Big Data Might Tell Retailers Which Consumers to Charge More

(Bloomberg View) -- My first graduate microeconomics class didn’t start with any grand theories or mathematical proofs. It started with a demonstration. The professor marched us down to a computer lab, divided us into groups of six, and had us participate in a simulated market. Amazingly, the price of the virtual commodity we were trading quickly converged to a single value and stayed there. This was a powerful demonstration of the so-called law of one price, the economic principle that markets produce a single price for a given commodity or good.

Many sellers would love to change that law if they could. If merchants could raise prices for customers who were willing to pay more, they could make a lot more profit. Suppose there are two potential customers for my product -- one who would buy it for $100 and another who would be willing to pay $150. If I have to charge them both the same price, I’ll charge $100, so that I can sell two units and make $200. The second customer, who would have been willing to pay a higher price, ends up getting a bargain. In economics lingo, that bargain is called a consumer surplus.

But if I could charge each a different price, I’d charge the first customer $100 and the second one $150, for a total of $250. In this case, the second customer wouldn’t get a bargain. That’s called price discrimination.

Merchants already do a fair amount of price discrimination. A car dealer will try to figure out how much you’re willing to pay by talking to you while you take a test drive, and try to persuade you to pay more. Movie theaters offer discounts to seniors and students -- people who are likely to have less money -- while charging higher prices for working-age adults. The examples are many and ingenious. But there are limits on these shenanigans. When prices are posted publicly, as with a price tag at a store, everyone sees what everyone else is paying. That makes it a lot harder to charge different prices to different customers.

But in the age of the internet, the price-tag system becomes a lot more tenuous. Shopping online, it’s only me and my computer. When I see a posted price, I tend to assume others are seeing the same, but is that really true?

Blogger James Plunkett has a good rundown of recent evidence on this question. Various online retailers are occasionally caught charging different prices to different customers, including Amazon.com Inc., which said it abandoned the practice back in 2000. Researchers found in 2014 that human shoppers got worse bargains on a number of websites, compared to an automated browser that didn’t reveal its identity.

There are a bunch of ways companies can figure out how much each consumer is willing and able to pay. Some are easy -- charge more to people using an expensive Apple Inc. computer, or to people using certain types of email accounts. An online retailer can also look at people’s product search histories to steer them toward higher-priced products.

This kind of price discrimination, based on income, isn’t very harmful to society -- it can even be good, since it involves charging poor people less and allows them to buy things they otherwise couldn’t afford.

But the real money would come from big data. The more data a merchant gets about a customer -- where they live, what they buy, what websites they visit, etc. -- the better they can predict how much they’d be willing to pay for a certain product. Conceivably, with enough data, merchants could figure out the maximum amount each customer is willing to pay, and then charge them that. If all merchants did this -- which seems more likely in a world dominated by a few big players like Amazon -- then competition wouldn’t result in prices being driven down. No shopper would ever get a bargain. It wouldn’t matter that poor people would potentially be able to buy more things; merchants might charge them so much that they’re not really any better off. And price discrimination is especially bad when combined with monopoly power -- if large, dominant online retailers get into differential pricing, consumers could suffer a great deal.

A 2015 report by the Obama White House found little evidence of this extreme form of price discrimination -- at least as of three or four years ago, online merchants were sticking to more traditional, less pernicious methods. That might be due to legal rather than technological constraints. In the U.S., the Robinson-Patman Act forbids charging different customers different prices unless it’s justified by different costs or by competition. But few people are going to sue an online retailer over a five-cent difference in the price of a roll of paper towels. And if a bunch of different retailers did price-discriminate, any that got sued might be able to convince a court that it needed to do it to keep up with the competition.

So as technology advances, regulation will need to keep up. Regular randomized experiments should be used to test large online retailers for evidence of personalized pricing. The government should allow price discrimination based on broad income classes -- like giving discounts to seniors or students. But if big companies are caught trying to use big data to act like car dealers and squeeze customers for every last penny, they should be punished accordingly.

This column does not necessarily reflect the opinion of the editorial board or Bloomberg LP and its owners.

Noah Smith is a Bloomberg View columnist. He was an assistant professor of finance at Stony Brook University, and he blogs at Noahpinion.

To contact the author of this story: Noah Smith at nsmith150@bloomberg.net.

To contact the editor responsible for this story: James Greiff at jgreiff@bloomberg.net.

For more columns from Bloomberg View, visit http://www.bloomberg.com/view.

©2018 Bloomberg L.P.