The Cobra Effect

How to Get More Cobras: A Practical Guide to Government Incentives

Some years ago, the British government in India supposedly discovered that Delhi had too many cobras.  This was a reasonable conclusion.  One cobra in the pantry is probably too many, particularly if you are reaching behind the flour sack in search of the baking powder.

The British decided to solve the problem with the one tool governments have always trusted: other people’s money.  They announced that anyone delivering a dead cobra would receive a cash bounty.

At first, the plan worked splendidly.  Dead cobras arrived by the basketful.  Some official in Delhi undoubtedly prepared a lengthy report in triplicate announcing that the Cobra Reduction Initiative had exceeded expectations and recommending that everyone involved receive a promotion.

Then the locals began breeding cobras.

This was not the outcome the British had intended, but it was the outcome they had purchased.  The government was paying for dead cobras, and the most efficient way to obtain dead cobras was no longer wandering around in the weeds hoping to meet one.  It was raising them at home.

Eventually, the authorities discovered the cobra farms and canceled the bounty.  The breeders, now stuck with cages full of poisonous merchandise for which there was no longer a government market, released the snakes.  Delhi supposedly ended up with more cobras than it had before.

This is known as the cobra effect: an economic policy designed to solve a problem creates an incentive that makes the problem worse.

It is a wonderful story.  It may not be true.  But we shouldn’t let that bother us too much—a hell of a lot of the things we believe aren’t real.  We can take solace in the fact that the principle described is all too real.

The British really did pay bounties for cobras, but historians have found no solid contemporary evidence of a great Delhi cobra-breeding industry.  An 1887 inquiry by the Bombay Natural History Society concluded that breeding large numbers of cobras in captivity for bounties was highly improbable.  The elaborate tale about the cobra farms and the mass release of snakes appears to have grown in the retelling.

That is unfortunate because economics seldom supplies us with a story containing bureaucrats, poisonous snakes, entrepreneurial villagers, and a conclusion suitable for printing on a coffee mug.

Fortunately, the French stepped in with rats.

The Great Hanoi Rat Enterprise

In 1902, the French colonial government in Hanoi became alarmed about the city’s rat population.  Rats carried plague, crawled through the new sewer system, and generally behaved like rats.  The French therefore offered a bounty for every rat killed.

Officials did not want thousands of dead rats piled on their desks, which was understandable.  They required people to turn in only the rat’s tail as proof of death.

Somewhere in Hanoi, a fellow looked at a rat, looked at a pair of shears, and founded an industry.

Soon, French officials began noticing healthy rats running through the city sans tails.  The rat catchers had realized that killing a rat eliminated a productive asset.  Cutting off its tail, collecting the bounty, and releasing it allowed the animal to breed more rats, each equipped at birth with another government check attached to its rear end.

There were also reports of rats being brought into Hanoi and of people deliberately raising them.  The French had wanted fewer rats.  They had accidentally established a municipal rat-farming program.

Unlike the Delhi cobra story, the Hanoi rat hunt is well-documented.  So, if strict historical accuracy concerns you, call it the rat effect.  Unfortunately, that sounds less like an economic principle and more like a discussion about Congress.

People Respond to Incentives, Not Speeches

The cobra effect rests upon one of the simplest principles in economics: people respond to incentives.

They do not necessarily respond to what officials intended, what the regulation’s preamble says, or what was written on the PowerPoint slide at the committee meeting.  They respond to what gets them paid, promoted, fined, fired, or left alone.

The government wanted fewer cobras, but it paid for carcasses.  The French wanted fewer rats, but they paid for tails.  In both cases, officials confused the thing they could count with the result they actually wanted.

This happens regularly because the real objective is usually difficult to measure.

A school wants educated students, but education is hard to measure, so it rewards test scores.  Teachers naturally spend more time teaching what appears on the test.  In extreme cases, weak students are encouraged to stay home on examination day, or someone quietly improves their answer sheets afterward.  The scores rise, the superintendent holds a press conference, and the children remain about as educated as they were on Monday.

A bank wants profitable, satisfied customers, but customer satisfaction is hard to enter on a spreadsheet.  The bank therefore rewards employees for opening accounts.  Wells Fargo employees, crushed beneath aggressive sales quotas, opened millions of accounts customers had not requested.  Management asked for more accounts and received more accounts.  The fact that many were fraudulent was apparently an unnecessary detail omitted from the quarterly report.

Hospitals want patients treated promptly, but that is complicated.  Administrators therefore measure waiting times.  At some Veterans Affairs facilities, employees manipulated appointment dates and scheduling records so the official waiting times looked shorter.  The veteran was still sitting at home waiting for an appointment, but the computer was feeling much better.

This is the great comfort of modern administration.  If reality refuses to improve, there is usually a number somewhere that can be adjusted.

That is the essence of the cobra effect. It is not an argument that government action, regulation, or incentives never work. It is an argument that good intentions are not a substitute for understanding how people will respond.

