical markets), have more than one plant and/or supply more than one good or ser-
vice. However, the terms muln'product and multimarket do not always mean the same thing, given that a
multiproduct firm may participate in the market for a differentiated good, selling two different varieties.
2. For example, if I want to buy fresh milk but, at the moment of the purchase I consider the price too high,
or I find out that there is only UHT milk left, I am likely to consider buying UHT milk instead.
3. The key for entering the market for supplying cartons for long-shelf-life food products lies in the ability to
supply aseptic packaging machines for these cartons. A machine producer must not only have an adequate
sterilization technique for the cartons, but must also be able to incorporate this technique into a reliable filling
machine capable not only of working continuously and reliably at high speeds but also of maintaining an aseptic-
sterile environment in dairy conditions. These are the technical barriers to entry for production of machines.
4. For a definition of the term, see R611er and Tombak (1990). However, a framework in which firms choose
between multiproduct and single-product activity is not appropriate for the ease studied here because Elopak
is restricted to participate in the nonaseptic sector alone due to an exogenously given nonavailability of the
necessary know-how for aseptic packaging. Similar works on multiproduct firms with symmetric production
possibilities, like Lal and Matutes (1989), are, therefore, not immediately applicable to the case under study.
5. We assume that the interaction of the two products in consumer preferences and in the cost function of firm
A is expressed in a term of the form ~'qAl(qa2 + qB2) and (3qalqA2, respectively.
6. That is,

°P1 < o, OP2 < 0 and OP2 < O.

Oqal OqA2 OqB2

7 r --OqA1 0~q/lp and --0qA1 < ~q/ , for i = A2, B2.

8. We use the annexes of the EC Commission Decision, Tetra Pak II, presented in Appendix 2.
Since the decision affects all European countries, we use cross-section data that refer to the EC Member
States for the year 1985. This implies a total of twelve observations for each variable. In Appendix 2, variables
q,41 and qA2 denote Tetra Pak's sales of cartons in the aseptic and nonaseptie sectors, respectively, in millions
of packages. Variable qn2 notes the sales of Elopak in the nonaseptic sector of cartons.
94 GARCIA GALLEGO AND GEORGANTZIS

The data set cannot be used to estimate the parameter of the demand relation between products 1 and 2.
However, since it is reasonable that aseptic and nonaseptic products are--as considered by the Commission--
demand substitutes, we assume h < 0. Therefore, and given that no information is available to estimate the
value of parameter 3, we look for the sign{X - 3}.
Expression (9) can also be written as follows:

qB2 = qa2 -F ~ ' ~ qA1.

Lf2 J

The sales of Elopak in the nonaseptic sector--qB2--are used as the endogenous variable, while Tetras sales
in both sectors--qA 2 and qAl--are the explanatory variables of the empirical model. A simple regression anal-
ysis leads us to the following estimated model:

qB2 = 0-89qA2 -- 0"0044qA1,

(2.77) (-0.04)

with R 2 = 0.673 and t-statistics in parentheses. Then, the parameter that multiplies the explanatory variable
qa2 is significantly different from zero and very close to unity, as it is predicted by the theoretical model.
We should also point out that Garc{a (1994) finds a low correlation between the outputs of Tetra Pak in
the two sectors (aqalqa2 = 0.2), which indicates that the estimation of this simple model should not suffer
from multicollinearity. In any case, the correlation of outputs of different firms in the same sector may result
from the correlation of beth to the size of national sales.
Finally, the coefficient of qal is not significantly different from zero.

