Price Discrimination Through
Price Discrimination Through
Price Discrimination Through
Png
D. Hirshleifer
University of California, Los Angeles
0021-939818716003-0003$01.50
1. For a general review of the literature on coupons, see Vilcassim and Wittink (1985).
Price Discrimination 367
and
Assume that the tourist demand and the local demand are such that
the following two assumptions hold.
ASSUMPTION 1.
arg max [IIp(p)] < r.
ASSUMPTION 2. IIe(.) is twice differentiable and G ( p ) < 0.
4. The discriminatory effect of such discounts is the focus of Shilony (1977), Rosen-
thal (1980), and Varian (1980).
370 Journal of Business
= [ Q ~ P+
) Q t l b - c)dF(p) = neb.,) + nt(p,), (5)
Pm
It is likely that the firms in such a situation will appreciate that there
are gains from coordination. In a static model, they may exploit exter-
nal devices to achieve a correlated eq~ilibrium.~ That is, they may
adopt strategies that specify moves contingent on external signals. For
instance, one strategy might be, Set list price at p , if the closing Dow
Jones the previous day is even and at r otherwise. The best response to
such a strategy is, Set a list price at r if the closing Dow Jones the
previous day is even and at p , ~ t h e r w i s e . ~
There is a multiplicity of possible correlated equilibria. First, there
are many possible external signals on which the equilibrium may be
based. Second, for any given signal, there are multiple equilibria that
differ in the division of profits between the two sellers. We shall focus
attention on a symmetric equilibrium, in which the external signal is
such that, with probability one-half, one firm sets list price p , and the
other list price r and, with probability one-half, the roles are reversed
between the two firms. In this equilibrium, the expected profit of a firm
will be
6. In a setting of repeated plays, a seller may play a strategy of the form, Price atp, on
odd dates and at r on even dates. The best response to this is the strategy, Price at r on
odd dates and at p, on even dates. By alternating their prices, the two firms can achieve
an equal division of profit. A similar practice occurred in the Great Electrical Equipment
Conspiracy of 1955-56: General Electric, Westinghouse, I-T-E, and Allis-Chalmers
coordinated their bids to supply large turbine generators by the phases of the moon (see
Sultan 1974, p. 39).
7. For the original reference on the subject of correlated strategies, see Aumann
(1974).
372 Journal of Business
The effective price to the locals is the minimum price quoted by the n
firms. The distribution of the minimum price is
pr[min(pi, . . . , P,) PI = 1 - pr[pi > p , for all i = 1, . . . , n]
nation. The larger the number of firms, the smaller is the pressure on
any firm to set a low price.*
In equilibrium, the profit for a firm if it sets list price p E [p,, r] is
given by
H(p : F,) =
I
+ nt(p) if P E [P,,
Srp,,,) Qe(x)(x ,- c ) . dGn(x)
r),
+ [ - Qtlne(r)l . Q~(P)(P- c)
+ IIt(r) i f p = r ,
where Gn(.) is the distribution of the minimum of the list prices of the
other n - 1 firms. In the proof of proposition 3, it is shown that the
expected profit from the strategy F,(.),
(9)
Hence
Hence
The analysis shows clearly not only that is it possible for price dis-
crimination to exist in the presence of competition but indeed that
there are situations in which the seller makes use of its competitors to
effect discrimination. Discrimination is effected when one firm lists a
low price: the competing sellers then list a higher price so as to extract
more profit from the tourists while effectively charging the lower price
to locals.
One of the major results is that, starting from a nondiscriminating
monopoly, the introduction of competition that enables discrimination
through offers to match price would lead to a reduction in total sales.
Moreover, the entry of additional firms would lead all sellers to raise
their list price and thus further reduce total sales. If sellers coordinate
their pricing, they will increase their profit by increasing their sales to
local customers. The essential reason for these outcomes is that each
firm would like to discriminate by setting a high list price for tourists
and taking advantage, through its offer to match, of a lower list price
elsewhere to charge a lower list price to local customers. Without
coordination, the larger the number of competing sellers, the less the
pressure on any particular firm to be the low-price firm. With coordina-
tion, the firms can economize on setting the low list price-only one
need do so at any instant for all firms to benefit.
