1

"A Primer on Foreclosure. Patrick Rey and

Jean Tirole. Handbook of Industrial

Organization (2007)"

A summarizing exposition

Alexandrina I. Scorbureanu

Universite de Cergy-Pontoise

March 13, 2008

1 What is foreclosure?

In general, market foreclosure represent market practices that reduce buyers

access to a supplier (upstream foreclosure) and/or limit the supplier's access

to a buyer (downstream foreclosure). Some of the tools used to achieve

market foreclosure include:

1. a buyer that purchases a supplier or set up his own production unit so

as to manufacture the intermediate good internally;

2. a supplier that signs exclusive-dealing with his buyers;

3. a manufacturer that makes his good incompatible to complementary

goods sold by other manufacturers.

There are two major types of foreclosure: in the _rst type, one of the sec-

tors (up- or downstream) is monopolized (in this case foreclosure practices

include exclusive dealing and competition reduction). In the second, neither

sector is monopolized (foreclosure increases monopolization of one of the sec-

tors). In this chapter, authors consider two particular cases of the _rst-type

foreclosure: vertical and horizontal integration.

As intended in this paper, foreclosure is a _rm's restriction of output in

one market through the use of market power in another market. It refers to a

dominant _rm's denial of proper access to an essential good that it produces,

with the intent of extending monopoly power from that bottleneck segment

of the market to an adjacent segment (a potentially competitive one). When

the bottleneck good (for instance infrastructure, software, etc) is used as an

input by a potentially competitive downstream industry, or when it is sold

directly to customers (who use it together with other complementary goods).

Foreclosure may be complete (refusal to deal, extravagant prices or technical

complementary integration between goods) or partial. It also may be:

1. vertical - arises when a _rm controls an essential input for the poten-

tially competitive industry; this _rm can alter competition by denying

or limiting access to its input,

2. horizontal - when the bottleneck good is sold directly to the _nal con-

sumer and the _rm bundles the potentially competitive good to the

bottleneck one.

Reassuming, some instruments used by the forecloser are:

(a) Integration of the forecloser _rm to other _rms in the comple-

mentary segment,

(b) Refusing to cooperate to put competitors in disadvantage (economies

of scope or scale in the same market),

(c) Granted exclusivity to a subset of _rms producing selected goods,

(d) Second-degree price discrimination (by loyalty programs to all or

rebates based on the rate of growth of purchases) and third-degree

price discrimination (charging di_erent cost-adjusted prices to dif-

ferent customers), as well as "mixed bundling" (conditional dis-

counts on complementary goods).

A number of solutions have been considered, namely:

2

(a) structural policies such as divestitures and line of business restric-

tions (but with high transaction costs) that may allow the joint

ownership by all competitors of an essential facility;

(b) access price control when antitrust authorities compare the price

of access with some measure of its cost (di_cult to measure em-

pirically marginal costs),

(c) access quantity control within an exclusivity contract, some amount

of each operator's capacity must be allocated to new entrants,

(d) price linkage between access charges, for instance the e_cient

component pricing rule that links the integrated monopolist's ac-

cess and retail prices, to avoid margin squeezes,

(e) common carrier policies that means the turning of vertical struc-

ture of the industry upside down (referred to the semantic meaning

of naming "upstream" and "downstream" operators),

(f) disclosure requirements the requirement for contracts of interme-

diary goods to be made public for the sake of transparency.

2 Vertical foreclosure

The "leverage" concept argued that there is a single source of monopoly

pro_t, and that a bottleneck monopolist can earn the entire monopoly

pro_t without extending its market power to related segments (vertical

integration cannot increase pro_tability of merging _rms). For exam-

ple, a bottleneck holder faces a commitment problem similar to that of

a durable-good monopolist (see Coase's durable good analysis): once

it has contracted with a downstream _rm for access to its facility, it

is tempted to o_er access to other _rms as well (even if their competi-

tion to the other _rm will reduce monopolist's pro_ts). Nevertheless,

the positioning (downstream or upstream) of the _rm is not aleatory

and it has some important consequences on the monopolist's power. A

comparison between exclusive contracts and commitment problem lead

us to two major problems:

3

(a) upstream bottleneck's pro_t is smaller, the larger the number of

downstream _rms, and

(b) for a given number of downstream _rms, the upstream pro_t is

smaller, the more substitutable the downstream units.

