The Science of Tipping

Why do we tip, and does it make any sense? Tuck professor Laurens Debo creates a modeling framework to find some answers to these economically significant questions.

According to a recent study reported in USA Today, Americans these days spend nearly $500 per year on tipping at restaurants and bars. More than one-quarter of those surveyed for the study said they often felt forced to tip more than they wanted to, which isn’t surprising given that tipping percentages have steadily increased over the last hundred years. And yet, tipping is a completely voluntary behavior. Moreover, even though it accounts for more than $50 billion in the U.S. economy and is the primary source of income for millions of restaurant workers, it’s a behavior we know very little about. 

Laurens Debo, professor of operations management and area chair of the operations and management science faculty research group, teaches the MBA course Management of Service Operations.

In a new working paper, Tuck professor Laurens Debo, together with Ran I. Snitkovsky of Coller School of Management, Tel Aviv University, tackle the tipping conundrum from an analytical point of view. The paper, “A modeling framework for tipping in the presence of a social norm,” details a mathematical model that can help explain why people tip and why, perhaps, tipping isn’t the most efficient way to pay servers for their work. 

Economists have long posited that a purely rational actor would not leave a tip. That’s because a tip is paid after the service has been provided and it’s not a mandatory charge to the customer. In theory, tipping makes sense for repeat customers, but empirical studies have failed to support that line of reasoning. 

Without a sound economic justification for tipping, researchers in social psychology, sociology and hospitality management have delved into the behavioral motivations for the practice. This literature, mostly based on surveys, has found two main motivations for tipping: expressing gratitude and conforming to the social norm. The technical term in the literature for expressing gratitude is “reciprocity.” Indeed, in lab experiments where people are asked to split a sum of money with another person that must accept any proposal, they usually don’t keep all of it for themselves. Instead, they tend to voluntarily give up 30 percent of the total amount. Offers even tend to increase if the initial sum depends on the other person’s performance in an earlier stage of the experiment. In short, we seem to be hardwired to be more generous than we need to be. 

As it turns out, we’re also hardwired to conform to social norms. The norms for tipping are spread through word of mouth, via books and guides, and even at the bottom of the check, which lists the suggested tip amounts. Once people know the norm, researchers contend we incur a “disutility” when we act against that norm. To avoid that psychological pain of not fitting in, we tip according to the norm. “If everybody tips 20 percent,” Debo says, “I don’t want to be the person who tips 5 percent, unless the service really stinks.” 

In the paper, Debo and Snitkovsky build a modeling framework around these two known tipping motivations. The model predicts that tipping will increase gradually until it settles on a long-run equilibrium, so that a social norm of tipping is formed. In this equilibrium, customers are divided into two: those who find the server’s work valuable and tip generously to show their appreciation, and the rest who tip lesser amounts just to avoid negative feelings. What’s interesting is that the model can be used to answer counterfactual questions about tipping. For example, it predicts that if some customers decide to tip more, say, because they think waiters deserve a higher pay, or because they feel stronger needs to conform, it will bump up the norm and raise the tipping equilibrium for everyone. “This is what excites us about the model,” Debo says. “It can help us identify and study the long-run equilibrium and make sense of its drivers.”

Empirically, there is very little variation in tips that can be explained by service quality, and our model corroborates that.
— Laurens Debo, Professor of Operations Management

In the second part of the paper, Debo and Snitkovsky apply the modeling framework to a few important issues around tipping. First, they test the conjecture that tipping rewards good service. People often claim to reduce their tip when service is bad or increase it when service is especially good. But “empirically, there is very little variation in tips that can be explained by service quality,” Debo says, “and our model corroborates that.” The reason is the power of the social norm. Even when service is sub-par, patrons don’t want to incur the psychological cost of tipping below the norm, so they still give a good tip. Restaurant managers think tipping is a good way to motivate servers to give great service, but the authors find that those managers would be better off training them or incentivizing them some other way. 

A second context for Debo’s model is the legal framework around pay for tipped workers. The federal minimum wage for non-tipped workers is $7.25 per pour, and $2.13 per hour for tipped workers. If employees don’t reach $7.25 per hour between their minimum wage and tips, the employer must make up the difference. That gap—$5.12—is called the “tip credit.” But the tip credit varies by state, with some states requiring employers to pay tipped workers the full non-tipped-worker minimum wage before tips (essentially, a $0 tip credit) and others hewing to the federal minimum. “If the tip credit is low,” Debo explains, “that means whatever you tip goes to the workers. This is great for workers but bad for employers, because employers don’t get to use tips to subsidize their labor costs.” The model, Debo says, indicates that if society wanted to abolish tipping, getting rid of the tip credit would be a good place to start.

But do we want to abolish tipping? For one, there are better ways to improve service quality than tips and, secondly, it exerts uncomfortable social pressure on consumers. Yet the model predicts that tipping is unlikely to die out on its own. The reason? “Tipping is in essence a form of price skimming,” Debo says. “It allows firms to extract rents from customers that can be used to retain workers, which is important when labor is mobile, exactly the nature of many tipped jobs.”