Wednesday, June 5, 2013

The tip of the iceberg

Should tipping be banned? That is the topic of a recent Freakonomics podcast. Listen to the whole thing.

Tipping expert Michael Lynn says if he had his druthers he would outlaw tipping. I found the reasoning for this conclusion lacking. Most of his research is based on survey data, which by its nature comes with a full shaker of salt. Of course many of the results of the survey research agrees with conventional wisdom priors, and this is a rare instance when I tend to agree with conventional wisdom. Hard to say that the surveys tell us much when they simply tell us what we already believed.

Confirmation bias is seductive. For example, we are told that physically attractive females earn better tips especially from men. What isn't so clear is if assuming they could earn more, the female servers actually deliver superior service to that target audience. Or perhaps they are take the game theory to the next level. Since the men are a sure thing, the female servers focus their attention on other patrons. The men appreciate them just the same while the other patrons get superior service. In this case tips rise from all patrons when compared to the alternative--non-attractive servers for the men and standard or inferior service for other patrons.

The major case Lynn finds against tipping is that it is de facto racially, et al. discriminatory assuming the survey results match reality. Here is where the reasoning is poor. Association with an outcome potentially undesirable, such as black waiters making less tips than white waiters when other factors besides race are supposedly held constant, is then construed as being the undesirable behavior per se. It is as if tipping were the cause of the discrimination rather than simply a correlated symptom of the problem. This is the logic of the disparate impact doctrine. Unfortunately, eliminating tipping even if it is truly used in a discriminatory manner including simply having a discriminatory result doesn't eliminate racial discrimination or disparate impact.

A bigot can exert harm in ways that may be more harmful if the tool of tipping is removed. This is true if the discrimination is done consciously (disparate treatment) or unconsciously (disparate impact). Without the ability to choose a restaurant blind to the color of the staff knowing the bigot can always tip less if the server is of an "undesirable" race (from the perspective of the bigot), the bigot may be lead to only consider restaurants that have low to no proportion of "undesirable" races employed within. In this thinking tipping is a more subtle tool for exerting bigoted behavior. Take away the tool, and the bigot is left with only more blunt means of acting out the bigotry.

Throughout the podcast there seemed to be an air of confusion about what the purpose of tipping is and a dismissiveness in the sense that a practice while long standing and ubiquitous was nevertheless illogical.

The purpose of tipping is to properly align incentives and minimize principal agent problems. It is not a gift. It is not a requirement. A tip should be understood as part of the compensation withheld until service is rendered to be delivered directly from the patron with potential variability depending on quality. It is an emergent solution to a knowledge and cooperation problem.

Some more thoughts:

  • Those eligible for tips should be those with an ability to perform above or below the standard. 
  • "They work for tips" is not adequate reasoning for how much one should tip. That is simply the definition of the service arrangement. 
  • Inflation does not imply that the rate used for a standard tip should change, say from 15% as the old norm to 18% as the new norm. That is really bad math.
  • For tipping to be effective, one must be willing to differentiate. At the least that means tipping a minimum amount (perhaps 10%) subject to upward revision if service is excellent. More desirably it means a willingness to tip zero for horrible service or negative (complaining to the manager) and a willingness to tip very well for excellent service.
  • Tipping based on a percentage of the cost of goods is generally fine, but there is a floor and ceiling on this causing the rate to become an S curve rather than a straight line. If I order a Coke at the bar while my friends drink beer, I don't simply tip 15% on the $2.00 soft drink. Likewise, an expensive dinner for my wife and I where the before-tip bill is $300 might only rate a $50 tip even if service is excellent.