From the standpoint of formulating marketing strategy, the analysis
indicates that a firm with several outlets should not impose a uniform
price on all its outlets. In this regard, our conclusion concurs with a
result of Salop (1977). Salop analyzed a model in which consumers
differed in their cost of information about prices at various outlets and
showed that a firm with several outlets would increase expected profit
by randomization of prices. The model of this paper shows, that the
firm should not give its outlets independence in pricing: it will max-
imize its profit by arranging that, at any time, only a single firm chosen
at random sets a low list price.
A possible extension would be to examine the implications of allow-
ing price-beating as well as price-matching offers. It may be conjec-
tured that this would lead to competitive pressures to have high list
prices since under price beating it may be the high-list-price firm that
captures the local patronage by undercutting the lower-priced competi-
tor.
Another interesting line of further inquiry is to analyze the incentive
for a low-price seller to advertise his price-for it is such information
that enables his competitors to discriminate. (There is one "discount"
electronics store in Los Angeles that maintains a strict policy of not
quoting prices over the telephone.) Another avenue for future research
is to study the seller's choice between alternative methods of price
discrimination. This could build on the existing work on various
methods of discrimination, which has treated each method in isolation.
Price Discrimination 377
Appendix
Proof o f lemma I . Suppose that the strategy of firm j is @. Consider the
strategy of firm i.
First, consider list prices pi 5 r. Suppose that firm j does not offer to match.
Hence, if firm i also does not offer to match, its sales to locals will depend on
the realized price of firm j. In particular, if pi < p j , firm i will capture all the
local customers, while ifpi > pi, it will receive no local customers. The sales of
firm i to tourists will be Q , regardless. Thus
n ( p i : @) =
1
2 Q e ( ~ i ). ( p i - C ) + Q f . ( p i - C ) i f pi < p j ,
Q e ( ~ i .) ( p i - C ) + Qt . ( p i - c ) ifpi = p j ,
Qt . ( p i - C ) if pi > pj.
If firm i does offer to match, it will capture all the local customers i f p i < pj,
but if pi 2 p j , firm i will receive exactly half the local customers-because they
will take advantage of the offer to match. Thus profit will be
2 Q e ( ~ i ). ( p i - C) + Q t . ( p i - C ) i f p i < pj,
Qe(pi) . ( p i - C ) + Qt . ( p i - C ) if pi = p j ,
Q e ( ~ i .) ( p i - C ) + Qt . ( p i - C ) i f pi > pj.
By offering to match, firm i can increase its profit in the event that pi > pj.
Hence, if firm j does not offer to match, it is weakly dominant for firm i to offer
to match.
The next case to consider is that in which firm j does offer to match. In this
case, firm i will receive exactly half the local customers when pi < pj; it will not
capture the whole local business. If firm i does not offer to match, its profit will
be
Q e ( ~ i ') ( p i - C ) + Q t . ( p i - C ) i f p i < p j ,
Q e ( ~ i .) ( p i - C ) + Q f . ( p i - C ) i f p i = pj,
Qt ( p i - C ) if pi > pj.
If firm i does offer to match, its profit will be
Q e ( ~ i ') ( p i - C ) + Qt . ( p i - C ) i f pi < pj,
Q e ( ~ i .) ( p i - C ) + Qf . ( p i - C ) i f p ; = pj,
Q e ( ~ i ') ( p i - C ) + Qt . ( p i - C ) i f p i > pj.
Again, it is weakly dominant for firm i to offer to match.
A similar argument provides the proof for the case that pi > r. The only
difference is that then firm i will sell nothing to tourists.
Proof o f lemma 2. In lemma 1, it was proved that all firms will offer to
match. Here, the style of proof will be to suppose that the strategy of one firm
is given by some distribution function cP, with p sfmin[p : @(p) > 01, and then
to consider the expected profit n ( p : @) to the remaining firm of alternative list
prices p.
a) For any P < P,,
W P : @) = [1 - @(PI]
while
378 Journal of Business
Hence
W p m : @) - n ( p : @) = - @(prn)I . [M~rn)
+ nt(~rn)I
Now p, = arg max [IIe(x) + H,(x)]; hence the first term in ( A l ) is positive.