The study of this problems lead the authors to derivate three major

policy implications:

(a) it is important whether the more competitive of two complemen-

tary sectors lies upstream or downstream (prices are lower if the

bottleneck owner lies upstream),

(b) non-discrimination laws may have a perverse e_ect of restoring the

monopoly power (when an upstream bottleneck practices foreclo-

sure by discrimination among competitors, o_ering to all com-

petitors the same commercial conditions forces bottleneck to sell

further units at the same high price, that helps it to commit),

(c) ECPR (e_cient component pricing rule) has scarce e_ect on un-

regulated markets. It is a partial rule that provides a link between

access and _nal prices (the higher the _nal price, the higher the

access price can be).

A simple example of (1 monopoly X 2 retailers) in a two-stage game

framework is built and we are reformulating the main results. Provided

that the vertical structure of industry's monopoly output is formed by:

Qm = argmax f(P(q) 􀀀 c)qg

pm = P(Qm)

_m = (pm 􀀀 c)Qm = [P(Qm)]2 􀀀 cQm:

(1)

The interaction between _rms is described in the following: PLAYER STAGE I STAGE II EQUILIBRIUM

U MC = c; (q1(T1)); (q2(T2))

D1 T1(_) MC = 0

Rev1 = q1P(q1 + q2)

D2 T2(_) MC = 0

Rev2 = q2P(q1 + q2)

Consumer q = D(p); p = P(q)

Sub-cases related to observability hypothesis:

(a) Commitment, observability, credibility. Both tari_s T1; T2 o_ered

by U are observed by both D1 and D2.U exerts his full market

power such as to extract all Di's pro_t. Nevertheless, contract

may be secret or can be privately renegotiated. If we hav

"A Primer on Foreclosure. Patrick Rey and

Jean Tirole. Handbook of Industrial

Organization (2007)"

A summarizing exposition

Alexandrina I. Scorbureanu

Universite de Cergy-Pontoise

March 13, 2008

1 What is foreclosure?

In general, market foreclosure represent market practices that reduce buyers

access to a supplier (upstream foreclosure) and/or limit the supplier's access

to a buyer (downstream foreclosure). Some of the tools used to achieve

market foreclosure include:

1. a buyer that purchases a supplier or set up his own production unit so

as to manufacture the intermediate good internally;

2. a supplier that signs exclusive-dealing with his buyers;

3. a manufacturer that makes his good incompatible to complementary

goods sold by other manufacturers.

There are two major types of foreclosure: in the _rst type, one of the sec-

tors (up- or downstream) is monopolized (in this case foreclosure practices

include exclusive dealing and competition reduction). In the second, neither

sector is monopolized (foreclosure increases monopolization of one of the sec-

tors). In this chapter, authors consider two particular cases of the _rst-type

foreclosure: vertical and horizontal integration.

As intended in this paper, foreclosure is a _rm's restriction of output in

one market through the use of market power in another market. It refers to a

dominant _rm's denial of proper access to an essential good that it produces,

with the intent of extending monopoly power from that bottleneck segment

of the market to an adjacent segment (a potentially competitive one). When

the bottleneck good (for instance infrastructure, software, etc) is used as an

input by a potentially competitive downstream industry, or when it is sold

directly to customers (who use it together with other complementary goods).

Foreclosure may be complete (refusal to deal, extravagant prices or technical

complementary integration between goods) or partial. It also may be:

1. vertical - arises when a _rm controls an essential input for the poten-

tially competitive industry; this _rm can alter competition by denying

or limiting access to its input,

2. horizontal - when the bottleneck good is sold directly to the _nal con-

sumer and the _rm bundles the potentially competitive good to the

bottleneck one.

Reassuming, some instruments used by the forecloser are:

(a) Integration of the forecloser _rm to other _rms in the comple-

mentary segment,

(b) Refusing to cooperate to put competitors in disadvantage (economies

of scope or scale in the same market),

(c) Granted exclusivity to a subset of _rms producing selected goods,

(d) Second-degree price discrimination (by loyalty programs to all or

rebates based on the rate of growth of purchases) and third-degree

price discrimination (charging di_erent cost-adjusted prices to dif-

ferent customers), as well as "mixed bundling" (conditional dis-

counts on complementary goods).

