A parent offers a child five dollars for every book completed.
Reading increases immediately.
So does the child’s interest in books with large type, wide margins and an unexpectedly generous relationship between pictures and profit.
The parent optimized reading volume.
The child optimized revenue.
Both are behaving rationally.
This is why incentives are such dangerous little sentences. They sound like instructions:
Read more.
But they are actually environments:
Here is the new reason to read, the unit that counts and the game worth learning.
Incentives work
Let us begin with the inconvenient fact.
Rewards and penalties often change behaviour.
Pay people to perform a clear, verifiable task and more of it may happen. Fine a prohibited action and less of it may happen. Use a temporary reward to help establish a habit and the behaviour may continue.
An essay about perverse incentives can easily become a collection of clever stories told by people whose salary is also an incentive.
The interesting question is narrower:
When does an incentive increase the visible action while weakening the value underneath it?
A reward points a flashlight
The first effect is attention.
If a call centre rewards calls handled per hour, employees notice call length.
If a software team rewards bugs closed, bugs become countable units. One complicated failure may be less attractive than six tiny ones. Preventing bugs becomes a particularly poor career strategy because prevented bugs have the irritating habit of not appearing in reports.
The incentive designer sees the desired outcome.
The participant sees the scoring rule.
This overlaps with the problem of metrics, but the mechanism is different. A metric tells us what is visible. An incentive tells us what visible thing is worth pursuing.
Together, they can turn a proxy into a sport.
A reward can rewrite the story
Imagine asking a neighbour to help move a sofa.
They agree.
Afterwards, you offer them four dollars.
Four dollars is not nothing.
It is also somehow worse than nothing.
Before the payment, the action belonged to friendship. After the payment, it belongs to a very disappointing labour market.
Psychologists and economists call one version of this motivation crowding: an external reward can sometimes weaken other reasons for acting, especially where the activity was already interesting, moral or identity-based.
A classic meta-analysis found that expected tangible rewards could reduce later intrinsic motivation under some conditions, while positive feedback could increase it. The effects depended heavily on the type of reward, the task and how the reward was delivered. (Deci, Koestner & Ryan, 1999)
So “rewards destroy motivation” is too simple.
Sometimes a reward is information:
You are getting better.
Sometimes it is control:
Do this because I own the prize.
Sometimes it changes the category:
Apparently this is not what generous neighbours do. It is what extremely affordable movers do.
A fine can become a price
The most famous example involves day-care centres in Israel.
Researchers introduced a fine for parents who collected children late. Late pickups increased rather than decreased. One interpretation is that the fine changed lateness from a social violation—someone is waiting because of me—into a service with a price. When the fine was later removed, the higher level of lateness persisted during the study period. (Gneezy & Rustichini, 2000)
It is a wonderfully portable story.
Possibly too portable.
It involved ten centres, and many retellings leap from that setting to a universal law of human motivation. The study shows that pricing a norm can change its meaning. It does not show that every fine invites more offending.
An expensive parking ticket is still capable of an educational experience.
The design matters. So does the size.
Rewards teach people what the designer really values
Organizations communicate through speeches.
They communicate more convincingly through compensation.
A company may say:
Put the customer first.
Then pay salespeople entirely on this quarter’s revenue.
The incentive explains which sentence has enforcement.
This can produce behaviour nobody officially requested:
- unsuitable products sold to meet commission;
- cooperation withheld because another team would receive credit;
- attendance without contribution because presence is rewarded;
- “bugs fixed” created by dividing one issue into several;
- referrals that become less trustworthy once friends know money is attached.
The participant is not necessarily violating the culture.
They may be reading it accurately.
Incentives also select people
Suppose a community activity is done by volunteers who care about the mission.
Add a small payment.
Participation may rise. It may also attract people for whom the payment is central and make the original volunteers wonder whether their contribution is now being valued at nine dollars an hour.
In one field experiment on blood donation, offering payment reduced donation among women in the sample, while allowing the payment to be donated to charity removed that reduction. There was no comparable effect among men. One study in one setting is not a universal rule, but it shows how the framing of a reward can matter as much as its amount. (Mellström & Johannesson, 2008)
An incentive does not merely move the existing population.
It can change who enters, who stays and what they think everyone else is doing there.
When incentives are a good fit
The case is strongest when:
- the desired behaviour is clear;
- quality is easy to verify;
- side effects appear quickly;
- intrinsic motivation is weak or irrelevant;
- participants cannot cheaply game the proxy;
- and the reward does not contradict the social meaning of the act.
Paying someone to clear snow is not an assault on their inner love of shovelling.
A deposit on a bottle can align private convenience with public cleanup.
A temporary reward may help someone begin an unpleasant but beneficial routine.
The fact that incentives carry stories does not mean we should stop using them.
It means the story is part of the design.
Three questions before adding the prize
Before rewarding a behaviour, I would ask:
-
What exactly can the person maximize? Not what we hope they maximize. What does the rule literally reward?
-
What does the reward imply? Does it signal appreciation, control, distrust, unpleasantness or permission?
-
What happens when the reward disappears? Has the person built skill and habit, or learned that the behaviour is irrational without payment?
Perhaps there is a fourth:
If people become very good at this incentive, will we still like the result?
Children are unusually helpful test cases because they find loopholes before the implementation meeting has ended.
Adults simply call the loopholes “quarterly strategy.”
What I still cannot figure out
An incentive must be noticed to work.
But once noticed, it can become the explanation for the behaviour—even when duty, curiosity or care was already present.
Before rewarding an action, should we first ask what story the reward will tell people about why the action matters?
Until the next strange question,
Osagie
My current hunch: Incentives fail when they reward an observable fragment of the goal and accidentally replace the reason people cared about the whole.
Most likely reason I am wrong: We may romanticize intrinsic motivation; in many settings, a plain, well-priced incentive outperforms vague appeals to purpose.