Making*
by
*
We thank Ian Harper, Donald Robertson and Philip Williams for helpful discussions. Of course,
responsibility for all views expressed lies with the authors.
**
Department of Economics, Department of Economics and Melbourne Business School, respectively.
All correspondence to Joshua Gans, Melbourne Business School, 200 Leicester Street, Carlton,
Victoria 3053; E-mail: J.Gans@mbs.unimelb.edu.au. The latest version of this paper is available at
http://www.mbs.unimelb.edu.au/home/jgans.
2
I. Introduction
1
There is a large economics literature that considers this general problem and potential solutions.
Baron (1989) and Laffont and Tirole (1992) provide useful survey of this literature.
3
2
The ACCC may consider cost savings as a ‘public benefit’. However, it has stated that if cost savings
are retained by the relevant firms and not passed on to consumers then they are likely to be given less
importance in evaluating the net public benefit or detriment from a merger, compared with the same
situation where the cost savings are passed on to consumers. See the ACCC’s merger guidelines
(Australian Competition and Consumer Commission, 1996), paragraphs 6.43 and 6.44. In the parlance
of economics, this means that a merger is more likely to be authorised if it not only raises social
welfare but also ensures that some of the welfare gain is passed on to consumers.
4
3
Mergers examined under section 50 will contravene the TPA if they have the (likely) effect of
substantially lessening competition in a substantial market. However, mergers dealt with under the
authorisation process will be allowed notwithstanding an anticompetitive effect so long as it can be
shown that the merger will produce a net public benefit. The main benefits likely to be given weight
by the ACCC are cost efficiencies and factors leading to an increase in international competitiveness.
Therefore, the types of undertakings parties will make under each process are quite different.
4
Section 87B undertakings can also be accepted by the ACCC in the exercise of all its powers (except
Part X) of the TPA.
7
5
See Australian Competition and Consumer Commission (1996) paragraph 7.11. Clearly behavioural
undertakings raise regulatory issues despite being offered by the relevant firm(s). Prior to the ACCC
making a decision about a merger, firms will often be relatively willing to offer undertakings.
However, once the ACCC has decided not to oppose a merger and the merger has been implemented,
firms’ incentives to implement the undertakings are sharply reduced and a significant degree of
monitoring by the ACCC is often required. However, the same problems also arise to some degree with
structural undertakings. For example, if the undertaking involves asset sales, it may be difficult for the
ACCC to guarantee that the assets actually sold are the same as specified in the undertaking. Further,
the sale will usually only occur if the firm receives a reasonable price. But this may be a source of
dispute.
6
See Australian Competition and Consumer Commission (1996) paragraph 7.10.
8
assets it would acquire after the merger, thereby avoiding the anti-
competitive effects of the merger.
The Ampol-Caltex merger also involved a structural undertaking. The
ACCC concluded that the proposed merger would breach section 50 by
reducing the number of oil companied from five to four. The parties offered
undertakings that had the effect
inter alia of facilitating import competition
that would not otherwise have occurred. In this instance the undertaking to
increase import competition was regarded as balancing the reduced
competition in the domestic market.
Despite the strong statements in the merger guidelines, the ACCC will
accept behavioural undertakings so long as they are coupled with structural
undertakings. In theCaltex-Ampol merger ,the parties agreed not to deal
exclusively, thereby relieving concerns about the vertical aspects of the
merger. The parties also guaranteed supply of petrol at a competitive price
during a transition phase. Similarly in the
Westpac-Bank of Melbourne
merger the parties agreed to provide access to third parties of its ATMs at
agreed upon access prices.
Behavioural undertakings have also been used in authorisations to
ensure that there are sufficient public benefits from a merger. In the Davids-
Composite Buyers case, the ACCC accepted undertakings regarding the terms
and conditions between Davids and retail consumers. Presumably, this
undertaking was accepted to guarantee that private benefits, in the form of
cost savings to Davids, were (at least in part) passed on to retail consumers.
These undertakings were later withdrawn and the Tribunal authorised the
merger without such undertakings.
While the ACCC largely dismisses behavioural undertakings as
unworkable, they do not consider the potential benefits that might flow from
behavioural undertakings. As we demonstrate, these benefits might be
substantial.
9
7
Suppose that market (inverse) demand is given by P = 10 – (q 1 + q2). Firm 1 has a (constant)
marginal cost of 0 while 2’s is c. The firms initially compete as Cournot duopolists. If c > 25/11and
merger rationalises production to 1’s assets, then a merger raises the sum of consumer and producer
surplus. This is because the merger reduces average industry costs be a sufficient amount to overcome
deadweight losses associated with the merged firm’s market power. Such efficiency improvements
from the removal of smaller firms can also occur in contestable markets (see Gans and Quiggin, 1997).
8
While Demsetz (1974) notes that a merger may remove an inefficient competitor, Williamson (1968)
notes that a merger can create a new more efficient firm.
9
The assumption of constant marginal costs is made without loss in generality. All the arguments
below extend to the case of increasing (or indeed, decreasing) marginal costs, although this would
complicate the graphical analysis.