When a Fine Becomes a Fee

The cobra effect does not always require fraud.  Sometimes perfectly respectable people make the situation worse simply by behaving rationally.

Economists once studied Israeli day-care centers where some parents habitually arrived late to collect their children.  The centers introduced a fine, expecting parents to become more punctual.

Late pickups increased.

Before the fine, parents who arrived late felt guilty because they were inconveniencing the staff.  After the fine, they no longer viewed lateness as inconsiderate.  It had become a service available for purchase.

“I am not keeping the teacher late,” the parent could reason.  “I am buying twenty minutes of additional childcare.”

The fine did not forbid the behavior.  It established a market price.

Even after the fine was removed, lateness remained elevated.  The old social obligation had been destroyed.  Once you convert shame into a financial transaction, it can be difficult to get the shame back.

This is worth remembering whenever someone proposes a small fine for objectionable behavior.  If the fine is lower than the value of committing the offense, it may function as a license.

A wealthy man does not necessarily see a $25 parking ticket as punishment.  He may see it as valet parking provided by the city.

Mexico City Solves Pollution by Selling More Cars

Mexico City once attempted to reduce air pollution by prohibiting cars from operating one weekday according to the last digit of their license plates.

The theory was that people would use public transportation, share rides, or remain home contemplating the environmental damage caused by their selfishness.

Many families bought a second car with a different license number.

Since poorer families could not afford a shiny new second automobile, they often purchased an older, dirtier one.  The policy intended to reduce driving encouraged additional automobile ownership.  Studies found no convincing improvement in air quality.

This is what happens when policymakers assume that human beings will stand quietly where they have been placed.  Human beings instead go around the rule, under it, through it, or down to a used-car lot.

The regulation applied to a particular license plate.  It did not apply to the trip the driver still wanted to make.

Carbon Credits With a Tail Attached

The same problem has appeared in environmental policy.

HFC-23 is an extraordinarily powerful greenhouse gas produced as a byproduct of manufacturing another chemical.  Carbon-credit programs paid factories to destroy HFC-23 rather than release it into the atmosphere.

Destroying the gas was beneficial.  The difficulty was that the credits could become extremely valuable.  A factory might earn more from destroying the waste than from selling the original product.

At that point, producing the chemical created more waste, and more waste created more credits.  A program intended to eliminate a pollutant risked making pollutant production profitable.

This is the cobra effect wearing a laboratory coat.

Whenever the government pays people to dispose of something, it ought to ask whether somebody will begin manufacturing it.  If Washington offered $100 for every broken toaster, the country would soon contain forty-seven new toaster factories, all producing defective appliances with patriotic efficiency.

The Difference Between a Target and a Goal

Economists often connect the cobra effect with Goodhart’s law: “When a measure becomes a target, it ceases to be a good measure.”

A measurement can tell us something useful until people have a reason to manipulate it.

The number of arrests may once have indicated police activity.  Attach promotions to arrest totals, and officers may concentrate on easy, minor arrests rather than serious crimes.

The number of academic publications may once have suggested scholarly productivity.  Make careers depend upon publication, and professors divide one useful article into six smaller ones, establish journals nobody reads, and produce research proving that additional research is urgently needed.

The number has not become false.  It has become useless.

This does not require wicked people.  Ordinary people adjust to the rules placed before them.  If you reward quantity, they produce quantity.  If you reward speed, they sacrifice quality.  If you punish reported mistakes, they stop reporting mistakes.

Then management announces that errors have fallen to zero.

How Not to Breed Cobras

Avoiding the cobra effect requires asking a question rarely heard at government meetings:  How could a reasonably clever scoundrel take advantage of this?

Better yet, invite several reasonably clever scoundrels into the planning meeting.  They aren’t hard to find, every organization already employs a few.  They know where the weaknesses are.

Policymakers should measure genuine outcomes rather than convenient activities.  They should examine what happens outside the measured category.  They should conduct small trials before launching enormous programs.  They should use independent audits, limit rewards, and plan what happens when the program ends.

Most importantly, they should stop assuming that citizens, employees, businesses, teachers, doctors, and rat catchers will respond as obedient figures in an economic diagram.

They will respond to the actual incentives.

The cobra effect is not an argument that government should never act, that regulations never work, or that every public employee is hiding a cage of snakes behind his desk.  Markets create perverse incentives too, as Wells Fargo demonstrated with considerable enthusiasm.

It is an argument for humility.

Good intentions are pleasant things.  They look handsome in speeches, fit comfortably into campaign literature, and require no particular knowledge of economics.  But good intentions do not repeal human behavior.

Before adopting any bounty, subsidy, quota, fine, or performance target, we should ask what intelligent people will do when they pursue it literally and selfishly.  Can they collect the reward without solving the problem? Can they hide the problem, rename it, move it somewhere else, or manufacture more of it?

If so, the policy may not eliminate the cobras.  It may merely teach people how to raise them for profit.