It’s been a tough couple years for restaurants. When the economy goes south, dining out is something relatively easy to cut from the family budget. Throw in rising food prices, and you gotta brutal business. Thus it is not surprising that firms are looking to save a little cash on the labor front. The Wall Street Journal had an interesting story on steps Chili’s Grill and Bar (a division of Brinker International) is taking to reduce the amount of labor it needs in its restaurants (Chili’s Feels Heat to Pare Costs, Jan 28). Some of these steps are straight out of our core Ops class in terms of reducing the amount of time resources need to spend with each unit flowing through the process:
With costs in mind, Chili’s studied how its kitchen operations could be more efficient. It determined employees were spending too much time performing solo tasks, such as hand-mashing potatoes in a skillet or minding ribs in the oven.
Now it is testing in 10 restaurants a new combination oven that replaces the skillet and the smoker. The chain wouldn’t disclose the ovens’ cost but said that they can smoke ribs and cook bacon at three times the current capacity. That, in turn, frees employees to do other things until the timer buzzes.
Other steps include experimenting with conveyor belt ovens for burgers and quesadillas and shifting food prep from line cooks to (presumably less expensive) prep cooks. All of these are fine and good and don’t necessarily affect the customer. The customer doesn’t see what happens in the kitchen and could probably care less about who dices up the onions. The more interesting part is what they are doing in the front of the house.
Last year, the chain shifted the way tables are served. Now, waiters work in pairs in zones and table bussers have been eliminated. Chili’s said it noticed that dedicated bussers were slowing service because they held onto tubs used to remove dirty plates, making servers reliant on them to clear tables. If the busser didn’t clean fast enough, new customers couldn’t be seated. Without bussers— who now have been gone since July—servers no longer have to share tips with them and have an incentive to turn over tables faster.
“We took a big chunk out of our labor model at the front of the restaurant without compromising the guest experience,” Ms. Valade says.
So I discussed this model of waiter management with some Kellogg econ PhD students some time back. (I believe that they had actually talked to someone at Brinker. That is, many of these changes have been in the works for a while and are not all in response to the impending changes in health care regulation even though that is how Mr. Muroch’s paper pitches things.) There is an interesting trade off here. We basically have an agency problem. The restaurant manager cannot monitor everything that each server does so having them dependent on tips gives them an incentive to put in effort and treat the customer well. As soon as you pool tips, that creates a problem. If Pablo puts in a lot of effort and gets the customer to bump up the tip a couple of bucks, he now has to share that with Mallesh. That implies that Pablo should not put in as much effort as he would if the tip was all his. Or at least that is what a standard model would predict.
So what is the trade off? The trade off arises because there are aspects of the tips that are beyond the control of the waiters. For example, the waiters have no control over whether they end up with a party of teetotalers who don’t buy booze and run up their tab. Or they don’t completely control whether a given table will turn two or three times this evening. That creates variability in their earnings but that variability is dampened out as they have a slice of more tables. Less variability would mean higher utility (assuming they are risk averse) and thus the firm can pay a lower base salary. (If you doubt that servers are that dependent on tip income, check out Steve Dublanica’s recent book Keep the Change: A Clueless Tipper’s Quest to Become the Guru of the Gratuity.)
There are a couple of other points here that might tip the scales toward pooling the waiters. For one, service might actually improve even if the waiters don’t work as hard. One can think of the waiters as a queuing system with random arrival of requests for attention (e.g., foods up for table 5, need more drinks at table 12, etc.). There are economies of scale in queuing systems so doubling the number of servers and doubling the number of requests should result in faster service of requests as long as effort doesn’t drop too much.
The second point is that waiters might not be able to shirk that much. Going back to Pablo and Mallesh, if Pablo is a total slacker, Mallesh will notice even if the manager does not. This is a repeated game and Mallesh may demand a different partner or even a different shift if working with Pablo is too much hassle. Developing a reputation as a loser and consequently being left off high volume shifts is probably more than enough discipline to keep the likes of Pablo in line.