A number of solutions have been considered, namely:

2

(a) structural policies such as divestitures and line of business restric-

tions (but with high transaction costs) that may allow the joint

ownership by all competitors of an essential facility;

(b) access price control when antitrust authorities compare the price

of access with some measure of its cost (di_cult to measure em-

pirically marginal costs),

(c) access quantity control within an exclusivity contract, some amount

of each operator's capacity must be allocated to new entrants,

(d) price linkage between access charges, for instance the e_cient

component pricing rule that links the integrated monopolist's ac-

cess and retail prices, to avoid margin squeezes,

(e) common carrier policies that means the turning of vertical struc-

ture of the industry upside down (referred to the semantic meaning

of naming "upstream" and "downstream" operators),

(f) disclosure requirements the requirement for contracts of interme-

diary goods to be made public for the sake of transparency.

2 Vertical foreclosure

The "leverage" concept argued that there is a single source of monopoly

pro_t, and that a bottleneck monopolist can earn the entire monopoly

pro_t without extending its market power to related segments (vertical

integration cannot increase pro_tability of merging _rms). For exam-

ple, a bottleneck holder faces a commitment problem similar to that of

a durable-good monopolist (see Coase's durable good analysis): once

it has contracted with a downstream _rm for access to its facility, it

is tempted to o_er access to other _rms as well (even if their competi-

tion to the other _rm will reduce monopolist's pro_ts). Nevertheless,

the positioning (downstream or upstream) of the _rm is not aleatory

and it has some important consequences on the monopolist's power. A

comparison between exclusive contracts and commitment problem lead

us to two major problems:

3

(a) upstream bottleneck's pro_t is smaller, the larger the number of

downstream _rms, and

(b) for a given number of downstream _rms, the upstream pro_t is

smaller, the more substitutable the downstream units.

The study of this problems lead the authors to derivate three major

policy implications:

(a) it is important whether the more competitive of two complemen-

tary sectors lies upstream or downstream (prices are lower if the

bottleneck owner lies upstream),

(b) non-discrimination laws may have a perverse e_ect of restoring the

monopoly power (when an upstream bottleneck practices foreclo-

sure by discrimination among competitors, o_ering to all com-

petitors the same commercial conditions forces bottleneck to sell

further units at the same high price, that helps it to commit),

(c) ECPR (e_cient component pricing rule) has scarce e_ect on un-

regulated markets. It is a partial rule that provides a link between

access and _nal prices (the higher the _nal price, the higher the

access price can be).

A simple example of (1 monopoly X 2 retailers) in a two-stage game

framework is built and we are reformulating the main results. Provided

that the vertical structure of industry's monopoly output is formed by:

Qm = argmax f(P(q) 􀀀 c)qg

pm = P(Qm)

_m = (pm 􀀀 c)Qm = [P(Qm)]2 􀀀 cQm:

(1)

The interaction between _rms is described in the following: PLAYER STAGE I STAGE II EQUILIBRIUM

U MC = c; (q1(T1)); (q2(T2))

D1 T1(_) MC = 0

Rev1 = q1P(q1 + q2)

D2 T2(_) MC = 0

Rev2 = q2P(q1 + q2)

Consumer q = D(p); p = P(q)

Sub-cases related to observability hypothesis:

(a) Commitment, observability, credibility. Both tari_s T1; T2 o_ered

by U are observed by both D1 and D2.U exerts his full market

power such as to extract all Di's pro_t. Nevertheless, contract

may be secret or can be privately renegotiated. If we have the

following situation for instance:

q2 = Qm

2 ; T2 = pmQm

2

q1 = argmaxq fq [P(Qm=2 + q) 􀀀 c]g = RCournot > Qm=2

with : P0 < 0;

(2)

Therefore there is an incentive to secretly contract between U and

D1.

(b) Secret contracts At the _rst stage U o_ers secret contracts to each

Di. The equilibrium is characterized by the Cournot quantities,

prices and pro_ts:

q1 = argmaxq f[P(q + q2) 􀀀 c] qg = RCournot(q2); (8)q

q1 = q2 = qCournot = RCournot(qCournot) > Qm=2

p1 = p2 = pCournot = P(2qCournot) < pm

_U = (pCournot 􀀀 c)2qCournot = 2_Cournot < _m

_D1 = _D2 = 0:

(3)

This result puts emphasis on the commitment problem faced by

the monopoly supplier (a credibility problem prevents him to gain

the monopoly outcome). In the case there are n downstream com-

petitors, the symmetric passive conjecture equilibrium is given by

q = RCournot((n 􀀀 1)q), where q is output per downstream _rm.