10
10
Our conclusions in this paper about the desirability of minimum quantity undertakings rest on three
relatively weak assumptions about firm interaction. First, we assume that if there is an increase (no
change) in the total output of one subset of firms then the best response by all other firms leads to an
increase (no change) in total industry output ceteris paribus. In other words, if a subset of firms raises
their output, other firms may lower output but not to such a degree that total industry output falls. This
is a standard assumption and is satisfied, for example, by a standard Cournot model. Also, if there is no
change in the output of a subset of firms then, in the absence of any other change, total industry output
is unchanged. This means that simply the act of merger with no change in the output of the merged
firms cannot change the behaviour of other firms in the industry. This will normally be satisfied so long
as firms make independent strategic decisions.
Secondly, we assume that there is a unique well-defined equilibrium in the industry both pre-
merger and post-merger. This assumption is for convenience as it enables us to ignore issues of
multiple equilibria.
Thirdly, we assume that if a subset of firms face an increase in production costs then its
equilibrium output cannot increase ceteris paribus. This rules out the intuitively implausible case
where a merger raises the merged firms’ costs, the merged firms increase output, industry output rises
and consumer prices fall, but social welfare falls because of the rise in the merged firms’ costs.
The diagrammatic analysis presented in this paper is based on a standard Cournot model but
this is purely for exposition.
11
A is simply a transfer among firms and does not affect industry producer surplus.
11
Given the potential for cost reductions, the regulator faces a difficult
problem when evaluating a merger. A particular merger proposal may be
predominantly motivated by a reduction in competition (B – D – A) or by cost
reductions for the firms (E). However, evidence on cost reductions comes
from the merging firms themselves. If those firms know that a regulator is
more likely to authorise a merger if there is favourable evidence on cost
reductions then the firms are likely to present such evidence. Unfortunately,
this information may be distorted and unreliable. While the merging firms
know the true underlying motivation for the merger and the expected extent
of any cost reductions that are likely to be achieved, there is an asymmetry of
information between the regulator and firms. The regulator may try and
reduce this asymmetry by conducting further investigations but it will
inevitably need to rely on information provided by parties who have a vested
interest in the merger. It will be difficult for the regulator to verify
information provided to it. After all, the regulator is not trying to determine
cost reductions that have already been achieved but rather the merging firms’
beliefs about potential savings and the true reason behind the merger. So the
regulator must look upon the information provided by parties with a sceptical
eye and potentially might refuse some mergers that would be socially
beneficial.12
The regulator may try and gather other information about the
proposed merger that is less open to distortion and manipulation. Swan
(1995) argues that competitors’ responses to the merger provide relevant
information to regulators.
As a matter of logic, rivals should unambiguouslysupportmergers which are
anti-competitive but will do their best toprevent mergers which promote
synergies and cost reductions. To the extent that the merger has elements of
both anti-competitive effects and synergistic cost reductions, the net balance
between these two offsetting effects should determine the attitude of rivals
towards the merger. Rivals are likely to have a better idea than any group
outside the industry, no matter how knowledgeable, as to whether a
proposed merger is on balance pro- or anti-competitive in its effect. The more
that synergistic cost reductions outweigh anti-competitive effects, the more
likely rivals are to oppose the merger. (Swan, 1995, pp.88-89; the italicsin
original)
12
See Milgrom and Roberts (1986) for a discussion of the optimality of such scepticism.
12
13
Similar, more specific analyses are provided by Levin (1990) and McAfee and Williams (1992).
13
14 In
case that any privately profitable merger is socially desirable as well.
effect, the Farrell and Shapiro test evaluates the net benefits to agents other
than the merging firms. These are the other firms in the industry (who get F +
A) and consumers who lose (F + B + C). This external effect is positive if A > B
+ C.
The Farrell and Shapiro test reduces the regulator’s reliance on
information supplied by the merging firms because the components A, B and
C do not directly depend on the degree of cost savings. A relates to the
aggressiveness of competing firms’ responses to output reductions. B
depends on the market share of the merging firms. C relates to the elasticity of
market demand. Farrell and Shapiro (1990) provide some simple tests that are
based only on pre-merger market shares and demand elasticity that indicate
when a privately profitable merger will be socially profitable. These tests,
however, suffer from several difficulties. First, they require specific
knowledge of market and technological conditions. Second, they are only
sufficient conditions. Hence, some socially profitable mergers may not pass
the tests. Finally, the tests rely on private profitability. Regulatory rules based
15
on them do not ensure that every socially profitable merger takes place.
The Farrell and Shapiro approach offers a way for regulators to trade-
off the anti-competitive and cost reducing effects of a merger. But it is far
from perfect. As we will discuss in the next section, effective undertakings
may help regulators to evaluate mergers without having to obtain detailed
knowledge of the industry structure and of the magnitude of potential cost
reductions.
14
A merger is privately profitable if E – D > A – B and socially beneficial if E – D > C. The private
condition is more stringent than the social condition if A – B > C.
