Internal Versus External Growth in Industries With Scale

Manuscript Click here to access/download;Manuscript;2018.09.19 MNSW.
tex
Copyright The University of Chicago 2019. Preprint (not copyedited or formatted).

Please use DOI when citing or quoting. DOI: 10.1086/704069
Internal versus External Growth in Industries with Scale

Economies: A Computational Model of Optimal Merger Policy∗
Ben Mermelstein Volker Nocke
Bates White University of Mannheim, NBER, and CEPR
Mark A. Satterthwaite Michael D. Whinston
Northwestern University M.I.T. and NBER
October 3, 2018
Abstract
We study merger policy in a dynamic computational model where firms can reduce
costs through investment or through mergers. Firms invest or propose mergers according
to the profitability of these strategies. An antitrust authority can block mergers at some
cost. We examine the optimal policy for an antitrust authority who cannot commit to its
future policy and approves mergers as they are proposed. We find that the optimal policy
can differ substantially from a policy based on static welfare. In general, antitrust policy
can greatly affect firms’ investment behavior, and firms’ investment behavior can greatly
affect the optimal antitrust policy.
∗
We thank the editor and referees, John Asker, Dennis Carlton, Alan Collard-Wexler, Uli Doraszelski, Ariel
Pakes and Patrick Rey for their comments, as well as seminar audiences at Bates White, Chicago, ECARES,
Harvard, NYU, Penn, Stanford, Tilburg, Toulouse, UCLA, the 2012 Northwestern Searle Antitrust conference,
the 2012 SciencesPo Workshop on Dynamic Models in IO, the 2013 CRESSE Conference, the 2013 NBER IO
Summer Institute, the 2013 EARIE conference, the 2013 SFB-TR15 Meeting, and the 2017 Paris ICT Confer-
ence. Mermelstein and Satterthwaite thank the General Motors Research Center for Strategy and Management
at Northwestern University’s Kellogg School of Management for financial support. Nocke gratefully acknowl-
edges financial support from the European Research Council (ERC Starting Grant 313623) and the German
Research Foundation (DFG) through CRC TR 224. Whinston thanks the National Science Foundation and
the Toulouse Network for Information Technology for financial support. The views expressed in this article are
solely those of the authors and do not necessarily reflect the opinions of Bates White or its clients. We thank
Ruozhou Yan for research assistance.
Journal of Political Economy

Downloaded from www.journals.uchicago.edu by Australian Catholic University on 04/21/19. For personal use only.
1 Introduction
Most analyses of optimal horizontal merger policy in the economics literature are static and
focus on the short-run price effects of mergers.1 But many real-world mergers occur in markets
in which dynamic issues are a central feature of competition among firms. As a result, antitrust
authorities are regularly confronted with the need to consider likely future effects of a merger
on an industry’s evolution when deciding whether to approve the merger.2
In this paper, we study optimal merger policy in a dynamic setting in which investment
plays a central role, as the presence of economies of scale presents firms with the opportunity
to lower their average and marginal costs through capital accumulation. These scale economies
are also the source of merger-related efficiencies, as a combination of firms’ capital through
merger lowers average and marginal costs. In such a setting, an antitrust authority’s merger
approval decisions must weigh any increases in market power against the changes in productive
efficiency caused by a merger. Approval of the merger will lower production costs immediately
by increasing the scale of the merged firm (“external growth”), which may mean that there is
an immediate increase in welfare. However, if the merger is rejected, the firms that wished to
merge might instead invest individually to gain scale and lower their costs over time (“internal
growth”). Moreover, rivals’ investments may change as a result of the merger, altering their
efficiency and pricing. Finally, while approval or disapproval of a particular merger may affect
welfare, merger policy can alter firms’ pre-merger investment behaviors, since those behaviors
may be affected by the likelihood that mergers will be approved in the future.
As one example, consider the 2011 attempted merger between AT&T and T-Mobile USA.3
The merger would have combined the network infrastructure of the two firms. Proponents of
the merger argued that this combination would greatly improve both firms’ service, creating
a more potent rival to Verizon. Opponents countered that the merger would increase market
power, and that absent the merger the two firms would each have incentives to independently
increase their networks. Thus, the Federal Communications Commission and Department of
Justice faced the question of whether the merger would result in a sufficient efficiency im-
provement (which in this case would be realized on the demand side through enhanced service
quality) to offset the increase in market power, taking into account not only any immediate
service improvement but also any induced change in the merging firms’ future investments.
Moreover, the merger would also likely change the investments of the merging firms’ rivals,
Verizon and Sprint, and possibly potential entrants. Lastly, the investments of firms like
T-Mobile could in the future be affected by their expectations of whether mergers such as
this would be approved. Similar issues are present in the currently proposed Sprint/T-Mobile
USA merger, where the central question is whether the merger would enhance competiton by
1
For example, see the classic papers by Williamson (1968) and Farrell and Shapiro (1990).
2
The U.S. Horizontal Merger Guidelines, for example, devote considerable attention to discussions of entry,
investment, and innovation.
3
See Pittman and Li (2013) and DeGraba and Rosston (2014), and the references therein.
1
creating a stronger third firm.4

Our model builds on the computational literature on industry dynamics, pioneered by Pakes
and McGuire (1994) and Ericson and Pakes (1995), but with some important differences that
make the model more attractive for studying mergers. In that literature, each firm can add one
unit of capital in each period, so a merger reduces the investment opportunities both for the
merging firms and for the economy. We modify the investment technology to make it merger
neutral, so that mergers do not change the investment opportunities that are available in the
market. Our investment technology also allows for significantly richer investment dynamics,
as firms can increase their capital stocks by multiple units, and new entrants can choose
endogenously how many units of capital to build when entering.
In addition, we introduce the possibility for firms to merge, as well as an antitrust author-
ity who can block proposed mergers. The decision to propose a merger is endogenous and
determined through a bargaining process. We model the authority as a player who cannot
commit to its future policy.5,6 Perhaps surprisingly, issues of policymakers’ time consistency
have received scant attention in the antitrust literature. We consider both maximization of
discounted expected consumer surplus (“consumer value”) and discounted expected aggregate
surplus (“aggregate value”) as possible objectives of the authority, and refer to the policy that
emerges as a Markov perfect policy.
We begin in Section 2 by describing our model. In each period, firms first bargain over
merger proposals. If a merger is proposed, the authority decides whether to allow it and, if
so, a new entrant arrives with no capital. Then, the incumbent firms compete in a Cournot
fashion. Finally, firms – including any new entrants – decide on capital investment.
In Section 3 we study duopoly markets. A significant challenge in studying optimal
merger policy is the lack of a well-accepted canonical model of bargaining in the presence of
externalities. While a relatively small share of markets are duopoly markets, and mergers to
monopoly are rarely proposed and approved, a significant advantage of examining the behavior
of our model in such settings is that the merger bargaining process we adopt for these settings
– bilateral Nash bargaining – is well-accepted and easily understood. Throughout most of
the section we focus on a single market parameterization so that we can describe equilibrium
firm behavior and its interaction with antitrust policy in detail; we discuss afterwards how
4
See, for example, “T-Mobile and Sprint: How Fewer Competitors Could Increase Competition,” New York
Times, July 30, 2018. In the EU, similar examples include the Hutchison and Orange Austria, the Hutchison
and Telefonica Ireland, the Telefonica Germany and EPlus, the TeliaSonera and Telenor, the Hutchison 3G and
Telefonica UK, and the H3G Italy and Wind merger cases.
5
Despite the existence of merger guidelines in many jurisdictions, antitrust authorities may choose not
to follow them when confronted with particular mergers. This was widely viewed to be the case in the years
following the release of the 1992 DOJ/FTC Merger Guidelines in the United States. For example, in announcing
the release of the 2010 DOJ/FTC Horizontal Merger Guidelines, then Assistant Attorney General Christine
Varney commented that “The revised guidelines better reflect the agencies’ actual practices.” (August 19, 2010
press release)
6
In the Online Appendix, we also study the optimal commmitment policy.
2
outcomes vary across a wide range of parameters. When no mergers are allowed, this market
spends most of the time in duopoly states and a merger would often increase current-period
aggregate surplus.
Our analysis first examines how firm behavior responds when all mergers are allowed or
when the antitrust authority implements a static policy that considers only welfare effects in
the current period. Not surprisingly, the steady state when all mergers are allowed involves a
monopoly or near-monopoly market structure much more often than when mergers are pro-
hibited. It also involves a lower average level of capital. This arises because total investment is
lower in monopoly and near-monopoly states. Investment behavior also changes when mergers
are allowed. Particularly striking is significantly greater investment by small firms in states
in which one firm is very dominant, a form of “entry for buyout” [Rasmusen (1988)]. Their
investments, made in anticipation of being acquired, are done at high cost and substitute for
lower cost investment by larger incumbents, dissipating a great deal of both industry profit and
aggregate surplus. Because in this market a merger would increase current-period aggregate
surplus in many states, firm behavior with a static aggregate surplus-based policy is essentially
equivalent to when all mergers are allowed. In contrast, a static consumer surplus-based policy
allows almost no mergers.
We then endogenize merger policy by identifying the Markov perfect policy. With a con-
sumer value objective, the Markov perfect policy basically allows no mergers, just as with the
static consumer surplus criterion. With an aggregate value objective, however, the Markov
perfect policy allows many fewer mergers than the optimal aggregate surplus-based static
policy. The reason is that the inefficient entry for buyout behavior greatly reduces the an-
titrust authority’s desire to approve mergers. The resulting policy significantly reduces the
frequency of monopoly and near-monopoly states, and increases both consumer and aggregate
value compared to allowing all mergers or following the static aggregate surplus-based policy.
Strikingly, it nevertheless results in a lower steady state aggregate value than prohibiting all
mergers, or equivalently, having an antitrust authority who seeks to maximize either consumer
value or current-period consumer surplus.
In Section 4, we turn our attention to triopoly markets using a variant of the bargain-
ing model of Burguet and Caminal (2015), a model of merger bargaining in the presence of
externalities with a number of desirable features. We first confirm that our earlier duopoly
results in Section 3 are robust to the possibility of entry of a third firm. We then consider
two ways of increasing demand from that considered in Section 3 that lead to markets that
spend much of the time as a triopoly when mergers are not allowed. Interestingly, the effects
of allowing mergers differ markedly between these two markets. In one market this results in
a merger to duopoly, followed by a stable duopoly that almost never attracts entry. In the
other, entry of a third firm occurs with regularity, followed by a merger of the entrant with
the smaller of the two incumbents, and a repeat of this cycle. Because mergers confer large
positive externalities on non-merging firms, allowing mergers creates strong investment incen-
3
tives in this second market for the duopolist incumbents as they seek to position themselves to
be the beneficiary of these externalities. The strong investment by incumbents also reduces
the entry for buyout incentives of potential entrants. With the harm arising from entry for
buyout either not present or reduced, the aggregate value-based Markov perfect policy is quite
permissive in both of these markets; for example, it always allows a merger by symmetric firms
who are smaller than their non-merging rival. Overall, compared to not allowing mergers this
Markov perfect policy lowers steady state aggregate value in the first market, but leads to
little change in aggregate value (despite a strong reduction in consumer value) in the second.
Section 5 concludes and summarizes our insights.
The paper is related to several strands of literature. The first is theoretical work on dy-
namic merger policy, most notably Nocke and Whinston (2010).7 In that paper, the dynamics
arise from merger opportunities occurring stochastically over time; there is no investment.
Another relevant theoretical literature studies the welfare effects of mergers in static models
with investment [Motta and Tarantino (2018), Bourreau, Jullien, and Lefoulli (2018), Federico,
Langus, and Valletti (2018)].
A second related strand of literature examines mergers in computational dynamic models
of industry equilibrium with investment.8 The closest paper to ours is Gowrisankaran (1999)
who introduces an endogenous merger bargaining game into the Pakes-McGuire/Ericson-Pakes
framework and examines industry evolution when firms can choose whether, when, and with
whom to merge. The adopted investment technology implies that a merger significantly reduces
the merging firms’ abilities to make future investments, making it unattractive for modeling
mergers. There are no scale economies; instead, merger-related efficiencies are assumed to be
one-time random benefits. Finally, the model includes a complicated bargaining process whose
general properties are unknown; when specialized to the case of two firms, however, it gives the
smaller firm the right to make a take-it-or-leave-it offer to the larger firm.9 Hollenbeck (2017)
builds on the approach in our paper, but examines instead settings with investment in quality
in an industry with differentiated product price competition. Unlike our paper, he simply
compares the outcomes arising if all mergers are allowed to those if a static consumer surplus-
based policy is instead followed. Finally, Jerziorski (2015) studies the radio broadcasting
industry. He specifies a dynamic model of endogenous mergers with a particular random
proposer bargaining process and endogenous station format repositioning investments, and
conducts an empirical exercise to estimate his model’s parameters. He then simulates the
7
Nilssen and Sorgard (1998), Motta and Vasconcelos (2005), and Matushima (2001) analyze static models
of competition where two mergers between two non-overlapping pairs of firms can take place sequentially.
8
Berry and Pakes (1993), Cheong and Judd (2000), and Benkard, Bodoh-Creed, and Lazarev (2010) examine
the effects of one-time mergers on industry evolution.
9
In unpublished work, Gowrisankaran (1997) introduces antitrust policy into the Gowrisankaran (1999)
model. Specifically, he examines the effect of commitments to Herfindahl-based policies that block mergers if
they result in a Herfindahl index above some maximum threshold and finds little effect of varying the threshold
on welfare.
4
effects of commitments to four specific counterfactual merger policies. He does not examine
optimal policy.
Given our focus on only duopoly and triopoly markets, motivated by the plethora of possible
approaches to bargaining with externalities with more than two firms, we regard the paper as
only a first step in studying optimal merger policy in industries where investment is a central
concern. Our results show how optimal policy in dynamic settings with investment can differ
in significant ways from what would be statically optimal and provide insights into the factors
that affect optimal merger policy in such environments.
2 The Model
We study a dynamic industry model in which a set of n ≥ 2 firms, I ≡ {1, ..., n}, may invest in
capacity, or alternatively merge, to increase their capital stocks and harness scale economies.
The model follows in broad outline Pakes and McGuire (1994) and Ericson and Pakes (1995),
but with some important differences in its investment technology, as well as in the introduction
of mergers and merger policy. We focus on symmetric Markov perfect equilibria of our model.
Within each period, the sequence of events is as shown in Figure 1: The firms begin each
period observing each others’ capital stocks K ≡ (K1 , ..., Kn ), the model’s state variable,
which affect the firms’ production costs. The firms then bargain over which merger, if any, to
propose to the antitrust authority. If no merger agreement is reached, the firms proceed to the
Cournot competition phase with their current capital levels. If a merger agreement is reached,
the merger partners propose their merger to the antitrust authority, who may then decide to
block it. If the merger is allowed, the firms combine their capital, and a new entrant appears
with no initial capital stock. Following these merger bargaining, merger decision, and entry
phases, the active firms engage in Cournot competition given their capital stocks, and earn
profits on their sales. Following this Cournot competition stage, the firms choose their capital
investments. Finally, depreciation may make obsolete some of a firm’s capital. The resulting
capital levels after depreciation become the starting values in the next period.
We begin by describing the demand and production costs the firms face (the latter as a
function of their capital stocks), and the static Cournot competition that occurs in each period.
We then detail how merger bargaining works, the merger policy of the antitrust authority, and
the investment, entry, and depreciation processes. The Online Appendix contains a more
formal description of our model and computational methods.
2.1 Static Demand, Costs, and Competition

In each period, active firms produce a homogeneous good in a market in which the demand
function is Q(p) = B(A − p). The production technology, which requires capital K and labor
θ
L, is described by the production function K β L(1−β) , where β ∈ (0, 1) is the capital share
5
Figure 1: Sequence of events in a single period
and θ > 1 the scale economy parameter. Normalizing the price of labor to be 1, for a fixed
level of capital, this production function gives rise to the short-run cost function
Q1/[(1−β)θ]
C(Q|K) =
K β/(1−β)
with marginal cost (1/[(1−β)θ])−1

1 Q
CQ (Q|K) = .
(1 − β)θ K β/(1−β)
With this technology, a merger that combines the capital of two identical firms reduces
both average and marginal cost if their joint output remains unchanged. This effect will be
the source of merger-related efficiencies in our model. Letting R measure the extent of this
cost reduction, we have

CQ (2Q|2K) C(2Q|2K)/2Q 1
( 1−θ
θ )
R≡ = =2 1−β
.
CQ (Q|K) C(Q|K)/Q
Note in particular that the marginal cost reduction depends on the scale economy parameter
θ and capital share β, but is independent of the output level (and hence demand). In our
computations we will focus on a case in which β = 1/3 and θ = 1.1.10 Given these values, R
is 0.91; that is, a merger of two equal-sized firms results in a 9% efficiency gain.
In each period, active firms engage in Cournot competition given their capital stocks (a
firm with no capital produces nothing), resulting in profit π(Ki , K−i ) for a firm with capital
stock Ki when the vector of its rivals’ capital stocks is K−i ≡ (K1 , ..., Ki−1 , Ki+1 , ..., Kn ).11
10
A capital coefficient of 1/3 is routinely assumed in the macroeconomic literature; see, for example, Jones
(2005). The scale economy parameter of 1.1 leads to plausible efficiency gains and is selected so that mergers
are statically aggregate surplus increasing in a substantial proportion of industry states.
11
A firm’s short-run cost function is strictly convex if (1 − β)θ < 1, in which case there is a unique Cournot
equilibrium if the demand function is weakly concave. In our analysis, these conditions are satisfied.
6
2.2 Mergers and Bargaining

A merger Mij , which involves the combining of the merging firms’ capital, is feasible between
any pair ij ∈ J ≡ {ij|i, j ∈ I, i 6= j} of firms. Proposing merger Mij for approval to the
antitrust authority involves a cost φij , which is drawn i.i.d. (both across pairs of firms as well
as over time) from a continuous distribution function Φ with support [φ, φ]. We introduce
these proposal costs primarily for technical reasons to ensure existence of (pure strategy)
equilibrium; in the real world, they may represent legal costs.12 As shown in Figure 1, the
antitrust authority may block a merger proposal, while if the firms’ merger is allowed a new
entrant appears in the market with zero capital.13
Let V (Ki , K−i ) denote the interim expected net present value (“continuation value”) of
firm i when it has Ki units of capital and the vector of its rivals’ capital levels is K−i at the
start of the Cournot competition stage (see Figure 1). If the capital stocks prior to the merger
stage are K, then the bilateral value gain from merging, gross of any proposal cost, is
∆ij (K) ≡ V (Ki + Kj , K−ij , 0) − (V (Ki , K−i ) + V (Kj , K−j )), (1)
where K−ij is the vector of capital levels of firms other than i and j. The first term in (1) is the
joint interim value if the merger takes place; the second term is the sum of the disagreement
payoffs.
The probability that merger Mij gets approved when proposed is denoted aij (K). The
expected change in the merger partners’ joint value from proposing merger Mij can thus be
written as
Sij (K, φij ) ≡ aij (K)∆ij (K) − φij .
The expected externality of the proposal of merger Mij on an outsider k 6= i, j is given by
Xkij (K) ≡ aij (K) V (Kk , Ki + Kj , K−ijk , 0) − V (Kk , K−k ) ,

where K−ijk is the vector of capital levels of firms other than i, j and k.
In each period, at most one merger can be proposed for approval to the antitrust authority.
Firms bargain under complete information about which merger to propose (if any) and how
to split the surplus, given the vector of merger approval probabilities, (aij (K))ij∈J , the vector
12
See Doraszelski and Satterthwaite (2010) for a discussion of introducing random private payoffs as a means
of ensuring existence.
13
The immediate arrival of a new entrant following a merger can also be thought of as being the result of a
structural remedy imposed by the antitrust authority, which involves the transfer of know how to a firm outside
the industry, permitting this firm to become a new entrant. The 2011 U.S. Department of Justice Antitrust
Division Policy Guide to Merger Remedies states: “Structural remedies generally will involve [...] requiring
that the merged firm create new competitors through the sale or licensing of intellectual property rights.” For
the case n = 2, we have also analyzed in the Online Appendix the case in which the probability of entry is less
than one. We also analyze there a case in which only the incumbent manager-owners possess the knowledge of
how to operate a firm in this industry so that the new entrant is one of these owners and obtain similar results.
7
of continuation values in the absence of a merger, (V (Ki , K−i ))i∈I , the vector of surpluses,
ij
(Sij (K, φij ))ij∈J , and the matrix of externalities, Xk (K) . For our purposes,
ij∈J ,k∈I,k6=i,j
the outcome in state K of a generic bargaining process can be summarized by the vectors
(ϑij (K))ij∈J and (V (Ki , K−i ))i∈I , where ϑij (K) is the probability that merger Mij gets pro-
posed and approved in state K and V (Ki , K−i ) is the beginning-of-period value of firm i (prior
to the realization of proposal costs).
In Section 3, we focus on the case of two firms (n = 2) and assume Nash bargaining. In
that case, merger M12 gets proposed if and only if the bilateral surplus S12 (K, φ12 ) is positive.
The probability that the merger occurs in state K is therefore given by
ϑ12 (K) = a12 (K)ψ12 (K),
where ψ12 (K) ≡ Φ(a12 (K)∆12 (K)) is the probability of the merger being proposed. Firm i’s
beginning-of-period value in state K includes its possible share of any merger surplus, and
equals
1 φ +
Z
V (Ki , K−i ) = V (Ki ; K−i ) + S (K, φ12 )dΦ(φ12 ),
2 φ 12
+
where S12 (K, φ12 ) ≡ max{0, S12 (K, φ12 )}. In Section 4, we explore situations with three
firms using an adaptation of the bargaining process of Burguet and Caminal (2015), which we
describe there.
2.3 Merger Policy

The antitrust authority has the ability to block mergers. Blocking a proposed merger Mij
involves a cost bij ∈ [b, b] drawn each period in an i.i.d. fashion from a distribution H. We
introduce these blocking costs primarily for technical reasons to ensure existence of (pure
strategy) equilibrium; in the real world, they may represent the opportunity costs of an in-
depth merger investigation (which is required for blocking a merger but not for approving a
merger) or possible litigation costs.
In our analysis, we focus on a situation in which the antitrust authority cannot commit
to its policy.14 In that case, in any state K, it will decide whether to block a merger by
comparing the increase in its welfare criterion from blocking to its blocking cost realization
bij . As welfare criteria, we will consider both consumer value (CV) and aggregate value (AV),
the expected net present values of consumer surplus and aggregate surplus, respectively. A
Markovian strategy for the antitrust authority is a state-contingent and history-independent
threshold bbij (K) describing the highest blocking cost at which it will block merger Mij in
a given state K. Equivalently, this can be translated into a merger acceptance probability
14
In the Online Appendix, we also consider the case of an antitrust authority who can commit to a deter-
ministic policy (aij (·))ij∈J that specifies whether a proposed merger would be approved (aij (K) = 1) or not
(aij (K) = 0) in each state K.
8
aij (K) ∈ [0, 1]. As we previously noted, we call the equilibrium policy that emerges a Markov
perfect policy (MPP). In practice, an antitrust authority may well lack an ability to commit
to its future approval policy. While the Department of Justice and Federal Trade Commission
in the U.S. periodically issue Horizontal Merger Guidelines, which may partially commit these
agencies, over time their actual policy often comes to deviate substantially from the Guidelines’
prescriptions.
2.4 Investment, Entry, and Depreciation