The commitment problem becomes more severe, the larger the

number of downstream _rms (increasing competition). Also, the

same result is reached if we allow for downstream product di_eren-

tiation. The retail prices are di_erent, respectively p1 = P1(q1; q2)

and p2 = P2(q1; q2) and the equilibrium of the overall game is

still a Cournot equilibrium (in which downstream _rms face a

marginal cost c). The result is that the ratio of Cournot industry

pro_t over monopoly pro_t increases with the degree of di_eren-

tiation and the attractiveness of monopoly power is stronger the

more substitutable are retail products. In this situation, foreclo-

sure aims to reestablish monopoly power (U has an incentive to

alter downstream market structure using techniques as: exclusive

dealing, integration with downstream _rms, price oor).

The empirical experimental evidence that tests the foreclosure the-

ory yield that non-integration with public o_ers and vertical inte-

gration lead the monopoly outcome whereas non-integration with

secret o_ers result in Cournot outcome. Others _nd only partial

support for this theory (see Martin et al.2001). The _eld studies

results do not show relevant evidence of foreclosure e_ects (impact

of vertical mergers on downstream rivals and end users) but claim

that vertical integration may help solving commitment problems

of upstream monopolies. Three are the tested hypotheses:

i. retail _rms (rivals) receive less input from or pay a higher

price to the upstream monopolist _rm U;

ii. if D2 is publicly traded, then its stock price gets lower when

merger U 􀀀 D1 is announced (if U does not extract all the

rent from downstream units);

iii. _nal customers su_er from a merger (decrease in welfare is

measured by a decrease in stock price or an increase in future

price of _nal good).

6

2.1 Vertical foreclosure: Policy implications

The Coasian pricing problem is more likely to arise when monopo-

list bottleneck market is situated upstream. From the consumer or

total welfare perspective it is preferable to put the more competitive

sector downstream and let consumers deal directly to the competing

operators. Additionally, non-discrimination laws aim to protect con-

sumers from abuses of dominant position. In the context just described

above, these non-discrimination laws have adverse e_ects on all con-

sumers and total welfare, because they eliminate opportunistic behav-

ior of U and allow it to fully exercise its monopoly power. If U of-

fers the non-discriminatory two-part tari_ T(qi) = _m + cqi (wholesale

price=marginal cost and _xed fee=monopoly pro_t), an equilibrium

will exist if the coordination between U 􀀀D1􀀀D2 exists. The compet-

itive sector will gain zero-pro_ts and U will gain the monopoly pro_t.

If U does not consider the impact of a decrease in output on the down-

stream _rms pro_ts (and it has a quasi-concave objective function,

T(q) = wq and maximizes _U = (w(Q) 􀀀 c)Q), the result leads to a

choice of Q < Qm.

2.2 Restore of monopoly power: vertical integra-

tion and exclusive dealing

Vertical integration leads to the exclusion of the non-integrated retail

_rm, given that there is no other potential supplier for D2. The in-

troduction of an alternative supplier ^U does not a_ect _nal prices and

quantities or the structure of the production, but it produces a change

in the pro_t sharing among U and retailers. We re-write the two-stages

game with two alternative suppliers U and ^U : PLAYER STAGE I STAGE II EQUILIBRIUM

U UC = c; q1(T1); q2(T2)

^U

UC = ^c > c; q1( ^ T1); q2( ^ T2)

D1 T1; ^ T1 MC = 0

Rev1 = q1P(q1 + q2)

D2 T2; ^ T2 MC = 0

Rev2 = q2P(q1 + q2)

Consumer q = D(p); p = P(q)

From the _rst two lines of the game above we observe that U is more

e_cient than ^U , therefore it will potentially supply both D1 and D2,

although under more favorable conditions for the retailers due to the

competition with ^U 1.

