Tuesday, 27 May 2014

Is there a simple trick to reduce over-eating at buffet restaurants?

I recently read an interesting 2013 paper from the journal Applied Economics, by Erez Siniver of COMAS (in Israel), Yosef Mealem of Netanya Academic College (Israel) and Gideon Yaniv of COMAS (ungated version here). The authors are attempting to explain why people overeat at buffets. It's interesting because, as the authors state:
Traditional economics suggests that the entry price constitutes a sunk cost that should be ignored when deciding how much to eat; only incremental costs and benefits matter. Hence, since the marginal cost is zero, eating should rationally cease at the point of fullness, where the marginal utility from eating falls to zero.
In other words, if people are rational they will eat at the buffet until they are full. It shouldn't matter how much they have paid to enter the buffet, because that cost is sunk (it cannot be recovered) and so it shouldn't come into the decision about how much to eat. Of course, most people aren't fully rational and one of my favourite field experiments of all time, conducted by a student of Richard Thaler and described briefly here, showed that when people at a pizza restaurant were given a full refund before ordering they ate less.

The Siniver et al. paper investigates a related question: whether the time of paying (before or after the meal) affects the amount that buffet patrons eat. Again, the time of paying should not affect a rational person's decision on how much to eat at the buffet. Specifically, they:
...report on the results of two experiments destined to test this conclusion, where the time of paying for the buffet serves as a proxy for the quality of treatment. In the experiments, conducted in collaboration with a sushi restaurant on campus, we offered students and staff an all-you-can-eat sushi buffet during four hours at lunch time. Half of the participants were asked to pay for the buffet before eating, and half were told that they would pay after they finished eating.
The argument is that customers who are asked to pay before the meal feel less trusted by the restaurant (it's as if the restaurant is worried they will eat and run without paying). Discontented customers (those who are asked to pay beforehand) will eat more because the more contented customers (those who pay afterwards) reach the point where they have 'gotten their money's worth' at a lower quantity of consumption. The authors find:
Controlling for other explanatory variables, such as gender, food quality, drink consumption and body mass index (BMI), sushi consumption exhibits a statistically significant decline of about 4.5 units from the paying-before to the paying-after group, thereby supporting our hypothesis.
The reduction by 4.5 units of sushi is about a 20% reduction in consumption, which is quite substantial. What this shows clearly once again is that people are not fully rational - the time of paying shouldn't affect the amount that they eat, but it does. So a simple policy nudge might be able to affect eating behaviour.

The policy implication that the authors draw is that if governments want to reduce the contribution of buffets to over-eating and obesity, they should implement a policy that buffet restaurants must have their customers pay after the meal rather than beforehand. Restaurants receive an addition benefit from having their customers pay after the meal - it reduces the number of required transactions and therefore reduces the costs to the restaurants. Indeed, most buffet restaurants I have been to charge separately for drinks, which means having the customers pay after the meal rather than before the meal is required anyway.

How far can the idea of restaurants trusting customers, and this resulting in reduced food consumption, be extended? If customers have a self-payment option on exit from the restaurant (where they swipe their credit card and pay without needing to deal with restaurant staff, similar to the self-service lanes at supermarkets), would that reduce consumption even further? Or, does that create a backlash because it reduces the perceived quality of service? Could make for some interesting follow-up research.

No comments:

Post a Comment