In Pakes and McGuire (1994) and Ericson and Pakes (1995) a firm chooses in each period
how much money to invest, with the probability of successfully adding one unit of capital
increasing in the investment level. We depart from this technology because in a model of
mergers it would impose a significant inefficiency on mergers. In particular, it would restrict
the merged firm to adding one unit of capital each period while, if they had not merged, the
firms could have each added one unit of capital for a total addition of two units.15 Instead, we
specify an investment technology that is merger neutral at a market level. By that we mean
that a planner who controlled the firms and wanted to achieve at least cost any fixed increase
in the market’s aggregate capital stock would be indifferent about whether the firms merge.
With this assumption we isolate the market-level technological effects of mergers fully in the
scale economies of the production function. These technological effects on production costs,
combined with firms’ behavioral responses in investment, will determine the efficiency benefits
of mergers in our model.
We imagine that there are two ways that a firm can invest. The first is capital augmentation:
each unit j of capital that a firm owns can be doubled at some cost cj ∈ [c, c] drawn from
a distribution F . The draws for different units of capital are independent and identically
distributed. Thus, for a firm that has K units of capital, there are K cost draws. Given these
draws, if the firm decides to augment m ≤ K units of capital it will do so for the capital units
with the cheapest cost draws. Note that capital augmentation is completely merger neutral:
when two firms merge, collective investment possibilities do not change.
The second is greenfield investment: a firm can build as many capital units as it wants
at a cost cg ∈ [c, cg ] drawn from a distribution G. Greenfield investment allows a firm whose
capital stock is zero to invest, albeit at a cost that exceeds that of capital augmentation. We
also choose the range of greenfield costs [c, cg ] to be small so that this investment technology is
approximately merger neutral. (It would be fully merger neutral if cg = c; in our computations
we introduce uncertain greenfield investment costs to ensure existence of equilibrium.)
As we noted earlier, our model allows for entry. In contrast to Pakes and McGuire (1994)
and Ericson and Pakes (1995), we endow an entrant with the same investment technology as
15
Alternatively, if the merged firm kept both investment processes we would need to keep track, as a separate
state variable, of how many investment processes a firm possesses, which has no natural bound.
9
incumbents. The entrant, however, starts with no capital, so it must use greenfield investment.
Note that with our assumptions investment opportunities will be (approximately) merger
neutral at the market level. The assumptions also imply the following two properties of
investment costs at the firm level:
1. Holding the firm’s current capital stock K fixed, the expected per unit cost of adding
∆K units of capital is increasing in the investment size ∆K.16
2. Holding the firm’s investment size ∆K fixed, the expected investment cost is decreasing
in the size of its current capital stock K.17
Both properties are consistent with the large literature on capital adjustment costs that
Abel (1979) and Hayashi (1982) initiated.18 The second property is also in line with the large
(theoretical and empirical) literature on entry in industrial organization, where it is commonly
assumed that potential entrants have to incur a setup cost before entering, implying that new
entrants have to incur higher costs than incumbents if they want to add the same amount of
productive capital.19,20
Put together, the capital augmentation and greenfield investment processes allow for sig-
nificantly richer investment dynamics than in the typical dynamic industry model. Firms can
expand their capital by multiple units at a time through either investment method, and new
entrants can decide endogenously how far to jump up in their capital stock.
Capital also depreciates: in each period each unit of capital has a probability d > 0 of
becoming worthless (including for any future capital augmentation). Depreciation realizations
are independent across units of capital. This depreciation process is also merger neutral.21
Finally, the firms discount the future according to discount factor δ < 1.
16
For a firm with no capital, the unit cost of adding ∆K units of capital is constant as such a firm has only
access to the greenfield technology.
17
This implies that investment opportunities – while merger neutral at the market level – are not merger
neutral at the firm level.
18
Most of the literature on capital adjustment costs assumes that adjustment costs are a convex function of
the proportional change in the firm’s capital stock. With that formulation, the cost of a given sized increase
in capital is strictly decreasing in firm size. More recent work, such as Cooper and Haltiwanger (2006), has
introduced non-convex components of adjustment costs. But even in those models small firms have very large
investment costs.
19
In the Online Appendix, we also examine the effects of requiring a minimum scale of greenfield investment.
20
New entrants may also face higher financing costs than established firms. Indeed, there are many empirical
studies finding a positive effect of cash flow on investment, pointing to credit constraints; see the influential
paper by Fazzari, Hubbard and Petersen (1988) and the survey by Bond and van Reenen (2007). However,
these findings cannot easily be mapped into our model as cash flow is likely to be related to retained earnings
– which in turn depend on own past capital stocks as well as the rivals’ past capital stocks.
21
This is in contrast to Ericson and Pakes (1995), Gowrisankaran (1997, 1999) and many other papers in
the computational IO literature. There, depreciation is modeled as a perfectly correlated industry-wide shock,
following which each firm loses one unit of capital, independently of its size. That is, these papers assume that
the expected depreciation rate is decreasing with firm size.
10
In our computations firms will be restricted to an integer number of possible capital levels,
with the maximal capital level K chosen to be non-binding. We define S ≡ {0, 1, 2, ..., K} to
be the admissible values of Ki and S n = S × ... × S to be the state space.
3 Merger Policy in Duopoly Markets
In this section we study duopoly markets. While a relatively small share of markets are duopoly
markets, and mergers to monopoly are rarely proposed and approved, a significant advantage of
examining the behavior of our model in such settings is that the merger bargaining process we
adopt for these settings – bilateral Nash bargaining – is well-accepted and easily understood.
Throughout most of this section we focus on a single market parameterization so that we can
describe equilibrium firm behavior and its interaction with antitrust policy in detail; we discuss
afterwards how outcomes vary across a wide range of parameters. We begin in Section 3.1 by
describing the parameters of the market we focus on. In Section 3.2 we examine how firms’
behaviors and market performance depends on merger policy, and in Section 3.3 we study the
Markov perfect antitrust policy and its positive and normative features. In Section 3.4 we
turn to outcomes for other parameters.
3.1 Parameterization
In most of this section, we describe the results for a market in which demand is Q (p) = B(A−p)
with (A = 3, B = 26), while firms’ production functions are Cobb-Douglas with capital share
parameter β = 1/3 and scale parameter θ = 1.1 (recall that a merger between two equal-sized
firms then lowers marginal and average costs by 9 percent at fixed outputs).
Table 1 gives a sense of this market’s static properties with its strong economies of scale
and linear demand. It shows the static Cournot equilibrium outcomes for three different states:
(1, 0), (10, 0), and (5, 5). The comparison between the (1, 0) and (10, 0) monopoly states shows
the effects of the scale economies on marginal cost. It also shows for state (1, 0) the effect
of linear demand when price is high and quantity small: demand is quite elastic causing a
small price-cost markup. Aggregate surplus in the monopoly (10, 0) state is almost identical
to that in the duopoly (5, 5) state because the strong scale economies almost exactly offset
the inefficient monopoly pricing. The distribution of the surplus, however, tilts strongly away
from consumers and towards producers.
Turning to investment costs, the capital augmentation cost for a given unit of capital is
independently drawn from a uniform distribution on [3, 6], while the greenfield investment
cost cg is drawn from a uniform distribution on [6, 7].22 Firms’ discount factor is δ = 0.8,
22
The large spread of the capital augmentation cost distribution reflects empirical results showing large
variation in firms’ costs within an industry. See, for example, Bernard, Eaton, Jensen, and Kortum (2003) and
Syverson (2004). The Online Appendix includes an extension with a smaller variation in firms’ costs.
11
corresponding to a period length of about 5 years. We chose this to reflect the time to build
new capital. Each unit of capital depreciates independently with probability d = 0.2 per
period. We take the state space to be {0, 1, . . . , 20}2 , so each active firm can accumulate up
to 20 units of capital. In this market (and the ones considered in Section 3.4) firms almost
never end up outside of the quadrant {0, 1, . . . , 10}2 ; we allow for capital levels up to 20 so
that we can calculate values for mergers and avoid boundary effects. We assume that proposal
and blocking costs are uniformly distributed on [0,1].23
We focus on these parameter values to highlight the tension between the goals of achieving
cost reductions immediately through a merger, preventing increased exercise of market power,
and maintaining desirable investment behavior.
Finally, as noted at the end of Section 2, we assume that merger bargaining, which occurs
between the two active firms, is described by the bilateral Nash bargaining solution.
3.2 Investment and Merger Incentives under Fixed Merger Policies

In this section we examine the Markov perfect equilibrium for three types of fixed merger
policies: (i) the case in which mergers are prohibited—the “no-mergers-allowed” case, (ii) the
case in which firms are permitted to merge in any state in which it is profitable for them to do
so—the “all-mergers-allowed” case, and (iii) the case of “static” merger policy in which mergers
are blocked if and only if they would result in lower current period welfare. In the third case,
we consider both current period consumer surplus and aggregate surplus as possible welfare
measures. For each policy, we report its long-run steady state distribution over the state
space S 2 , the consumer, incumbent, entrant, and aggregate values it generates (the discounted
expected value of consumer, incumbent, entrant, and aggregate surpluses, respectively), the
investment incentives it creates, and the frequency of mergers it induces.
3.2.1 Equilibria with No-Mergers-Allowed
We begin by examining the equilibrium when no mergers are allowed.24 Figure 2(a) shows the
beginning-of-period steady state equilibrium distribution under a no-mergers-allowed policy.
(The other panels of Figure 2 show the steady state distributions for other cases discussed
below.) The first column of Table 2 lists some measures of the no-mergers equilibrium.25 As
can be seen, under the no-mergers policy the industry spends most of its time in duopoly states
23
We know of no empirical literature on proposal and blocking costs. We chose these wide spreads to help
ensure convergence of the numerical algorithm. See Doraszelski and Satterthwaite (2010).
24
We have assembled the data that we have generated into large Excel workbooks that each contain for each
equilibrium, first, a detailed description of the equilibrium strategies of the firms and, for Markov perfect merger
policies, of the antitrust authority and, second, a full set of performance statistics. These workbooks are posted
on the web as part of our Online Appendix. They enable the reader to explore our results much as we have
explored them.
25
For comparison, in the first-best solution (with price equal to marginal cost and aggregate value-maximizing
investment), the aggregate value is 164.7 and the average total capital level is 10.6.
12
in which both firms are active, but also spends roughly 18 percent of the time in monopoly
states. If the industry finds itself in a monopoly state, it can stay there a long time. For
example, Figure 3(a) shows the one-period transition probabilities starting from state (5, 0);
it illustrates the weak entry behavior that allows this monopoly persistence – in fact, starting
in state (5, 0), the probability that the industry is a monopoly five periods later is 0.84.
There are two cost-based reasons why it is so hard for an entrant starting in state (5, 0)
to catch up. First, the entrant pays much more per unit of capital purchased: the large firm
can add a unit of capital using the lowest of its five cost draws from the uniform distribution
on [3, 6], whereas the entrant draws from the uniform distribution on [6, 7]. Second, the large
firm’s scale economies gives it a marginal cost of 1.70 when setting a monopoly price of 2.35. If
the potential entrant should enter with two units of capital, then at state (5, 2) the dominant
firm sells quantity 14.6 at price of 2.18 with marginal cost 1.62. The entering firm sells 6.7
units with marginal cost 1.92. Profits are 14.5 and 5.1, respectively.
3.2.2 Equilibria with All-Mergers-Allowed
Under an all-mergers-allowed policy equilibrium is quite different. Figure 2(b) shows both the
beginning-of-period steady state equilibrium distribution under an all-mergers-allowed policy,
as well as the probability that a merger actually happens in each state. Shading shows states
in which mergers occur with a darker shade representing a higher probability of a merger
happening; cells in which mergers never occur are unshaded: for example, a merger happens
with probability 1 in state (3, 3), with probability zero in state (2, 2), and with probability
0.59 in state (2, 3). Observe that firms do not always merge in non-monopoly states. The
reason is that if both firms’ capital stocks are low, then merging attracts a new entrant who
dissipates the merger’s gains.
The second column of Table 2 shows the properties of the all-mergers-allowed equilibrium.
Mergers happen 37.7% of the time, which results in the market being in a monopoly state (at
the time of Cournot competition) 86.0% of the time, and in a near-monopoly 99.1% of the
time. As a result of allowing mergers, average output falls from 22.2 to 19.2, while the average
price rises from 2.15 to 2.26. Average total capital falls from 8.0 to 7.0. Not surprisingly, the
change in policy leads to substantial negative changes in consumer value, which falls from
48.1 to 35.8. More surprisingly, average incumbent value falls even though the firms are
now allowed to merge whenever they want. This is despite firms’ success in raising price,
reducing quantity, and limiting total capital. Even once one accounts for future entrants’
value, producer value (the sum of incumbent and entrant values) barely rises. Combined with
the reduction in CV, aggregate value falls from 117.5 to 105.8.
To explore the reasons behind these results, consider first the reduction in total capital.
Allowing mergers does two things. First, it changes the states in which investments are taking
place by moving the market to monopoly and near-monopoly states. Second, firms’ investment
13
Figure 2: Beginning-of-period steady state distribution of various equilibria. The height of each
pin indicates the steady state probability of that state. The shading of each cell reflects the
probability of a merger happening in that state (darker grey represents a higher probability).
Figure 3: One-period transition probabilities from state (5,0).
14
policies change. Table 3 summarizes these effects. Holding investment behavior fixed, average
capital addition decreases when weighted by the all-mergers-allowed steady state rather than
the no-mergers steady state. However, holding the steady state weighting fixed, average capital
addition increases when investment behavior is that of the all-mergers-allowed equilibrium
rather than the no-mergers equilibrium. Together, these opposite effects reduce the average
capital addition moving from the no-mergers-allowed policy to all-mergers-allowed.
What drives the increased investment incentive? If a merger is certain to occur next period,
a firm i’s marginal return to investment is ∂V (Ki , Kj )/∂Ki + (1/2)∂∆ij (Ki , Kj )/∂Ki where
∂∆ij (Ki , Kj )/∂Ki is the marginal effect of Ki on the gain from merger as defined in (1).26
Each firm is in a state where ∂∆ij (Ki , Kj )/∂Ki is positive 97.5% of the time in the no-mergers
steady state and 100% of the time in the all-mergers-allowed steady state; the fact that a firm’s
gains from a merger are increasing in its capital stock tends to make allowing mergers increase
investment incentives.27
In the all-mergers-allowed equilibrium the steady state distribution is concentrated in
monopoly and near-monopoly states. The increased investment incentive is particularly large
and detrimental to producer value in such states. An entrant with zero capital frequently
invests in the hope of being bought out: there is a great deal of “entry for buyout” behavior
[Rasmusen (1988)].28 For example, Figure 3(b) shows the one-period transition probabilities
in state (5, 0) when all mergers are allowed, which can be compared to Figure 3(a) where
no-mergers are allowed. The probability that the entrant invests and has non-zero capital
after depreciation is 0.57 in the former case, versus 0.04 in the latter. Further, the probability
of a merger is 0.49 in the first period after the entrant invests when all mergers are allowed,
and 0.85 within two periods. Figure 3(b) also shows that the entrant’s increased investment
lowers the incentive of the incumbent to invest in state (5, 0).
This entry for buyout behavior reduces producer value as the entrants’ investments are
made at high cost and displace lower cost investments by the incumbent monopolist. Fig-
ure 4 illustrates the destructiveness of this behavior for producer value. It shows for each
state the change in the row firm’s (firm 1) beginning-of-period value that a switch from a
no-mergers-allowed to an all-mergers-allowed policy induces. In most states the row firm’s
value is enhanced but in monopoly states in which the monopolist has at least three units of
26
This abstracts away from the discrete nature of capital additions.
27
The change from the no-mergers-allowed to the all-mergers-allowed policy also changes the interim value
function V (·).
28
While we are unaware of any formal empirical studies that document the frequency of entry for buyout
behavior, Rasmusen (1988) gives a number of examples of entry for buyout in homogeneous goods industries. In
the literature on start-ups, acquisition is considered to be one of the primary ways of capturing a start-up’s value
[see, for example, Gans and Stern (2003)]. Although start-ups frequently introduce product innovations and
do not literally fit our homogeneous-goods model, we can reinterpret the capital in our model to be “knowledge
capital” and the resulting cost reductions enabled to be consumer value enhancements that increase the firm’s
profit.
15
Figure 4: Beginning-of-period value of the row firm (firm 1) in the all-mergers-allowed equi-
librium minus its value in the no-mergers equilibrium. Negative numbers are in parentheses.
capital, the monopolist’s value falls dramatically. This behavior is also highly detrimental for
aggregate value: In both the no-mergers and all-mergers-allowed equilibria, dominant firms
generally have insufficient incentives, while entrants have excessive incentives.29 The entry
for buyout phenomenon therefore causes a shift in investment away from the dominant firm,
whose incentives are already insufficient, toward the entrant, whose incentives are excessive.
3.2.3 Equilibria with Static Policies
We next consider optimal static merger policy, as in Williamson (1968) and Farrell and Shapiro
(1990). These policies block a merger if and only if it decreases welfare (either consumer
surplus or aggregate surplus, depending on the criterion) due to production and consumption
in the period the merger occurs.30
Mergers lower consumer surplus in all but state (1, 1), so the static consumer surplus-based
policy is essentially equivalent to allowing no mergers.
In contrast, Figure 5 shows that many mergers increase aggregate surplus. In general,
these tend to be states in which the total capital in the industry is not more than 10, though
in some asymmetric states with total capital above 10 there is also a gain.31 The gains in
aggregate surplus are generally smaller the larger is the total capital in the industry.32 An
29
The Online Appendix contains tables showing the difference between firms’ investment incentives and the
benefits to social welfare from investment.
30
Another possible benchmark is the second-best dynamic problem where the planner controls firms’ merger
decisions as well as their investment decisions, but not their output decisions. This benchmark is analyzed in
the Online Appendix. It turns out this second-best merger policy is very similar to the optimal static aggregate
surplus-based policy.
31
The only exception is state (5, 5) where the static gain in aggregate surplus is approximately zero.
32
To understand this result, observe that the change in aggregate surplus from a merger in a symmetric state
is approximately
∆Q ∆Q ∆ACM
Q (P − M C) − 1 + ACM ,
Q Q ACM
where (P − M C) is the premerger price-cost margin, ACM is the average cost if no merger occurs but the
16
Figure 5: Static change in aggregate surplus from a merger in the (A = 3, B = 26) market.
Negative numbers are in parentheses.
increase in the asymmetry of capital positions, holding total capital fixed, has varying effects
on the static gains in aggregate surplus from a merger. This gain gets smaller with increased
asymmetry at low levels of total capital, but grows larger with increased asymmetry at greater
levels of total capital.
Figure 2(c) shows the beginning-of-period steady state equilibrium distribution under the
aggregate surplus-based static merger policy and Table 2 shows equilibrium performance statis-
tics under this policy. As can be seen in the figure and table, the outcome with the aggregate
surplus-based static policy is very close to the all-mergers-allowed outcome.
3.3 Equilibria with Markov Perfect Merger Policy

We now introduce an optimizing antitrust authority who cannot commit to its future policy,
determine its Markov perfect policy, and examine the outcome it induces. In this setting the
antitrust authority, like each of the firms, is a player in a dynamic stochastic game; Markov
perfection requires that in each state the policy survives the one-stage-deviation test.33
As with the static consumer surplus-based policy, the Markov perfect policy outcome when
the antitrust authority seeks to maximize consumer value (“CV”) is essentially equivalent to
the no-mergers-allowed outcome (see Table 2). For the rest of this section, we therefore focus
on an authority who seeks to maximize aggregate value (“AV”) .
For an antitrust authority following the AV criterion, neither the no-mergers-allowed nor
the all-mergers-allowed policy survive the one-stage-deviation test given the firm behavior they
output level changes to its post-merger level, and ∆ACM is the change in average cost at the post-merger
output level due to the combination of capital. At larger capital levels, (P − M C) and |∆Q/Q| are both
greater, (∆ACM /ACM ) is unchanged, and ACM is smaller, making the sign of the effect on aggregate surplus
more likely to be negative for an output-reducing merger. For example, (P − M C) is 0.32 at state (2, 2) and
0.45 at state (4, 4), (∆Q/Q) is −0.065 at (2, 2) and −0.126 at (4, 4), and ACM is 21% lower at (4, 4) than at
(2, 2).
33
In the Online Appendix we discuss as well the case in which the antitrust authority can commit to its future
policy.
17
induce: assuming future behavior following the no-mergers equilibrium, the antitrust authority
would allow many mergers in a one-stage-deviation; assuming future behavior following the
all-mergers-allowed equilibrium, the antitrust authority would allow very few mergers.
Figure 2(d)’s shading shows the probability a merger occurs in various states under the
Markov perfect policy. The policy differs markedly from all of the policies we have previously
considered. The authority approves a proposed merger with positive probability in near-
monopoly states in which min{K1 , K2 } = 1, as well as in states (2, 2), (3, 2), and (2, 3). Given
this policy, mergers are proposed with probability one in all of these states, except in state
(1, 1), where a merger is never proposed, and in states (2, 1), and (1, 2), where a merger is
proposed with less than full probability. This policy induces an even higher merger probability
following entry than the all-mergers-allowed policy: For example, the probability of a merger
is 0.69 in the first period after entry in state (5,0), compared to 0.49 in the all-mergers-allowed
equilibrium. Firms are more likely to merge in the first period under the Markov perfect policy
because if the entrant grows further they are unlikely to be allowed to merge in the second
period.
Figure 2(d) also shows the steady state distribution arising under the Markov perfect
policy, while Table 2 shows its performance statistics. The industry is in a monopoly state at
the Cournot competition stage 49.4% of the time, and in near-monopoly states 55.8% of the
time. Compared to the steady state induced when no mergers are allowed, the economy spends
much more time in such states. In addition, the average aggregate capital level is lower (7.7 vs.
8.0). The reason is the shift in the steady state distribution toward more asymmetric states,
in which investments are lower. However, because a new entrant and the incumbent are not
always allowed to merge, monopoly states are less frequent and average capital is greater than
under the all-mergers-allowed and static aggregate surplus-based policies.
The Markov perfect policy with the AV criterion is much better for consumers and ag-
gregate value than allowing all mergers or following the static aggregate surplus-based policy.
However, it results in a level of steady state AV that is about 3% lower than with the no-
mergers policy: AV is 113.6 compared to 117.5 when no mergers are allowed.34 Firms are only
slightly better off – harmed again by the entry for buyout behavior the merger policy induces
– while consumers are much worse off: CV is 43.3 (vs. 48.1) and producer value is 70.4 (vs.
69.4). Consumers are harmed both from the monopoly pricing and the reduction in capital.
Strikingly, observe that a commitment to maximizing CV or to the static consumer surplus-
based policy would be better here for aggregate value than the policy that results when the
antitrust authority seeks to maximize AV but cannot commit.35
34
The finding that the Markov perfect policy with the AV criterion performs worse than the no-mergers policy
but better than the all-mergers-allowed policy holds not only for the steady state averages of AV and CV but
also for a “new” industry: as shown in Table 2, at state (0,0) the AV (resp. CV) value of the Markov perfect
policy is 35.5 (25.6), that of the no-mergers policy 36.7 (30.3), while that of the all-mergers-allowed policy is
only 34.0 (23.9).
35
This conclusion is reminiscent of Lyons (2002), but arises for different reasons.
18
3.4 Results for Other Demand Parameters

Up to this point we have limited our discussion to a single market parameterization. While
this focus allowed us to discuss in detail the outcomes and strategies that arise in this case,
it naturally leaves open the question of how our results extend to other market conditions.
Here we examine the extent to which several of the features of the equilibria discussed extend
across a wider range of demand parameters.36
We first examine how the no-mergers-allowed and all-mergers-allowed equilibria differ.
Figure 6(a) reports on the difference in aggregate value between these two policies for linear
demand functions Q(p) = B(A − p), where A is the choke price and B is the market size
parameter (e.g., number of consumers). The figure depicts contour lines showing the demand
parameters at which the aggregate value difference, (AVN o − AVAll )/AVN o , achieves a given
percentage value. Also shown in the figure are three dots. The middle one is the (A = 3, B =
26) market that our discussion above focused on. The other two dots represent a “smaller”
and a “larger” market whose equilibria we discuss in greater detail in the Online Appendix,
paralleling our discussion above of the (A = 3, B = 26) market. In the figure, dashed lines
show markets that spend 5%, 20%, and 60% of the time in monopoly states when no mergers
are allowed. Market parameters to the Northeast in the figure are large markets with low levels
of monopoly, while markets to the Southwest are small markets with high monopoly levels.
As can be seen in the figure, aggregate value with no mergers allowed is greater than with all
mergers allowed provided that the market is large enough, with aggregate value approximately
equal for these two merger policies for markets in which the no-mergers-allowed equilibrium
spends about 70% of the time in monopoly states.
For the same range of demand parameters, Figure 6(b) shows the percentage difference in
entry probabilities in the no-mergers-allowed and all-mergers-allowed equilibria, [Pr(Entry)All −
Pr(Entry)N o ]/ Pr(Entry)All ].37 Consistent with the entry for buyout we observed earlier, the
level of entry is always weakly greater in the all-mergers-allowed equilibrium, although the
difference declines to zero in very large markets where the probability of entry rises to one
under either merger policy.
36
In the Online Appendix we also consider the effect of varying the production scale parameter θ. The results
show similar patterns to those we discuss here, with outcomes closely related to the percentage of time spent
in monopoly states when no mergers are allowed.
We also examine there the following modeling extensions: allowing the probability of entry following a merger
to be less than one; modifying our greenfield investment technology (used primarily by entrants) to require a
minimum scale of investment greater than one unit of capital; reducing the gap of investment costs faced by
incumbents and entrants; having bargaining power proportional to capital stocks; allowing a planner to control
investment behavior and merger decisions taking as given only Cournot competition; and assuming new entrants
are the owners of the firms purchased in mergers.
37
Pr(Entry)x is calculated by weighting the probability of entry in each monopoly state under merger policy
x by the probability of that state in the all-mergers-allowed equilibrium. In the Southeast region of the figure,
the no-mergers equilibrium has no entry in states that arise with positive probability in the all-mergers-allowed
equilibrium, leading the percentage difference in entry probabilities to be 100%.
19
Figure 6: (a) Contour lines of the percentage difference between the steady state aggregate
value of the no-mergers and all-mergers-allowed equilibria, (AVN o −AVAll )/AVN o . (b) Contour
lines of the percentage difference between the steady state aggregate value of the no-mergers
and all-mergers-allowed equilibria, (AVN o − AVAll )/AVN o .
Figure 7 focuses on the Markov perfect policy. Figure 7(a) shows the percentage dif-
ference in aggregate value between the Markov perfect policy and the no-mergers-allowed
equilibrium, (AVM P P − AVN o )/AVM P P . In small markets, the Markov perfect policy leads
to higher aggregate value than when no mergers are allowed. Similar to the comparison be-
tween the no-mergers and all-mergers-allowed policies, the no-mergers policy outperforms the
Markov perfect policy provided the market is large enough. However, for the largest mar-
kets in the Northeast corner, the Markov perfect policy leads to the same equilibrium as
the no-mergers policy because mergers are never consummated. Figure 7(b) shows the same
AV comparison but relative to the outcome with the static aggregate surplus-based policy,
(AVM P P − AVStatic )/AVM P P . The figure shows that the Markov perfect policy outperforms
the static aggregate surplus-based policy provided the market is large enough.
4 Merger Policy in Triopoly Markets

