Do Restaurants Blacklist Low-Spending Customers?

Ian Ayres

Stephen Dubner has an interesting post on Super Crunchers over at the Freakonomics blog. He ask:

do restaurants turn down reservations for customers who have proven themselves to spend too little money?

Dubner is clear that this is likely to be mostly an issue for high-end restaurants that are slot constrained.

But the 50 comments to the post are really interesting. They mention that the real action may not be in reservation discrimination based on profit predictions, but in service discrimination based on tipping predictions.

Several servers are confident in their ability to make tipping predictions:

I know how to stereotype pretty well the non-regulars and tourists that come in. I don’t always do it intentionally, but when I am busy tables that have been good to me in the past or that I think will tip me well are the ones that I devote my time to.

One of the big risks here is that stereotyping will be used as excuse to provide poorer service to African Americans.

Fred Vars, Nasser Zakariya and I did a study of taxi-cab tipping that found among other things that African-American passengers left tips that were about 50% less than those left by white passengers.

Database analysis might actually be able to help. Programs like Open Table and even slight tweaks to credit card data could provide feed backing on tipping histories. Restaurants run the bill first (so they know the price of the food without the tip) and then servers have to send them the tip amount once the customer fills it in. So it would be possible for Visa to figure out who are the good and bad tippers.

If patrons let restaurants run their cards at the beginning of meals or at the time they make a reservation, it would be possible for servers to make much more accurate assessments of likely tips. I've spoken with several African-Americans who feel that it is very difficult for them to signal that they are good tippers (and don't feel free to leave a low tip even when the service is bad). But statistical tipping histories would allow servers not to rely on crude sterotypes which may all too often devolve to questions of race and sex.

But allowing customers to signal their tipping tendencies may not be such a good thing. Another commentator proudly suggested the following strategy:

If you know you are going to be frequenting a place I would highly advise getting to know at least one or two servers and throwing an extra 5% on the table. It normally isn’t much money, but being known as someone who is nice and tips better then average will mean you get treated wonderfully even if you only come in for a quick dessert or always split entrees.

But I worry that such a tipping arms race is imposing costs on fellow patrons. It's a bit like the joke of the two hikers who are being chased by the bear. One puts on running shoes -- not to be faster than the bear, but to be faster than the other hiker.

A larger issue is the extent to which Super Crunching aids or hinders racial equality. In finance, the move toward less discretionary pricing has disabled some of the most blatant forms of race discriination. But there is also the possibility that firms will use data mining to engage in a kind of virtual redlining -- where they discriminate on a bundle of non-racial factors that correlate very closely with race.
Posted 1:56 PM by Ian Ayres [link]