If U and D1 integrate, the result leads to a reduction in the supply for

D2 which faces a higher opportunity cost (^c > c):D2 will buy from ^U

and the equilibrium quantities correspond to the asymmetric Cournot

duopoly:

q1 = RCournot(qCournot

2 )

q2 = RCournot(qCournot

1 ) _ argmaxq f(P(q + q1) 􀀀 ^c)qg

with : 􀀀1 < R0Cournot(q) < 0

RCo^urnot(q) < RCournot(q)

implying : 2qCournot < qCournot

1 + qCournot

2

_U+D1 = _Cournot

1 + (^c 􀀀 c)qCournot

2

_D2 = _Cournot

2

equil : c1 = c < c2 = ^c;

(4)

Thus D2 obtains lower pro_ts than _U+D1 through integration, that

1more precisely, U will supply both _rms with the same qCournot but for the payment

_Cournot 􀀀maxq

e the

following situation for instance:

q2 = Qm

2 ; T2 = pmQm

2

q1 = argmaxq fq [P(Qm=2 + q) 􀀀 c]g = RCournot > Qm=2

with : P0 < 0;

(2)

Therefore there is an incentive to secretly contract between U and

D1.

(b) Secret contracts At the _rst stage U o_ers secret contracts to each

Di. The equilibrium is characterized by the Cournot quantities,

prices and pro_ts:

q1 = argmaxq f[P(q + q2) 􀀀 c] qg = RCournot(q2); (8)q

q1 = q2 = qCournot = RCournot(qCournot) > Qm=2

p1 = p2 = pCournot = P(2qCournot) < pm

_U = (pCournot 􀀀 c)2qCournot = 2_Cournot < _m

_D1 = _D2 = 0:

(3)

This result puts emphasis on the commitment problem faced by

the monopoly supplier (a credibility problem prevents him to gain

the monopoly outcome). In the case there are n downstream com-

petitors, the symmetric passive conjecture equilibrium is given by

q = RCournot((n 􀀀 1)q), where q is output per downstream _rm.

The commitment problem becomes more severe, the larger the

number of downstream _rms (increasing competition). Also, the

same result is reached if we allow for downstream product di_eren-

tiation. The retail prices are di_erent, respectively p1 = P1(q1; q2)

and p2 = P2(q1; q2) and the equilibrium of the overall game is

still a Cournot equilibrium (in which downstream _rms face a

marginal cost c). The result is that the ratio of Cournot industry

pro_t over monopoly pro_t increases with the degree of di_eren-

tiation and the attractiveness of monopoly power is stronger the

more substitutable are retail products. In this situation, foreclo-

sure aims to reestablish monopoly power (U has an incentive to

alter downstream market structure using techniques as: exclusive

dealing, integration with downstream _rms, price oor).

The empirical experimental evidence that tests the foreclosure the-

ory yield that non-integration with public o_ers and vertical inte-

gration lead the monopoly outcome whereas non-integration with

secret o_ers result in Cournot outcome. Others _nd only partial

support for this theory (see Martin et al.2001). The _eld studies

results do not show relevant evidence of foreclosure e_ects (impact

of vertical mergers on downstream rivals and end users) but claim

that vertical integration may help solving commitment problems

of upstream monopolies. Three are the tested hypotheses:

i. retail _rms (rivals) receive less input from or pay a higher

price to the upstream monopolist _rm U;

ii. if D2 is publicly traded, then its stock price gets lower when

merger U 􀀀 D1 is announced (if U does not extract all the

rent from downstream units);

iii. _nal customers su_er from a merger (decrease in welfare is

measured by a decrease in stock price or an increase in future

price of _nal good).

6

2.1 Vertical foreclosure: Policy implications

The Coasian pricing problem is more likely to arise when monopo-

list bottleneck market is situated upstream. From the consumer or

total welfare perspective it is preferable to put the more competitive

sector downstream and let consumers deal directly to the competing

operators. Additionally, non-discrimination laws aim to protect con-

sumers from abuses of dominant position. In the context just described

above, these non-discrimination laws have adverse e_ects on all con-

sumers and total welfare, because they eliminate opportunistic behav-

ior of U and allow it to fully exercise its monopoly power. If U of-

fers the non-discriminatory two-part tari_ T(qi) = _m + cqi (wholesale

price=marginal cost and _xed fee=monopoly pro_t), an equilibrium

will exist if the coordination between U 􀀀D1􀀀D2 exists. The compet-

itive sector will gain zero-pro_ts and U will gain the monopoly pro_t.