In this section, we extend our framework by introducing a third firm. The key novelty in the
triopoly case is that a bilateral merger may now induce an externality on the non-merging firm,
which in turn introduces some new investment incentives not present in our earlier duopoly
markets. Our analysis here should be viewed as giving a glimpse of the new effects this can
introduce, as we do this for one particular three-party bargaining process among many possible
ones. Triopoly markets also allow us to study optimal policy toward mergers that combine two
20
Figure 7: (a) Contour lines of the percentage difference between the steady state aggregate
value of the MPP-AV and no-mergers equilibria, (AVM P P − AVN o )/AVM P P . (b) Contour lines
of the percentage difference between the steady state aggregate value of the MPP-AV and
static aggregate surplus-based policy equilibria, (AVM P P − AVStatic )/AVM P P .
weaker (i.e., lower capital stock) firms who face a stronger rival, an issue that arises frequently
in merger cases (such as the AT&T/T-Mobile USA and Sprint/T-Mobile USA mergers).
We first examine the robustness of our previous two-firm results to the possibility of a
third firm. We show that the (A = 3, B = 26) market that we studied in Section 3 is a
“natural duopoly” in the sense that a third firm does not wish to enter when no mergers
are allowed, although when mergers are allowed the entry for buyout motive sometimes leads
a third firm to enter temporarily. Nonetheless, our previous conclusions continue to hold.
We then examine merger policy in two “natural triopoly” markets, where when no mergers
are allowed the market usually has three firms with positive levels of capital. The presence
of externalities on non-merging firms introduces a new effect on incumbent investment that
impacts optimal merger policy significantly in one of these markets.
To proceed, we consider a three-firm version of the general model of Section 2. The
bargaining stage in each period is a static version of the bargaining protocol in Burguet and
Caminal (2015): One firm, say i, is randomly selected as the proposer, with each firm equally
likely to be selected. The proposer chooses which of its two rivals to invite for merger negotia-
tions. Suppose firm i invites j 6= i. If firm j accepts the invitation, then these two firms enter
merger negotiations. Otherwise, firm j invites firm k 6= i, j. If firm k accepts, then j and k
enter bilateral merger negotiations. If it rejects the invitation, then no merger takes place this
period. Bilateral merger negotiations are such that each party is equally likely to be selected
to make the other a take-it-or-leave offer. If the offer is accepted, then the merger is proposed
21
to the authority; if it is rejected, then no merger occurs in that period. So, conditional on two
firms entering bilateral merger negotiations, the expected payoffs coincide with those in the
Nash bargaining solution between those two firms. An attractive feature of this bargaining
process is that, no matter which firm is selected as the proposer, each of the three mergers is
feasible.38 As we will see below, another attractive feature is that, generically, the bargaining
process has a unique equilibrium for given continuation values.39
Recall that (for the case of three firms)
∆ij (K) ≡ V (Ki + Kj , Kk , 0) − (V (Ki , K−i ) + V (Kj , K−j ))
denotes the joint gain firms i and j get from merging, gross of proposal costs, relative to if no
merger occurs, and that
Sij (K, φij ) ≡ aij (K)∆ij (K) − φij
is the expected bilateral surplus of firms i and j from entering merger negotiations in state
+
K (after the realization of the proposal cost φij ), and that Sij (K, φij ) ≡ max{0, Sij (K, φij )}.
In the following, we will sometimes say that merger Mij is more profitable than merger Mik
+ +
if Sij (K, φij ) > Sik (K, φik ). Note, however, that this notion of profitability ignores the ex-
ternality that i and j impose on firm k when entering merger negotiations, which equals
I{S + (K,φij )>0} Xkij (K).
ij
+
Note also that the “profitability” of a merger between two firms i and j, Sij (K, φij ),
depends on continuation values. Thus, a merger can be “unprofitable” because it is better
for one or both of the firms not to merge in the hopes of benefiting should its rivals merge in
the next period.
The following proposition completely characterizes the equilibrium outcome of the merger
process:
Proposition 1 Suppose firm i is selected as the proposer in state K. Then:
+ + +
(i) If Sjk (K, φjk ) > max{Sij (K, φij ), Sik (K, φik )}, then firm i invites either firm j or firm k,
and merger Mjk gets proposed.
38
For example, a simpler random proposer bargaining process in which a proposer is chosen in each period
who can make a take-it-or-leave-it merger offer to either of the other firms would have the disadvantage that
one of the three mergers would end up being impossible in each period. If there is a clearly most profitable
merger, with probability 1/3 the only way for it to happen would be for no merger to occur today in the hope
that (with a 2/3 probability) it can happen in the next period. One might think that it is possible to avoid
this problem by allowing multiple rounds in each period, with a new proposer chosen randomly in each round
should a deal not yet be reached. However, when we experimented with such a procedure we found that cycles
could arise in which the equilibrium outcome depended drastically on how many rounds were allowed.
39
At the same time, there are also features that one might view as less attractive. For example, once an
invitation to negotiate is accepted, a firm that is negotiating cannot use the possibility of striking a deal with
the excluded firm to improve its deal.
22
+ + +
(ii) If Sij (K, φij ) > Sik (K, φik ) ≥ Sjk (K, φjk ), then firm i invites firm j, and merger Mij
gets proposed.
+
(iii) If Sij +
(K, φij ) > Sjk +
(K, φjk ) > Sik +
(K, φik ) and Sij (K, φij )/2 > Xijk (K), then firm i
invites firm j, and merger Mij gets proposed.
+
(iv) If Sij +
(K, φij ) > Sjk +
(K, φjk ) > Sik +
(K, φik ) and Sij (K, φij )/2 < Xijk (K), then firm i
invites firm k, and merger Mjk gets proposed.
+ + +
(v) If Sij (K, φij ) = Sjk (K, φjk ) = Sik (K, φik ) = 0, then no merger occurs.
In case (i), even though firm i is the proposer, merger Mjk must be the outcome as both
+
firms j and k prefer their half of the surplus from merger Mjk , Sjk (K,φjk ), to what they can
get bargaining with firm i. In contrast, in case (ii), both j and k prefer to split the bargaining
surplus available in a deal with firm i to bargaining with each other. So firm i can get either
merger Mij or merger Mik , and prefers the former. In cases (iii) and (iv), firm i will get merger
Mij if it proposes bargaining with firm j (j prefers to split surplus with firm i rather than
with firm k), but can induce the second-most profitable merger Mjk by proposing to bargain
with firm k (k prefers to bargain with j); which of these options firm i prefers depends on
comparing its split of the bargaining surplus with firm j to the externality it experiences when
merger Mjk happens. In case (v), no merger has a positive surplus, so no merger occurs.40 A
formal proof is found in the Online Appendix. Observe that merger Mjk will happen in two
circumstances when firm i is the proposer: when Mjk is the most profitable merger, and when
Mjk is more profitable than Mik and firm i gains more when merger Mjk occurs than its half
of merger Mij ’s surplus.
4.1 Allowing a Third Firm in the (A = 3, B = 26) Market

When no mergers are allowed in the (A = 3, B = 26) market, introducing a third firm has
almost no impact as triopoly states are very rare: states with three active firms are visited
only about 0.5 percent of the time. That is, the two-firm equilibrium outcome we studied
in Section 3 when no mergers are allowed approximates well the outcome of a “free entry”
equilibrium. (Equilibrium statistics when we allow a third firm are displayed in the Online
Appendix.) In this sense, it is a “natural duopoly” market.41
40
We assume that firms i and j do not merge if aij (K)∆ij (K) − φij = 0; however, in case (v) this is a measure
zero event — generically we have aij (K)∆ij (K) − φij < 0 for all i and j. Note as well that since the surplus
measures the bilateral gain from a merger relative to no merger occuring, one reason that a merger may have
negative surplus is that one or both firms anticipate the possibility of experiencing a positive externality should
rivals merge in a subsequent period.
41
The same is true in the “small” (A = 3, B = 22) and “large” (A = 3, B = 30) markets discussed in the
Online Appendix.
23
When mergers are allowed, however, the prospect of entry for buyout can lead both a second
and a third firm to enter for the prospect of being acquired.42 When this happens a subsequent
merger induces a duopoly rather than a monopoly outcome. As a result, introducing the
possibility of a third active firm reduces the steady state frequency of a monopoly state from
86.0% to 85.4%, but increases the probability of a merger from 37.7% to 50.8%.43
Despite these quantitative changes in the equilibrium outcome, our previous insights carry
over to the three-firm case. Allowing all mergers induces entry for buyout and implies that the
industry spends much more time in a monopoly state: the steady state frequency of monopoly
increases from 18.0% under the no-mergers policy to 85.4% under the all-mergers-allowed
policy. Since firms invest on average less in such monopoly states than in more symmetric
states, the average total capital level is lower under the all-mergers-allowed policy.44 Because
the industry spends more time in monopoly and firms invest less, consumers are much worse
off when all mergers are allowed: consumer value decreases from 48.3 to 38.0. Despite the
large increase in the frequency of monopoly, firms collectively do not gain much from allowing
all mergers because of the distortions in firms’ investment behavior associated with entry for
buyout. As a result, average AV falls from 117.6 to 107.7 when all mergers are allowed.
Turning to the Markov perfect policy, when the authority uses the AV criterion the steady
state probability that the industry finds itself in a duopoly state pre-merger is high (77.6%),
and the likelihood of triopoly is low (2.2%). In duopoly states, the antitrust authority approves
a merger only when at least one of the incumbents is very small, and is more restrictive than
would be statically optimal; a merger happens 13.3% of the time versus 50.2% if a static
aggregate surplus-based policy were instead followed. The performance measures (such as
CV, AV, merger frequency, probability of monopoly) under the Markov perfect policy with
AV criterion all lie between those of the no-mergers policy and the all-mergers-allowed policy.45
In particular, the simple commitment policy of never allowing a merger induces again a higher
42
For example, with two firms the probability that the entrant invests in state (5,0) was 58%. With three
firms, in state (5,0,0), the probability that each entrant invests is only slightly smaller, namely 51%. This
implies that, starting from state (5,0,0), the probability that there are three firms with capital at next period’s
merger stage is equal to 16.7% (the probability that both entrants invest times the probability that neither
entrant’s capital depreciates). At the same time, the probability that no entry (and therefore no merger) occurs
in state (5,0,0) is considerably lower than in state (5,0), namely 35.0% rather than 53.6%.
Note, however, that our restriction to symmetric strategies implies that in state (5, 0, 0) either both firms
invest or neither do. There could be an asymmetric equilibrium in which the entrants invest asymmetrically,
perhaps even with one entrant not investing at all.
43
With three firms, the industry does not spend any time in a triopoly state at the output competition stage.
44
The average total capital level decreases from 8.0 to 7.6 when all mergers are allowed.
45
The only exception is that the average total capital level under the Markov perfect policy is larger than in
the no-mergers policy. As we have seen in the two-firm case, allowing mergers tends to lead to more investment
state-by-state but increases the relative frequency of monopoly states in which investment is lower. With only
two firms, the second effect outweighs the first. Here, the first effect outweighs the second because introducing
a third firm reduces the frequency of monopoly states under the Markov perfect policy (for the same reason as
it does under the all-mergers-allowed policy).
24
average AV than the Markov perfect policy with the AV criterion.
4.2 Two Natural Triopoly Markets

We now examine merger policy in two “natural triopoly” markets. In the first, we propor-
tionally increase market size by examining a market with (A = 3, B = 70), while in the second
we increase the choke price by setting (A = 4, B = 20); all other parameters remain the same
as in Section 3. In these markets the industry spends, respectively, 75.5% and 99.4% of the
time in a triopoly state when no mergers are allowed. The (A = 3, B = 70) market spends
long periods of time in roughly symmetric triopoly states with each firm having about 9 units
of capital, but in the unlikely occurrence that one firm depreciates to zero the industry stays
in duopoly for a long time, only generating entry if one of the duopolist’s capital depreciates
to a very low level.46 In the (A = 4, B = 20) market, on the other hand, the depreciation
of one firm to zero capital is soon followed by entry and a return to symmetric triopoly with
each firm having roughly 5 units of capital. Indeed, in the (A = 4, B = 20) market there is a
positive probability of entry against symmetric duopolists (i.e., a third firm with zero capital
investing a positive amount) as long as the incumbents each have less than 11 units of capital;
in contrast, in the (A = 3, B = 70) market an entrant will not enter unless the incumbents
each have 3 units of capital or less.
4.2.1 Merger Bargaining with Three Firms
To understand some of the effects of merger policy that we observe in these markets it is useful
to first examine some features of the merger bargaining process. To do so, we look at the case
in which all mergers are allowed.
Tables 4 and 5 show the probability that a merger occurs, the bargaining surplus, and the
merger externality in symmetric states that are around the typical triopoly states in these
markets.47 As in our duopoly markets, a merger occurs with certainty in these symmetric
triopoly states unless the capital stocks are low. Notably, however, the gain for the firm
not involved in the merger far exceeds the gain for the firms that merge, especially in the
(A = 4, B = 20) market.
Next, consider asymmetric states. In the all-mergers-allowed steady states of these mar-
kets, mergers tend to happen when there are two large incumbents and a small recent entrant.
Generally, a merger between the two largest firms generates a negative surplus for the merger
partners, as it leads to further entry. But a merger between one of the incumbents and the
entrant is worthwhile. Given the large positive externalities on the non-merging firm in these
46
We increase the state space to allow up to 30 units of capital for each firm for this market.
47
In these symmetric states each firm has a two-thirds chance of being involved in a merger should one occur.
In line with Proposition 1, in any of these symmetric states, a merger will be proposed if Sij (K, 0) is larger
than φij for some merger Mij . Recall that Sij (K, 0) incorporates any change in continuation payoff, including
expected future externalities, to the merging firms due to their merger.
25
markets, when each large firm is the proposer the resulting merger is between the small firm
and the other incumbent.48 Thus, the relative likelihoods of the two possible mergers is deter-
mined by the preference of the smaller firm, who prefers to split surplus with the incumbent
that generates the largest merger surplus (net of the proposal costs).
In the (A = 4, B = 20) market, the firms that merge when all mergers are allowed are
highly likely to be the two smallest ones. For example, if instead of being at state (5,5,5) the
firms are at (5,5,6), then the likelihood that the smaller firms merge is 67%, while 33% of the
time one of the smaller firms merge with the larger firm. At state (5,5,7) the likelihood of the
two smallest firms merging rises to 97%, and it is 100% at state (5,5,8). Similarly, at (1,7,8),
the likelihood that the two smallest firms merge is 89%, and this increases to 100% at (1,7,9).
Overall, conditional on a merger occurring in a state with a unique largest firm, the steady
state likelihood that the two smallest firms merge is 97%. The average merger surplus when
the smallest two firms merge is 1.26, while the average externality on the largest firm is 3.06.
As we will see, the desire to be the firm capturing this externality is an important driver of
investment by incumbents in this market.
The effect of asymmetry is much less pronounced, however, in the (A = 3, B = 70) market.
For example, at state (2,12,13) the likelihood that the two smallest firms merge is 51%, and
only increases to 52% at state (2,12,15). Moreover, in some cases in this market, it is the
largest firm that is most likely to merge with the smallest firm: for example, at state (2,7,12)
a merger involving the smallest firm is certain to occur, but it is with the largest firm 56% of
the time. Overall, conditional on a merger occurring in a state with a unique largest firm, the
steady state likelihood that the two smallest firms merge in this market is 49%.
The different effects of asymmetry in these two markets appears to be related to their very
different likelihoods of entry and subsequent mergers, which we discuss in the next subsection.
In the (A = 4, B = 20) market, the states in which entry occurs lead to situations in which
a merger in the current period is fairly likely to lead to further entry and mergers. Hence,
a merger today that changes the identity of the largest firm is fairly likely to change who
benefits from merger externalities in the following periods, making the smaller incumbent
value a merger highly when the incumbents’ capital stocks are close. In contrast, in the
(A = 3, B = 70) market, the states following entry are highly likely to be ones in which the
probability of near-term future mergers is low, making the relative surpluses created by mergers
with each of the incumbents fairly unaffected by any effects on future merger bargaining.
The presence of these externalities can also lead to implications for stock price responses to
merger announcements, as firms involved in a merger can experience negative returns because
the market has learned that the firm will not be benefitting from a merger externality. For
example, in the all-mergers-allowed equilibrium of the (A = 4, B = 20) market, the average
48
The chosen incumbent proposer makes an offer to the other incumbent, who then invites the small firm to
bargain.
26
percentage change in joint value for the two merging firms upon announcement is -0.5%.49
Conditional on the two merging firms being unequal in size, the average percentage change in
value for the larger merging firm is -0.6%, while it is 0.1% for the smaller merging firm.50 The
small value change for the small firm occurs because in most states in which mergers occur
a merger is certain to occur and involve the small firm. In such cases, the only surprise can
involve which firm it merges with, and often this is fully anticipated as well.
4.2.2 Effects of Fixed Merger Policies
Table 6 reports the outcomes in these two markets when all mergers are allowed. As in
our previous two-firm analysis, in both markets switching from the no-mergers-allowed policy
to the all-mergers-allowed policy has a significant negative impact on steady state consumer
value, here due to the industry spending much less time in triopoly and more time in duopoly
states when all mergers are allowed. However, in other respects, when all mergers are allowed
these two markets display some important differences from each other and from the outcomes
we discussed in Section 3.
In the (A = 3, B = 70) market, mergers almost never happen when all mergers are allowed.
The reason is that the industry settles into a roughly symmetric duopoly in which each firm
has roughly 12 units of capital, and in which entry is very unlikely.51 For entry to happen with
positive probability one or both incumbents needs to experience a great deal of depreciation.
For example, with symmetric incumbents entry only starts to have a positive probability once
both have depreciated to 8 units of capital or less. As in Section 3, the prospect of entry
for buyout incents entry (when no mergers are allowed entry would not happen unless both
symmetric incumbents had less than 3 units of capital), but still not enough for entry to be
more than a rare occurrence. Compared to the no-mergers-allowed steady state, the shift
from triopoly to duopoly states reduces investment, causing the level of capital and AV to fall,
although in contrast to the duopoly situation in Section 3 producer value does increase.
Allowing all mergers in the (A = 4, B = 20) market instead leads to a 27.5% likelihood of
entry. This happens because the industry converges to a duopoly outcome in which two active
firms each have roughly 8 units of capital and in each period there is approximately a 25%
49
This calculation assumes that the market knows the firms’ capital stocks but not the realization of proposal
costs – so value changes occur only when mergers occur, due to proposal cost realizations that were not fully
anticipated given the capital state K.
50
In such cases, the large firm has on average 6.96 units of capital, while the small firm has 2.07 units of
capital. Positive value changes occur for the large firm only 1.5% of the time and only when the observed
merger was the only merger with positive probability but that probability was less than one. Negative returns
occur for the small firm only 2.4% of the time and only when there are positive probabilities of it merging with
each of two differently-sized larger firms.
51
For example, starting from a symmetric state in which three firms each have 5 units of capital, a merger
is certain to happen; after that, entry is very unlikely and the two remaining firms converge over a number of
periods to having roughly 12 units of capital each.
27
chance of an entrant building one unit of capital.52 When this happens, the entrant is acquired
by one of the two incumbents, investments again lead to a situation in which the two firms each
have roughly 8 units of capital, and the process repeats itself. In contrast to what we have seen
in the duopoly markets of Section 3 and the (A = 3, B = 70) market, producer value increases
enough to make steady state AV almost identical to that when no mergers are allowed, and the
average total capital level is higher (14.3 rather than 14.0). The greatly improved relative AV
performance of the all-mergers-allowed policy is largely driven by the incumbents’ investment
responses to the bargaining externality. When there are two incumbent firms with capital
levels that are not too far apart and an entrant with no capital, this externality incents these
incumbents to invest more in an attempt to become the largest firm in the industry, and then
benefit from its rival acquiring the entrant. Moreover, this enhanced incumbent investment
incentive also curbs the amount of entry for buyout behavior. For example, in state (0, 7, 7),
the vector of expected investments is (0.6, 1.2, 1.2) and (0.3, 1.7, 1.7) under the no-mergers and
all-mergers-allowed policies, respectively.53
In both markets a static aggregate surplus-based policy allows many mergers. For example,
in the (A = 3, B = 70) market, when a merger is proposed between symmetric firms the
merger is always approved if the firms are each smaller than the non-merging rival, and is
often approved even if they are larger than the rival. The static policy is not quite so lenient
in the (A = 4, B = 20) market, but is still very permissive: for example, when a merger is
proposed between symmetric firms facing a rival with 10 units of capital, the merger will be
approved if and only if the merging firms each have no more than 5 units of capital. As a
result, the outcomes in these markets when the authority follows a static aggregate-surplus
based policy is very close to that when all mergers are allowed. In contrast, a static consumer
surplus-based policy essentially allows no mergers in these markets.
4.2.3 Markov Perfect Merger Policy
Because entry against duopolists is very unlikely in the (A = 3, B = 70) market, with an AV
criterion the Markov perfect policy’s treatment of a proposed merger of two of three active
firms is largely unaffected by any entry for buyout concerns. Without such concerns, this
Markov perfect policy is quite permissive.54 For example, like the static aggregate surplus-
52
For example, starting from a symmetric state in which three firms each have 5 units of capital, a merger is
certain to happen; after that the two remaining firms converge over a number of periods toward having roughly
8 units of capital each, until another entry event occurs.
53
Still, much of the entrant’s investment incentive comes from the prospect of being acquired: for example,
in state (0, 7, 7) it would not invest at all if the incumbents were following their all-mergers-allowed investment
policies but no mergers were allowed.
54
The remaining factors tend to favor a permissive policy here: First, combining two firms’ capital stocks
increases aggregate surplus in many states. Second, the combination reduces investment costs since investment
costs are decreasing in firm size. Third, since investment by small firms in triopoly is often socially excessive
(much as in duopoly), mergers of small firms that reduce their investments can be beneficial for aggregate value.
28
based policy it always allows two symmetric firms to merge when each is no larger than their
non-merging rival, and often allows them to merge even when they are larger. Given this
leniency, the steady state outcome under the Markov perfect policy with an AV criterion is
essentially identical to that when all mergers are allowed. As with allowing all mergers, the
Markov perfect policy therefore lowers steady state AV compared to a policy of allowing no
mergers.55,56
In the (A = 4, B = 20) market, entry happens but the investment incentives noted above
curtail the extent of the investment inefficiency following a merger. Overall, the Markov perfect
policy with an AV criterion is again fairly lenient and allows two symmetric firms to merge
whenever they are each smaller than their non-merging rival. Indeed, in this market it is
much more lenient than the static aggregate surplus-based policy. As a result, the steady
state outcome under the Markov perfect policy with the AV criterion is again very similar to
that under the all-mergers-allowed policy.57
In summary, firms respond very differently in these two triopoly markets to policies allowing
mergers, driven by the differing likelihoods of entry in these markets. In both markets, however,
the negative effects of entry for buyout are either not present or limited, resulting in an AV-
based Markov perfect policy that is fairly lenient, a contrast to our finding for duopoly in
Section 3. Like the duopoly case, the resulting steady state outcome is not only bad for
consumers but fails to raise AV relative to allowing no mergers, although the difference is
minimal in the (A = 4, B = 20) market.
5 Conclusion
We have studied optimal merger policy in a dynamic industry model in which scale economies
can be achieved through either investment (“internal growth”) or merger (“external growth”).
In such a setting, an antitrust authority’s merger approval decisions must weigh any increases
The Markov perfect policy is, however, somewhat more stringent than the static policy when the non-merging
firm is not large; for example, when the nonmerging rival has 4 units of capital, the Markov perfect policy
allows a merger of symmetric firms as long as they have no more than 6 units of capital, while the static policy
would allow the merger even if they each have 10 units of capital. The two policies become quite similar when
the non-merging firm has more than 8 units of capital.
55
Observe in Table 6, however, that starting at state (0,0,0) the Markov perfect policy yields a higher AV
than allowing no mergers. The Markov perfect policy outcome has all three firms’ capital quickly reach a point
where a merger does not result in entry, then a merger occurs, followed by a future with a very low likelihood
of entry. Note that the antitrust authority’s lack of commitment in this market is not very important because
once the first merger occurs, future mergers are rare.
56
As in the (A = 3, B = 26) duopoly market, in both of these triopoly markets the Markov perfect policy
with a CV criterion yields a steady state equilibrium equivalent to the no-mergers-allowed policy.
57
In contrast to the results for the (A = 3, B = 70) market and the (A = 3, B = 26) duopoly market of
Section 3, here a commitment to the static aggregate surplus-based policy induces a slightly higher steady state
aggregate value than the no-mergers-allowed policy, the all-mergers-allowed policy, and the Markov perfect
policy.
29
in market power against the changes in productive efficiency caused by a merger, which are
affected not only by the immediate cost reductions of the merging parties due to their increased
scale, but also the investments of both the merging parties and rivals following the merger.
Moreover, an antitrust authority that is able to commit to its policy must also consider how
firms’ investment behavior is affected by the prospect of future merger approvals.
To shed light on this complicated problem we have developed and computationally solved
a dynamic model in which forward-looking Cournot firms invest in capital to produce a ho-
mogeneous product. Our model has three significant innovations relative to previous com-
putational dynamic industry models. First each firm in each period can flexibly decide how
many additional units of capital it wishes to purchase. Second, this investment technology is
(approximately) merger neutral in the sense that the investment opportunities available in the
market are unchanged following a merger, offering a much more attractive setting for studying
merger policy than the original Ericson and Pakes (1995)/Pakes and McGuire (1994) model.
Third, we introduce an antitrust authority as an active, maximizing player that cannot commit
to its future merger approval policy. Because of the time inconsistency difficulties that arise
in dynamic games the authority is unable to achieve as high a level of welfare as an authority
that can commit would be able to achieve.58
In much of our main analysis, we have focused on markets with two firms so as to be
able to use the familiar and well-accepted Nash bargaining solution. In addition, we have
studied markets with three firms, using one particular model of multi-firm bargaining with
externalities. Our analysis of these markets provides insights into the factors affecting optimal
merger policy when investment behavior and firm scale is a critical determinant of welfare,
and shows how optimal policy in dynamic settings with investment can differ in significant
ways from what would be statically optimal. Specifically, we make five key observations:
First, the desirability of approving a merger can depend importantly on the investment
behavior that will follow if it is or is not approved. However, this involves more than just the
behavior of the merging firms, as the investment behavior of outsiders to the merger (here, new
entrants) can have significant welfare effects. In particular, when entrants (or, more generally,
small firms) have higher investment costs than large established incumbents, entry for buyout
behavior can impose significant welfare losses and make merger approvals much less attractive
for an antitrust authority.
Second, in the other direction, investment behaviors can be greatly influenced by firms’
beliefs about future merger policy. Importantly, when the antitrust authority adopts a less
restrictive policy, this may spur entry for buyout behavior by firms seeking to be acquired.
Third, the inability to commit may be costly for an antitrust authority. In fact, in cases in
which aggregate value is the true social objective, it can often be better to endow the antitrust
authority with a consumer value objective (which roughly corresponds to the objective of most
58
In the Online Appendix, we have also analyzed the case of an antitrust authority that can commit to its
future merger approval rule.
30
antitrust authorities, including the U.S. and EU).

