Why we work
Often I think our modern math and neuroscience just tell us what we already know. Here’s Max Weber writing about a puzzle of agricultural output in 1905 in “The Protestant Ethic:”
a man who at the rate of 1 mark per acre mowed 2.5 acres per day and earned 2.5 marks, when the rate was raised to 1.25 marks per acre mowed, not 3 acres, as he might easily have done, thus earning 3.75 marks, but only 2 acres, so that he could still earn the 2.5 marks to which he had become accustomed. The opportunity of earning more was less attractive than that of working less.
Despite this, Colin Camerer et al. noted in 1997 that “The standard economic prediction is that a temporary increase in wages should cause people to work longer hours.” Camerer et al. used data on New York City cab drivers to test this premise, built on the economic truism that people will substitute labor for leisure (that is, work more when it pays better, and work less when it pays less). They found that many cab drivers, especially those with less experience, have “negative wage elasticity.” That is, they work less on good days, rather than building up cash reserves that will tide them over and allow them to work less on days when customers are hard to come by. In the jargon of behavioral economics, this is because of ‘narrow bracketing,’ where drivers bracket their time day-by-day, rather than taking a longer-term perspective, and also due to loss aversion (the cabbies set daily targets and don’t want to fall short on bad days, so, having knocked off early on their good days, on their bad days they must drive more).
Camerer’s paper also makes reference to agricultural studies, and notes that such problems of narrow bracketing and loss aversion could be endemic to the kind of people who seek out piece-rate work like farming and cab-driving. But the authors say this seems unlikely. [The paper is “Labor Supply of New York City Cab Drivers: One Day at a Time,” Camerer, Colin F., Linda Babcock, George Lowenstein and Richard H. Thaler, as published in “Choices, Values and Frames, ed. by Daniel Kahneman and Amos Tversky, c. 2000]
The wonder is that the standard model would ignore the observations of actual behavior. Camerer, by the way, is not using particularly complicated math in his study; I wonder if it will work against the adoption of such ideas.