Next, nE(.) is concave, and I&(.) is linear; therefore the function [n,(.)+ n , ( . ) ]
is concave. The function attains a maximum at p = p,. Hence for p 5 x 5 p,
we have &(x) + n , ( x ) 2 n e ( p ) + n t ( p ) . But n,(p,) > n , ( x ) since p, > x.
Therefore &(x) + n r ( p m )> neb)+ n t ( p ) ,and hence the second term in ( A l )
is positive.
The third term in ( A l ) is
1 [nt(p,) - n t ( p ) l d @ ( x ) =
while
r :) = 1 - ( 1 [ ( r ) + ( r ) ]+ [ne(x) + n,(r)]d@(x).
Thus
n ( r : @) - n ( p : @) = @(p)l +
b d@(x)'
I . [ n e ( r ) + n,(r)]
Price Discrimination 379
The function &(.) is concave and attains a maximum at p p < r < p . Hence,
for x E [r, p ] , n t ( r ) 2 &(x). Thus the first and second terms in ( A 2 ) are
positive. The third term is St; n,(r)d@(x) > 0; therefore n ( r : @) > H ( p : @).
Proof o f proposition 1. 1; lemma 1, it was shown that each firm will couple
its list price with an offer to match. To determine the equilibrium strategies, it
remains to solve for the list prices.
a) We first prove that, in pure-strategy equilibrium, the list price of each firm
will be either p , or r. From lemma 2, we know that any strategy with support
outside [p,, r] is weakly dominated and hence not an equilibrium strategy.
Therefore, in pure-strategy equilibrium, pi E [p,, r ] , i = 1 , 2.
Let firm 2 price at p2 E [p,, r ] , and study the pricing decision of firm 1.
Consider those p l > p,. Notice that, for such values of p l , the locals will take
advantage of firm 1's offer to match; hence its effective price to local customers
will be p*. Since both firms charge an identical price to locals, each will receive
half the local demand, that is, Qe(p2).
For all p 1 E [ p 2 ,r ] , the profit of firm 1 is
~ I ( P:I~ 2 =) Qt . ( P I - c ) + Qe(p2) + ( p 2 - c),
which is less than Q , . ( r - c ) + Q e ( p z ). ( p 2 - c). Moreover, for allpl > r , the
profit of firm 1 is
~ I ( P: I~ 2 =) + Qe(p2) . ( ~ -2 c),
0
which is less than Q , . ( r - c ) + Q e ( p 2 ). ( p 2 - c). Therefore the best response
that satisfies the constraint p l > p2 is p 1 = r.
Consider those p % p 2 . For such values of p ,, the effect of firm 2's offer to
match will be that the price to locals will be p l at both firms. Hence the profit to
firm 1 will be
n i ( ~:i~ 2 =) Qt . ( P I - C ) + Q e ( ~ 1 ) ( P I - c).
'
This is the same program as that for the nondiscriminating monopolist with the
addition of the constraint pl 5 p2. Now the unconstrained solution for the
nondiscriminating monopolist is p,. But p2 E [p,, r ] , and hence min ( p 2 ,p,)
= p,. Thus the best response for firm 1 is to set list price pl = p,.
By a similar argument beginning with some price pl for firm 1, it may be
demonstrated that firm 2 will choose a list price of either p, or r. This com-
380 Journal of Business
pletes the proof that, in pure-strategy equilibrium, the list price of each firm
will be either p, or r.
b) Now suppose that firm 2 sets p2 = r. By the definition of p,,
for all p . This implies that n l ( p m : r) 2 n l ( r : r) and hence that firm 1 will set
P I = Pm.
Suppose that seller 1 sets p l = p,. Now, profit of seller 2
and
n z ( r : ~ r n )= Qt . (r - C) + Qe(p,) . ( p , - c).
Since p, < r , it follows that n 2 ( r : p,) > n2Vm:f,). Therefore seller 2 will set
p2 = r. Thus strategies of a list price p , coupled with an offer to match for firm
1 and a list price r coupled with an offer to match for firm 2 are an equilibrium.
