Business & Organizations · People & management
When incentives backfire at work
Tie pay to a number and people will chase the number. Work out why a measure breaks once it becomes a target, and how pay can be designed around the problem.

What you will learn
- Explain why a measure loses its value once it becomes a target
- Spot surrogation, when a measure starts to stand in for the goal
- Describe the sales pressure behind Wells Fargo's unauthorized accounts
- Weigh the evidence on who was to blame for the fake accounts
Lesson 1 of 4 · 5 min
Why a number stops working once people are paid to hit it
In 1902, French officials in Hanoi wanted fewer rats in the city's sewers, so they paid a bounty of 1 cent per rat tail. Tails poured in. Then officials began noticing live rats with no tails.
Rat catchers were cutting off the tails and letting the rats go, so they could breed. Inspectors found farms outside the city raising rats for their tails. The bounty was canceled, and the rat problem was worse than before.
Workplaces do this with less drama. Say a warehouse counts the orders each packer packs per hour. At first the count is a fair guide to who does good work. Then the warehouse starts paying a bonus on it. Now a packer can raise the count by rushing, and a high number no longer separates good work from careless work.
This is Goodhart's law, named after the British economist Charles Goodhart, who made the point about monetary policy in 1975. Its best-known wording is "When a measure becomes a target, it ceases to be a good measure."
The American social scientist Donald Campbell saw the same thing in schools. When test scores become the goal of teaching, he wrote in 1976, "they both lose their value as indicators of educational status and distort the educational process in undesirable ways."
That's two harms. The number stops being a reliable guide, and the work it tracked gets bent toward the number. In U.S. classrooms, critics point to teaching to the test and outright cheating.
A third harm is quieter. People who bend their decisions to improve a measure often come to believe the measure shows true performance better than it really does. This is called surrogation, because the measure ends up standing in for the goal.
None of this is a case against measuring. The usual reading of Campbell is to keep measuring, put qualitative indicators beside the numbers, and take concrete steps to limit manipulation.
Try to recallA sales director starts ranking reps by the number of calls they make each day. Six months later the average count has doubled, and the director reports that the team is twice as effective. What mistake might the director be making?
The director may be falling into surrogation, treating the call count as if it were true performance. Once reps are ranked on calls, they can raise the count with short or pointless calls, so a doubled count doesn't have to mean doubled sales.
Surrogation has a famous real case. Under heavy pressure to meet sales goals, employees at the bank Wells Fargo opened millions of accounts customers never asked for. A 2019 analysis by a management professor and a doctoral student read that scandal as surrogation.
So when someone is paid to hit a number, ask what the cheapest way to raise it would be. If the cheapest way doesn't serve the goal, expect people to find it, as the rat catchers did.
Lessons in this topic
Why a number stops working once people are paid to hit it
A number that tracks performance well can stop working once people are paid to hit it, and it can fool the people chasing it too.
How sales goals at Wells Fargo produced fake accounts
See how a bank's push to sell every customer more products ended in millions of accounts nobody asked for, and sort out what the record shows about blame.
Why paying for what's easy to count can backfire
Work out why paying for the part of a job that's easy to count can pull effort from the rest, and when a fixed salary does better.
How to pay for effort without paying for luck
Pay tied to raw results rewards luck along with effort, and comparing a firm with similar firms filters some of that luck out.
Three quick stories from this topic
Small fees, giant fines
Wells Fargo's fake accounts earned $2 million in fees and cost it billions in penalties
Employees opened millions of accounts without customers' consent to hit sales goals, and only about 85,000 ever ran up fees. By the end of 2018 the bank had paid nearly $3 billion in settlements. In 2020 it agreed to pay $3 billion more.
Why open them at all
New accounts counted toward sales goals and earned bonuses for the employees who opened them. Employees described expectations as high as 20 products a day.
The number kept rising
The bank promoted its products-per-customer figure to investors. The SEC later said that figure was inflated by accounts that were unused, unneeded or unauthorized.
The rat tail bounty
Hanoi paid a bounty for rat tails in 1902 and ended up with a worse rat problem
French officials paid 1 cent per tail. Rat catchers cut the tails off and released the rats to breed, and inspectors found farms outside the city raising rats for their tails. Pay for a count, and people find the cheapest way to raise it.
Goodhart's law
"When a measure becomes a target, it ceases to be a good measure." The law is named after the British economist Charles Goodhart, who made the point about monetary policy in 1975.
The same thing at work
Pay packers a bonus per order and they can raise the count by rushing. A high number then no longer separates good work from careless work.
When a plain salary wins
A Nobel-winning model shows that a plain salary can work better than performance pay
Bengt Holmström and Paul Milgrom showed in 1991 that when pay rides on the parts of a job that are easy to count, effort drains from the parts that aren't. Teachers paid on test scores might spend too little time on creativity and independent thinking.
Teachers and test scores
If teachers' pay depends on test scores, they might spend too little time on creativity and independent thinking. A fixed salary would balance their effort better.
The prize
Holmström shared the 2016 Prize in Economic Sciences with Oliver Hart for their contributions to contract theory.
Sources
- Goodhart's law, Wikipedia
- Campbell's law, Wikipedia
- Great Hanoi Rat Massacre, Wikipedia
- Wells Fargo cross-selling scandal, Wikipedia
- Wells Fargo to Pay $500 Million for Misleading Investors About the Success of Its Largest Business Unit (press release 2020-38), U.S. Securities and Exchange Commission
- The Prize in Economic Sciences 2016. Popular science background: Contract theory, The Royal Swedish Academy of Sciences, NobelPrize.org





