Multiproduct Monopoly, Commodity Bundling, and Correlation of Values

Author(s): R. Preston McAfee, John McMillan and Michael D. Whinston

Reviewed work(s):
Source: The Quarterly Journal of Economics, Vol. 104, No. 2 (May, 1989), pp. 371-383
Published by: Oxford University Press
Stable URL: http://www.jstor.org/stable/2937852 .
Accessed: 03/02/2013 21:48

Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at .
http://www.jstor.org/page/info/about/policies/terms.jsp

.
JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of
content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms
of scholarship. For more information about JSTOR, please contact [email protected].

Oxford University Press is collaborating with JSTOR to digitize, preserve and extend access to The Quarterly
Journal of Economics.

http://www.jstor.org

This content downloaded on Sun, 3 Feb 2013 21:48:36 PM

All use subject to JSTOR Terms and Conditions
 MULTIPRODUCT MONOPOLY, COMMODITY BUNDLING,
AND CORRELATION OF VALUES*

R. PRESTON McAFEE
JOHN MCMILLAN
MICHAEL D. WHINSTON

I. INTRODUCTION
Through what selling strategy can a multiproduct monopolist
maximize his profits when his knowledge about individual consum-
ers' preferences is limited? One possibility, extensively studied in
the context of a single-good monopoly, is to use quantity-dependent
pricing as a means of discriminating among customers with dif-
fering tastes (see, for example, Oi [1971] and Maskin and Riley
[1984]).
An alternative technique for price discrimination, first sug-
gested by Stigler [1968] and analyzed further by Adams and Yellen
[1976], is for the monopolist to package two or more products in
bundles rather than selling them separately.' Through a series of
examples Adams and Yellen illustrate that bundling can serve as a
useful price discrimination device, even when all consumers' will-
ingnesses to pay for each of the goods individually are unaffected by
whether they are also consuming the other product. A typical
example is illustrated in Figure I (adapted from Figure IV in Adams
and Yellen), where there are two goods, three consumers (AB,C)
who consume at most one unit of each good (with reservation values
for each good that are independent of whether the other good is
consumed), and zero costs of production. There, a bundle offered at
a price of 100 fully extracts all potential surplus, which would be
impossible pricing the goods independently. Unfortunately,
though, these authors do not provide any general characterization
of the circumstances in which bundling is actually a multiproduct
monopolist's optimal strategy. Their examples, however (such as
Figure I), create the impression that the profitable use of bundling

*This note bundles results previously offered independently in McAfee and

McMillan [1985] and Whinston [1986]. We thank Ignatius Horstmann and Gerry
Koumatos for comments. The third author also thanks the Federal Trade Commis-
sion for its hospitality during the time that this work was done and the NSF for
financial support (SES-8618775).
1. Examples are film distributors' bundling of assorted movies for sale to
exhibitors [Stigler, 1968] and IBM's bundling of maintenance and programming
services with computers [Scherer, 1980, p. 319].

? 1989 by the President and Fellows of HarvardCollege and the MassachusettsInstitute of

Technology.
The Quarterly Journal of Economics, May 1989

This content downloaded on Sun, 3 Feb 2013 21:48:36 PM

All use subject to JSTOR Terms and Conditions
372 QUARTERLY JOURNAL OF ECONOMICS

Valuationof
good 2
100

100 Valuationof good I

FIGURE I

is somehow tied to the presence of a negative correlation be-

tween the reservation values of the goods across the population of
consumers.2
In this note we investigate the conditions under which bun-
dling is an optimal strategy in the Adams and Yellen [1976] model.
Our analysis distinguishes between cases where the monopolist can
and cannot monitor the purchases of consumers. For each of these
cases we provide and interpret sufficient conditions for bundling to
dominate unbundled sales. One implication of these conditions is
that bundling is always an optimal strategy whenever reservation
values for the various goods are independently distributed in the
population of consumers (regardless of the monopolist's ability to
monitor purchases). In addition, when purchases can be monitored,
bundling dominates unbundled sales for virtually all (in a sense
made precise below) joint distributions of reservation values.
The analysis here is related to two other recent papers.
Schmalensee [1984] considers the optimality of bundling in the
Adams and Yellen model for the special case of a joint normal
distribution of reservation values. Using both analytical and
numerical technqiues, Schmalensee's results illustrate, among

2. For an expression of this view in the literature, see Schmalensee [1982],

p. 71.