If U does not consider the impact of a decrease in output on the down-

stream _rms pro_ts (and it has a quasi-concave objective function,

T(q) = wq and maximizes _U = (w(Q) 􀀀 c)Q), the result leads to a

choice of Q < Qm.

2.2 Restore of monopoly power: vertical integra-

tion and exclusive dealing

Vertical integration leads to the exclusion of the non-integrated retail

_rm, given that there is no other potential supplier for D2. The in-

troduction of an alternative supplier ^U does not a_ect _nal prices and

quantities or the structure of the production, but it produces a change

in the pro_t sharing among U and retailers. We re-write the two-stages

game with two alternative suppliers U and ^U : PLAYER STAGE I STAGE II EQUILIBRIUM

U UC = c; q1(T1); q2(T2)

^U

UC = ^c > c; q1( ^ T1); q2( ^ T2)

D1 T1; ^ T1 MC = 0

Rev1 = q1P(q1 + q2)

D2 T2; ^ T2 MC = 0

Rev2 = q2P(q1 + q2)

Consumer q = D(p); p = P(q)

From the _rst two lines of the game above we observe that U is more

e_cient than ^U , therefore it will potentially supply both D1 and D2,

although under more favorable conditions for the retailers due to the

competition with ^U 1.

If U and D1 integrate, the result leads to a reduction in the supply for

D2 which faces a higher opportunity cost (^c > c):D2 will buy from ^U

and the equilibrium quantities correspond to the asymmetric Cournot

duopoly:

q1 = RCournot(qCournot

2 )

q2 = RCournot(qCournot

1 ) _ argmaxq f(P(q + q1) 􀀀 ^c)qg

with : 􀀀1 < R0Cournot(q) < 0

RCo^urnot(q) < RCournot(q)

implying : 2qCournot < qCournot

1 + qCournot

2

_U+D1 = _Cournot

1 + (^c 􀀀 c)qCournot

2

_D2 = _Cournot

2

equil : c1 = c < c2 = ^c;

(4)

Thus D2 obtains lower pro_ts than _U+D1 through integration, that

1more precisely, U will supply both _rms with the same qCournot but for the payment

_Cournot 􀀀maxq

(P(q + qCournot) 􀀀 ^c)q

_

since each retailer can also buy from ^U at some

price ^p > ^c.

proves to bene_t integrated _rms. Vertical integration maintains pro-

duction e_ciency while it lowers consumer surplus and total welfare

(and the higher the cost of bypassing the bottleneck monopolist, the

larger the negative impacts on consumers and welfare). Vertical inte-

gration is more pro_table if ^c is higher.

Some policy solutions came along to limit the negative impact on wel-

fare of the vertical integration leading to foreclosure. With or without

vertical integration it is still desirable to ensure that the most com-

petitive sector faces _nal consumers. In the vertical integration case

with no bypass, it technically does not matter if the monopolist sector

is upstream or downstream, but by de_nition, there is no incentive to

integrate if the monopolist is situated downstream (in which case it

does not exist a commitment problem). In the case with possible by-

pass, the position of monopolist does matter (if it is downstream, the

less e_cient alternative supplier cannot be shut down and this results

in productive ine_ciency; there is also an indi_erence of U whether to

integrate with D1 or not).

Assuming the vertical integration between the upstream monopoly and

the downstream retailer, the equilibrium outcome without ECPR also

satis_es ECPR (it does not impose constraints on foreclosure, therefore

it is expected to perform a function it was not designed for). Under

the hypothesis of a single monopolist that integrates with D1, o_ering

a linear ECPR-compatible access price w2 < pm 􀀀 0 = pm to D2. The

revenue of _rm D2 that buys q2 intermediary units and transforms them

into a _nal good is: [P(Qm + q2) 􀀀 w2] q2 < [P(Qm) 􀀀 w2] q2 = 0. A

negative pro_t for D2 imposes a situation of no-viable activity for this

retailer, as the authors state.

