In Praise of Tipping
One of the most common justifications for government intervention is an alleged "market failure." There are countless mainstream economic models that purport to show the inefficiency of unregulated markets, which can ostensibly be improved by political means.
These analyses, however, often abstract away from the real-world problems of dispersed knowledge and high transactions costs. Upon closer examination, one often finds that the market's "failure" is in fact the best way humanly possible to deal with a complex situation. And nothing better demonstrates this resilience of voluntary, decentralized exchange than the custom of tipping.
Tipping, though commonplace, is at first glance a curious phenomenon. Why isn't the tip for a waiter, say, included in the price of the meal? Or, going the other way, why aren't employees of fast food restaurants tipped? If one assumes that all economic actors possess perfect knowledge, tipping seems to be a quaint and inefficient practice.
But when we realize that market participants do not possess the same knowledge, the practice become more intelligible. For example, a restaurateur would find it very difficult to keep tabs on his entire staff. As such, it wouldn't be worth the trouble to tailor the wages of waiters and waitresses to their courtesy with customers.
The simple market solution is to allow the customers themselves to evaluate the service of the waiter or waitress. This not only reduces monitoring costs for the employer, but allows for the satisfaction of possibly different preferences among customers. Indeed, empirical studies, as well as casual observation, suggest that customers (especially male ones) reward physically attractive servers with higher tips, and thus tipping can be understood as a discreet way of channeling the most appropriate personnel into this line of work.
The phenomenon of tipping, broadly conceived, is the real-world market's response to the "missing markets" postulated in formal economic models. When transactions costs prevent the establishment of a formal institution, spontaneous exchanges still occur to the benefit of both parties.
For example, a bar patron who doesn't wish to wait between drinks may give unusually high tips to ensure that the bartender is particularly attentive. (This avoids the need to set up, say, various zones in the bar, with one section charging higher prices for drinks but guaranteeing quick refills.) Street musicians receive tips ("donations") proportionate to their quality, as judged by their customers. I personally have found (when ordering Chinese takeout) that something so simple as allowing the vendors to "keep the change" means that I will be given preference over other customers (who in this case happened to be quite rude to the workers) upon my next visit.
To take other examples, one of my friends gives outrageous tips when leaving his car in a parking garage, on the theory that if and when the attendants decide to take a joyride, it won't be with his vehicle. I have read of a cab driver who makes thousands of dollars more in tips every year by keeping his taxi spotless and having the latest newspapers available for his customers. Finally, consumers who are egalitarians in their hearts but "know better" in their heads can indulge their generous impulses by rewarding those individuals who possess initiative and a good work ethic.
The above theory of the "function" of tipping not only provides a plausible explanation, but serves to clarify some of the debates over the custom. In recent times, there has been a drive to increase the range of professions receiving a tip (often at the behest of unions). But tipping only serves a purpose in professions where there is a high degree of subtlety in the performance, and where the productivity of employees is not easily monitored.
It would be rather pointless, for example, to tip the cashier at a fast-food restaurant, since he or she does not have much discretion over the quality of the meal. I would tend to side with tradition when it comes to which professions are tipped, and view with suspicion any move to bring the practice to a new one. In any event, as Steven Landsburg points out, increased income through tipping would in the long run be offset by lower base wages (at least in a competitive labor market).
This insight also shows the folly of arguing for a higher (or lower) percentage when deciding upon a tip. As with money--which fulfills its economic function regardless of the quantity of cash in the economy--tipping serves its purpose so long as everyone's expectations are coordinated. If the standard tip for a meal is 15 percent, then this is taken into consideration when diners choose a restaurant and waiters choose a profession. To switch to a new standard of 10 or 20 percent would only introduce uncertainty during the transition period, when prices and base wages adjusted. The purpose of restaurant tipping is to allow diners to deviate from the standard percentage, to either punish poor service or reward excellent service.
Before closing, I should point out that one of the major "paradoxes" in the mainstream treatment is why people should ever tip when they don't expect to return to a given restaurant (or other business). After all, the voluntary nature of tipping seems to offer a classic opportunity for "free riding."
However, I think this sort of question would be akin to asking why most people don't steal money from little children, even when there are no police officers in sight. It results from a crude homo economicus approach to human behavior, which is definitely useful in certain applications, but not as a general theory of human action.
The custom of tipping is an excellent demonstration of the market's ability to solve real-world problems of dispersed knowledge and transactions costs in order to ensure the maximum satisfaction of consumer preferences. Normally taken for granted, tipping is in fact a beautiful illustration of the power of voluntary, decentralized exchange.
 See Thomas McQuade (2000), "The Anatomy of Tipping," for a thorough discussion.
 This is not to suggest that mainstream economics is incapable of modeling the arguments of this article. See for example Andrew Schotter (1979), The Economics of Tipping and Gratuities: An Essay in Institution-Assisted Microeconomics (New York University Starr Center Report #79-19).