This content downloaded on Sun, 3 Feb 2013 21:48:36 PM

All use subject to JSTOR Terms and Conditions
 MONOPOLY, BUNDLING, AND VALUES 373

other things, that bundling can be optimal even when the correla-
tion between reservation values in the population is nonnegative.
Spence [1980], in an elegant paper, considers the multiproduct
quantity-dependent pricing problem. Though much of Spence's
focus is on computational aspects of the problem, in an appendix he
provides an example with an independent distribution of prefer-
ences for two goods in which the multiproduct seller's optimal
strategy differs from the outcome that would arise with two single-
product monopolists engaged in quantity-dependent pricing. Given
the differences between Spence's example and our model, his
findings provide an interesting complement to the results here.3
In Section II we briefly describe the model. In Section III we
present our results, distinguishing between cases where monitoring
is and is not possible. Finally, in Section IV we briefly discuss the
implications of our results for models of multiproduct oligopoly.

II. THE MODEL

The model is essentially that of Adams and Yellen [1976]. A
multiproduct monopolist sells two products, goods 1 and 2, which
are produced at constant marginal costs of cl and c2, respectively.4
Each consumer desires at most one unit of each good, demands each
independently of his consumption of the other, and is characterized
by his reservation values for each of the two goods, (v1,v2) > 0.
Reservation values for each of the two goods are jointly distributed
in the population according to the distribution function F(v1,v2). In
contrast to Adams and Yellen, we assume that this distribution
possesses no atoms, and represent its density function by f(v1,v2).
Let gi(vi Iv3) and hi(vi) denote the conditional and marginal densi -
ties derived from f(.,.); Gi(viIv1) and Hi(vi) denote the conditional
and marginal distribution functions (for i = 1,2). Also, in order to
avoid trivial outcomes, we assume that for each good i there exists a

3. In particular, in Spence's example consumer preferences have the form

0xu(x) + Oyu(y) - P, where x and y are the quantities of the two goods purchased, P is
the total payment, Oxand Oyare iid on {0, 01, and u(.) is increasing, differentiable, and
concave. Thus, the monopolist in Spence's example possesses the ability to
completely separate the four consumer types (quantities can be varied continu-
ously). Here, on the other hand, the monopolist faces a continuum of consumer
types, but has a much cruder ability to segment consumers (since he can offer only
three different bundles for sale: one unit of x, one unit of y, and a bundle of one unit
of each). We discuss the relationship between our results and Spence's example
further in footnotes 10 and 16.
4. Neither the assumption of two goods nor that of constant marginal costs is
actually critical for the results to follow. The assumptions are made to ease the
exposition and facilitate comparison with the existing literature.

This content downloaded on Sun, 3 Feb 2013 21:48:36 PM

All use subject to JSTOR Terms and Conditions
374 QUARTERLY JOURNAL OF ECONOMICS

positive measure of consumers who have vi > ci. Last, resale by

consumers is assumed to be impossible.
The monopolist can choose one of three pricing policies. First,
he could simply price each commodity separately. Second, he could
offer the goods for sale only as a bundle, with a single bundle price.
Third, he could offer to sell either separately or bundled, with a
price for the bundle that is different from the sum of the single-good
prices. Following Adams and Yellen, we shall refer to the two kinds
of bundling as pure bundling and mixed bundling. We can, however,
immediately rule out pure bundling as a (uniquely) optimal strate-
gy, because mixed bundling is always (weakly) better: mixed bun-
dling with a bundle price PB and single-good prices P1 = PB - C2 and
P2 = PB - c1 always yield profits at least as high as pure bundling
with price PB, and typically does better [Adams and Yellen, p. 483].5
Thus, we now turn to a comparison of the profits obtainable from
mixed bundling and unbundled sales.

