Do incentives matter?

Daniel wonders whether or not incentives can be harmful, citing an earlier post where I said they mattered.

Firstly, having been a consultant, I recommend taking what consultants say with a grain of salt. Secondly, incentives don't go away if you don't measure them -- people have always and will always want more money, more prestige, the world ordered according to their whims, more recognition, more love, more fun etc. etc. Rules just determine what dimensions people compete along, but they do not eliminate the human urge to compete. When the government used to set airfares, airlines simply competed by offering more amenities (non-price competition). When you eliminate differences in job compensation, employees begin competing for non-compensation perks. When you eliminate the competition to take other people's stuff, they compete by making more of their own stuff.

On the job, measuring human performance can be harmful. Firstly, people will perform to the measurements, which may not be optimal for the company. Secondly, measuring itself is costly, and you need to make sure the system is not being gamed. Thirdly, there might be large elements of luck in the job, meaning performance may not be closely tied to effort. Fourthly, you need to pay employees more for them to choose more variable pay (would you prefer $50,000 for sure, or a 50/50 chance of $0 or $100,000?) But if you don't measure how someone is performing, you don't reward the right people enough (so they leave), you don't get rid of the wrong people (who aren't performing), and you don't know where people are struggling with their jobs, so can't help them. These are real costs and risks too.

So in the context for measuring job performance, optimal pay structures turn out to depend a lot on the specific circumstances. But the optimal pay structure merely channels people's competitive tendencies in the right direction, it does not make those tendencies go away. People respond to incentives.