References

Banmol, W., R. Willig, and J. Panzar. (1982). Contestable Markets and the Theory of Industry Structure. San
Diego: Harcourt.
Bernheim, D., and M. Whinston. (1990). "Multimarket Contact and Collusive Behavior." Rand Journal of
Economics 21(1), 1-26.
Bresnahan, T.E (1989). "Empirical Studies of Industries with Market Power." In R.D. WlUig and R. Schmalensee
(eds.), Handbook of Industrial Organization (vol. 2, oh. 17). Amsterdam: North Holland. New York: Har-
court, Brace, Jovanovich.
Bulow, J.I., J.D. Geanakoplos, P.D. Klemperer. (1985). "Mulfimarket Oligopoly: Strategic Substitutes and Com-
plements" Journal of Political Economy 93(3), 488-511.
EC Commission. (1988). "Tetra Pak I " EC Commission Decision, O.J. L 272/27, July 26.
EC Commission. (1991). "Tetra Pak lI." EC Commission Decision, O.J. L 72/1, July 24.
Ferndndez de Araoz, A. (1993). "El Problema del Oligopolio en el Derecho de la Competencia: Reflexiones
en torno al Mercado, la libre Competencia y la Geometn'a" Rev/sta General de Derecho 589-590, 10057-10131.
Garcfa, A. (1994). "Insights from Experimentation in Oligopoly and Evidence from the Real World,' Ph.D.
thesis, European University Institute, Florence.
Georgantzis, N. (1994). "Short Run and Long Run Cournot Equilibria in Multiproduct Industries." EUI Work-
ing Paper, ECO 94/23.
George, K. (1990). "UK Competition Policy: Issues and Institutions." In J. Lesourne and H. Sonnenschein (eds.),
Competition Policy in Europe and North America: Economic Issues and Institutions. Chur: Harwood Academic
Publishers.
George, K., and A. Jacquemin. (1990). "Competition Policy in the European Community." In J. Lesourne and
H. Sormenschein (eds.), Competition Policy in Europe and North America: Economic Issues and Institutions.
Chur: Harwood Academic Publishers.
Crerosld, E A., and A. Jacquemin. (1984). "Dominant Firms and Their Alleged Decline?' International Journal
of Industrial Organization 2(1), 1-29.
MULTIPRODUCT ACTIVITY AND COMPETITION POLICY 95

Gyselen, L. (1990). "Abuse of Monopoly Power Within the Meaning of Articles 86 of the EEC Treaty: Recent
Developments?' In Barry Hawk (ed.), Annual Proceedings of the Fordham Corporate Law 1992 and EEC/US
Competition and Trade Law Ardsley-on-Hudson. New York: Transnational Juris Publications Incorporated.
Kauper, T. (1990). "Dominant Positions and Their Abuse Under Article 86." In Barry Hawk (ed.), Annual
Proceedings of the Fordham Corporate Law 1992 and EEC/US Compeation and Trade Law Ardsley-on-Hudson.
New York: Transnational Juris.
Kim, T., L.H. R/~ller, and M.M. Tombak. (1992). "Strategic Choice of Flexible Production Technologies and
Welfare Implications: Addendum et Corrigendum?' Journal of Industrial Economics 40(2), 233-235.
Lal, R., and C. Matutes. (1989). "Price Competition in Multimarket Duopolies." Rand Journal of Economics
20(4), 516-537.
Martin, S. (1993). Advanced Industrial Economics. Oxford: Blackwell.
Martin, S. (1994). "Oligopoly Limit Pricing: Strategic Substitutes, Strategic Complements" EUI Working Paper,
ECO No. 94/15.
Ordover, A. (1990). "Economic Foundations of Competition Policy." In J. Lesourne and H. Sonnenschein (eds.),
Competition Policy in Europe and North America: Economic Issues and Institutions. Chur: Harwood Academic.
Parker, P., and L.H. R6ller. (1994). "Collusive Conduct in Duopoties: Mulfimarket Contact and Cross-Ownership
in the Mobile Telephone Industry." CEPR Discussion Papers No. 989.
Rfller, L.H., and M.M. Tombak. (1990). "Strategic Choice of Flexible Production Technologies and Welfare
Implications" Journal of Industrial Economics 38(4), 417-431.