III. RESULTS

In considering possible mixed bundling strategies, it is impor-

tant to distinguish between cases where the monopolist can and
cannot monitor purchases. In the latter case he is effectively
constrained to offer a bundle price PB which is no larger than the
sum of individual goods prices, P1 + P2, for otherwise consumers
would never buy the bundle. On the other hand, if the monopolist
can monitor purchases, then he faces no such constraint since he
can always prevent consumers from purchasing both good 1 and
good 2 separately.6
In what follows, we employ one additional assumption on the
distribution of preferences: that gi(Pi s) is continuous in Pi at Pi*
for all s, where Pi' is the optimal nonbundling price for good i.7

5. In principle, the monopolist could also use a randomized strategy, making

the allocation of the goods to a buyer stochastic. We shall ignore this possibility in
the text, but see footnote 7 for a further discussion of this point.
6. The literature on single-good quantity-dependent pricing (e.g., Maskin and
Riley [1984]), for example, implicitly assumes that purchases can be monitored since
the monopolist does not worry about consumers purchasing multiple subquantities
(though this problem disappears when quantity discounts exist just as it disappears
when bundle discounts are used here).
7. It can be shown that a simple pricing policy of this sort is the optimal
single-good selling strategy as long as the single-good profit function is concave in
price. In particular, in that case no randomized selling procedure can improve
profits. McAfee and McMillan [1988] also derive a sufficient condition for the
optimal multiproduct sales policy to be nonstochastic. Their condition actually
implies the condition (1) that we derive below, which ensures that bundling
dominates unbundled sales.

This content downloaded on Sun, 3 Feb 2013 21:48:36 PM

All use subject to JSTOR Terms and Conditions
 MONOPOLY, BUNDLING, AND VALUES 375

Obviously, one sufficient condition for this to be true is that f (v1,v2)

is continuous, but we prefer to state the assumption in this
nonprimitive way in order to emphasize that the set of cases in
which the results apply is really much larger than that.
We begin by establishing a general sufficient condition (valid
regardless of whether purchases can be monitored) for bundling to
dominate unbundled sales.

PROPOSITION1. Let (P*,P2*) be the optimal nonbundling prices.

Mixed bundling dominates unbundled sales if8

(1) fS {[1 - G2(P2* Is)] - g2(P2* Is)(P2* - C2)} hl(s)ds

+ (P* - > 0.
ci) [1 - G2(P2* IPr)]hl(Pr)
Proof of Proposition 1. The proof proceeds by arguing that a
local improvement can be made regardless of whether purchases
can be monitored if condition (1) holds.
Suppose that the inequality in (1) is satisfied. First, introduce a
bundle whose price is PB = Pi* + P2* (and leave the single-good
offers unchanged). Clearly, profits are unchanged since the bundle
is irrelevant due to its pricing. Now consider a small increase in the
price of good 2 to P2 = P2* + E,where e> 0. Note that the ability to
monitor purchases is irrelevant here since PB < Pi* + P2. The
resulting pattern of purchases, which is depicted in Figure II, is
given by
Good 1 only:
(i) V -P* 0
(ii) V2 - P2* ' ?

Good 2 only:
(i) V2-P2*- 2 0
(ii V - P* + e '< 0

Bundle:
(i) V1 + V2 -Pi* -P2* >- 0
(ii) V2-P2* 2 0

(iii) V - P* + E 2- 0.

8. Schmalensee [1984] utilizes an analogous condition in his analysis of the

special case of a Gaussian distribution of reservation values.