Fourth, the optimal antitrust policy for maximizing aggregate value in our model can differ
significantly from the optimal static policy that considers a merger’s effects only at the time it
would be approved, although it may be either more or less permissive than the static policy.
Finally, externalities on rivals arising from mergers in markets with more than two firms
can have significant effects on firms’ investment incentives and thereby shape the antitrust
authority’s optimal policy.
At a more general level, the existing literature on antitrust policy has largely neglected
issues relating to investment or firm entry and exit that are inherently dynamic. In a world in
which the antitrust authority cannot commit fully to its future actions, analyzing such issues
requires modeling the authority as a player who acts dynamically. The present paper is a
first step in doing so in a truly dynamic setting. By proposing a merger-neutral investment
technology that allows for complex multi-unit investment choices, and yet is tractable, it also
contributes to the computational industrial organization literature more generally.
References
[1] Abel, A. (1979), Investment and the Value of Capital. New York: Garland Publishing.
[2] Benkard, C. L., A. Bodoh-Creed, and J. Lazarev (2010), “Simulating the Dynamic Effects
of Horizontal Mergers: U.S. Airlines,” unpublished working paper.
[3] Bernard, A., J. Eaton, J. B. Jensen, and S. Kortum (2003), “Plants and Productivity in
International Trade,” American Economic Review 93: 1268-1290.
[4] Berry S. and A. Pakes (1993), “Applications and Limitations of Recent Advances in
Empirical Industrial Organization: Merger Analysis,” American Economic Review 83:
247-252.
[5] Bond, S. and J. van Reenen (2007), “Microeconometric Models of Investment and Em-
ployment, in: Handbook of Econometrics, vol. 6A, eds. J. J. Heckman and E. E. Leamer,
Amsterdam: North Holland.
[6] Bourreau, M., B. Jullien, and Y. Lefouilli (2018),“Mergers and Demand-Enhancing Inno-
vation,” TSE Working Paper, no. 18-907.
[7] Burguet, R. and R. Caminal (2015), “Bargaining Failures and Merger Policy,” Interna-
tional Economic Review 56: 1019-1041.
[8] Cheong, K.-S., and K. L. Judd (2000), “Mergers and Dynamic Oligopoly,” unpublished
working paper.
31
[9] Cooper, R. W. and J. C. Haltiwanger (2006), “On the Nature of Capital Adjustment
Costs,” Review of Economic Studies 73: 611-633.
[10] DeGraba, P. and G. L. Rosston (2014), “The Proposed Merger of AT&T and T-Mobile:
Rethinking Possible,” in: The Antitrust Revolution: Economics, Competition, and Policy,
6th edition, ed. J. Kwoka, New York: Oxford University Press.
[11] Doraszelski, U. and M. Satterthwaite (2010), “Computable Markov-perfect Industry Dy-

namics,” RAND Journal of Economics 41: 215-43.
[12] Ericson, R. and A. Pakes (1995), “Markov-perfect Industry Dynamics: A Framework for
Empirical Work,” Review of Economic Studies 62: 53-82.
[13] Farrell, J. and C. Shapiro (1990), “Horizontal Mergers: An Equilibrium Analysis,” Amer-
ican Economic Review 80: 107-126.
[14] Fazzari, S. M., R. G. Hubbard, and B. C. Petersen (1988), “Financing Constraints and
Corporate Investment,” Brookings Papers on Economic Activity 1988: 141-195.2.
[15] Federico G., G. Langus, and T. Valletti (2018), “Horizontal Mergers and Product Inno-
vation, ” International Journal of Industrial Organization 59: 1-23.
[16] Gans, J. and S. Stern (2003), “The Product Market and the Market for ‘Ideas’: Commer-
cialization Strategies for Technology Entrepreneurs,” Research Policy 32: 333-350.
[17] Gowrisankaran, G. (1997), “Antitrust Policy Implications of a Dynamic Merger Model,”

unpublished working paper.
[18] Gowrisankaran, G. (1999), “A Dynamic Model of Endogenous Horizontal Mergers,”

RAND Journal of Economics 30: 56-83.
[19] Hayashi, F. (1982), “Tobin’s Marginal q and Average q: A Neoclassical Interpretation,”

Econometrica 50: 213-224.
[20] Hollenbeck, B. (2017), ”Horizontal Mergers and Innovation in Concentrated Industries,”

unpublished working paper.
[21] Jeziorski, P. (2015), “Empirical Model of Dynamic Merger Enforcement – Choosing Own-
ership Caps in U.S. Radio,” unpublished working paper.
[22] Jones, C. I. (2005), “The Shape of Production Functions and the Direction of Technical
Change,” Quarterly Journal of Economics 120: 517-549.
[23] Lyons, B. (2002), “Could Politicians be More Right than Economists? A Theory of Merger
Standards,” CCP Working Paper 02-01, University of East Anglia.
32
[24] Motta, M. and E. Tarantino (2018), “The Effect of Horizontal Mergers, When Firms
Compete in Prices and Investments,” unpublished working paper.
[25] Motta, M. and H. Vasconcelos (2005), “Efficiency Gains and Myopic Antitrust Authority
in a Dynamic Merger Game,” International Journal of Industrial Organization, 23: 777-
801.
[26] Nilssen, T. and L. Sorgard (1998), “Sequential Horizontal Mergers,” European Economic
Review 42: 1683-1702.
[27] Nocke, V. and M. D. Whinston (2010), “Dynamic Merger Review,” Journal of Political
Economy 118: 1200-1251.
[28] Pakes, A. and P. McGuire (1994), “Computing Markov Perfect Nash Equilibrium: Nu-
merical Implications of a Dynamic Differentiated Product Model,” RAND Journal of
Economics 25: 555-589.
[29] Pittman, R. and Y. Li (2013), “AT&T and T-Mobile: Economies as an Antitrust Defense
Applied,” Journal of Competition Law and Economics 9: 49-63.
[30] Rasmusen, E. (1988), “Entry for Buyout,” Journal of Industrial Economics 36: 281-300.
[31] Syverson, C. (2004), “Market Structure and Productivity: A Concrete Example,” Journal
of Political Economy 112: 1181-1222.
[32] Williamson, O. E. (1968), “Economies as an Antitrust Defense: The Welfare Tradeoffs,”

American Economic Review 58: 18-36.
33
Table 1: (A = 3, B = 26) Market Static Equilibria

State (1, 0) (10, 0) (5, 5)
Marginal Cost (M C) 2.56 1.32 1.54
Price (P ) 2.78 2.16 2.02
P ÷ MC 1.09 1.63 1.32
Total Quantity 5.67 21.80 25.40
Total Profit 5.12 26.00 22.80
Consumer Surplus 0.62 9.14 12.40
Aggregate Surplus 5.74 35.12 35.16
34
Table 2: Performance Measures for the (A = 3, B = 26) Market under Various Policies
No-Mergers/
Performance Measure 59 All-Mergers Static-AS MPP-AV
Static-CS/
MPP-CV
Avg. Consumer Value 48.1 35.8 35.9 43.3
Avg. Incumbent Value 69.4 68.1 68.5 69.9
Avg. Entrant Value 0.0 1.9 1.8 0.5
Avg. Blocking Cost 0.0 0.0 0.0 -0.1
Avg. Aggregate Value 117.5 105.8 106.2 113.6
Avg. Price 2.15 2.26 2.26 2.19
Avg. Quantity 22.2 19.2 19.2 21.0
Avg. Total Capital 8.0 7.0 7.0 7.7
Merger Frequency 0.0% 37.7% 37.9% 16.1%
% in Monopoly 18.6% 86.0% 88.0% 49.4%
% min{K1 , K2 } ≥ 2 75.7% 0.9% 0.7% 44.2%
State (0,0) CV 30.3 23.9 24.1 25.6
State (0,0) AV 36.7 34.0 34.1 35.5
59
All values are ex ante (beginning-of-period) values except % in Monopoly and % min{K1 , K2 } ≥ 2 (showing
the percentages of the time that industry capital is in each type of state) which are at the Cournot competition
stage. “No mergers” and “All Mergers” refer to the no-mergers-allowed and all-mergers-allowed policies, re-
spectively. “Static CS” and “Static AS” refer, respectively, to the equilibria under the optimal static consumer
surplus-based and aggregate surplus-based merger policies. “MPP CV” and “MPP AV” refer, respectively, to
the equilibria when the antitrust authority cannot commit (resulting in a Markov perfect policy) under con-
sumer value and aggregate value welfare criteria. “State (0,0) CV” and “State (0,0) AV” are the values of CV
and AV, respectively, for a new industry that starts with no capital.
35
Table 3: Average Capital Addition in the (A = 3, B = 26) Market

Steady State Distribution
No-Mergers All-Mergers-Allowed
Investment No-Mergers 2.0 1.5
Behavior All-Mergers-Allowed 2.2 1.8
36
Table 4: Merger Bargaining Outcomes in Symmetric States

when All Mergers are Allowed [(A = 3, B = 70) Triopoly Market]
Capital Stock for Each Firm 3 4 5 6 7 8 9 10
Pr(Merger) 0% 61% 100% 100% 100% 100% 100% 100%
Bargaining Surplus Sij (K, 0) -2.64 0.27 5.73 7.86 8.71 9.19 9.59 9.98
ij
Externality Xk (K) -1.13 5.12 10.04 12.47 14.06 15.41 16.66 17.85
37
Table 5: Merger Bargaining Outcomes in Symmetric States

when All Mergers are Allowed [(A = 4, B = 20) Triopoly Market]
Capital Stock for Each Firm 1 2 3 4 5 6 7 8
Pr(Merger) 0% 12% 100% 100% 100% 100% 100% 100%
Bargaining Surplus Sij (K, 0) -1.75 0.04 2.03 3.52 4.03 4.64 5.20 5.42
ij
Externality Xk (K) 0.02 1.47 6.91 10.42 12.67 14.56 15.93 17.06
38
Table 6: Performance Measures for the Triopoly Markets

(A = 3, B = 70) (A = 4, B = 20)
No-Mergers/ No-Mergers/
All-Mergers/ All-Mergers/
Performance Measure 60 MPP-CV/ Static-AS MPP-CV/ Static-AS
MPP-AV MPP-AV
Static-CS Static-CS
Avg. Consumer Value 209.9 180.0 180.2 161.9 145.1 146.4
Avg. Incumbent Value 199.3 221.7 221.7 126.7 143.0 142.2
Avg. Entrant Value 0.0 0.0 0.0 0.0 0.5 0.7
Avg. Blocking Cost - - - - - -
Avg. Aggregate Value 409.2 401.7 401.7 288.7 288.6 289.3
Avg. Price 1.91 1.99 1.99 2.20 2.30 2.29
Avg. Quantity 76.6 71.0 70.9 36.0 34.0 34.2
Avg. Total Capital 26.1 23.2 23.2 14.0 14.3 14.6
Merger Frequency - 0.4% 0.4% - 27.5% 29.4%
% in Monopoly 0.0% 0.0% 0.0% 0.0% 0.0% 0.0%
% in Duopoly 24.5% 99.6% 99.6% 0.6% 97.9% 96.7%
% with 1 AV/AS
0.0% 0.0% 46.0% 0.0% 5.8% 1.0%
Enhancing Merger
% with 2 AV/AS
0.7% 0.4% 0.0% 0.0% 23.8% 31.7%
Enhancing Mergers
% with 3 AV/AS
74.8% 0.0% 0.4% 0.0% 0.0% 0.3%
Enhancing Mergers
State (0,0,0) CV 136.4 132.3 132.5 111.0 111.5 111.8
State (0,0,0) AV 143.4 147.6 147.8 134.4 124.3 125.7
60
All values are ex ante (beginning-of-period) values, except the % in Monopoly, % in Duopoly, and State
(0,0,0) CV and AV which are at the Cournot competition stage.
39

Copyright The University of Chicago 2019. Preprint (not copyedited or formatted).(b) All-mergers-allowed
(a) No-mergers-allowed
0.1 0.1
0.08 0.08
0.06 0.06
0.04 0.04
0.02 0.02
0 0
10 10
9 9
8 8
7 10 7 10
6 9 6 9
5 8 5 8
7 7
4 6 4 6
3 5 3 5
2 4 2 4
3 3
1 2 1 2
0 1 0 1
0 0
(c) Static-AS (d) MPP-AV
0.1 0.1
0.08 0.08
0.06 0.06
0.04 0.04
0.02 0.02
0 0
10 10
9 9
8 8
7 10 7 10
6 9 6 9
5 8 5 8
7 7
4 6 4 6
3 5 3 5
2 4 2 4
3 3
1 2 1 2
0 1 0 1
0 0

(a) No-mergers-allowed (b) All-mergers-allowed
0.4 0.4
0.3 0.3
0.2 0.2
0.1 0.1
0 0
8 8
7 7
6 2 6 2
5 5
4 4
3 1 3 1
2 2
1 1
0 0 0 0



(a) Aggregate value (b) Entry probability

3.2 3.2
9%
10%
3.15 3.15
20%
3.1 3.1 30%

20% 5% 5%
mo mo mo
A (demand choke price)

nop nop nop
oly oly 20% oly 60%
3.05 3.05 mo
nop
9% oly
90%
3 3
6%
3%
2.95 2.95 99%
0%
60% 60%
2.9 mon 2.9 mon
opo opo
ly ly
99
%
2.85 2.85
2.8 2.8
20 22 24 26 28 30 32 20 22 24 26 28 30 32
B (size of market) B (size of market)

(a) MPP-AV and No-mergers (b) MPP-AV and Static-AS

3.2 3.2
6%
MPP-AV equilibrium is the same as 9%

3.15 3.15
the no mergers equlibrium
3.1 3.1
20% 5% 0% 20% 5%
mo mo mo mo

nop nop nop nop
oly oly oly oly
3.05 3.05
9%
3% 6%
3 3
0% 0%
2.95 2.95
60% 60%
3% mon mon
2.9 opo 2.9 opo
ly ly
2.85 6% 2.85
9%
2.8 2.8
20 22 24 26 28 30 32 20 22 24 26 28 30 32
B (size of market) B (size of market)

Supplementary/Online-Only Appendix Click here to access/download;Supplementary/Online-Only
Appendix;2018.08.09 MNSW - Online Appendix.pdf

Online Appendix - Internal versus External Growth in

Industries with Scale Economies: A Computational Model of
Optimal Merger Policy
Ben Mermelstein Volker Nocke
Bates White University of Mannheim, NBER, and CEPR
Mark A. Satterthwaite Michael D. Whinston
Northwestern University M.I.T. and NBER
August 9, 2018
1 Formal Model, Bargaining, and Computation

In this section, we provide a more detailed description of parts of the main paper: the model
presented in Section 2, the calculations of merger bargaining outcomes, the proof of Proposition
1, and our computational algorithm.
1.1 Formal Model Description

We follow the timing displayed in Figure 1 and the notation established in Section 2 of the
main paper. Let the number of …rms be n 2, the set of …rms be I = f1; ; ng, and the
industry state be the vector of their capital stocks K = (K1 ; ; Kn ). Firms are restricted
to an integer number of possible capital levels, with the maximal capital level K chosen to
be non-binding. Since a …rm may have zero capital, let S f0; 1; 2; :::; Kg be the admissible
values of Ki and let S n be the state space. The industry’s state at the beginning of a period is
its ex ante state while its state just after the entry stage and before the Cournot competition
stage is its interim state.
The logic of backward induction guides our presentation of the model. The …rms take their
environment and the antitrust policy faij ( )gij2J as given, where J fijji; j 2 I; i 6= jg is
the set of pairs of …rms, and aij (K) is the probability that the authority approves merger Mij
when proposed in ex ante state K. Therefore we …rst derive conditions for their symmetric
Markov perfect equilibrium behavior given that their goal is to maximize the expected net
present value (ENPV) of their future cash ‡ows. We then turn to the antitrust authority’s
problem of maximizing welfare. We consider authorities that vary in their goals and their
ability to commit.
1
Firm i’s ex ante value function in ex ante state K, Vi (K), is the beginning-of-period ENPV
of its future cash ‡ows. Similarly, …rm i’s interim value Vi (K) gives the ENPV of its future
cash ‡ows starting from interim state K. The transition from the ex ante state to the interim
state is the outcome of the merger bargaining game between …rms and, if a merger has been
proposed, the merger approval decision of the antitrust authority. The transition from the
interim state to next period’s ex ante state is the outcome of …rms’investment decisions and
the subsequent (stochastic) depreciation of capital.
Throughout we assume that the …rms play Markov perfect equilibrium strategies that the
antitrust policy and the speci…ed merger protocol induce. To understand the …rms’ Markov
perfect equilibrium, …x the antitrust authority’s merger approval policy and let fVi ( )gi2I be
the value functions that give the ENPV of the …rms’ future cash ‡ows at the beginning of
period t + 1 as a function of the ex ante state K. Given these value functions, each …rm i
uses backward induction to calculate, for each interim state K0 2 S n ; its optimal period-t
investment decision, which must be a best reply to its competitors’investment policy choices.
Given this Nash equilibrium in investment policies conditional on the beginning of period t + 1
value functions fVi ( )gi2I ; each …rm can calculate for all interim states K0 its interim values
V i (K0 ) conditional on fVi ( )gi2I .
Based on this vector of interim values and given the antitrust authority’s approval policy,
the …rms negotiate over mergers. These negotiations, conducted in accordance with the pro-
tocols speci…ed in Section 1.2 below, determine for each ex ante state K the probability of
each possible merger Mij being proposed, as well as the ex ante values fV^i ( )gi2I in period
t. If fV^i ( )gi2I = fVi ( )gi2I , then the ex ante value functions, the interim value functions,
the investment functions, and the equilibrium merger bargaining outcomes together form a
Markov perfect equilibrium for the industry with respect to the …xed merger policy.
We now present the model and our notion of Markov perfect equilibrium in more detail.
Following the logic of backward induction, we begin by describing …rms’investment policies.
1.1.1 Firms’Investment Policies
At the investment stage, each …rm i, after privately learning the Ki independent draws of its
capital augmentation costs (c1 ; :::; cKi ) and the single independent draw of its green…eld cost
cg , unilaterally decides how many units of capital (if any) to add.1
Firm i’s investment policy is denoted i ( jK) : f0; 1; ; K Ki g S n ! [0; 1]. Prior
to the realization of its cost draws, policy i gives the probability i (ki jK) of …rm i adding
ki 2 f0; 1; :::; K Ki g units of capital in interim state K. Recall that at the end of each period
each unit of capital depreciates with probability d, so if …rm i enters the depreciation stage
1
Each …rm also decides on the quantity it produces. This decision is embedded in …rm i’s single-period pro…t
function (Ki ; K i ) because we assume competition in the product market is static Cournot.
2
with Ki units of capital, then the probability it exits the stage with Ki0 units of capital is
( 0 0
Ki
0 Ki0 (1 d)Ki dKi Ki if Ki0 2 f0; 1; :::; Ki g
(Ki jKi ) = : (1)
0 otherwise
Given that …rm i follows investment policy i ; the probability of …rm i in interim state K
leaving the period with Ki0 2 S units of capital is therefore given by the transition function
KXKi
0
i (Ki jK) = i (mjK) (Ki0 jKi + m): (2)
m=0
Consider now …rm i’s optimal investment policy for a given realization c~ of its (Ki + 1)-
length vector of cost draws. Let cKi ( j~c) denote the resulting cost function where cKi (ki j~c) is
the minimum cost to add ki units of capital with cost draws c~. Let CKi be the set of possible
cost draws c~ and let hKi be the associated density that the distributions F and G of the cost
draws determine. For a given draw c~; cost function cKi ( j~ c), ex ante value function Vi ( ); and
rival transition functions i (induced by rival investment policies i ), …rm i chooses ki so
as to maximize its expected continuation value minus its investment cost:
2 3
X Y
max cKi (ki j~
c) + (Ki0 jKi + ki ) 4 j (Kj jK)5 Vi (K );
0 0
ki 2f0;1;:::;K Ki g
K0 2S n j6=i
where < 1 is the discount factor that the …rms and the antitrust authority use. Let ki denote
the solution to this optimization problem (which, generically, is unique) and de…ne !(ki j~c; K)
to be the indicator function with value 1 if ki = ki and 0 otherwise. Firm i’s investment policy
therefore is Z
(k
i i jK) = !(ki j~
c; K)hKi (~
c)d~
c; (3)
CKi
for ki 2 f0; 1; :::; K Ki g: This gives rise to …rm i’s expected investment cost in interim state
K: Z X
Eci (K) = !(ki j~
c; K)cKi (ki j~
c)hKi (~
c)d~
c: (4)
CKi
ki 2f0;1;:::;K Ki g
Firm i’s interim value in state K is its static pro…t less its expected investment cost plus
its ENPV in the continuation game; that is,
2 3
X Y n
V i (K) = (Ki ; K i ) Eci (K) + 4 j (Kj jK)5 Vi (K );
0 0
(5)
K0 2S n j=1
where (Ki ; K i ) is …rm i’s single-period pro…t from static Cournot competition in the product
market.2
2
Note that the static pro…t function is symmetric in that it depends only on the …rm’s own capital stock Ki
and the vector K i of its rivals’capital stocks, and any permutation of K i does not a¤ect the …rm’s pro…t.
3
1.1.2 Merger Bargaining and Merger Outcomes
We now fold backwards to the merger bargaining, merger approval, and entry stages. If no
merger occurs in ex ante state K, then the interim state remains the same as the ex ante
state. If merger Mij occurs, then with probability 1/2 the industry transits to interim state
K0 in which …rm i becomes the merged …rm with capital stock Ki0 = Ki + Kj , an entrant with
capital Kj0 = 0 replaces …rm j, and all other …rms’capital stocks remain unchanged. With the
complementary probability 1/2 …rm j becomes the merged …rm and …rm i is replaced by the
entrant. This probabilistic transition rule, in conjunction with the restriction to symmetric
equilibrium strategies (as de…ned below), ensures that the steady state distribution over S n
is symmetric. The …rms, seeking to maximize their ENPVs, negotiate what mergers (if any)
occur in accordance with the protocols de…ned in Section 2 for the model with two …rms, and
Section 4 for the model with three …rms.
Given the authority’s approval policy faij ( )gij2J and the interim value functions V i ( ) i2I ,
the …rms play subgame perfect strategies in the bargaining stage. As a general matter,
given an extensive form merger bargaining protocol3 , the antitrust authority’s approval pol-
icy faij ( )gij2J , and interim values V i ( ) i2I , we can solve for subgame perfect equilibrium
bargaining strategies, f i ( )gi2I . The outcome arising from these strategies determines the
probability ij (K) that each possible merger Mij is proposed in a given state K, each …rm
i’s ex ante expected proposal costs denoted by E i [ jK], and the …rms’ex ante values in that
state, fVi (K)gi2I . As well, these merger proposal probabilities and the antitrust authority’s
approval policy together determine the transition probability T0 (K; K0 ) from ex ante state
K to interim state K0 . Here, we treat these calculations as a black box. In Section 1.2 of
this Online Appendix we explicitly present for n = 2 and n = 3 the essential details of these
calculations for the merger protocols that we use in the main text. As an illustration, when
n = 3, the formula for the ex ante value of …rm i in ex ante state K is
8 9
1< X =
jk
Vi (K) = V i (K) + E i [ jK] + (K)a (K) (K) + jk (K)Xi (K): (6)
2: ;
2fij;ikg
Here, Xijk (K) aij (K)[V (Ki ; Kj + Kk ; 0) V (Ki ; K i )] is the externality of the proposal
of merger Mjk on outsider …rm i, and so the last term on the right-hand side of (6) is the
expectation of the externality imposed on …rm i from Mjk . The interim value V i (K), which is
the …rst term on the right-hand side, is …rm i’s disagreement value in the bargaining with other
…rms. The second term is …rm i’s half of the expected merger gains (net of expected proposal
costs) from mergers involving …rm i. Note that this formula de…nes a mapping from interim
3
Recall that the two-…rm Nash bargaining process speci…ed in Section 2 (and used in Section 3) can equiva-
lently be represented as a non-cooperative bargaining game in which one of the two …rms is randomly selected
to make a take-it-or-leave-it o¤er to the other, so that it is nested in the three-…rm Burguet-Caminal bargaining
protocol speci…ed in Section 4.
4
values V i ( ) i2I to ex ante values fVi ( )gi2I . More generally, given a bargaining protocol
and a merger approval policy, the equilibrium bargaining strategies give rise to a mapping V
from interim values to ex ante values:
fVi ( )gi2I = V( V i ( ) i2I