In the exclusive dealing case vertical integration may also yield social

bene_ts (not only social costs). These can be evaluated by investi-

gating alternative strategies available to foreclosure (such as exclusive

dealing or exclusive supply contracts 2) and their relative costs. An

exclusive dealing may represent a perfect substitute for vertical inte-

2see P.Rey and J.Tirole. A primer on Foreclosure. IO Handbook 2007. p.2176

gration (given that if vertical integration is prohibited, by an exclusive

agreement between U 􀀀D1, U commits himself not to sell to D2). That

is to say that a policy that prohibits vertical integration but allows for

exclusionary agreements (that may become socially less desirable be-

cause its rigid constraints), is of no use. Exclusive dealing is pro_table

in a context where we consider the alternative supplier ^U does not

impose any competitive constraint and he is less e_cient than U that

gets the monopoly pro_t with exclusive dealing (and the Cournot pro_t

in other case). Instead, by auctioning an exclusive deal, U can earn

_Excl

Cournot(^c) = _Cournot

1 􀀀 maxq

n

[P(qCournot

1 (^c) + q) 􀀀 ^c]q

o

= (c 􀀀 ^c)q.

O_ering exclusivity or not yield zero pro_t when the second supplier is

equally e_cient (c = ^c), and are more pro_table when ^U is less e_cient

c < ^c. If there is no alternative supplier but the retailers produce a

di_erentiated good, the integrated _rm U􀀀D1 may still want to supply

D2 whereas an exclusive agreement with D1 would lead to the exclusion

of D2 (ine_ciency and reduction in welfare).

To conclude, exclusive dealing yields less pro_t to U than vertical inte-

gration. Secondly, the prohibition of vertical integration without norms

on exclusive dealing leads to a socially less desirable outcome (it reduces

the choice available to _nal consumers, by excluding rivals). Further

subjects to be developed are indicated by the authors, namely:

(a) private incentives to support not-exclusivity. Independent users of

intermediate goods may diminish investments that approach them

to the upstream bottleneck, or to a competitive-technology sector

(this choice is made when they anticipate the monopolist's fore-

closure, because of an existing vertical integration). Competition

protects investment in situations in which it is di_cult to write a

long-term e_cient contract. Therefore, an monopolist that has to

lower speci_c investments, does not want to compete in the future

with a favored downstream user;

(b) the Coasian approach (Cestone and White 2003) that is applied

beyond industrial markets, for instance to intermediary's owner-

ship of equity;

(c) contract with externalities (Segal 1999) situations in which a prin-

cipal contracts with multiple agents and one of the contracts has

externalities on other agents. General results are obtained on the

type of trade between the principal and agents (secret contracts

and public commitments) as a function of the type of externalities;

(d) alternative conjectures such as the passive conjecture from the

Cournot situation in which the monopolist produces to order.

There is a strategic interdependence between for instance U and

D1, when the contract signed with D1 a_ects conditions that U

would like to o_er to D2 (that is the competitor of D1). This

interdependence creates problems like non-concavity (that make

disappear pure-strategy with passive beliefs 3.);

(e) bidding games that are situations in which downstream rival retail-

ers bid (causing externalities on each other) for the input supplied

by the upstream monopolist (that chooses how much to supply,

eventually). On the contrary, if retailers determine quantities and

o_ers are public, they can protect against opportunistic behavior

of the rival by choosing a exible contract (adapt purchases to the

terms in rival's contracts).

3 Horizontal foreclosure

It refers to a situation in which a _rm U is present in two _nal markets

A (monopoly segment for _rm U) and B (the competitive segment

for _rm U). In this context it could appear a foreclosure situation

if U forecloses competitors in market B to link the bottleneck good

A to its own o_er on B (case invoked especially when A and B are

complements). Nevertheless, this situation is not pro_table for the _rm

U given that if it decides to foreclose rivals and become monopolist on

both markets (by bundling products A and B) it obtains a lower pro_t

than in the situation in which it keeps unbundling and _x a price such

3because the gain from simultaneously changing contracts o_ered to D1 and D2 may

exceed total gains from modi_cation of just one contract.

11

as to extract the pro_t of rivals on market B. If the _rm is integrated

and present on the two markets, it will be more likely to invest in B

(given that any increase in competition for B stimulates consumers to

pay for the monopolized product A). The same situation discourages

rivals to invest in B.