This content downloaded on Sun, 3 Feb 2013 21:48:36 PM

All use subject to JSTOR Terms and Conditions
376 QUARTERLY JOURNAL OF ECONOMICS

V 2 onlyf
~~~Good Supportof f (,

221d l B

0 ~ ~~~~~~~~~~~o onlon y

Pt-E P* V1

FIGURE II

Since the actions of and profits from consumers with v1 > P* are
unaffected by this change in P2, we need only focus on the change in
profitability from sales to those with v1 < P*. Profits from this
group as a function of e are given by

= (P* ?e-C2) p {f f(s,q)ds} dq

+ (P* + P* - C- C2) {4;+prS f(sq)dq }ds.

P - 'E +P*-s

Differentiating this expression with respect to e and taking the limit

as -e 0 yields the expression in condition (1) implying that 4'(O)> 0
so that a strict improvement in profits would be possible through
bundling.
Q.E.D.
Condition (1) is derived by starting at an initial position with
the optimal independent goods pricing prices (P*,P*) and a bundle
price equal to the sum of these prices, and then marginally increas-
ing the price of good 2. In principle, one could also consider
marginally raising the price of good 1 or marginally lowering the

This content downloaded on Sun, 3 Feb 2013 21:48:36 PM

All use subject to JSTOR Terms and Conditions
 MONOPOLY, BUNDLING, AND VALUES 377

price of the bundle from the initial position. It is not difficult to

show, however, that either of these changes gives rise to a marginal
change in profits that is identical to that in the expression in
condition (1) (this follows from the first-order conditions for P* and
P*), and so there is no reason to consider these conditions indepen-
dently.
One case of special interest is that of independently distributed
reservation values. Condition (1) of Proposition 1 leads immedi-
ately to the following result:
COROLLARY 1. If v1 and v2 are independently distributed, then bun-
dling dominates unbundled sales.
Proof of Corollary 1. For the case of independently distributed
reservation values, condition (1) reduces to (note that hi(P) =
gi(P Is) for all (P,s) and i = 1,2):

(2) H1(Pr) {[1 - H2(P )] - h2(P*) (P* - C2)}

+ (P* - cl) hl(Pr) [1 - H2(P2*)] > 0.

But, if P* is the optimal unbundled price for good 2, then the first
term in (2) is equal to zero, so that (1) reduces to
(3) (Pr - ci) h1 (Pr)[1 - H2(P*)]> 0.
Now, by the assumptions of no atoms and existence of a positive
measure of valuations above cost, (Pr - c1) [1 - H2(P*)] > 0. Also,
under our continuity assumption it must be that h1(Pr) > 0 (again,
from the nonbundling first-order condition). Thus, condition (3)
holds, and a local gain from bundling is possible.
Q.E.D.
It is worthwhile to consider the independence case graphically.
In Figure II one can see three first-order effects of locally raising
P2.9 First, there is a direct price effect from raising revenues
received from consumers in the set {(v1,v2)lvl < P*, V2 2 P2*}.
Second, sales of good 2 fall by the measure of area (abcd). Third,
sales of good 1 increase by the measure of area (defg) due to
consumers switching from purchasing only good 2 to purchasing the
bundle. The sum of the first two of these effects is exactly the local
gain if the monopolist was able to slightly raise the price of good 2
only to consumers with valuations less than P*. With indepen-
dence, however, this local gain must be zero at the optimal price P*2

9. In the case of independentdistributionsthe set {(v1,v2)I f(v1,v2) > 01 would

actuallybe rectangular.

This content downloaded on Sun, 3 Feb 2013 21:48:36 PM

All use subject to JSTOR Terms and Conditions
378 QUARTERLYJOURNALOFECONOMICS

because the monopolist desires the same good 2 (unbundled) price

regardless of a consumer's level of v1 (this follows from the fact that
the distribution of reservation values for good 2 is independent of
the value of vj). Thus, the net effect of moving to mixed bundling
when reservation values are independently distributed is positive.
It is also instructive to consider graphically (Figure III) the
effect of marginally lowering PB by E (recall that this change in
profits is numerically identical to that of raising P2). To first-order,