) (7)
Consequently, (7) and (5) together implicitly de…ne the Bellman equation for the ex ante values
fVi ( )gi2I .
1.1.3 Markov Perfect Equilibrium
De…nition of Markov perfect equilibrium. Given the authority’s merger policy faij ( )gij2J ,
if the merger bargaining strategies f i ( )gi2I constitute a subgame perfect equilibrium of the
bargaining protocol given the interim values V i ( ) i2I , and if the …rms’ investment poli-
cies f i ( )gi2I , the …rms’ ex ante value functions fVi ( )gi2I ; and the …rms’ interim values
V i ( ) i2I satisfy equations (3), (5), and (7), then the collection (f i ( )gi2I ; f i ( )gi2I ;
fVi ( )gi2I ; V i ( ) i2I ) constitutes a Markov perfect equilibrium that policy faij ( )gij2J in-
duces.
Restriction to symmetric equilibria. Our models focus on symmetric environments,
in which a …rm’s static pro…t and investment cost distribution depend only on its capital level
Ki and the vector of rival capital levels K i . As well, any permutation of its rivals’ capital
stocks leaves …rm i’s static pro…t and investment cost distribution unchanged. In addition,
merger proposal costs and merger blocking costs are independent of the identities of the …rms
proposing a merger. Finally, the bargaining protocols we specify are symmetric in the sense
that a …rm’s opportunities do not depend on its identity.4
Given these symmetric environments, we restrict attention to symmetric (Markovian)
merger approval policies. A merger approval policy faij ( )gij2J is symmetric if for any state K,
there exists a single-valued function a( ) such that we can write aij (K) = a((Ki ; Kj ); K ij ) =
0
a((Ki ; Kj )p ; Kp ij ), where (Ki ; Kj )p is any permutation of the capital stocks Ki and Kj of the
0
two merging …rms, and Kp ij is any permutation of the capital stock vector of their rivals.
In addition, we restrict attention to Markov perfect equilibria for the …rms in which a …rm’s
investment policy and value function are symmetric, as are the merger proposal probability
functions arising from the merger bargaining protocol’s subgame perfect equilibrium. For-
mally, …rm i’s investment policy i is symmetric if i (ki jK) = (ki jKi ; K i ) = (ki jKi ; Kp i ),
where Kp i is any permutation of its rivals’ vector of capital stocks.5 A similar condition
de…nes symmetry for …rm i’s ex ante and interim value functions Vi ( ) and V i ( ): The equi-
4
For example, in the non-cooperative implememtation of the two-player Nash bargaining solution used in
Section 3, each …rm has a 1/2 probability of being the proposer.
5
Observe that a symmtric investment policy function gives rise to a symmetric capital stock transition
function for the …rm, satisfying i (Ki0 jK) = (Ki0 jK) = (Ki0 jKi ; Kp i ) for any permutation p.
5
librium outcome of the merger bargaining induces symemtric merger probability functions if
p 0
ij (K) = (Kij ; K ij ) = (Kij ; Kp ij ) for all permutations p and p0 .6
State transition matrix and industry steady state distribution. The investment
and depreciation stage transitions ( ) combined with the merger bargaining, merger approval,
and entry stage transitions T0 ( ) determine the transitions from the ex ante state in one period
to the ex ante state at the start of the next period. For example, when n = 2, given symmetric
merger policy a ( ) and a symmetric Markov perfect equilibrium that it induces, the probability
that the industry transitions from state K at the beginning of period t to state K0 at the
beginning of period t + 1 is
T K; K0 = (1 a12 (K) 12 (K)) (K10 jK1 ; K2 ; ) (K20 jK2 ; K1 ; )

1
+ a12 (K) 12 (K)f (K10 j0; K1 + K2 ; ) (K20 jK1 + K2 ; 0; )
2
+ (K10 jK1 + K2 ; 0; ) (K20 j0; K1 + K2 ; )g:
To calculate welfare measures and statistics of the industry’s dynamics we need the long-
run, steady state distribution that results from implementation of merger policy a ( ).
For example, consider again the case in which n = 2. Let : S 2 ! f1; 2; : : : ; (K + 1)2 g
be an invertible mapping that maps the two-dimensional matrix of states K into a vector of
2
states. Then, for every pair of states fK; K0 g 2 S 2 S 2 , de…ne the K + 1 (K + 1)2
transition matrix T ^ to have element T^ (!; ! 0 ) = T 1 (!) ; 1 (! 0 ) at row ! (the state
at the beginning of the period) and column ! 0 (the state at the beginning of the next period)
where state ! = (K) and state ! 0 = (K0 ).
Let P^ be a length (K + 1)2 row vector whose elements are non-negative and sum to one,
^ is a probability distribution on the state space S 2 transformed by : If P
i.e., P ^ T ^ = P;^ then
^ is a steady state distribution that the policy a ( ) induces over the industry’s state space. If
P
^ is unique, then, for any probability vector P,
P
^ = lim P T
P ^ ^ T
^ ^; (8)
t!1 | T {z T
}
t times
i.e., no matter what the initial probability distribution P on states is, the industry converges
^ 7 Rewrite P
to the steady state distribution P. ^ as a K + 1 (K + 1) matrix P where its
element in row (K1 + 1) and column (K2 + 1),
P (K1 ; K2 ) P^ [ (K)] ; (9)
is the steady state probability of the industry being in state K:

6
A merger bargaining protocol can be said to be symmetric if given any symmetric interim value functions and
any symmetric merger approval rule it induces symmetric merger proposal probability functions and symmetric
ex ante value functions.
7
In our model we cannot guarantee that, for some positive integer t; every element of T^t is positive, i.e., we
cannot guarantee that T^ is a regular Markov transition matrix. If it were regular, then P^ would be unique.
6
^ and P are more complex than these n = 2 examples,

While for n > 2 the formulas for T
their construction has the same basic structure.
1.1.4 Antitrust Policy and Welfare Metrics
In this section we specify the distinct choice problems that a “commitment” authority faces
and that a “no-commitment”authority faces. We then de…ne a variety of consumer value and
aggregate value welfare metrics that the antitrust authority may use as its objective function
W . Throughout the discussion ; ; V; V are the symmetric policy functions of the Markov
perfect equilibrium that a ( ) induces the …rms to follow.
Optimal commitment policy. The antitrust authority commits to a pure action aij (K) 2
f0; 1g for each possible merger Mij in each state K 2 S n so as to maximize either (i) ex ante
welfare W (K0 ) in a speci…c state K0 2 S n or (ii) some measure W of “average”ex ante welfare
across all states K 2 S n : For example, if n = 3; a policy to encourage the development of an
infant industry might maximize W (0; 0; 0): On the other hand, a general purpose policy for
mature industries might maximize average steady state welfare W SS where the ex ante wel-
fare W (K) of each state K 2 S n is weighted by its steady state probability P (K) : Observe
that the infant industry objective is a weighted average with weight one placed on ex ante
welfare in state (0; 0; 0). Therefore de…ne Z (fW (K)gK2S n ) to be whatever weighted average
the antitrust authority selects as its objective.
Let A be the class of admissible commitment policies a0 ( ) : Restricting A is necessary
because, even in the computationally easiest case of n = 2; the class of all possible symmet-
(K+1)K
ric commitment policies contains 2 2 elements. For K = 20, this makes the problem
computationally intractable. The optimal commitment policy a ( ) is therefore
a ( ) = arg max Z (fW (K)gK2S n ) (10)

a0 ( )2A
where the value of Z ( ) implicitly varies with the Markov perfect equilibrium that a0 ( ) induces
the …rms to play.
If the merger bargaining strategies f i ( )gi2I constitute a subgame perfect equilibrium
of the bargaining protocol given the interim values V i ( ) i2I , the …rms’ investment poli-
V i ( ) i2I satisfy equations (3), (5), and (7) for all states K 2 S n , and the merger approval
policy a ( ) satis…es (10), then the collection ; ; V; V and a( ) are respectively a Markov
perfect equilibrium for the industry and an optimal commitment policy for the “commitment”
antitrust authority.
Markov perfect policy. In this case, the antitrust authority acts instead as an additional
player that, unable to commit, makes its approval decision in every state K so as to maximize
its welfare criterion going forward, given the …rms’ Markov perfect equilibrium play in the
continuation game. The resulting policy a ( ) and the …rms’ equilibrium actions together
7
determine the welfare criterion’s ex ante and interim values, W (K) and W (K) for each state
K 2 S n.
A given merger policy a ( ) is a Markov perfect merger policy if at every state it satis-
…es the one-step deviation principle when the …rms play the industry Markov perfect equi-
librium ; ; V; V that approval policy a ( ) induces. Given that …rms i and j have pro-
posed to merge in state K and that the authority’s realization of its random blocking cost
is bij , welfare in the event that the authority approves the merger is W (K1 : : : ; Ki 1 ; Ki +
Kj ; : : : ; Kj 1 ; 0; Kj+1 ; : : : ; Kn ) W (Ki + Kj ; 0; K ij ) while welfare in the event that it blocks
it is W (K) bij : The authority chooses the maximum of the two, approving the merger if and
only if
bij W (Ki + Kj ; 0; K ij ) W (K) ij W (K) :
This results in a state-dependent, history-independent threshold bbij (K) that bij must ex-
ceed for the antitrust authority to approve merger Mij in state K:
bbij (K) = ij W (K) (11)
We call this a Markov perfect merger policy because in each period the antitrust authority
maximizes anew.
Recall that the blocking cost bij is a random variable with distribution H whose realization
is private to the antitrust authority. The …rms only know bbij (K) and H: Given the authority’s
decision rule, this means that in each state K …rms know that the probability of merger Mij
being approved is
aij (K) = 1 H(bbij (K)): (12)
If the merger bargaining strategies f i ( )gi2I constitute a subgame perfect equilibrium
of the bargaining protocol given the interim values V i ( ) i2I , the …rms’ investment poli-
V i ( ) i2I satisfy equations (3), (5), and (7) for all states K 2 S n , and the merger approval
policy a ( ) satis…es (12) for all states K, then the collection ; ; V; V and a are respec-
tively a Markov perfect equilibrium for the industry and a Markov perfect policy for the
“no-commitment” antitrust authority.
Consumer surplus, producer surplus, and aggregate surplus. For the several
de…nitions of welfare and cost measures that follow we restrict the analysis to the n = 2 case
because, as with the state transition matrix and the industry steady state distribution, the
formulas for the general case with n > 2 are complicated and contribute little insight.
If the ex ante state is K 2 S 2 and no merger occurs, then the consumer surplus realized is
CS (K), where Z 1
CS (K) D(s)ds;
P (Q(K))
where D( ) is the industry demand function, P ( ) D 1 ( ) is the inverse demand function,

and Q(K) is the total quantity in the Cournot equlibrium at state K. If merger M12 occurs
8
in the period, then the consumer surplus realized is CS (K1 + K2 ; 0). The expected consumer
surplus at the ex ante state K is therefore
ECS(K) = [1 a12 (K) 12 (K)] CS (K) + a12 (K) 12 (K)CS(K1 + K2 ; 0)
where a12 (K) 12 (K) is the probability merger M12 occurs. Similarly, expected producer sur-
plus at ex ante state K 2 S 2 is
EP S(K) = [1 a12 (K) 12 (K)] P S (K) + a12 (K) 12 (K)P S (K1 + K2 ; 0)
where P S (K) = (K1 ; K2 ) + (K2 ; K1 ). Aggregate surplus is the sum of consumer surplus
and producer surplus: AS (K) = CS (K)+P S (K). Consequently, in ex ante state K expected
aggregate surplus is EAS (K) = ECS (K) + EP S (K).
Consumer value and aggregate value. We generalize these static criteria to their
dynamic analogues, CV and AV , whose values are the ENPVs of consumer welfare and of
aggregate welfare respectively. Aggregate welfare accounts not only for consumer and producer
surplus at the Cournot competition stage, but also for investment costs, merger proposal costs,
and blocking costs.
Ex ante consumer value, CV (K) ; is the ENPV of current and future expected consumer
surplus. Its Bellman equation is
X X
CV (K) = ECS(K) + T K; K0 CV (K0 ):
K10 2S K20 2S
Interim consumer value is, for all states K,

X X
CV (K) = CS (K) + (K10 jK1 ; K2 ; ) (K20 jK2 ; K1 ; )CV (K0 )
K10 2S K20 2S
because consumer surplus is realized at the Cournot competition stage after any proposed
merger has been consummated.
Ex ante aggregate value AV (K) has four components: consumer value CV (K), the sum
of the incumbent …rms’ex ante values V1 (K) + V2 (K), the ENPV of all future entrants’cash
‡ows EEV (K), and the ENPV of the antitrust authority’s blocking costs EBC (K). Note that
the sum V1 (K) + V2 (K) fully accounts for the incumbents’ expected merger proposal costs
and expected capital investment costs. But neither CV (K) nor V1 (K) + V2 (K) includes the
last two components, EEV (K) and EBC (K) : We discuss each in turn.
Consider the ENPV of future entrants’cash ‡ows, EEV (K). A new …rm 1 (with probability
0.5 it could be …rm 2 instead) comes into existence at the entry stage of each period in which
a merger occurs. This new …rm’s interim value is V 1 (0; K1 + K2 ) where K1 + K2 is the merged
…rm’s capital level. In the ex ante state K = (K1 ; K2 ) the Bellman equation of the ex ante
ENPV of all future entrants’cash ‡ows is
X X
EEV (K) = a12 (K) 12 (K)V 1 (0; K1 + K2 ) + T K; K0 EEV (K0 ):
K10 2S K20 2S
9
In interim state K, the ENPV is

X X
EEV (K) = (K10 jK1 ; K2 ) (K20 jK2 ; K1 )EEV (K0 ):
K10 2S K20 2S
Next consider the ENPV of the antitrust authority’s blocking costs, EBC (K) : This de-
pends on whether the authority commits or not. The …rst case is for a commitment authority
that selects a policy a12 ( ) that, in each ex ante state K, speci…es either “approve”or “block”
with certainty, i.e., a12 (K) 2 f0; 1g : Given this commitment, each …rm knows that expending
resources proposing a merger when a12 (K) = 0 is hopeless because the authority will block
with probability 1. Consequently the authority never has to block a proposal, incurs zero
blocking costs, and EBC (K) = 0 for all K 2 S 2 .
For the second, no-commitment case, as explained above, in each ex ante state K the
authority sets a threshold bb12 (K) such that it blocks a proposed merger if and only if the
realization of its private, randomly distributed blocking cost b is less than bb12 (K) : Conditional
on a merger being proposed, the expected blocking cost in state K is
Z b
b12 (K)
E[bjK] = b dH(b):
b
where H is b’s distribution function that has support b; b . The Bellman equation for the ex
ante ENPV of blocking costs in ex ante state K is8
X X
EBC(K) = (K)E[bjK] + T K; K0 EBC(K0 ):
K10 2S K20 2S
In interim state K its value is

X X
EBC(K) = (K10 jK1 ; K2 ) (K20 jK2 ; K1 )EBC(K0 ):
K10 2S K20 2S
Given these de…nitions, ex ante aggregate value in ex ante state K is
AV (K) = CV (K) + V (K1 ; K2 ) + V (K2 ; K1 ) + EEV (K) EBC(K) (13)
and interim aggregate value in interim state K is
AV (K) = CV (K) + V (K1 ; K2 ) + V (K2 ; K1 ) + EEV (K) EBC(K) (14)
with the caveat that EBC(K) = EBC(K) = 0 if the antitrust authority employs a commitment
merger policy.
8
When n > 2 the expected blocking costs in state K is the expectation over expected blocking costs for each
possible merger given the various mergers’proposal probabilities.
10
Steady State Welfare. Given ex ante welfare function W ( ) the steady state, ex ante
average welfare the antitrust authority achieves under policy a ( ) is
X X
W SS = P K0 W K0
K10 2S K20 2S
where, as de…ned in equation (9), P (K0 ) is the industry’s steady state probability of being in
state K0 :
1.2 Details regarding Calculation of Merger Outcomes for the n = 2 and

n = 3 Cases
This section presents the details of the merger negotation calculations for two cases: duopoly
industry states K in which two …rms have capital stocks that are non-zero and triopoly industry
states K in which three …rms have non-zero capital stocks.
Merger proposals: duopoly industry states. Ex ante state K is a duopoly state if
and only if two …rms have capital stocks of at least one unit. Sections 2 speci…es that Nash
bargaining determines the outcome of merger negotiations in duopoly states: For ij 2 J , recall
that
ij (K) V (Ki + Kj ; 0) [V (Ki ; K i ) + V (Kj ; K j )];
is the joint gain from merger Mij gross of the proposal cost ij , and that
Sij (K; ij ) aij (K) ij (K) ij
denotes the expected bilateral surplus of …rms i and j from merging, conditional on the proposal
+
cost realization ij , and that Sij (K; ij ) max 0; Sij (K; ij ) . The …rms propose their
+
merger only if this surplus is positive, i.e., Sij (K; ij ) > 0. Proposal costs ij are distributed
indepedently with distribution function ( ) : Consequently, the ex ante probability of merger
Mij being proposed in ex ante state K is
ij (K) (aij (K) ij (K)) (15)
and the ex ante probability of a merger occurring is #ij (K) aij (K) ij (K). Nash bargaining
over the gains from merging implies that …rm i’s ex ante value is
1
V (Ki ; K i ) = V (Ki ; K i ) + ij (K) aij (K) ij (K) E ij jK (16)
2
where the interim value V (Ki ; K i ) is …rm i’s disagreement value, the term in curly brackets
is the merging …rms’expected net gain from proposing a merger (which they divide equally),
and R a(K) ij (K)
d ( )
E ij jK
ij (K)
11
is the expected proposal cost conditional on the merger being proposed. Equation (16) gives a
formula for V (Ki ; K i ) in terms of V (Ki ; K i ) and ij (K) where ij (K) itself is a function
of interim values. Consequently equation (16) together with equation (5) for V (Ki ; K i )
implicitly de…ne the Bellman equation for the ex ante value V (Ki ; K i ).
Merger proposals: triopoly industry states. Ex ante state K is a triopoly state if
three …rms have positive capital stocks. Paralleling the discussion of mergers in a duopoly
state, to characterize mergers in triopoly states we must derive merger proposal probabilities
ij (K) and write a formula for each …rm’s ex ante value V (Ki ; K i ).
Under the static Burguet and Caminal bargaining protocol that guides our analysis in
triopoly states …rm i is chosen to be the proposer with probability 1/3 and proposal costs
ij ; ik ; jk are independently drawn from the cumulative distribution function whose
support is [ ; ]: Let the joint density of the costs be 3 ij ; ik ; jk on the domain =
[ ; ] :3
Proposition 1 implicitly partitions into …ve regions that determine what merger proposals
are made, if any, in state K: Let ~ = ~ ij ; ~ ik ; ~ jk be the realization of the proposal costs.
De…ne the function ~ ; K that outputs the merger, if any, that is proposed to the antitrust
i
authority given the realized proposal costs ~ and the ex ante state K:
8
> + ~ ~ ~
> Mjk if Sjk (K; jk ) > maxfSij (K; ; ij ); Sik (K; ik )g
>
>
> + ~ + ~ + ~
>
< Mij if Sij (K; ij ) > Sik (K; ik ) Sjk (K; jk )
i
~; K Mij if Sij +
(K; ~ ij ) > Sjk
+
(K; ~ jk ) > Sik
+
(K; ~ ik ) & Sij
+
(K; ~ ij )=2 > Xijk (K) :
>
>
>
> Mjk if Sij +
(K; ~ ij ) > Sjk
+
(K; ~ jk ) > Sik
+
(K; ~ ik ) & Sij
+
(K; ~ ij )=2 < Xijk (K)
>
>
:
M? if Sij +
(K; ~ ij ) = Sik
+
(K; ~ ik ) = Sjk
+
(K; ~ jk ) = 0
where M? represent no merger proposed.
On the domain fMij ; Mik ; Mjk ; M? g S 3 de…ne the indicator function i ( ; ~ ; Kj i )
to have value 1 if i ~ ; K = and 0 otherwise. Conditional on …rm i being the randomly
selected proposer, the probability that merger M 2 (Mij ; Mij ; Mjk ; M? ) will be proposed is
Z
i
(K) = ~ ~ ~
i ( ; ; Kj i ) 3 ( )d :
where, for example, if = Mij we write iij (K) rather than iMij (K) ; etc. The ex ante
probability of merger M 2 (Mij ; Mik ; Mjk ; M? ) occuring in ex ante state K is then
1 P3 i
= (K) : (17)
3 i=1
where the 13 coe¢ cient is the probability each …rm has of being selected proposer. Proposal
costs are incurred whenever a merger is proposed. Therefore, conditional on …rm i being the
random proposer, expected proposal costs of mergers in which i is involved, are
8
< 0 if = M?
E i [ jK] = P R
~
i ( ; ; Kj i )
~ ~
3 ( )d
~ otherwise
:
2fMij ;Mik g
12
where if = Mij we write ~ ij ; etc. Ex ante, total expected proposal costs across all three
…rms are Z
1 X
E [ jK] = i( ; ~ ; Kj i ) ~ 3(
~ )d ~
2
2fMij ;Mik ;Mjk g
where the 21 coe¢ cent corrects for double counting.