15
Ziss (1998) also demonstrates that when output decisions are delegated within a corporation, the
tests are limited in their applicability.
14
socially desirable, but in the absence of an undertaking, the merger will both
violate section 50 of the TPA and fail to be authorised by the ACCC. Then the
firms that wish to merge have an incentive to assuage the regulator’s concerns
by presenting a behavioural undertaking. The behavioural undertaking must
credibly signal the regulator that the merger will raise social welfare and, to
the degree that the regulator weights consumer surplus more highly than
firm profits, must signal that consumers will not lose from the merger.
Assume that the regulator can observe the pre-merger output of the
firms, q0. Recall that a reduction in social welfare can only occurQif1 < Q0. As
outside firms’ costs do not change following the merger, their market share
will be unchanged ifq1 = q0, and will tend to fall ifq1 > q0. This means that so
long as q1 ≥ q0, Q1 must be at least equal to Q
0. Hence, if the merger proposal
industry profits so firms that seek to merge when there are no cost savings
16
would not be willing to make such an undertaking.
In brief, a minimum quantity undertaking is a credible signal that the
merger involves expected cost savings and guarantees that those cost savings
will not be more than offset by a fall in consumer surplus due to diminished
post-merger competition.
There are several things to emphasise about the minimum quantity
undertaking. First, the regulator does not need to know anything about the
industry, the magnitude of∆, or the level of concentration in the industry. All
it needs to know isq0. While determiningq0 may be difficult in practice it will
often present the regulator with significantly less difficulty than determining
likely cost savings or potential competitive reactions.
Secondly, it is well known that socially beneficial mergers may not be
privately profitable in a Cournot oligopoly (Salant, Switzer and
Reynolds,
1983). This is because the merged firm loses A to other firms in the industry. If
A is large, it is possible that a socially beneficial merger is not privately
profitable.17 However, with the undertaking, the merged firm does not lose A
as a result of the merger. The minimum quantity undertaking aligns private
and social incentives and encourages mergers that are socially desirable but,
in the absence of a minimum quantity undertaking, would not have been
privately profitable. This is possible because the undertaking changes the
post-merger equilibrium in the industry. It can help the merged firms
maintain higher output when this is in both their own and society’s interest
but, in the absence of the undertaking, would not be credible.
Finally, note that the undertaking actually improves the efficiency of
the merger. Without an undertaking, the regulator could approve a merger if
16
Formally, to see that these conclusions follow from our three assumptions on firm behaviour
presented in footnote 12, the second assumption means that we can avoid multiple equilibria. If ∆ < 0,
then the quantity undertaking must bind by the third assumption, so the merged firms will make a loss.
The merger is socially and privately unprofitable. If ∆ ≥ 0 and the quantity undertaking does not bind,
then industry output must rise by the first assumption so the merger is socially profitable if it is
privately profitable. If the quantity undertaking binds, then by the first assumption, industry output will
be unchanged and the merger is socially desirable if it is privately profitable.
17
Note that the social benefits from a merger are greater than the private benefits if A – B > C.
16
18
To see why quantity rather than price undertakings directly address the competitive concerns,
consider the following simple example. Suppose a merged firm made a price undertaking then sold on
its operations to create another company. The ‘shell’ of the merged companies can trivially satisfy the
price restriction and simultaneously sell nothing. The new company would be free of the price
restriction and can set its production to maximise profit. This could not occur under a minimum
quantity restriction as the ‘shell’ would still have to meet the outcome target or face a penalty.
19
For example, the industry may move to a new equilibrium where the merged firm reduces production
but other firms raise production so that industry output and price remain unchanged. This outcome
satisfies the price undertaking but, to the degree that the merged firm has lower production costs, this
outcome represents a less efficient mix of production than under a quantity undertaking.
17
that the regulator can judge an appropriate quantity undertaking on the basis
of outputs over a number of years, it is unlikely that unilateral preparation for
merger would represent a major problem.
When using historic data to judge an appropriate quantity
undertaking, the regulator will need to allow for a variety of factors that may
alter output over time. These same issues arise when placing the quantity
undertaking in a dynamic context, and are discussed below.
20
For example, Armstrong, Cowan and Vickers (1994) presents a useful overview of price caps
including regulatory experience in the UK.
19
the merged firm’s ability to exploit market power but still leave it with the
ability to raise prices and limit output after the merger.
VII. Conclusion
Figure 1
P1
P0 F B C
A D
c
H E G
c-∆
q1 P(Q)
q0
Q1 Q0 Q
26
References
McAfee, R.P. and M.A. Williams (1992), “Horizontal Mergers and Antitrust
Policy,” Journal of Industrial Economics, 40 (2), pp.181-187.
Salant, S.W., S. Switzer and R.J. Reynolds (1983), “Losses from Horizontal
Merger: the Effects of an Exogenous Change in Industry Structure on
Cournot-Nash Equilibrium,” Quarterly Journal of Economics, 98 (1),
pp.185-199.