the monopolist loses e from consumers in area (abe), but gets
consumers in area (aefg) to switch from buying only good 1 to also
buying the bundle, and consumers in area (bcde) to buy the bundle
instead of good 2 only; the first two of the effects, however, add up
to zero due to the initially optimal level of P*. What is interesting to
note about this fact is that if we imagine a monopolist who can
quote a price for each good i depending on a consumer's level of v;,
the monopolist would have to lower P1 by Efor all v2> P* and P2 by e
for all v1 > P* in order to generate these same demand shifts, but at
a loss of 2 E from consumers in area (abeh). Thus, lowering PB is

Good2 only
V2 c b

ae
P*u
2 V d1ol
d

~
IE P1 VI
FIGUREIII

This content downloaded on Sun, 3 Feb 2013 21:48:36 PM

All use subject to JSTOR Terms and Conditions
 MONOPOLY,BUNDLING, AND VALUES 379

worthwhile here even though lowering prices in an independent

goods pricing regime would not be.10
Proposition 1, of course, implies that bundling is generally
optimal in a much broader range of cases than just independence.
The second term on the left side of inequality (1) corresponds to the
rectangular area (efgd) in Figure II: this is the extra profit from
consumers who buy both goods with bundling when without bund-
ling they would buy only one good. This effect will always tend to
generate gains from bundling. The first term in inequality (1),
though, may be either positive or negative (in the independence
case it is zero). We can, however, provide an intuitive sufficient
condition for this term to be nonnegative: imagine again that the
monopolist can observe vj but not vi, and let Pi (vj) be the monopo-
list's optimal price for good i conditional on knowing that a
consumer's valuation for good i is vj.1'Then, if P* (v1) is decreasing
in v1, the first term inequality (1) must be positive (since we know
that fo7{[1 - G2(P2* Is)] - g2(P2*Is) (P2* - c2)} h1(s)ds = 0) thereby
ensuring that bundling dominates unbundled sales (regardless of
whether purchases can be monitored).12
It might be thought that this condition-that those with low
values of v; should be charged a higher price for good i-is directly

10. The effects noted here and in the previous paragraph have interesting
parallels to those in Spence's [1980] example. There he shows that if the bundles for
types (0,0), (0,0), (0,0), and (0,) are (x,x), (y,z), (z,y), and (e,e), that z = e = the
first-bestlevel for type A,while x < x* < y, where x* is the level purchased by type 0
in the independent goods (quantity-dependent) pricing scheme. The motivation foir
setting y > x* there is that a small reduction in the charge to type (04) allows an
increase in profits from both types (0,0) and (O0,0)through an increase in their
consumption that is equal to what would arise in a single-good pricing problem with
just groups (0,0) and (0,0); an effect that closely parallels the effect here of lowering
PB. Likewise, the driving force behind having x < x * is that increasing the charges to
types (, 0) and (0,b) and lowering x from the single-good quantity dependent pricing
levels has no first-order effect, but relaxes the constraints between type (04) and
types (0,0) and (#0,0),thereby allowing profits to be raised by increasing the latter
groups consumption levels; an effect analogous to what occurs when we raise P2
marginally here.
11. This interpretation of condition (4) and what follows assumes that the
problem of picking an optimal price for good i given a value of v; is concave.
12. On the other hand, if the inequality in condition (1) is not satisfied, and if
profits are concave in (P1,P2,PB) on the set {(P1,P2,PB) IP1 + P2 ; PB1, then single-
good pricing is optimal when purchases cannot be monitored. To see this, note that
the gradient of profits at the single-good optimal prices (with PB = P* + P*), Vir*,
satisfies V7r* * AP < 0 for all AP such that AP1 + AP2 < APB. Unfortunately,
however, we have not found any general condition under which profits are concave in