The expected externality that …rm i 6= k; j realizes if merger Mjk occurs is
Xijk (K) ajk (K) V i (Ki ; Kj + Kk ; 0) Vi (Ki ; K i ) :
The ex ante value of …rm i in ex ante state K is therefore

8 9
1< X =
jk
V (Ki ; K i ) = V (Ki ; K i )+ E [ jK] + (K)a (K) (K) + jk (K)Xi (K)
2: ;
2fMij ;Mik g
(18)
where V (Ki ; K i ) is …rm i’s disagreement value, the second term is the expected merger gains
net of expected proposal costs that …rm i shares with its merger partners, and the last term is
the expectation of the externality …rm i realizes if merger Mjk occurs. Equation (16) gives a
formula for V (Ki ; K i ) in terms of V (Ki ; K i ) and ij (K) where ij (K) itself is a function
of interim values. Consequently equation (16) together with equation (5) for V (Ki ; K i )
implicitly de…ne the Bellman equation for the ex ante value V (Ki ; K i ).
1.3 Proof of Proposition 1

We begin with the following lemma.
Lemma 1 Suppose …rm i is selected as the proposer in state K. Further, suppose that …rm i
invites …rm j to enter merger negotiations. Then:
+ +
(i) If Sij (K; ij ) > Sjk (K; jk ), …rm j accepts the invitation and merger Mij gets proposed.
+ +
(ii) If Sij (K; ij ) < Sjk (K; jk ), …rm j declines the invitation and merger Mjk gets proposed.
+ +
(iii) If Sij (K; ij ) = Sjk (K; jk ) = 0, no merger gets proposed.
Proof. If …rm j accepts …rm i’s invitation, its expected continuation value is
1 +
V (Kj ; K j) + Sij (K; ij ):
2
If …rm j instead declines the invitation, it enters merger negotiations with …rm k, resulting in
an expected continuation value of
1 +
V (Kj ; K j) + Sjk (K; jk ):
2
13
+ +
Hence, …rm j strictly prefers accepting the invitation if Sij (K; ij ) > Sjk (K; jk ), and strictly
+ +
prefers declining it if the inequality is reversed. If Sij (K; ij ) = Sjk (K; jk ) = 0, no matter
whether …rms i and j or j and k enter into merger negotiations, no merger gets proposed as,
generically, both Sij (K; ij ) < 0 and Sjk (K; jk ) < 0 in that case.
Proof of Proposition 1.
+ + +
Part (i). Suppose Sjk (K; jk ) > maxfSij (K; ij ); Sik (K; ik )g. Lemma 1 implies that, no
matter whether …rm i invites …rm j or …rm k, that invitation gets declined, and merger Mjk
gets proposed.
+ + +
Part (ii). Suppose Sij (K; ij ) > Sik (K; ik ) Sjk (K; jk ). Lemma 1 implies that if
…rm i chooses to invite …rm j, then merger Mij gets proposed, yielding …rm i an expected
continuation value of
1 +
V (Ki ; K i ) + Sij (K; ij ):
2
+ +
If …rm i chooses to invite …rm k and Sik (K; ik ) > Sjk (K; jk ), then by the Lemma merger
Mik gets proposed, yielding …rm i an expected continuation value of
1 + 1 +
V (Ki ; K i ) + Sik (K; ik ) < V (Ki ; K i ) + Sij (K; ij ):
2 2
+ + +
If …rm i chooses to invite …rm k and Sik (K; ik ) = Sjk (K; jk ) = 0 (the case Sik (K; ik ) =
+
Sjk (K; jk ) > 0 generically does not occur), then by Lemma 1 no merger gets proposed,
yielding …rm i an expected continuation value of
1 +
V (Ki ; K i ) < V (Ki ; K i ) + Sij (K; ij ):
2
Hence, …rm i invites …rm j and merger Mij gets proposed.
+ + +
Parts (iii) and (iv). Suppose Sij (K; ij ) > Sjk (K; jk ) > Sik (K; ik ). From Lemma 1, if
…rm i chooses to invite …rm j, then merger Mij gets proposed, yielding …rm i an expected
continuation value of
1 +
V (Ki ; K i ) + Sij (K; ij ):
2
Similarly, if …rm i chooses to invite …rm k, then merger Mjk gets proposed, yielding …rm i an
expected continuation value of
V (Ki ; K i ) + IfS + (K; Xijk (K) = V (Ki ; K i ) + Xijk (K);

jk jk )>0g
+
where the equality follows from the fact that, by assumption, Sjk (K; jk ) > 0, implying that
merger Mjk would get proposed if …rm i were to invite …rm k (as …rm k would reject and invite
…rm j with whom …rm k has a larger and positive surplus). Hence, if Sij+
(K; ij )=2 > Xijk (K),
then …rm i invites …rm j and merger Mij gets proposed; if the inequality is reversed, then …rm
i invites …rm k and merger Mjk gets proposed.
Part (v) is immediate.
14
1.4 Computation
The algorithm that we use numerically to solve for equilibria is a version of the well-known
Pakes-McGuire (1994) algorithm. It is a straightforward iterative process. For a given merger
policy a ( ) the procedure works as follows. Pick an initial guess for the investment function
(0)
and the ex ante value function V (0) . Then compute an updated estimate of the investment
policy function (1) using equation (3). As this is a di¢ cult integral to evaluate, we use Monte
Carlo integration at each state K. Speci…cally, for a given vector c~ of random cost draws, the
ex ante value function V (0) , and the rival’s investment policy function (0) [which determines
the rival’s transition probabilities via equation (2)], we calculate …rm i’s optimal investment
decision ki for that instance of c~. Repeating this with many cost draws we use the proportion
of cost draws for which ki is optimal as our estimate of (1) ki jK1 ; K2 ; V (0) .
(1)
We then use equation (5) to calculate the interim value function V . Using this interim
value function and merger policy a ( ), we compute the merger proposal function (1) using
equation (15). Finally we calculate an updated ex ante value function V (1) using equation
(16).
Computation of the Markov perfect policy involves an additional step where we update
the antitrust authority’s merger policy a(1) ( ) using equation (12). We calculate W based on
the authority’s objective function and the state transitions induced by the …rms’investment
policy function (1) , merger proposal function (1) , and the authority’s intitial merger policy
a(0) ( ).
We iterate this process using the updated investment function (1) and the updated ex
ante value function V (1) as our starting point. We continue this iterative procedure until
V (`+1) V (`) " for some small " > 0.9
Computation for the model with three …rms is analogous except for updating the merger
proposal function . Because the solution to the bargaining process involves integrals which
are di¢ cult to evaluate, we use Monte Carlo integration, simulating proposal costs and the
selection of the intitial proposer.
A copy of the MatLab code and a document that describes the code and this algorithm in
more detail is available online.
2 Merger Policy in the Small and Large Markets

9
The distance metric we use combines absolute di¤erences and percentage di¤erences. For values less than
one we use absolute di¤erences, while for values greater than one we use percentage di¤erences. This is because,
for an " = 0:0001, we want a value of 0:001 and 0:0009 to be considered the same even though they have a
percentage di¤erence of 0:1 and we want a value of 1000 and 1000:1 to be considered the same even though
they have an absolute di¤erence of 0:1.
15
Figure 1: Static change in aggregate surplus for (a) the small market and (b) the large market.
Negative numbers are in parentheses.
In this section, we describe our results for the optimal merger policy in the small (A =
3; B = 22) and large (A = 3; B = 30) markets, and compare them to our results for the
intermediate (A = 3; B = 26) market found in the main paper. The static welfare e¤ects of
mergers are very similar in the three markets: in all of them only a merger in state (1; 1)
increases static consumer surplus, and in all of them, a merger in state (K1 ; K2 ) increases
static aggregate surplus unless both K1 and K2 are “large,”with the set of statically aggregate
surplus-increasing mergers being larger in larger markets. Figure 1 shows the set of aggregate
surplus-increasing mergers in the small and large markets.
Figures 2 through 7 show the steady state distributions and …ve-period transitions for
the small, intermediate and large markets when no mergers are allowed. When the antitrust
authority pursues instead an AV goal and cannot commit, the Markov perfect merger policy
results in mergers only in near-monopoly states in which the incumbent is su¢ ciently large.
The larger the market, the more restrictive is the antitrust authority in equilibrium. Figures 8
and 10 show the steady state distribution and probabilities that a merger happens in the small
and large markets, while Tables 1 and 2 provide some summary statistics of these equilibria.
The average merger probability is 30.6% in the small market, but only 3.0% in the large market
(versus 16.1% in the intermediate market). In the small market the industry is almost always
(98.6% of the time) in a monopoly state at the Cournot competition stage, compared to 49.4%
in the intermediate market, and only 8.2% in the large market. The equilibria involve larger
16
0.1
0.08
0.06
0.04
0.02
0
10
9
8
10
7 9
6 8
5 7
4 6
5
3 4
2 3
1 2
1
0 0
Figure 2: Beginning-of-period steady state distribution of the equilibrium generated with no

mergers in the small market. The height of each pin indicates the steady state probability of
that state.
capital levels as the market size grows.

Driving these di¤erences are the larger returns from capital additions that increased market
size provides. Figures 5 through 7 illustrate the strength of this e¤ect in the no-mergers-
allowed equilibria as the market size increases. In these …gures each arrow represents the
average movement over …ve periods starting in each state. The almost non-existent movement
toward duopoly from state (5; 0) in the small market evident in Figure 5, changes to robust
movement towards duopoly from state (5; 0) in the large market in Figure 7. Entry, without
the carrot of entry for buyout, is much more attractive and thefore a more e¤ective check on
monopoly in large markets. The antitrust authority therefore has an incentive to be more
aggressive in blocking mergers.
As in the intermediate market, if the antitrust authority pursues a CV goal and cannot
commit, the Markov perfect merger policies in the small and large markets are essentially
equivalent to the no-mergers policy.10 The same is true if it adopts the static consumer surplus-
based policy. In contrast, pursuing the static aggregate surplus-based policy is essentially
equivalent in outcome to allowing all mergers.
10
In the large market, the authority would approve mergers in states (1; 1), (2; 1), and (1; 2) but such mergers
are not value-enhancing for the …rms and therefore never proposed.
17
0.1
0.08
0.06
0.04
0.02
0
10
9
8
10
7 9
6 8
5 7
4 6
5
3 4
2 3
1 2
1
0 0
Figure 3: Beginning-of-period steady state distribution of the equilibrium generated with

no mergers in the intermediate market. The height of each pin indicates the steady state
probability of that state.
0.1
0.08
0.06
0.04
0.02
0
10
9
8
10
7 9
6 8
5 7
4 6
5
3 4
2 3
1 2
1
0 0
Figure 4: Beginning-of-period steady state distribution of the equilibrium generated with no

mergers in the large market. The height of each pin indicates the steady state probability of
that state.
18
10
0 1 2 3 4 5 6 7 8 9 10
Figure 5: Arrows show the expected transitions over 5 periods in the small market with no
mergers allowed.
10
0 1 2 3 4 5 6 7 8 9 10
Figure 6: Arrows show the expected transitions over 5 periods in the intermediate market with
no mergers allowed.
19
10
0 1 2 3 4 5 6 7 8 9 10
Figure 7: Arrows show the expected transitions over 5 periods in the large market with no
mergers allowed.
0.1
0.08
0.06
0.04
0.02
10
9
8
7 10
9
6 8
5 7
4 6
5
3 4
2 3
1 2
1
0 0
Figure 8: Beginning-of-period steady state distribution of the equilibrium generated by the

Markov perfect policy (AV criterion) in the small market. The height of each pin indicates
the steady state probability of that state. The shading of the cell re‡ects the probability of a
merger happening (with a darker grey representing a higher probability).
20
0.1
0.08
0.06
0.04
0.02
10
9
8
7 10
9
6 8
5 7
4 6
5
3 4
2 3
1 2
1
0 0

Markov perfect policy (AV criterion) in the intermediate market. The height of each pin indi-
cates the steady state probability of that state. The shading of the cell re‡ects the probability
of a merger happening (with a darker grey representing a higher probability).
Table 1: Performance Measures for the Small Market under Various Policies
No-Mergers/ Static- All- MPP- Comm.- Comm.-
Performance Measure 11
MPP-CV AS Mergers AV AV CV
Avg. Blocking Cost 0.0 0.0 0.0 0.0 0.0 0.0
Avg. Price 2.25 2.28 2.28 2.28 2.23 2.23
Avg. Quantity 16.5 15.8 15.8 15.9 16.9 16.9
Merger Frequency 0.0% 33.2% 33.9% 30.6% 6.8% 11.6%
% in Monopoly 58.2% 95.7% 95.2% 98.6% 68.6% 60.8%
% minfK1 ; K2 g 2 35.9% 0.1% 0.1% 0.3% 17.4% 32.3%
State (0,0) CV 24.0 18.9 18.8 19.4 21.8 24.1
State (0,0) AV 28.4 26.3 26.3 26.8 27.7 28.3
11
All values are ex ante (beginning-of-period) values except % in Monopoly and % minfK1 ; K2 g 2 which
are at the Cournot competition stage. “Static-CS” amd “Static-AS” refer, respectively, to the equilibria under
the optimal static consumer surplus-based and aggregate surplus-based merger policies. “MPP-CV”and “MPP-
AV” refer, respectively, to the equilibria when the antitrust authority cannot commit (resulting in a Markov
21
0.1
0.08
0.06
0.04
0.02
10
9
8
7 10
9
6 8
5 7
4 6
5
3 4
2 3
1 2
1
0 0
Markov perfect policy (AV criterion) in the large market. The height of each pin indicates
Table 2: Performance Measures for the Large Market under Various Policies
No-Mergers/ Static- All- MPP- Comm.- Comm.-
MPP-CV AS Mergers AV AV CV
Avg. Blocking Cost 0.0 0.0 0.0 0.0 0.0 0.0
Avg. Price 2.10 2.23 2.24 2.11 2.10 2.10
Avg. Quantity 27.0 23.0 22.9 26.7 27.0 27.0
% in Monopoly 2.3% 70.4% 68.4% 8.2% 2.3% 1.1%
% minfK1 ; K2 g 2 94.4% 3.6% 3.8% 87.9% 94.5% 95.5%
State (0,0) CV 36.4 30.0 29.9 35.5 36.5 36.4
State (0,0) AV 45.6 42.4 42.3 45.2 45.6 45.6
perfect policy) under consumer value and aggregate value welfare criteria. “Comm.-CV” and “Comm.-AV”
refer, respectively, to the equilibria when the antitrust authority commits to the optimal merger policy (within
the class described in Section 3 of this Online Appendix) for maximizing consumer value and aggregate value..
12
22
3 Commitment Policy
In our analysis in the main text we assumed that the antitrust authority, like each of the …rms,
acts as a player in a stochastic dynamic game, being unable to commit to its future policy. To
provide a benchmark for comparison, and also because it is of independent interest, we now
consider the optimal commitment policy, a state-dependent merger approval rule to which the
authority pre-commits before the game starts. For simplicity, we focus on the case n = 2.
We assume that the antitrust authority seeks to maximize the steady state level of expected
welfare, either CV or AV depending on the welfare criterion.13 In contrast to the Markov
perfect policy, the planner in the commitment case considers the impact his policy has on
…rms’strategies and, in particular, considers how …rms’investment behavior is a¤ected by the
prospects of future merger approvals.14 In our discussion, we will focus on the intermediate
market; the results for the small and large markets are summarized toward the end of the
section.
3.1 Feasible Policies

Formally, we assume that the antitrust authority pre-commits to a pure action aij (K) 2 f0; 1g
for each state K where aij (K) = 1 if the merger is approved and 0 if it is blocked. Observe
that there are 2100 possible deterministic symmetric merger policies. Thus, for computational
reasons, we restrict the space of admissible commitment policies to two classes.15
Her…ndahl-based policy. Under this type of policy, a proposed merger in state K is ap-
proved if and only if the induced change in the capital stock-based Her…ndahl index is below
a threshold H:
H(K) H([K1 + K2 ; 0]) H(K) H
where H(K) is the capital stock-based Her…ndahl index in state K and H is the authority’s
are at the Cournot competition stage. “Static-CS” amd “Static-AS” refer, respectively, to the equilibria under
the optimal static consumer surplus-based and aggregate surplus-based merger policies. “MPP-CV”and “MPP-
AV” refer, respectively, to the equilibria when the antitrust authority cannot commit (resulting in a Markov
perfect policy) under consumer value and aggregate value welfare criteria. “Comm.-CV” and “Comm.-AV”
refer, respectively, to the equilibria when the antitrust authority commits to the optimal merger policy (within
the class described in Section 3 of this Online Appendix) for maximizing consumer value and aggregate value.
13
This policy will generally di¤er from the policy that would be optimal given that the industry is starting
in a particular state (K1 ; K2 ). In addition to our primary analysis focusing on steady state welfare, we also
consider the commitment policy that maximizes the expected welfare of a “new” industry at state (0,0); see
footnote 19.
14
A second di¤erence is that under commitment the antitrust authority considers the impact its policy has
on proposal costs, while without commitment those costs are considered to be sunk at the time a merger is
reviewed. [A similar point arises in Besanko and Spulber (1992).]
15
The particular form these simple commitment policies take is partly motivated by which mergers are AV-
increasing as one-shot deviations. Note that to limit the number of feasible policies, we do not consider random
approval rules.
23
policy variable.16;17 For illustration, Figure 11(a) shows the policy H = 0:35 where states
with aij (K) = 1 are shaded (only states with maxfK1 ; K2 g 10 are shown), while Figure
11(b) shows the policy H = 0:2.
1. Capital-stock-based policy Under this type of policy, a proposed merger in state K is

approved if and only if K1 + K2 2 = (K; K) and minfK1 ; K2 g K i where K, K,
and K i are the authority’s policy variables. 18 Figure 11(c), for example depicts the
policy (K; K; K i ) = (4; 10; 1) where states with aij (K) = 1 are shaded (only states
with maxfK1 ; K2 g 10 are shown), while Figure 11(d) shows the policy (K; K; K i ) =
(10; 21; 1).
As observed earlier, under a commitment policy the antitrust authority never incurs any
blocking costs since if it commits to block a merger in state K the merger will not be proposed
in the …rst place.
3.2 Optimal Commitment Policy

In the intermediate market, the optimal commitment policy — for either a CV or AV standard
— is the Her…ndahl-type policy H = 0:225. For states in which each …rm has no more
than 10 units of capital, this policy involves approving a merger only when the smaller …rm
has one unit of capital and the larger …rm has at least seven units. Wherever a merger is
approved under this policy, it is also highly pro…table to the merging …rms and is proposed
with probability one. With mergers occurring only 3% of the time, this policy is fairly close
to the no-mergers-allowed policy.
Figure 12 shows the steady state distribution of the equilibrium induced by the optimal
commitment policy. Table 6 shows steady state averages of various performance measures for
this policy. The ability to commit leads to a 4% gain in AV compared to the Markov perfect
policy with the AV criterion, and a 2.5% gain in CV compared to the Markov perfect policy
with the CV criterion.19
16
To retain computational tractability we discretize the policy space: H 2 f0:075; 0:075 + ; 0:075 +
2 ; :::; 0:4 ; 0:4g, where = 0:025.
17
Because there are only two …rms, the post-merger Her…ndahl indices always equal one: H(K1 + K2 ; 0) =
H(0; K1 + K2 ) = 1, so H(K1 ; K2 ) = 1 H(K1 ; K2 ). Therefore a merger is approved if and only if
H(K1 ; K2 ) 1 H. Thus, under the Her…ndahl-based policy mergers are only approved if the beginning-of-
period Her…ndahl is su¢ ciently high.
18
To retain computational tractability we discretize the policy space: K 2 f2; 4; :::; 10; 12g, K 2
f6; 8; :::; 18; 20g and K i 2 f1; 2; :::; 6; 7g.
19
We also consider the optimal commitment policy for a new industry, which maximizes the welfare level (CV
or AV) at state (0,0). In searching for this policy, we identify …rst the state (0,0) welfare-maximizing policy
in the class of Her…ndahl-based or capital-stock-based commitment policies, and then allow the authority to
optimize fully for the states fKj0 Ki 4, i = 1; 2g. The rationale for the second step is that merger policy
at states with small capital levels is likely to be particularly important for maximizing welfare starting in state
(0; 0).
24
Figure 11: Panels (a) and (b) show Her…ndahl-based commitment policies, whereas panels (c)
and (d) show capital-stock-based commitment policies. (a) is H = 0:35, (b) is H = 0:2,
(c) is (K; K; K i ) = (4; 10; 1), (d) is (K; K; K i ) = (10; 21; 1). The shaded states are those in
which aij (K) = 1.
25
0.1
0.08
0.06
0.04
0.02
10
9
8
7 10
9
6 8
5 7
4 6
5
3 4
2 3
1 2
1
0 0
optimal commitment policy (AV and CV criteria) in the intermediate market. The height
of each pin indicates the steady state probability of that state. Cells in which mergers are
proposed and approved are darkly shaded.
26
0.4
0.3
0.2
0.1
10
9
8
7 4
6
5 3
4
2
3
2 1
1
0 0
Figure 13: Five-period transitions from state (5,0) under the optimal commitment policy. The
height of each pin indicates the probability of the industry being in that state. Cells in which
mergers are proposed and approved are darkly shaded.
Strikingly, even though mergers move the industry to a monopoly state, the industry
spends less time in a monopoly state (at the Cournot competition stage) with the optimal
commitment policy than under the no-mergers-allowed policy (14.3% vs. 18.6%), and capital
levels are higher (8.2 vs. 8.0). As can be seen in Figures 13 and 14, the reason there is
less monopoly is that the prospect of merger induces entrants to invest, but the limited set
of states in which mergers are allowed results in the industry often moving to symmetric
duopoly positions following these investments. Indeed, the probability that the industry is in
a monopoly state after …ve periods starting from state (5; 0) is much lower than under the
no-mergers policy: 0.45 vs. 0.84. The greater movement to symmetric, duopolistic states
from monopoly ones can also be seen by comparing Figure 15 to Figure 6.
While full commitment to a policy may be di¢ cult to achieve, an alternative is to endow
the antitrust authority with an objective that may not be the true social objective. In this
regard, note that the steady state level of AV under the Markov perfect merger policy when
the antitrust authority has a CV objective (essentially the no-mergers-allowed outcome) is
The optimal commitment policy starting from state (0,0) allows mergers in very few states. For the AV
objective, the authority allows mergers only in states K such that Ki 2 f1; 2g, i = 1; 2. However, as a merger
in state (1; 1) is never [and in states (1; 2) and (2; 1) only rarely] pro…table, this is almost equivalent to allowing
mergers only in state (2; 2). The resulting AV (resp. CV) level is 37.1 (26.6), whereas under the no-mergers
policy it is 36.7 (30.3). For the CV objective, the state (0,0) optimal commitment policy is a no-mergers policy.
27
0.4
0.3
0.2
0.1
0
8
7
6 2
5
4
3 1
2
1
0 0
Figure 14: Five-period transitions from state (5,0) under the no-mergers-allowed policy. The
height of each pin indicates the probability of the industry being in that state.
10
0 1 2 3 4 5 6 7 8 9 10
Figure 15: Arrows show the expected transitions over 5 periods under the optimal commitment
policy.
28
0.1
0.08
0.06
0.04
0.02
10
9
8
7 10
9
6 8
5 7
4 6
5
3 4
2 3
1 2
1
0 0
best commitment policy (CV criterion) in the small market. The height of each pin indicates
higher than that when it has an AV objective. Thus, when the antitrust authority cannot
commit, a CV-maximizing antitrust authority is better for AV in this market than an AV-
maximizing authority. This is consistent with a suggestion of Lyons (2002), but arises because
of the policy’s e¤ect on investment, rather than by inducing a socially more desirable choice
of merger partner as in Lyons (2002).
3.3 Commitment Policy in the Small and Large Markets