Similarly, it may be shown that the same strategies, reversed between the
firms, also are an equilibrium.
Proof o f proposition 2. Suppose that the distribution F is a symmetric
differentiable equilibrium strategy. Let S be its support and f the density func-
tion.
a) The first step is to prove that p, E S and r E S . Suppose that
def
min ( p : p E S ) p
=- > p,.
Then
W-P : F ) = ne(_~)
+ nt(_~).
Since p, < -
p,
for all p E ( p , , r ) , or
-
- - Q,ne(pm) + Q,(P, - C) = wpm: F ) .
n;(prn)
Thus, for all p E [ p , , r ] , II(p : F ) = n ( p m :F ) .
382 Journal of Business
In mixed-strategy equilibrium, the firm must be indifferent among all the list
prices in the interval [ p , , r ] , which implies that
Notice that this is identical to the equation for the symmetric equilibrium in the
case of two firms. Hence the solution must be
and therefore n ( p : G , ) = ( 2 / n ) [ n e ( p m )+ n r ( p m ) l .
Now G , ( p ) is the distribution of the minimum of the list prices of the other
(n - 1 ) firms. Thus
G , ( p ) = pr[minimum of ( n - 1 ) prices is no greater than p ]
= 1 - pr[all ( n - 1 ) prices are greater than p ]
= 1 - [1 - F , ( ~ ) ] ~ - ' .
References
Aumann, Robert J. 1974. Subjectivity and correlation in randomized strategies. Journal
of Mathematical Economics 1, no. 1 (March): 67-96.
Holt, Charles A., and Scheffman, David T. 1985. The effects of advance notice and best-
price policies: Theory with application to Ethyl. Working paper. Washington, D.C.:
Federal Trade Commission, Bureau of Economics, February.
Jeuland, Abel P., and Narasimhan, Chakravarthi. 1985. Dealing-temporary price
cuts-by seller as a buyer discrimination mechanism. Journal of Business 58, no. 3
(July): 295-308.
Price Discrimination 383
Klein, Benjamin, and Kenney, Roy W. 1985. Contractual flexibility. Working paper no.
388. Los Angeles: University of California, Los Angeles, Department of Economics,
December.
Lazear, Edward P. 1986. Retail pricing and clearance sales. American Economic Review
76, no. 1 (March): 14-32.
Narasimhan, Chakravarthi. 1984. A price discrimination theory of coupons. Marketing
Science 3, no. 2 (Spring): 128-47.
Rosenthal, Robert W. 1980. A model in which an increase in the number of sellers leads
to a higher price. Econometrics 48, no. 6 (September): 1575-79.
Salop, Steven. 1977. The noisy monopolist: Imperfect information, price dispersion and
price discrimination. Review of Economic Studies 44, no. 3 (October): 393-406.
Salop, Steven. 1986. Practices that (credibly) facilitate oligopoly coordination. In Joseph
E. Stiglitz and G. Frank Mathewson (eds.), New Developments in the Analysis of
Market Structure. Cambridge, Mass.: MIT Press.
Schmalensee, Richard. 1981. Output and welfare implications of monopolistic third-
degree price discrimination. American Economic Review 71, no. 1 (March): 242-47.
Shilony, Yuval. 1977. Mixed pricing in oligopoly. Journal of Economic Theory 14, no. 2
(April): 373-88.
Sobel, Joel. 1984. The timing of sales. Review of Economic Studies 51 (July): 353-68.
Sultan, Ralph G. M. 1974. Competition or Collusion. Vol. 1 of Pricing in the Electrical
Oligopoly. Boston: Harvard University, Graduate School of Business Administration,
Division of Research.
Varian, Hal R. 1980. A model of sales. American Economic Review 70, no. 4 (Septem-
ber): 651-59.
Varian, Hal R. 1985. Price discrimination and social welfare. American Economic Re-
view 75, no. 4 (September): 870-75.
Vilcassim, Naufel, and Wittink, Dick R. 1985. Manufacturers' use of coupons: Perspec-
tives and implications. Working paper. Ithaca, N.Y.: Cornell University, Johnson
Graduate School of Management.