this manner. In fact, the profit function in this problem appears to have inherent
nonconcavities. The usual sorts of monotonicity assumptions on f(., .), for example,
do not ensure concavity (e.g., note that the value of the shift of marginal consumers
on segment fh in Figure II from single-good to bundled sales may increase or de-
crease as P2 increases depending upon whether (PB - C1- C2)is larger or smaller
than (P1 - cl)).

This content downloaded on Sun, 3 Feb 2013 21:48:36 PM

All use subject to JSTOR Terms and Conditions
380 QUARTERLYJOURNALOFECONOMICS

tied to the presence of a negative correlation of reservation values.

This is not completely accurate, however. Defining Ei(Pi Iv) = Pi
{gi(PiIvj)1[1 - Gi(Pi vj)] to be the demand elasticity of good i
conditional on valuation v; for good j, it is easy to see that13

(4) sind
sign () = sign - P
dvj dvj
-signn d [1 GP*Iv
'I -gi(GPi*l
I~jvj)"
dgi(Pj v)_gPIv) dGi(P Iv
=sign f[l - Gi(PI Iv)] d i(PI ' v

Now, it can be shown that dGi(Pi Ivj/dvj > 0 (respectively, < 0) for
all Pi implies a strictly negative (respectively, positive) correlation
between v1 and v2.14But, the presence of the expression dgi(Pi Ivj)l
dvj in (4) (which cannot be signed a priori), indicates that the sign of
the first term in condition (1) cannot be tied solely to the level of
correlation between reservation values.
Our final result demonstrates that if purchases can be moni-
tored, then mixed bundling will dominate unbundled sales for
virtually any joint distribution of reservation values.
PROPOSITION2. Let (Pr ,P*) be the optimal nonbundling prices.
Suppose that the monopolist can monitor sales. Then bundling
dominates unbundled sales if

(5) f {[1 - G2(P* Is)] - g2(P* Is)(P* - c2)}h1(s)ds

(P* - ci) [1 - G2 (P2*IP*)]hi (Pr) ? 0.

?

Proof of Proposition 2. If (5) is violated because the expression

is positive, then Proposition 1 applies. Suppose then that the
expression in (5) is negative. Again introduce a bundle with price
PB = Pr + P*. Now, however, lower P2 slightly to P* - E, where e >

13. For a slightly different interpretation of the sufficient condition for bun-
dling to be profitable, note that the expression {[1 - G*(-) ]/gj(* )} commonly arises
in adverse-selection problems (see, for example, Myerson [1981], Maskin and Riley
[1984], McAfee and McMillan [1987]). Its expectation is the expected difference
between the first-order and second-order statistics, which is exactly the amount of
rent the buyer must be left with if he is not to understate his valuation. If this
informational rent decreases in the valuation of the other good, then the optimality
of bundling is ensured.
14. The covariance between vl and v2 can be written as f{J[v2 - E(V2)]
2(V2Ivl)dv2} [vl - E(vD)]hj(vj)dvj. If, for example, G2(. Ivj) is decreasing in vi, then
the expression in curly brackets is increasing in vl (see Milgrom [1981] ), and thus the
entire expression is positive.

This content downloaded on Sun, 3 Feb 2013 21:48:36 PM

All use subject to JSTOR Terms and Conditions
 MONOPOLY,BUNDLING, AND VALUES 381

Good2 only
V2 Bundle

P2-E
1only
Pane~~~~~~~~~~~Go

*
P* P +E VI
FIGUREIV

0. The resulting distribution of sales is depicted in Figure IV.

Analysis of the limit of the derivative of profits with respect to E as
e 0 then indicates that a gain in profits is available.15
Q.E.D.
It should be clear from condition (5) that independent goods
pricing will virtually never be an optimal sales strategy here when
purchases can be monitored.16 By providing an additional means for
the monopolist to segment groups of customers, bundling essen-
tially always raises profits in this case.

IV. OLIGOPOLY
It is reasonably straightforward to apply the results above to
the case of multiproduct oligopoly. For example, consider a duopoly

15. We omit the mathematical details; the point should be obvious from Figure
IV and is analogous to the argument in the proof of Proposition 1.
16. Spence also considers the use of bundling for nonindependent distributions
of preferences in his example. Interestingly, in contrast to our result here, he finds