If the products are relatively independent, this logic does not hold any

more (as pointed out by the Chicago School and criticized by the paper

of Whinston (1990)). A demonstration is provided by considering that

in the B market there is a potential entrant E and _rm U must choose

if bundle or not the two products A and B. the results is that bundling

allows U to discourage E to enter the market. Nevertheless, if entry

occurs, U has no more incentive to bundle the two goods (the use of

bundling or tying as an entry barrier, relies on a strong commitment,

eventually obtainable through technological choices, for example, by

making A incompatible with with competitive B versions). If there is

no independence between A and B, the exit of competitors from market

B damages to good A. To conclude, bundling intensi_es competition

(we focus on compatibility choices of competing _rms that each o_er

all components of a system, example of endogenous switch of costs).

When _rms opt for compatibility, the market-by-market competition

prevails, where _rms compete separately for each component.

Furthermore, we focus on the entry decisions if we consider risky projects

on the two markets (authors consider now that U is a monopoly on both

A and B perfect complementary markets). An investment in research

and development will allow _rm E to enter the market. If it succeeds

to enter both markets it will replace U and will get all the gain on

both markets, whereas if E enters only one market, its gain depends

on the bundling decision of U (since goods are complementary) and E

becomes competitor for U. By considering the probability of success

of E, the authors derive a conclusion on the riskiness of entry projects

for E (in the absence of risk bundling decision of U is irrelevant).

In the case of economies of scale and scope two periods decisions for

the two _rms are analyzed for the two markets A and B. The resulting

12

conclusion is that if bundling is not made, the entrance on market A

of E is pro_table, whereas entry on market B is pro_table only if it

generates pro_ts in both periods. Also, entry is pro_table if E enters

market B in period 2. By bundling the two goods together, U deters

the entrance of E and allows U to maintain a monopoly pro_t over all

the two periods.

Furthermore it is interesting to observe how investments in research and

development of adjacent market B in order to discourage competitive

e_orts by rival producers. According to a Chicago School argument,

_rm U has still incentive to innovate (even if it is forced, for the sake

of innovation competition, to share the resulting intellectual property

with some E) because improvements in the adjacent market bene_ts

the dominant _rms core activity. With antitrust intervention there

has been proved that no reinforcing of innovation can be made, and

moreover, it would damage to the intellectual property law (trade-o_

between the bene_ts of competition and the protection of innovation

from direct imitation). Competition in the B market brings product va-

riety and lower costs and prices. Therefore, it augments the value of the

bottleneck good and U's pro_t when the two goods are complementary

and not tied. Bundling and foreclosure must have e_ciency-objectives

and predatory intents. Motivations for bundling may be not related to

competition (distribution and compatibility cost savings, liability and

reputation, market segmentation and protection of intellectual prop-

erty, etc) therefore bundling may be used as a reasonable act by the

_rms.

4 Exclusive customer contracts and e_-

ciency arguments

Firms may use their market power (through long term exclusive ar-

rangements) in order to protect their position in the same market,

even in absence of interaction with related markets. Firms could "lock"

users through exclusive contracts (probably with the objective to ex-

tract some of the entrant's technological advantage). There are several

types of exclusionary techniques:

(a) penalty for breach agreements

(b) renegotiation they have an exclusionary impact given that they

rely on the assumption that U and D _rm cannot renegotiate their

contract, once the entrant E has made an o_er. From the welfare

point of view, exclusivity leads to over-investment relative to what

would be socially desirable, whenever some conditions are meet in

terms of c.d.f. of E's cost ([1􀀀 ^ F(_c)](_c􀀀c))+

R _c

c d ^ F(^c) < I < _c􀀀c:

Exclusivity contracts in which downstream customers commit to

purchase from an upstream supplier, may deter investments by

competing upstream suppliers.

In the rent extraction perspective, penalties for breach agreements are

used to force a more e_cient entrant to reduce its price. In the entry-

deterrence theory, penalties for breach arise a free-riding problem of

customers, when the entrant faces a large _xed cost for which it needs

a large market in order to become a real competitor.

E_ciency arguments for vertical foreclosure are:

(a) forbearance as a reward to investment or innovation

(b) free-riding by downstream units on the marketing expenses

(c) excessive entry

(d) monitoring bene_ts of vertical integration

(e) costly divestitures

(f) costly expansion of capacity or costs incurred in order to provide

access

(g) fear of being associated with inferior downstream partners that

could damage the _rms's reputation

(h) universal service

E_ciency arguments for tying:

14

(a) preventing ine_cient substitution

(b) metering

(c) signaling quality

5 Conclusion