Beruflich Dokumente
Kultur Dokumente
Howard Smith
University of Oxford
Introduction
Vertically related industries are endemic: many (if not most) rms do not sell
directly to consumers but rather to other rms.
Example: Wood pulp producer ! paper producer ! stationary wholesaler !
stationary retailer ! consumers.
The dierent production processes in the vertical chain may take place in different rms (vertical separation) or within the same rm (vertical integration).
Under vertical separation, one rm may impose vertical restraints (nonlinear
pricing, retail price maintenance, exclusive dealing, etc.) on another rm in the
vertical chain.
Questions:
marginal cost c U
marginal cost c D
Vertically
integrated
structure
p
Consumers
Qp
Vertical Integration
Suppose U and D are vertically integrated.
Hence, the marginal cost of producing the nal good is cU + cD .
The vertically integrated rm sets the monopoly price pM (cU + cD ), where
pM (c) arg max[p c]Q(p)
p
is the monopoly price of a rm with marginal cost c. The vertically integrated rms resulting prot is
M (c
+ cD ) =
pM (c
+ cD ) ( c U + c D )
Q pM (cU
+ cD ) :
U
Vertically
separated
structure
marginal cost c U
p
Consumers
Qp
marginal cost c D
max[p w cD ]Q(p);
p
max[w cU ]Q pM (w
w
+ cD ) :
b satises w
b > cU .
The equilibrium wholesale price w
b > cU , and since the monopoly (retail) price is increasing in the
Since w
marginal cost that the (retail) monopolist faces, the equilibrium (retail)
price for the nal good, pb, satises pb > pM (cU + cD ).
Remarks:
Vertical separation induces a higher price of the nal good (and therefore a lower output).
When rms are vertically separated, there is double marginalization
(Spengler, Journal of Political Economy, 1950): under imperfect competition, each rm adds its own prot margin at each stage of production. (What is worse than a monopoly? A chain of monopolies.)
Under vertical separation, there is an inherent vertical externality: each
rm does not take into account that an increase in its (wholesale or
retail) price reduces the prot of the other rm. This externality is
internalized under vertical integration.
Discussion
But this result relies heavily on the assumption of linear wholesale prices:
it would disappear if the upstream rm could use a two-part tari. In
that case, U would charge (w; F ) = (cU ; M (cU + cD )). The wholesale
price w = cU would induce D to set the monopoly price under vertical
integration pM (cU + cD ), and the xed (or franchise) fee F = M (cU +
cD ) would allow U to extract the entire rent of the vertically integrated
structure. Essentially, by oering the optimal two-part tari, U sells the
vertical structure to D who thereby becomes the residual claimant.
A two-part tari is not the only vertical restraint that would solve the
vertical externality. Alternatively, the upstream rm could impose retailprice maintenance (RPM) with p = pM (cU + cD ) and charge a linear
wholesale price w = pM (cU +cD )cD . The downstream rm would make
zero prot and the vertical structures joint prot would be M (cU + cD ),
as under vertical integration.
RPM is usually illegal but downstream competition is an alternative "restraint" for the upstream rm: if U sets w = pM (cU + cD ) cD then
sells to several Bertrand competitors, this results in p = pM (cU + cD ) .
Uncertainty Rey and Tirole (1986) compare the alternatives franchise fees
and RPM in the two cases where U is uncertain about (i) demand Q()
and (ii) downstream cost cD . D learns about these in time to set prices
but after the vertical contract is agreed.
Franchise fee: Since U is uncertain of M (cU + cD ), nonlinear wholesale prices that make D a residual claimant impose risk on D. If D is
risk neutral ith will accept xed
i fee equal to expected monopoly prot
i.e. F = E M (cU + cD ) If D is risk averse it will only accept a
lower F , which is costly to U:
The advantage of a xed fee (relative to RPM) is that prices are set
by the informed party (D) but the disadvantage is that reducing F is
costly to U:
Downstream pre-sale service free-riding may also result when pre-sale advice may be obtained from one seller but the good purchased from another.
For much of the 20th century, U.S. courts were hostile towards vertical
mergers and vertical restraints such as RPM (per se illegal), exclusive
dealing contracts (rule of reason), and tied sales.
From the 1950s, the Chicago School (Director, Posner, Bork, etc.) attacked legal practice:
Chicago School had profound impact on antitrust law/economics. In contrast to their opponents, the proponents of the Chicago School used formal
models (albeit simple models of monopoly and perfect competition).
Exclusive Contracts
Sequence of moves:
Assume that if E were to enter and sell to D, it would charge the price
p = cU as cU pM (cE ).
Assume also that E would enter in the absence of an exclusive contract:
(cU cE )Q(cU ) > F .
p Mc U
cU
cE
q
Qp Mc U
Z pM (c )
U
cU
Q(z )dz:
But
Z pM (c )
U
cU
and so U will not nd it protable to oer a suciently high payment that will
induce D to sign the exclusive contract.
Note: The Chicago argument does not rely on the assumption that U can make
a take-it-leave-it oer to D. The result is robust to any ecient bargaining
between U and D.
0
1
if p > v
if p v
The contract between U and D is now of the form (p; d), where p is
the price that D has to pay to U for one unit of the good and d is the
stipulated damage that D has to pay to U if D buys instead from E .
U has a rst mover advantagesets d before entrant can enter.
Sequence of moves:
Suppose E has decided to enter the market and has oered the good at
price pE . Then D will purchase from E if U s eective price p d is
not smaller than pE .
Eective price is the incremental cost of buying versus not buying from U
allowing for d
Note: In eect, the contract does not exclude E but allows the incumbent U
to extract all of the surplus from the more ecient entrant E .
Suppose now that there is uncertainty about E s marginal cost cE .
For example, suppose that v = 1, cU = 1=2, cE is uniformly distributed
on [0; 1], and F = 0.
In this case, the ecient solution would be that E enters if and only if
cE < 1=2.
(1)
Pr(cE < ) is the probability the entrant can price below and hence
enters
In a sense, partial exclusion is a side eect of the (optimal) contract between U and D. The objective of the contract is not to exclude but to
extract rents from E in order to maximize maximize the joint prot of U
and D at the expense of E .
The End