that a range of (positive) levels of correlation exists such that independent pricing is
optimal. This finding arises in Spence's model because of a discontinuity in the
derivative of profits with respect to a change in the charge to type (0, 0) at the
independent goods (quantity dependent) pricing outcome (recall footnote 10). Here,
no corresponding discontinuity exists.

This content downloaded on Sun, 3 Feb 2013 21:48:36 PM

All use subject to JSTOR Terms and Conditions
382 QUARTERLY JOURNAL OF ECONOMICS

comprised of firm A and firm B,'7 each of which produces a version

-of products 1 and 2. Consumer valuations for the goods are repre-
sented by four reservation values: (V1A,V1B,V2A,v2B). The firms
engage in simultaneous price choices (which may involve indepen-
dent goods pricing, pure bundling, or mixed bundling).
In this setting, given the pricing choice of its rival, each firm
acts as a monopolist relative to the demand structure induced by its
rival's prices. When will independent pricing arise as an equilib-
rium? Suppose that firm B is pricing its products independently at
prices (PlB,P2B). Then we can define each consumer's "pseudo-
reservation value" for firm A's two products as
V1 = VlA - maxf{OlvB - P1BI
V2 = V2A - max{Ov2 - P2B}-

If we let f(b1,b2) be the distribution of these induced reservation

values, it is not difficult to show that independent pricing can only
be a Nash equilibrium if condition (1) (or (5) if monitoring of
purchases is possible) fails to hold for the pseudo-reservation values
induced at the independent pricing Nash equilibrium prices. An
interesting implication of this fact is that independent pricing
cannot be a Nash equilibrium if the distribution of the reservation
values for product 1 is independent of the distribution for product 2
since in that case the induced pseudo-reservation values are them-
selves independent (so Corollary 1 applies).'8

UNIVERSITY OF WESTERN ONTARIO

UNIVERSITY OF CALIFORNIA,SAN DIEGO
HARVARD UNIVERSITY

REFERENCES
Adams, W. J., and J. L. Yellen, "Commodity Bundling and the Burden of Monop-
oly," this Journal, XC (1976), 475-98.
Maskin, E., and J. Riley, "Monopoly with Incomplete Information," Rand Journal
of Economics, XV (1984), 171-96.
McAfee, R. P., and J. McMillan, "Commodity Bundling by a Monopolist," mimeo,
University of Western Ontario, December 1985.
, and , "Auctions and Bidding," Journal of Economic Literature, XXV
(1987), 699-738.
, and , "Multidimensional Incentive Compatibility and Mechanism
Design," Journal of Economic Theory, XLVI (1988), 335-354.

17. The points made here easily extend to any finite number of firms.
18. The reservation values for product 1 are said to be independent of those for
product2 if the density of (v1A,v1B,v2A,v2B),0( ), can be written as 0v1A,v1B,v2A,v2B)
g(V1A V1B) - h(v2A,v2B) -

This content downloaded on Sun, 3 Feb 2013 21:48:36 PM

All use subject to JSTOR Terms and Conditions
 MONOPOLY,BUNDLING, AND VALUES 383

Milgrom, P., "Good News and Bad News: Representation Theorems and Applica-
tions," Bell Journal of Economics, XII (1981), 380-91.
Myerson, R. B., "Optimal Auction Design," Mathematics of Operations Research,
VI (1981), 58-73.
Oi, W. Y., "A Disneyland Dilemma: Two-Part Tariffs for a Mickey Mouse Monop-
oly," this Journal, LXXXV (1971), 77-90.
Scherer, F. M., Industrial Market Structure and Economic Performance(Chicago,
IL: Rand McNally, 1980).
Schmalensee, R., "Commodity Bundling by a Single-Product Monopolist," Journal
of Law and Economics, XXV (1982), 67-71.
, "Gaussian Demand and Commodity Bundling," Journal of Business, LVII
(1984), S211-30.
Spence, A. M., "Multi-Product Quantity-Dependent Prices and Profitability Con-
straints," Review of Economic Studies, XLVII (1980), 821-42.
Stigler, G. J., "A Note on Block Booking," in G. J. Stigler, ed., The Organization of
Industry (Homewood, IL: Richard D. Irwin, 1968).
Whinston, M. D., "Product Bundling and Correlation of Values," mimeo, Harvard
University, August 1986.

This content downloaded on Sun, 3 Feb 2013 21:48:36 PM

All use subject to JSTOR Terms and Conditions