We now brie‡y summarize our results for the optimal commitment policy in the small (A =
3; B = 22) and large (A = 3; B = 30) markets, and compare them to our results for the
intermediate (A = 3; B = 26) market.
If the antitrust authority pursues a CV goal, then the optimal commitment policy in all
three markets involves approving mergers only in near-monopoly states in which the incumbent
is su¢ ciently large. This policy is more restrictive the larger is the market, with the merger
probabilities ranging from 0.1% in the large market to 11.6% in the small market (see Tables
1 and 2). Figures 16 and 17 show the steady state distributions and optimal merger policy
for the small and large markets.
If the antitrust authority pursues an AV goal instead, its optimal commitment policy is
29
0.1
0.08
0.06
0.04
0.02
10
9
8
7 10
9
6 8
5 7
4 6
5
3 4
2 3
1 2
1
0 0
best commitment policy (CV criterion) in the large market. The height of each pin indicates
30
0.1
0.08
0.06
0.04
0.02
10
9
8
7 10
9
6 8
5 7
4 6
5
3 4
2 3
1 2
1
0 0
best commitment policy (AV criterion) in the small market. The height of each pin indicates
essentially to approve no mergers in the large market. In the small market, however, it does
approve mergers in states in which both …rms are su¢ ciently large (resulting in a merger
probability of 6.8%), which boosts …rms’ investment incentives (resulting in an almost 10%
higher capital level compared to the AV-maximizing Markov perfect policy). Figures 18 and 19
show the steady state distributions and optimal merger policies for the two markets. Observe
that the optimal commitment policy is more restrictive in larger markets even though the set
of states in which mergers increase static aggregate surplus is larger in larger markets.
Independently of whether the authority pursues a CV or AV objective, the advantage that
commitment has over no commitment is decreasing (both in absolute as well as in relative
terms) with the size of the market. For example, compared to the AV-maximizing Markov
perfect policy, the AV-maximizing commitment policy induces a steady state average AV that
is 6.7% higher in the small market but only 0.8% higher in the large market.
4 Extensions and Robustness

In this section, we investigate several extensions and robustness issues. Section 4.1 investigates
how changes in the ease of entry a¤ect the optimal merger policy. Section 4.2 examines the
31
0.1
0.08
0.06
0.04
0.02
10
9
8
7 10
9
6 8
5 7
4 6
5
3 4
2 3
1 2
1
0 0
best commitment policy (AV criterion) in the large market. The height of each pin indicates
32
e¤ects of reducing the di¤erence in investment costs between incumbents and entrants. Section
4.3 considers the equilibrium when a planner controls investment and merger decisions. Section
4.4 considers a modi…cation to the model where the entrant is the previously bought-out …rm’s
owner. Section 4.5 examines changing bargaining power from an equal weighting to a capital-
weighted bargaining power. Section 4.6 looks at the robustness of our results for various
production scale parameters. Finally, Section 4.7 looks at the robustness of our results for
various ranges of investment costs. Throughout this section, we focus on the duopoly case
(n = 2).
4.1 Ease of Entry

It is generally perceived that the potential anticompetitive e¤ects of horizontal mergers are
mitigated when entry into the industry is easy. For instance, the current (2010) U.S. Horizontal
Merger Guidelines (which are largely based on a consumer welfare standard) state:
A merger is not likely to enhance market power if entry into the market is so easy
that the merged …rm and its remaining rivals in the market, either unilaterally
or collectively, could not pro…tably raise price or otherwise reduce competition
compared to the level that would prevail in the absence of the merger. Entry is
that easy if entry would be timely, likely, and su¢ cient in its magnitude, character,
and scope to deter or counteract the competitive e¤ects of concern.
To study how the ease of post-merger entry a¤ects optimal merger policy and the resulting
performance of the industry, we extend the baseline model in two ways: …rst by introducing a
probability e 0 that a new entrant arrives at the entry stage whenever the current state of
the industry has a single active …rm, and second by introducing a minimum scale K g > 1 for
green…eld investment. We focus on the intermediate market and the AV criterion. Contrary
to the conventional view, we …nd that in both cases optimal merger policy may become more
permissive when entry becomes more di¢ cult.20
4.1.1 Timeliness of Entry
Consider, …rst, the timeliness of entry following a merger. Table 3 reports the performance
measures of the intermediate (A = 3; B = 26) market under the Markov perfect policy with an
AV welfare criterion for di¤erent levels of the entry probability e.21 Despite the ine¢ ciencies
20
For a similar observation in a static context, see Whinston (2007). While new entry is generally viewed as
being price-reducing and thus bene…cial to consumers, it may be excessive from an aggregate welfare point of
view [Mankiw and Whinston (1986)].
21
Formally, this requires extending the state space to S 0 f 1; 0; 1; :::; 20g2 , where Ki = 1 means that …rm
i is an entrant who has not yet arrived. The …rms’expected gain from merging is therefore now given by
G (K1 ; K2 ) = eV (K1 + K2 ; 0) + (1 e)V (K1 + K2 ; 1) [V (K1 ; K2 ) + V (K2 ; K1 )];
33
associated with entry for buyout, welfare declines as entry becomes less timely: the steady
state levels of CV and AV fall from 43.3 and 113.6, respectively, to 28.0 and 98.5 as e decreases
from 1 to 0. The reason for this …nding is that, as e decreases, the industry spends more and
more time in a monopoly state: the steady state probability of monopoly increases from 49.4%
at e = 1 to 100% at e = 0. This hurts consumers and society a lot in the short run (for a given
level of capital) but even more so in the long run because a monopolist has little incentive to
build capital in the absence of a threat of entry: the average total capital level decreases from
7.7 to 5.3 as e decreases from 1 to 0.
Table 3: Timeliness of Entry and Markov Perfect Policy Outcomes

(Intermediate Market, AV Criterion)
Performance Measure 22 e=1.0 e=0.8 e=0.6 e=0.4 e=0.2 e=0.0
Avg. Blocking Cost -0.1 0.0 0.0 0.0 0.0 0.0
Avg. Price 2.19 2.21 2.25 2.29 2.32 2.35
Avg. Quantity 21.0 20.6 19.6 18.5 17.7 16.9
% in Monopoly 49.4% 58.9% 83.6% 99.4% 100% 100%
% minfK1 ; K2 g 2 44.2% 35.3% 13.1% 0.1% 0.0% 0.0%
Table 3 also reveals that the frequency of mergers is non-monotonic in the timeliness of
post-merger entry: as e decreases from our base case of e = 1, the probability that a merger
occurs in a randomly selected period …rst increases (from 16.1% at e = 1 to 36.7% at e = 0:4)
and then decreases. As a merger is infeasible in states in which there is only one active …rm,
this steady state weighted merger probability is equal to the probability that there are two
active …rms times the probability of a merger conditional on two …rms being active, and is
bounded from above by the entry probability e.23 This explains why the merger frequency
converges to zero as the entry probability e becomes small.
where the …rst (second) term on the right-hand side is the probability of new entry (no new entry) occurring
times the continuation value of the merged …rm in that event.
22
All values are ex ante (beginning-of-period) values, while the performance measures in the last two rows
are at the Cournot competition stage.
23
The steady state weighted merger probability is maximized when the probability of a merger, conditional
on there being two active …rms, is equal to one. In that case, the probability that there are two active …rms
is equal to the entry probability e, implying that the steady state weighted merger probability is equal to e as
well.
34
To understand why the merger frequency increases as e decreases from 1 to 0.4, consider
the merger probability conditional on two …rms being active, which is the product of two
probabilities: the probability that the two active …rms propose a merger and the probability
that a proposed merger is approved.
Consider …rst states in which both …rms have at least one unit of capital. As e decreases,
mergers become more pro…table in such states as the merged …rm spends more time in a
monopoly state before a new entrant appears. Moreover, the AV-maximizing Markov perfect
policy tends to become less restrictive as e declines, re‡ecting the reduced entry for buyout
behavior. When the entry probability e is high, the Markov perfect policy approves mergers
only in states in which at least one of the …rms is su¢ ciently small (as we have seen for e = 1
in Section 3.3 of the main paper). As e decreases, this approval region increases. For example,
a proposed merger in state (3,3) is never approved if e 0:6 but always approved if e = 0:4.
Consider now states in which an entrant has arrived yet has no capital. When the entry
probability is one, the authority would always approve a proposed merger in such a state:
approving the merger has no e¤ect on AV, but blocking is costly. However, when e = 1, such a
merger would not be proposed as it is not pro…table.24 When the post-merger entry probability
is su¢ ciently small, such a merger becomes pro…table as the arrival of a new entrant following
a merger takes time, allowing the merging …rms to reap monopoly pro…ts in the meantime.
As e decreases, …rms are therefore more likely to propose mergers between an entrant and
an incumbent. At the same time, while the antitrust authority starts to block mergers, it
allows some proposed mergers. Hence, the probability of a merger between an entrant and an
incumbent becomes positive for e 0:6.25
4.1.2 Minimum Scale for Green…eld Investment
We now explore a di¤erent way in which entry may become more di¢ cult. Speci…cally,
we extend the model by introducing a minimum size for green…eld investment, K g , focusing
again on the intermediate market. As incumbents rarely use the green…eld technology, this
essentially amounts to introducing a minimum scale of entry.
Table 4 shows the same performance statistics for K g ranging from 1 (our base case) to 5.
As with reductions in the timeliness of entry, a larger minimum scale of green…eld entry raises
the likelihood of being in a monopoly state, and has a non-monotonic e¤ect on the probability
of merger. Similar to cases in which e approaches zero, as K g grows large the probability of
merger declines because the likelihood of post-merger entry grows small; nonetheless, as K g
24
When e = 1, a merger in state (K1 ; 0) or (0; K2 ) does not a¤ect producer value because the old entrant
gets immediately replaced by a new entrant. As the value of the new entrant is strictly positive, this implies
that the merger must decrease the joint continuation values of the merging …rms.
25
If the antitrust authority adopts a CV standard instead of an AV standard, the relationship between the
timeliness of entry and the steady state probability of a merger remains non-monotonic: as e decreases, the
merger frequency …rst increases and then decreases.
35
grows, both the set of states in which mergers are permitted by the antitrust authority and the
set of states in which mergers are proposed grow larger. However, in contrast to a reduction
in the timeliness of entry, aggregate value shows relatively small and non-monotonic changes
as K g rises. The reason for this di¤erence is that the aggregate capital in the market does
not fall as K g gets larger, in contrast to the case when e gets small. This occurs because
while the likelihood of entry grows smaller as K g grows, when entry does occur it is at a
larger scale, and the incumbent monopolist is incented to invest to reduce this possibility and
get better merger terms when entry does occur.
Table 4: Minimum Scale of Green…eld Investment and Markov Perfect

Policy Outcomes (Intermediate Market, AV Criterion)
Performance Measure 26 Kg = 1 Kg = 2 Kg = 3 Kg = 4 Kg = 5
Avg. Consumer Value 43.3 39.6 37.8 38.5 35.6
Avg. Incumbent Value 69.9 73.1 73.8 76.8 76.6
Avg. Entrant Value 0.5 0.3 0.2 0.1 0.0
Avg. Blocking Cost -0.1 -0.1 0.0 0.0 0.0
Avg. Aggregate Value 113.6 113.0 111.8 115.4 112.3
Avg. Price 2.19 2.22 2.24 2.23 2.26
Avg. Quantity 21.0 20.2 19.7 19.9 19.2
Avg. Total Capital 7.7 7.6 7.6 7.9 7.1
Merger Frequency 16.1% 16.4% 15.6% 8.8% 3.5%
% in Monopoly 49.4% 82.3% 97.0% 99.5% 99.8%
% minfK1 ; K2 g 2 44.2% 17.5% 2.6% 0.5% 0.2%
4.2 Entrant Investment E¢ ciency
In our analysis of the welfare e¤ects of various merger policies, “entry for buyout” plays
a prominent role. When mergers are allowed a new entrant’s private bene…t from investing
signi…cantly exceeds the incremental aggregate value that results from those investments, while
the incremental aggregate value from an incumbent’s investment exceeds its private bene…t
to the incumbent. As a result, the entrant invests too much and the incumbent invests too
little. The entrant’s high cost green…eld investment substitutes for the incumbent’s lower cost
investment done through capital augmentation and directly causes waste.
In practice, however, entrants’investments are not always less e¢ cient than incumbents’
investments, and may sometimes even be more e¢ cient.27 In this subsection, we explore
26
All values are ex ante (beginning-of-period) values, while the performance measures in the last two rows
27
Henderson (1993) provides evidence of this in the photolithographic alignment equipment industry where
several generations of entrants supplanted incumbents by more e¢ ciently using their knowledge capital.
36
this point by changing the model’s parameters to close the gap between the investment costs
entrants and incumbents face.
Focusing on the intermediate market, we examine whether this change largely eliminates
the waste that entry for buyout causes by studying the e¤ect of a change from the no-mergers-
allowed policy to the all-mergers-allowed and Markov perfect policies when the antitrust au-
thority’s criterion is AV maximization. Overall, we …nd that (i) entry for buyout behavior
continues to be prevalent, (ii) its social costs are greatly reduced; (iii) the antitrust authority
is much more willing to allow mergers in the Markov perfect policy; and (iv) with this change,
consumer value falls somewhat more when moving from no-mergers-allowed to the Markov
perfect policy.
Recall that capital augmentation each period enables a …rm with K units of capital, if
it wishes, to double each unit j at a cost cj drawn independently and uniformly from the
interval [c,c] : If it wants to more than double its current stock of capital, then it can purchase
additional green…eld units at constant unit cost cg , where cg is uniformly drawn from [c; cg ].
Let s = c c and sg = cg c be the spread of capital augmentation costs and green…eld
costs respectively. In the baseline industry analyzed in the previous sections the values are
c = 3; c = 6; cg = 7; s = 3; and sg = 1: To close the gap between entrant and incumbent
investment costs we reduce s to 1 and sg to 0:25. Since this change, if c were held …xed, would
reduce …rms’investment costs, leading to less monopoly and very di¤erent merger behavior,
we simultaneously raise c to 4:645, which keeps the frequency of monopoly unchanged when
no mergers are allowed. Thus, we have c = 4:645, c = 5:645, cg = 5:895; we refer to these
modi…ed parameter values as the “e¢ cient entry environment.”
Table 5 shows the results when we switch from our baseline environment to the e¢ cient
entry environment. The table reports the same performance statistics as before, with the
addition of one new measure: “Avg. Monop. to Merger Time.” This statistic measures the
expected number of periods the industry takes to transition from a monopoly state to a state in
which the incumbents merge.28 Comparing the two environments, we see that entry for buyout
behavior actually increases when we move to the e¢ cient entry environment; for example, when
all mergers are allowed, the monopoly to merger time falls from 2.6 to 2.1. However, the costs
of this behavior are greatly reduced: AV now falls only 0.6% when all mergers are allowed
(from 87.9 with no mergers to 87.4 when all mergers are allowed), compared to 10.0% in our
baseline case (from 117.5 with no mergers to 105.8 with all mergers allowed). Because of
the reduction in the ine¢ ciency of pre-merger investment behavior, allowing mergers is much
more attractive for the antitrust authority, and the Markov perfect policy results in far more
mergers in the e¢ cient entry environment: the probability of merger is now 42.6% in each
period, compared to only 16.1% in our baseline case. Indeed, the equilibrium is essentially
equivalent to the case in which all mergers are allowed. Finally, this increased merger activity
results in a much greater likelihood of the industry being in a monopoly state (79.4% of the
28
We use the steady state distribution over monopoly states as weights, and exclude state (0; 0).
37
time in the e¢ cient entry environment vs. 49.4% in our baseline case). As a consequence,
there is a somewhat greater reduction in consumer value when moving from no mergers being
allowed to the Markov perfect policy (a reduction of 13.6%, from 34.9 to 30.5 in the e¢ cient
entry environment, vs. a reduction of 10.0%, from 48.1 to 43.3).29
Table 5: Performance Measures for the E¢ cient Entry Environment

in the Intermediate Market
Baseline Environment E¢ cient Entry Environment
No- All- MPP- No- All- MPP-
Mergers Mergers AV Mergers Mergers AV
Avg. Entrant Value - 1.9 0.5 - 2.0 2.0
Avg. Blocking Cost - - -0.1 - - 0.0
Avg. Price 2.15 2.26 2.19 2.27 2.32 2.32
Avg. Quantity 22.2 19.2 21.0 18.9 17.7 17.7
% in Monopoly 18.6% 86.0% 49.4% 18.6% 79.4% 79.4%
% minfK1 ; K2 g 2 75.7% 0.9% 44.2% 68.6% 4.2% 4.2%
Avg. Monop.
- 2.6 6.1 - 2.1 2.1
to Merger Time
4.3 The Planner’s Solution
We consider the second-best dynamic problem where the planner controls both …rms’merger
decisions (independent of their private pro…tability) as well as their investment decisions (as-
suming the planner has perfect information about …rms’private cost draws), taking as given
only that, in every period, …rms compete in a Cournot fashion. The analysis provides a
29
The greater percentage reduction in consumer value in the Markov perfect policy compared to when no
mergers are allowed depends on market size. In results not reported here, we …nd that it remains true in the
large market, but in the small market there is no reduction in consumer value from allowing mergers in the
e¢ cient entry environment. The other e¤ects we report here for the intermediate market (continued entry for
buyout behavior, reduced cost of that behavior, and greater frequency of mergers) hold as well in the small and
large markets. Figures 26-28 in this Online Appendix show these e¤ects for a broader range of investment costs
for incumbents and entrants, focusing on the intermediate market. Speci…cally, for each combination of s 2 [1; 3]
and sg 2 [0:25; 1] we …nd the level c(s; sg ) at which the percentage of time in a monopoly state is 18.6% in the
no-mergers-allowed equilibrium. The …gures show, respectively, the average time from monopoly to merger
in the all-mergers-allowed equilibrium (Figure 26), (AVN o AVall )=AVN o (Figure 27), and the probability of
merger in the Markov perfect policy (Figure 28) for each economy (s; sg ; c(s; sg )).
30
38
0.12
0.1
0.08
0.06
0.04
0.02
12
11
10
9 12
8 11
7 10
6 9
8
5 7
4 6
3 5
4
2 3
1 2
0 1
0
Figure 20: Solution to the planner’s second-best problem (AV criterion) in the intermediate
market. The height of each pin gives the probability of the corresponding state in the steady
state generated by the planner’s optimal policy. The shading of the cells indicates the merger
probabilities, with a darker shading corresponding to a higher merger probability.
benchmark for how optimal merger policy would look absent concerns about the e¢ ciency of
investment behavior. In our analysis, we con…ne attention to the AV criterion.31
Figure 20 shows the steady state distribution for the solution of this second-best problem
in the intermediate market: the height of each pin gives the beginning-of-period probability of
the corresponding state in the steady state generated by this policy; the cells in which mergers
are approved are darkly shaded. As the planner controls not only merger decisions but also
…rms’investment decisions, the planner does not face a time inconsistency problem; i.e., the
solution is independent of whether or not the planner can commit to his future decisions.32
As Figure 20 shows, in the steady state generated by the planner’s solution, the industry
is always in a monopoly state. A merger is implemented in many states, unless these states
involve high capital levels for both …rms. In fact, the set of states in which mergers happen is
almost identical to the set of states in which a merger is statically aggregate surplus-increasing
31
The second-best solution is not well-de…ned for the CV criterion as consumers always bene…t from larger
capital stocks.
32
The existence of blocking costs is irrelevant for the solution to the second-best problem as it can never be
optimal from the planner’s point of view to propose a merger and subsequently block it in the event blocking
costs are su¢ ciently low.
39
(for reasons that will be discussed below).33
Table 6: Performance Measures for the Intermediate Market under Various Policies
No-Mergers/ All- MPP- Commitment
Performance measure 34 Planner
MPP-CV Mergers AV (CV and AV)
Avg. Consumer Value 48.1 35.8 43.3 49.3 39.2
Avg. Incumbent Value 69.4 68.1 69.9 68.8 82.1
Avg. Entrant Value 0.0 1.9 0.5 0.0 0.0
Avg. Blocking Cost 0.0 0.0 -0.1 0.0 0.0
Avg. Aggregate Value 117.5 105.8 113.6 118.1 121.3
Avg. Price 2.15 2.26 2.19 2.14 2.23
Avg. Quantity 22.2 19.2 21.0 22.5 20.1
Avg. Total Capital 8.0 7.0 7.7 8.2 8.1
Merger Frequency 0.0% 37.7% 16.1% 3.0% 0.0%
% in Monopoly 18.6% 86.0% 49.4% 14.3% 100.0%
% minfK1 ; K2 g 2 75.7% 0.9% 44.2% 78.8% 0.0%
State (0,0) CV 30.3 23.9 25.6 30.4 25.3
State (0,0) AV 36.7 34.0 35.5 36.7 41.8
The fact that in the second-best solution the industry is always in a monopoly state may be
surprising at …rst. After all, when mergers are not allowed the industry seems to be a workable
duopoly. The reason is closely related to the fact that mergers are frequently aggregate surplus
increasing, given our chosen parameters. To understand this point, suppose …rst that the
planner could not only control mergers but also costlessly undo previously approved mergers.
Suppose also that there were no merger proposal costs. What would the planner’s optimal
policy be in that case? In any state (K1 ; K2 ), the planner would optimally implement a merger
if and only if the merger increases static aggregate surplus as this is statically optimal and
also does not impede dynamic optimality as the planner controls investment, the investment
technology is merger neutral, and the planner can costlessly undo any previously approved
merger. In Figure 5 of the main paper we saw that a merger increases static aggregate surplus
in every state in which K1 + K2 10 (except in state (5; 5) in which the gain is approximately
zero) and, also, in several additional states in which K1 + K2 > 10. So, unless the planner
wants to spend a large amount of time in states with more than 10 units of capital, the steady
state generated by the planner’s policy will visit only monopoly states even if the planner
cannot undo previously approved mergers and there are proposal costs — which is what is
going on here.35 Finally, note that this reasoning also explains why the set of states in which
33
For comparison with the optimal merger policy (with and without commitment), performance measures of
the planner’s solution are provided in the last column of Table 6.
34
35
In the steady state generated by the planner’s solution, the industry is sometimes (8.3% of the time) in a
monopoly state with more than 10 units of capital, the joint frequencies of states (11; 0) and (0; 11) being 6.1%.
40
the planner implements mergers almost coincides with the set of statically aggregate surplus-
increasing mergers. They do not coincide fully because of the presence of merger proposal
costs, which the static criterion does not take into account.
The results in the small and large markets are very similar: in both markets, the AV-
maximizing second-best solution involves monopoly all of the time.
4.4 Entrant Identity

A key restriction in the duopoly model analyzed in Section 3 of the main paper is that no
more than two …rms can be active at any one time. Throughout this restriction has been posed
exogenously. Our baseline assumption is that the entering …rm after a merger is owned by an
entrepreneur who has never before been active within the industry. This assumption begs the
question as to why he did not enter previously before the merger took place.
An alternative to the exogenous restriction we have used is to assume that only two en-
trepreneurs have the necessary skill and knowledge set to compete in the industry. If that is
the case and both entrepreneurs are active in the industry, then the owner/manager of the
acquired …rm would become the new entrant following a merger. (We assume there is not a
“no-compete”clause in the acquisition agreement.) Equation (1) in the main paper giving the
joint value gain from merging then becomes
12 (K1 ; K2 ) V (K1 + K2 ; 0) + V (0; K1 + K2 ) V (K1 ; K2 ) + V (K2 ; K1 ) :
New to the de…nition is the entrant’s ex ante value V (0; K1 + K2 ). It must be included because
the entrepreneur who is bought out intends to re-enter. In other words, the two entrepreneurs
will agree to merge— one buying out the other— if it pays them jointly to create temporarily
a monopoly situation in the industry until that time the bought-out entrepreneur successfully
returns to the industry. Since V (0; K1 + K2 ) 0 this weakly increases the merger frequency
(holding the policy and value function constant). Table 7 shows a side-by-side comparison for
the intermediate market of the equilibria for these two di¤erent assumptions concerning entry.
When all mergers are allowed, this change increases the frequency of mergers. (Although note
that in the AV-maximizing Markov perfect policy the merger frequency ends up lower than
before.) Inspection shows that, overall, our results are not qualitatively di¤erent from our
earlier results.
But these are both states that are reachable by aggregate surplus increasing mergers.
41
Table 7: Performance Measures when the Bought Firm is the Entrant

in the Intermediate Market
New Entrant Bought is Entrant
No- All- MPP- No- All- MPP-
Mergers Mergers AV Mergers Mergers AV
Avg. Entrant Value - 1.9 0.5 - - -
Avg. Blocking Cost - - -0.1 - - -0.0
Avg. Price 2.15 2.26 2.19 2.15 2.26 2.17
Avg. Quantity 22.2 19.2 21.0 22.2 19.2 21.5
% in Monopoly 18.6% 86.0% 49.4% 18.3% 94.0% 35.9%
% minfK1 ; K2 g 2 75.7% 0.9% 44.2% 76.0% 0.1% 57.0%
Avg. Monop.
- 2.6 6.1 - 2.6 8.5
to Merger Time
4.5 Capital-weighted Bargaining Power

