Incentives as selection effects

June 2024

Show me the incentives, I'll show you the outcome.

We take that Charlie Munger quote as a truism; I think it is true but misleading. We naturally parse it as:

When I apply a new incentive to a target population, that same population changes their behavior.

But in my experience, that is not what happens. Instead:

When you apply a new incentive, you select for a new population that prefers the incentive.

Some examples:

  • We pay in equity so that employees feel and behave like owners. No, people that want to feel like part of something agree to be paid in equity.
  • We pay salespeople only if they hit their quota. No, people that like to be paid more when they hit their quota work in sales. The rest of us learns sales is not for us.
  • If our home state gives out tax breaks to the movie industry, we'll have locally produced movies. No, people that already produce movies elsewhere come to your state to film.

This has a few implications:

  • Use incentives when you want a particular outcome but you don't care who achieves it.
  • Incentives will be more effective when you allow new people to participate.

This is not true of all incentives. Many costs and monetary incentives work directly on the target population. For example, if you don't charge anything for water, people forget their sprinklers on. Similarly, when some product becomes cheaper, current buyers tend to buy more of it.

But for incentives inside companies, these are the exception. Most incentives by management (stock compensation, sales compensation, performance reviews, etc.) are trying to directly incentivize the current population to behave differently. They rarely do. They mostly select for people that are like those incentives.

Thank you to Brian Lui and Riley Turben for a discussion on this topic.