Arnold Kling writes “that price discrimination really deserves a lot more attention than it gets in the economics curriculum. A lot of “economic naturalist” sorts of questions are correctly answered by appealing to the concept of price discrimination”. Contrary to Arnold, I think price discrimination gets too much attention. Economists are quick to cite price discrimination and market power as an explanation when the same product sells for more than one price, when in fact cost differences are often driving the different prices.
Arnold’s George Mason colleague Russ Roberts, and coauthor John Lott, argued this case persuasively in a 1991 paper in which they use several case studies to show how variable prices that are commonly explained by price discrimination are actually better explained by cost differences. In the spirit of Roberts and Lott, I’ll try to offer a cost based explanation for “black friday” sales.
Stores must sell their goods at a prices that cover the wholesale cost of the individual goods as well as the overhead costs of the store, like labor and the building lease. The amount of additional price that must be charged for each good to cover overhead costs is a function of the average turnover of the goods sold. Ceteris paribus, the faster a store can sell its goods, the lower the average overhead costs of each good, and thus the less that must charge to cover the stores average total costs. If a store sell three times as much on black friday than they normally do, then the overhead costs are three times as small.
Arnold might counter that the discounts observed on black friday are too large to be accounted for by decreasing average overhead costs by a factor of 3 or 4, thus the cost theory cannot explain such deep discounts. However, not all goods are discounted. Assume that volume increases four-fold on black friday, and thus overhead costs decrease by a factor of four. If overhead costs normally add 4% to the total price of a good, then on black friday overhead costs only needs to add 1% to the total price of a good in order to cover costs. This means all goods can sell at a 3% discount. However, if the cost savings are used to discount only 10% of the goods in the store, then each discounted good can be discounted by 30%.
This explanation has the benefit of requiring no market power for the stores. Given the wide range of stores that discount on black friday, and my skepticism that so many everyday retailers have significant market power, I find the cost explanation more believable.

10 comments
Comments feed for this article
Wednesday ~ November 25th, 2009 at 8:41 am
kharris
Sorry, but I feel the need to ask. Are you citing a paper by the same John Lott who created a fictional female personae to heap praise on John Lott? The same John Lott who may have cooked data on gun violence to suit his thesis?
If so, then while it is entirely possible that the paper you cite is Nobel quality work, but we have reason to be dubious about with which Mr. Lott is associated.
Which leads to another question. Is it impossible to cite some more credible researcher on the same point? I would think the goal in writing a blog post, instead of just letting your scholarly paper stand for itself, is to spread your ideas beyond the camp of professional economists. Outside the camp of professional economists, going and reading a cited paper to discover its merits is not likely. Readers outside the profession are likely to judge the value of your work based on less time consuming indicators. Citing John Lott won’t make your scholarly work any better or worse (unless you rely on him when you shouldn’t within that work), but it seems to me highly likely to undermine you attempt to spread your ideas outside the profession.
Wednesday ~ November 25th, 2009 at 8:42 am
kharris
“..we have reason to be dubious about WORK with which Mr. Lott is associated.”
Sorry. I don’t edit good.
Wednesday ~ November 25th, 2009 at 1:09 pm
wcw
On-topic, I think you fall down immediately after “function of the average turnover of the goods sold,” when you reduce your period of analysis to a single day. In a deeply seasonal retail world, periodicity matters. The retail season is one year long, not one day. I don’t have to remind you that it’s called ‘black friday’ because it traditionally is the first day of the back-loaded retail year that your p/l finally goes in the black. Forgive me, but at this point I stopped reading. Your full argument may be very smart, but its utility will be restricted to a retail world that does not exist.
Off-topic, if only to avoid the inevitable sniping you should always avoid citing John Lott. Moreover, the one time I checked Lott’s work (by downloading his data set and doing some simple exploratory analysis; not his firearms work, fwiw), all my initial models pointed the opposite way of his conclusion. All of them. It was only using the specific model and the specific curvefitting method he chose that let me replicate his results even directionally. An author who publishes a single such result immediately does on my do-not-cite-ever list. Your mileage may vary.
Thursday ~ November 26th, 2009 at 1:10 am
Matt Matson
There is quite a bit of tension between a world in which companies use loss-leaders and a world in which companies price their products based on the average overhead cost for the day a product is sold.
I would also point out a building lease and, to a large extent, labor costs are fixed costs. A store will pay rent all year long and hire employees for at least the entire holiday season, regardless of how much is sold on one day in November.
Friday ~ November 27th, 2009 at 5:42 am
Price Discrimination, Again | Everyone Read It!
[...] Discrimination, Again Some pushback from Karl SmithAdam Ozimek. Stores must sell their goods at a prices that cover the wholesale cost of the individual goods as [...]
Sunday ~ November 29th, 2009 at 2:54 am
Bruce Wilder
So you think the expenditure of resources to create and maintain a scheme for price discrimination is evidence that price discrimination is not taking place. That’s kind of bizarre, really.
Sunday ~ November 29th, 2009 at 10:08 pm
nick
“Market power” is a poor way to describe what is required for price discrimination to work. All that is required is for discounts to be unpredictable and for the switching costs on a particular shopping trip to be too high. Both criteria are easily satisfied for the wealthier customers price discrimination targets, and indeed what we see are different discounts (“sales”) every week that shoppers can’t predict. The poorest customers wait for the sales and clip the coupons and compare prices between stores for the best discounts, while the wealthiest customers just go ahead and pay the high. Viola, price discrimination that requires the “market power” of high switching costs only on a single shopping trip for the wealthier customers being discriminated against.
Friday ~ December 4th, 2009 at 3:24 pm
Countercyclical Prices, Price Discrimination, and Black Friday « Modeled Behavior
[...] explaining how it can apply to Cyber Monday discounts as well. Arnold has previously responded to my theory of price discrimination by arguing that “the key issue here is that the cost of many products [...]
Friday ~ January 29th, 2010 at 7:29 pm
crisismaven
Hello, for all students (at least high-school and the undergraduates) and others with an interest or enrolled in economics I have started a blog which will comprise study literature in a more entertaining form than standard textbooks, see
CrisisMaven’s Economics Study Guide. While still in its initial stages it will be added to constantly and covers general ecenomics subjects while at the same time dealing with current topics from the news that provide a welcome backdrop to an elucidating chapter in economics as well, e.g. Of Mortgage Brokers, ARMs, Attrition and Marathons .
Thursday ~ February 11th, 2010 at 10:16 pm
Maybe Price Discrimination Does Explain Everything « Modeled Behavior
[...] likes to claim that economists underestimate the prevalence of price discrimination, while I have disagreed that economists tend to see price discrimination when costs are actually driving different prices [...]