A recent paper by Bergeman, Brooks and Morris (BBM) supposes a monopolist free to segment the market in any way she can (without worrying about arbitrage), and asks what is the achievable set of pairs of producer and consumer surplus? BBM gives a simple and satisfying answer to this question. This post attempts a short proof of their characterization.

A monopolist faces a market consisting of buyers with valuations . Order them so that . The number of buyers with valuation is and assume the buyers are divisble. A segmentation of the market is a partition of the buyers into upto markets with the property that the profit maximizing price in market is . If we let be the number of buyers with valuation in market , then any segmentation is characterized by the following:

Denote by the set of feasible segmentations. Let be the profit earned by the monopolist under the segmentation . The consumer surplus of buyers under the segmentation is

It is easy to see that . The upper bound follows from the segmentation that assigns all buyers with valuation to the market and no others. This corresponds to first degree price discrimination. It is also easy to see that . The lower bound comes from the segmentation that assigns all customers to market , where is the profit maximizing monopoly price without discrimination. BBM show the following:

**Theorem** is feasible iff and .

That , is straightforward. The hard part is to show two things.

1) For any such that there is a such that .

2)There exists an such that and .

To prove the first item (which BBM note in the paper is easy) on this list, call a segmentation upper triangular if for all . Note .

Let . We construct a new segmentation from by shifting the buyers in market with values below into market . As the profit maximizing price just to this portion of buyers is , moving them into market leaves the profit maximizing price in market unchanged. Formally:

for all and .

for all .

for all .

for all .

Under segmentation , both and increased in value, contradicting the initial choice of .

To prove the second item on the list, among all feasible segmentations such that , choose one that minimizes , say . Call lower triangular if for all . I show that must be lower triangular from which it follows that . The proof will be by induction on the number of distinct valuations.

The case is straightforward. Suppose first that . The following segmentation, as can be verified, does the trick:

where

If , the segmentation that assigns all buyers to market 2 will have the requiste property.

Now consider the case of arbitrary and suppose first that . Given an instance on valuations construct an instance on valuations by setting for all . It is easy to see that , i.e., the optimal monopoly profit with no discrimination remains unchanged. By the induction hypothesis there is a segmentation that is lower triangular. To conclude the argument we must show how to convert into a legitimate segmentation for .

for .

where for all and .

If the ‘s can indeed be chosen as specified, then, is lower triangular and the corresponding and . To verify that appropriate ‘s exist, it is enough to check that

which follows from the hypothesis that .

To conclude, suppose now that . Construct a new instance on valuations by setting for all and . Notice, . By the induction hypothesis there is a segmentation that is lower triangular. To conclude the argument we must show how to convert into a legitimate segmentation for .

for .

for all where , and .

for all .

If the ‘s can be chosen as specified then, is lower triangular, in and the corresponding and . Verifying that the appropriate ‘s exist, can be done in the same way as the previous case.

## 6 comments

January 21, 2014 at 2:49 am

“The Limits of Price Discrimination,” D. Bergemann, B. Brooks and S. Morris (2013) | A Fine Theorem[…] much to the regret of many of us at MEDS has recently moved on to a new and prestigious position, pointed out a clever paper today by Bergemann, Brooks and Morris (the first and third names you surely know, the second is a […]

January 21, 2014 at 5:09 pm

Noni MausaHelp help. For those of us who never got beyond “sigma” in calculus class, can you sketch out what these calculations tell us about monopolist behaviour?

January 21, 2014 at 8:55 pm

rvohraDear Noni

A good summary of the policy implications is provided by Kevin Bryan at

http://afinetheorem.wordpress.com/2014/01/21/the-limits-of-price-discrimination-d-bergemann-b-brooks-and-s-morris-2013/

Here is the relevant paragraph:

……..since information is always good for the seller (weakly) and ambiguous for the consumer, a policymaker should be particularly worried about bundlers providing information about willingness-to-pay that is expected to drastically lower consumer surplus while only improving rent extraction by sellers a small bit………….. It seems intuitive that precise information about consumers with willigness-to-pay below the monopoly price is unambiguously good for welfare, whereas information bundles that contain a lot of high WTP consumers but also a relatively large number of lower WTP consumers will lower total quantity sold and hence social surplus.

January 22, 2014 at 5:10 pm

Ben GolubThere seems to be a typo in the definition of upper-triangular.

January 22, 2014 at 5:45 pm

rvohraDear Ben

thanks v. much for pointing it out. Fixed it and another typo.

rakesh

January 24, 2014 at 11:27 pm

Somewhere else, part 111 | Freakonometrics[…] “The Limits of Price Discrimination” http://papers.ssrn.com/id=2288718 … via https://theoryclass.wordpress.com/… […]