Incentives Change Behavior

Shockingly little of our behavior stems from deeply-held moral convictions. And that’s a good thing! We shouldn’t have deeply-held moral convictions about most things, because it’s exhausting and alienating and inconsequential in most cases.

That leaves most of our behavior to be driven by the sequence of incentives that surround us. Have you ever picked one place to eat over another because it was closer? Worn a shirt you find slightly less comfortable because you receive more compliments when you do?

Of course. Everyone has incentives, they’re all competing, you’re creating them for other people, they’re creating them for you, and we’re navigating them all the time. But this effect is subtle, in the sense that water is subtle if you’re a fish. Because this is everywhere, you don’t really acknowledge it.

Still, there’s no excuse for being ignorant of it. People will change their behavior if the incentives change. It boggles my mind when people act like it doesn’t.

A few years ago, Philadelphia imposed this really high tax on sugary drinks sold in stores. It was absurd on the face of it, because the two stated goals of the program were entirely mutually exclusive: on the one hand, they said “this many sugary drinks get bought in Philly each year, so if we impose a 50-cent tax on each one, we’ll make X money in tax revenue for the city.” On the other hand, they said, “we want to discourage the consumption of sugary drinks, and this tax will help do that.”

See the problem? You can’t assume (with the first goal) that behavior will remain exactly the same while assuming (with the second goal) that it will change.

The two rules here are:

1. People will always change their behavior, in aggregate, when the incentives change. This is true even if you wouldn’t change your personal behavior, because you are not ‘the aggregate.’
2. Changing incentives changes behaviors, but it doesn’t necessarily change desires. People’s new behaviors will be whichever get them to their original desires most efficiently under the new incentive structure.

Here’s what actually happened in Philly, by the way – BOTH goals failed. The city didn’t gain much tax revenue (way less than the projections) AND the residents didn’t drink fewer sugary drinks. Instead, people just bought the same number of sugary drinks from stores just outside the city limits, and thus with markedly lower price points.

So, you want to understand and maybe predict human behavior? Okay, here’s the pattern. First, take a look at the action that a group of people are taking (all this stuff works way better with groups where you can average out the weird idiosyncrasies, but this method isn’t worthless when used on individuals – just be careful in your evaluations and put error bars on your predictions). Figure out first what desire they’re trying to meet. The desire is almost never the action. People aren’t buying sugary drinks because they’re below a certain price point. They’re buying sugary drinks because they want them. They’re incentivized to buy them from city bodegas because that was the most efficient method. So if you were smart, you’d look for the “path of least resistance” between the people and the desire that you’ve identified. That will be the new thing they do.

In sales organizations, sometimes leadership makes the silly decision to cap commissions. So at a certain point in the month, quarter or year, you can keep making sales but you won’t get paid for them anymore. In those organizations, you may be shocked to discover, the good salespeople tend to not work very hard in the last stretch of the commission period. They delay. They talk to their clients and prospects, sure, but they drag their feet getting contracts and closing deals. In sales they call this “sandbagging,” because they’re basically trying to delay commissionable deals until the start of the new commission period.

That’s a lot of wasted effort from the company’s perspective. But what else can you expect? Incentives change behavior. Salespeople don’t make sales because that’s what they want. They make sales because they want money. They’ll do whatever is the most efficient thing that gets them money. If you change what that thing is with rules, you’ll create a new “most efficient thing” – and people will do that.

Back when the British Empire ruled India, there was a time when cobras were a real menace. They were responsible for a lot of injuries and deaths in the cities (often of kids) and so the ruling Brits tried to solve the problem by putting out a bounty, paying a certain amount of money for every dead cobra people turned in. What actually happened was that people started secretly breeding cobras because it was a great way to make money. When the Brits caught on, they halted the bounty – and everyone dumped their cobras, which resulted in the population being higher than it was before.

And so here we add rule #3:

3. There is always more than one incentive.

See, the people really did dislike cobras. The reason they weren’t just killing them already (even without the bounty) is that cobras are dangerous! You can get really hurt trying to kill one, and it often might be better to just try to shoo it out of your house with a broom… where it can plague your neighbor. So creating an incentive to face that danger wasn’t, by itself, a bad decision. People really did desire to lead cobra-free lives. But they had an even stronger desire, which was to lead a wealthy life. That created conflicting incentives, and the stronger one won out. And when the bounty was ceased, people had all these cobras in boxes and stuff. They could have killed them instead of letting them go, but… remember that thing about cobras being dangerous and tricky to kill? It’s easier to tip over a box and run.

People will spend a lot of time “gaming the system” as long as they perceive that there are gains for doing so. This creates a lot of wasted effort. For example, Americans spend collectively 6 billion hours doing their taxes every year. As much as possible, most people are trying to game the system. That doesn’t mean they’re cheating or anything! It just means they’re following incentives… in fact, they’re doing what the tax code wants them to do!

Remember, every tax (or tax break) is ostensibly designed to create or change a behavior in the first place. The government wants revenue, but ultimately all sorts of social engineering motivations come into play. Cigarettes get taxed because politicians hope that somehow revenue will be raised and people will smoke less. They’re probably dishonest about one of those hopes, but even still, if you smoked less because of the tax, people could hardly claim that you were cheating on your taxes just because you ultimate paid less to the state coffers.

So when people spend 6 billion hours figuring out what they can claim, remember that it’s because the rules (at least claim to) encourage them to do certain things or not do other things. That’s the incentive.

But of course, oh those unintended consequences. 6 billion wasted hours. More cobras than ever. Extra driving – and extra pollution & congestion – to get that Coke. Always a side effect. Always something you didn’t see.

Choice architecture is hard. Designing incentive structures is hard. Predicting how people will react to them is hard – but predicting that people will react to them, in one way or another, is dead simple.

They will.