In the main paper, we have assumed that …rms split the surplus from merging equally when
n = 2. Here, we explore the case where the surplus division in Nash bargaining is proportional
to the merging …rms’capital stocks, i.e., in state (K1 ; K2 ), …rm i gets a share Ki =(Ki + K i ).
When a …rm expects to merge in the future, capital-weighted bargaining power provides
it with an additional incentive to add capital, holding …xed the rival’s investment. Consider,
for example, state (5; 5) under the all-mergers-allowed policy. Moving from equal bargaining
weights to capital-weighted bargaining power increases each …rm’s expected investment from
1.0 to 1.4. In monopoly states, however, the entrant faces a countervailing incentive because
(i) it will capture only a small fraction of the surplus from merging and (ii) the incumbent
invests more than under equal bargaining weights. As a result in monopoly states in which the
incumbent has more than …ve units of capital, which represents nearly all of the steady state
at the investment stage, the change in the division of bargaining power decreases the entrant’s
expected investment. The distortion in …rms’investment incentives due to entry for buyout is
thus mitigated when the division of bargaining power is proportional to …rms’capital stocks.
Table 8 provides the performances measures for the intermediate market under the no-
mergers, all-mergers-allowed and AV-oriented Markov perfect policies. As before, the no-
mergers policy achieves the highest average aggregate value while the all-mergers-allowed
policy performs worst. However, because of the changed investment incentives under the
capital-weighted division of bargaining power, the latter policy does not perform quite as
36
42
badly as before: Compared to the case of equal bargaining weights, the average capital stock
is considerably larger (7.5 instead of 7.0), resulting in a higher average aggregate value (111.2
instead of 105.8).
Because of the improved investment incentives in monopoly states under the capital-
weighted division of bargaining power is that the AV-oriented Markov perfect policy allows
mergers in a much larger set of states. In fact, the approval probability is less than 50% only
in states in which both …rms have at least …ve units of capital. As a result, the average merger
frequency increases from 16.1% to 29.6%, which is not much lower than the 32.1% merger
frequency when all mergers are allowed. The performance of the Markov perfect policy is
therefore close to that of the all-mergers-allowed policy: the average AV is 112.1, compared to
111.2 under the latter policy (and 113.6 under the Markov perfect policy with equal bargaining
weights).
Table 8: Performance Measures for the Intermediate Market

under Various Policies and Capital-weighted Bargaining Power
Performance Measure 37 No-Mergers All-Mergers MPP-AV
Avg. Consumer Value 48.1 37.2 37.3
Avg. Incumbent Value 69.4 72.9 74.0
Avg. Entrant Value 0.0 1.1 0.8
Avg. Blocking Cost 0.0 0.0 0.0
Avg. Aggregate Value 117.5 111.2 112.1
Avg. Price 2.15 2.25 2.24
Avg. Quantity 22.2 19.6 19.6
Avg. Total Capital 8.0 7.5 7.5
Merger Frequency 0.0% 32.1% 29.6%
% in Monopoly 18.6% 96.0% 49.4%
% minfK1 ; K2 g 2 75.7% 0.2% 44.2%
State (0,0) CV 30.3 25.4 25.9
State (0,0) AV 36.7 35.1 35.3
4.6 Outcomes for Various Scale Parameter Values

In Section 3.4 of the main paper, we examined the extent to which several of the features of
the equilibria in our small, intermediate, and large markets extend across a wider range of
demand parameters B and A. Here we do a similar analysis across demand parameter B, the
size of the market, and production parameter , the scale parameter. Our analysis in this
section shows the same patterns as are shown in the main paper. It suggests that changing
37
43
the production scale parameter leads to similar comparative statics as changing the demand
function choke price A.
Figure 21 reports on the di¤erence in aggregate value between the no-mergers-allowed
and all-mergers-allowed equilibria. The …gure depicts contour lines showing the parameters
at which the aggregate value di¤erence (AVN o AVAll )=AVN o achieves a given percentage
value (each contour line is labelled). Also shown in the …gure are three dots representing
the parameters of our small, intermediate, and large markets, as well as dashed lines showing
markets that spend 5%, 20%, and 60% of the time in monopoly when no mergers are allowed
(these are roughly the monopoly percentages in our large, intermediate, and small markets).
As can be seen in the …gure, aggregate value with no mergers allowed is greater than with all
mergers allowed provided that the market is large enough. This pattern is nearly identical to
the pattern seen in the main paper.
Figure 22 shows the percentage di¤erence in entry probabilities in the no-mergers-allowed
and all-merger-allowed equilibria, [Pr(Entry)All Pr(Entry)N o ]= Pr(Entry)All ].38 Consistent
with the entry for buyout we observed earlier, the level of entry is always weakly greater in the
all-mergers-allowed equilibrium, although the di¤erence declines to zero in very large markets
where the probability of entry rises to 1 under either merger policy.
Figure 23 shows the probability of a merger occurring under the Markov perfect policy.
We see the same pattern as we saw in the main paper. Moving from the Southwest corner, the
probability of merger increases as the market gets larger. Continuing in the same direction,
however, the probability of merger begins decreasing once the market is large enough that the
Markov perfect policy of the antitrust authority allows fewer mergers.
Figure 24 shows the percentage di¤erence in aggregate value between the Markov perfect
policy and the no-mergers-allowed equilibrium (AVM P P AVN o )=AVM P P . Again, we see the
same pattern as we saw in the main paper. In small markets, the Markov perfect policy leads to
higher aggregate value than when no mergers are allowed. The no-mergers policy outperforms
the Markov perfect policy provided the market is large enough. However, for the largest
markets in the Northeast corner, the Markov perfect policy leads to the same equilibrium as
the no-mergers policy because mergers are never consummated.
Figure 25 shows the same AV comparison but relative to the outcome with the static
aggregate surplus based policy, (AVM P P AVStatic )=AVM P P . The …gure shows that the Markov
perfect policy outperforms the static agrgegate surplus based policy provided the market is
large enough.
38
Pr(Entry)x is calculated by weighting the probability of entry in each monopoly stateunder merger policy
x by the probability of that state in the all-mergers-allowed equilibrium.
44
60%
1.14 State
mon
space
opo
1.13 potentially
binding
ly
1.12
Production scale
1.11
0%
6%
9%
1.1
3%
–6%
–3%
1.09
5%
1.08
m
20%
ono
–3%
1.07
poly
mon
opo
1.06 12%
0%
ly
1.05
15 20 25 30 35 40
B demand (size of economy)
Figure 21: Contour lines of the percentage di¤erence between the steady state aggregate value
of the no-mergers and all-mergers-allowed equilibria, (AVN o AVAll )=AVN o .
60%
1.14 State
mo
space
nop
potentially
5%
1.13
oly
binding
20%
m ono
1.12
mo
pol
nop
Production scale
1.11
oly
60
30
90
%
%
1.1
1.09
1.08
1.07
90%
1.06
1.05
15 20 25 30 35 40
Figure 22: Contour lines of the percentage of entry probabilities between the no-mergers and
all-mergers-allowed equilibria, [Pr(Entry)All Pr(Entry)N o ]= Pr(Entry)All ].
45
60%
1.14 State
mon
space
opo
1.13 potentially
binding
ly
1.12
30%
30
Production scale
%
1.11
1.1
10%
1.09
20
5%
20%
%
1.08
m
10%
20%
ono
1.07
poly
mon
opo
1.06
ly
1.05
15 20 25 30 35 40
Figure 23: Contour lines of the steady state probability of merger in the MPP-AV equilibria.
60%
1.14 State
mon
space
opo
1.13 potentially
binding
ly
1.12
Production scale
0%
1.11
1.1
6%
–3%
1.09
5%
3%
1.08
m
6%
20%
ono
0%
1.07
poly
mon
–3%
3%
opo
1.06
ly
0%
1.05
15 20 25 30 35 40
of the MPP-AV and no-mergers equilibria, (AVM P P AVN o )=AVM P P .
46
60%
9%
1.14 State
mon
space
opo
1.13 potentially
binding
ly
1.12
Production scale
1.11
1.1
0%
0%
1.09
3%
9%
5%
6%
1.08
m
20%
ono
1.07
poly
mon
opo
1.06
ly
1.05
15 20 25 30 35 40
of the MPP-AV and static-AS policy equilibria, (AVM P P AVStatic )=AVM P P .
4.7 Outcomes for Various Ranges of Investment Costs

In this section we examine outcomes for a broader range of investment costs for incumbents and
entrants, focusing on the intermediate market. Speci…cally, for each combination of s 2 [1; 3]
and sg 2 [0:25; 1] we …nd the level c(s; sg ) at which the percentage of time in a monopoly
state is 18.6% in the no-mergers-allowed equilibrium. Recall that our baseline intermediate
economy corresponds to (s; sg ) = (3; 1) and our e¢ cient entry environment discussed in Section
4.2 corresponds to (s; sg ) = (1; 0:25). These points represent, respectively, the Northeast and
Southwest corners of the contour plots below.
Figure 26 shows the average time from monopoly to merger in the all-mergers-allowed
equilibrium. As can be seen, there is quicker entry for buyout (i.e. average time from monopoly
to merger goes down) when the spread of augmentation draws decreases and the entrant’s
investments become more e¢ cient relative to the incumbent’s. Figure 27 shows the di¤erence
in aggregate value between the no-mergers-allowed and all-mergers-allowed equilibria. As
noted in Section 4.2, increasing entrant e¢ ciency (moving in the Southwest direction) helps
to mitigate the reduction in aggregate value from allowing mergers. Figure 28 shows the
probability of merger in the Markov perfect policy. The probability of merger increases as we
increase entrant e¢ ciency due to the fact that the antitrust authority allows more mergers
when entrants are more e¢ cient.
47
2.8 2.7
2.6
2.6 2.5
s (spread of augmentation draws)
2.4
2.2 2.4
1.8
1.6
2.3
1.4
1.2 2.2
1
0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
s (spread of greenfield draws)
g
Figure 26: Contour lines showing the steady state weighted average time from monopoly to
merger in the all-mergers-allowed equilibrium. The minimum augmentation draw, c, is set as
a function of s and sg to achieve 18.6% monopoly in the no-mergers equilibrium.
48
2.8
9%
2.6
2.4
2.2
8%
2
7%
1.8
1.6 6%
5%
1.4 4%
1.2
3%
2%
1
0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
g
of the no-mergers and all-mergers-allowed equilibria, (AVN o AVAll )=AVN o . The minimum
augmentation draw, c, is set as a function of s and sg to achieve 18.6% monopoly in the
no-mergers equilibrium.
49
3
10%
2.8
15% 20%
2.6 25% 30%
35%
2.4 40%
45%
2.2 50%
1.8
1.6
1.4 %
40
1.2
1
0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
g
Figure 28: Contour lines of the steady state probability of merger in the MPP-AV equilibria.
The minimum augmentation draw, c, is set as a function of s and sg to achieve 18.6% monopoly
in the no-mergers equilibrium.
50
5 Multiplicity of Equilibria
Dynamic stochastic games with in…nite horizons generally have multiple equilibria when players
are patient. Within the context of the Ericson and Pakes (1995) model of computable Markov
perfect equilibria, Besanko et al. (2010) develop a homotopy-based method for tracing out
paths on the equilibrium manifold and systematically …nding points in the parameter space
for which multiple equilibria exist.39 It does not, however, provide a guarantee that it will …nd
all equilibria.
The homotopy technique depends on di¤erentiating the equations that implicitly de…ne
the model’s equilibria. This requirement makes it, as a practical manner, infeasible to apply
to our merger model because a key step in numerically solving for equilibria is a Monte Carlo
integration. Numerically di¤erentiating this integral with reasonable accuracy is not possible
with the computing power to which we have access. Consequently we implemented a cruder
search for multiple equilibria that may fail to …nd cases of multiplicity that the homotopy
technique would …nd if it were feasible.40
The idea is straightforward. Along lines through the parameter space, we calculate se-
quences of equilibria using the equilibrium values of one equilibrium as the starting points
for the next equilibrium computation. For example, one line we search is where the demand
parameter B 2 f11; 12; 13; :::; 40; 41g and all other parameters are …xed. We start the equilib-
rium calculations from both ends of the line and use the equilibrium values calculated for a
particular B as the initial values for calculating the equilibrium at the next B: If equilibrium
multiplicity exists along the line, then the equilibrium values for a particular B reached from
the line’s left end may not equal the equilibrium values for that same B reached from the
line’s right end.
We performed this test for 93 total lines of three types. In the …rst type of line, we
vary the demand parameter B 2 f11; 12; 13; :::; 40; 41g while …xing the demand parameter
A 2 f1:5; 1:6; 1:7; :::; 4:4; 4:5g. All other parameters are …xed at their standard values. In
the second type of line, we vary the scale parameter 2 f1:05; 1:053; 1:057; :::; 1:147; 1:15g
while …xing the demand parameter B 2 f11; 12; 13; :::; 40; 41g. In the third type of line, we
vary the augmentation cost spread parameter s 2 f1; 1:13; 1:27; :::; 4:87; 5g while …xing the
minimum augmentation cost parameter c 2 f2; 2:1; 2:1; 4:9; 5g. For the no-mergers policy we
…nd multiplicity for some parameter values. For the all-mergers-allowed policy and the Markov
perfect policy based on the AV criterion we …nd no multiplicity, but we do …nd regions in which
our algorithm failed to calculate an equilibrium for the Markov perfect policy.41
39
See Borkovsky, Doraszelski, and Kryukov (2010, 2012) for further discussion and illustration of how to use
this homotopy technique.
40
We thank Uli Doraszelski for suggesting this technique to us.
41
The regions where we cannot …nd Markov perfect policy equilibria are where the equilibria transition from
being almost entirely monopoly in the steady state to being only 85% monopoly. At this point, small changes in
the antitrust authority’s policy result in large changes in equilibrium behavior so it is di¢ cult to …nd a Markov
51
Each instance of multiplicity that we …nd for the no-mergers policy has a common structure.
The distinguishing strategic di¤erence in the two equilibria is the investment behavior at state
(1,0) and in some cases state (2,0). Total investment is approximately the same, but in one
equilibrium, the incumbent invests more, and in the other equilibrium, the entrant invests
more. Each …rm wants to have an aggressive investment policy if the other …rm has a passive
investment policy, and a passive policy if the other …rm has an aggressive policy. Almost
certainly a third equilibrium exists that is unstable and not computable with our algorithm.42
An example of the no-mergers multiplicity is when (B; A) = (33; 2:8) where the investment
at state (1,0) is the di¤erence in the equilibria. In “equilibrium 1” the incumbent builds, in
expectation, 2.0 units of capital while the entrant builds 1.1 units of capital. In “equilibrium
2” the behavior reverses: the incumbent builds 1.2 units of capital while the entrant builds
2.2 units of capital. Table 9 shows that the performance measures for these two equilibria are
quite close since investment behavior in state (1,0) does not have much impact on steady state
behavior.
Finally, we point out that we …nd no multiplicity for our baseline parameters.
Table 9: Performance Measures for Two Pure Equilibria

under No Mergers at (B; A) = (33; 2:8)
Performance measure 43 Equil. 1 Equil. 2
Avg. Consumer Value 30.9 31.0
Avg. Incumbent Value 68.3 68.3
Avg. Aggregate Value 99.2 99.4
Avg. Price 2.20 2.20
Avg. Quantity 19.9 19.9
Avg. Total Capital 6.4 6.4
% in Monopoly 79.3% 78.9%
% minfK1 ; K2 g 2 19.3% 19.7%
6 Additional Tables and Figures Referenced in the Main Paper

Table 10 displays the equilibrium statistics of our primary duopoly market parameterization
from the main paper, only allowing for a third …rm. Comparing to Table 2 in the main paper,
it can be seen that this market is a “natural duopoly” in that even when three …rms are
allowed, the no-mergers steady state measures are very similar to when only two …rms are
perfect policy which is the best response to the …rm behavior it induces. We believe there are equilbria in this
region but that they are very unstable and our algorithm can not …nd them.
42
See Besanko et al. (2010, section 3.2) for a discussion of the inability of Pakes-McGuire-like algorithms to
compute unstable equilibria.
43
52
allowed. One can also see that our insights from the study merger policy in the duopoly case
carry over to the triopoly case.
Table 10: Performance Measures for the (A=3,B=26) Market

under Various Policies (Allowing a Third Firm)
No-Mergers/
Performance Measure 44 All-Mergers Static-AS MPP-AV
Static-CS/
MPP-CV
Avg. Consumer Value 48.3 38.0 38.1 47.1
Avg. Incumbent Value 69.3 68.9 69.4 69.6
Avg. Entrant Value 0.0 0.8 0.8 0.1
Avg. Blocking Cost 0.0 0.0 0.0 -0.0
Avg. Aggregate Value 117.6 107.7 108.2 116.7
Avg. Price 2.15 2.24 2.24 2.16
Avg. Quantity 22.2 19.8 19.8 21.9
Avg. Total Capital 8.0 7.6 7.6 8.1
Merger Frequency 0.0% 50.8% 50.2% 13.3%
% in Monopoly 18.0% 85.4% 86.9% 31.3%
% in Duopoly 81.5% 14.6% 13.1% 68.7%
State (0,0,0) CV 32.7 26.0 26.2 29.9
State (0,0,0) AV 34.7 31.9 32.1 34.9
Figure 29 shows the di¤erence between the private and social incentives to invest when all
mergers are allowed. The socially insu¢ cient incentive for incumbent …rms to invest and the
socially excessive incentive for entrants to invest results in the detrimental e¤ects of entry for
buyout seen in the main paper.
44
All values are ex ante (beginning-of-period) values except % in Monopoly and % in Duopoly (showing the
percentages of the time that industry capital is in each type of state) which are at the Cournot competition stage.
“No-mergers” and “All-Mergers” refer to the no-mergers-allowed and all-mergers-allowed policies, respectively.
“Static-CS” and “Static-AS” refer, respectively, to the equilibria under the optimal static consumer surplus-
based and aggregate surplus-based merger policies. “MPP-CV” and “MPP-AV” refer, respectively, to the
equilibria when the antitrust authority cannot commit (resulting in a Markov perfect policy) under consumer
value and aggregate value welfare criteria. “State (0,0) CV” and “State (0,0) AV” are the values of CV and
AV, respectively, for a new industry that starts with no capital.
53
Figure 29: Private incentive of the row …rm (…rm 1) to invest minus the social incentive for
the row …rm to invest in the intermediate market with all mergers allowed. Negative numbers
are in parentheses.
References
[1] Besanko, D., U. Doraszelski, Y. Kryukov, and M. Satterthwaite. (2010). “Learning-by-
Doing, Organizational Forgetting, and Industry Dynamics.” Econometrica 78: 453-508.
[2] Besanko, D., and D. F. Spulber (1993), “Contested Mergers and Equilibrium Antitrust
Policy,” Journal of Law, Economics, and Organization 9: 1-29.
[3] Borkovsky, R. N., U. Doraszelski, and Y. Kryukov. (2010), “A User’s Guide to Solving
Dynamic Stochastic Games Using the Homotopy Method,”Operations Research 58: 1116-
1132.
[4] Borkovsky, R. N., U. Doraszelski, and Y. Kryukov. (2012), “A Dynamic Quality Lad-
der Duopoly with Entry and Exit: Exploring the Equilibrium Correspondence Using the
Homotopy Method,” Quantitative Marketing & Economics 10: 197-229.
[5] Ericson, R. and A. Pakes (1995), “Markov-perfect Industry Dynamics: A Framework for
Empirical Work,” Review of Economic Studies 62: 53-82.
[6] Lyons, B. (2002), “Could Politicians be More Right than Economists? A Theory of Merger
Standards,” CCP Working Paper 02-01, University of East Anglia.
[7] Mankiw, N. G. and M. D. Whinston (1986), “Free Entry and Social Ine¢ ciency,” RAND
Journal of Economics 17: 48-58.
[8] Whinston, M. D. (2007), “Antitrust Policy toward Horizontal Mergers,” in: Handbook of
Industrial Organization, vol. 3, eds. M. Armstrong and R. H. Porter, Amsterdam: North
Holland.
54

Internal Versus External Growth in Industries With Scale

Uploaded by

Copyright:

Available Formats

You might also like

Internal Versus External Growth in Industries With Scale

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Internal Versus External Growth in Industries With Scale

Uploaded by

Copyright:

Available Formats

Manuscript Click here to access/download;Manuscript;2018.09.19 MNSW.

Copyright The University of Chicago 2019. Preprint (not copyedited or formatted).

Internal versus External Growth in Industries with Scale

Journal of Political Economy

creating a stronger third firm.4

2.1 Static Demand, Costs, and Competition

Figure 1: Sequence of events in a single period

2.2 Mergers and Bargaining

The expected externality of the proposal of merger Mij on an outsider k 6= i, j is given by

Xkij (K) ≡ aij (K) V (Kk , Ki + Kj , K−ijk , 0) − V (Kk , K−k ) ,

ϑ12 (K) = a12 (K)ψ12 (K),

2.3 Merger Policy

2.4 Investment, Entry, and Depreciation

3 Merger Policy in Duopoly Markets

3.2 Investment and Merger Incentives under Fixed Merger Policies

3.2.1 Equilibria with No-Mergers-Allowed

3.2.2 Equilibria with All-Mergers-Allowed

Figure 3: One-period transition probabilities from state (5,0).

3.2.3 Equilibria with Static Policies

3.3 Equilibria with Markov Perfect Merger Policy

3.4 Results for Other Demand Parameters

4 Merger Policy in Triopoly Markets

∆ij (K) ≡ V (Ki + Kj , Kk , 0) − (V (Ki , K−i ) + V (Kj , K−j ))

Proposition 1 Suppose firm i is selected as the proposer in state K. Then:

4.1 Allowing a Third Firm in the (A = 3, B = 26) Market

average AV than the Markov perfect policy with the AV criterion.

4.2 Two Natural Triopoly Markets

4.2.1 Merger Bargaining with Three Firms

4.2.2 Effects of Fixed Merger Policies

4.2.3 Markov Perfect Merger Policy

antitrust authorities, including the U.S. and EU).

[11] Doraszelski, U. and M. Satterthwaite (2010), “Computable Markov-perfect Industry Dy-

[17] Gowrisankaran, G. (1997), “Antitrust Policy Implications of a Dynamic Merger Model,”

[18] Gowrisankaran, G. (1999), “A Dynamic Model of Endogenous Horizontal Mergers,”

[19] Hayashi, F. (1982), “Tobin’s Marginal q and Average q: A Neoclassical Interpretation,”

[20] Hollenbeck, B. (2017), ”Horizontal Mergers and Innovation in Concentrated Industries,”

[32] Williamson, O. E. (1968), “Economies as an Antitrust Defense: The Welfare Tradeoffs,”

Table 1: (A = 3, B = 26) Market Static Equilibria

Table 3: Average Capital Addition in the (A = 3, B = 26) Market

Table 4: Merger Bargaining Outcomes in Symmetric States

Table 5: Merger Bargaining Outcomes in Symmetric States

Table 6: Performance Measures for the Triopoly Markets

Journal of Political Economy

(c) Static-AS (d) MPP-AV

Journal of Political Economy

(a) No-mergers-allowed (b) All-mergers-allowed

Journal of Political Economy

Journal of Political Economy

Journal of Political Economy

(a) Aggregate value (b) Entry probability

3.1 3.1 30%

A (demand choke price)

Journal of Political Economy

(a) MPP-AV and No-mergers (b) MPP-AV and Static-AS

MPP-AV equilibrium is the same as 9%

A (demand choke price)

Journal of Political Economy

Copyright The University of Chicago 2019. Preprint (not copyedited or formatted).