A reward for rat tails
In 1902, the French colonial government of Hanoi, in what is now Vietnam, had a problem. Plague was spreading in the region, and rats carry the fleas that spread it. So the city paid people to kill rats. To avoid handling piles of dead rats, officials paid a small reward for each rat tail.
The historian Michael Vann studied the city's records. On some days, more than 20 000 rats were killed. Then officials began to see rats in the streets with no tails. People were catching rats, cutting off the tails, and letting the rats go to breed. Investigators also found rat farms in villages outside the city.
Nobody in this story was foolish. The officials wanted fewer rats. The people responded sensibly to the rule they were given. But the rule paid for tails, not for fewer rats. So, in effect, it rewarded keeping rats alive.
For eight lectures, the story in this course has been one direction: the market fails, and the government steps in to fix it. A tax corrects pollution. A rule forces disclosure. A public body provides what the market will not. This lecture asks the uncomfortable question that comes next. What if the fix itself fails?
When the cure is worse than the disease
Recall what market failure means. A free market produces the wrong amount of something, compared with what is best for society as a whole. Pollution is overproduced. Public goods are underproduced. Hidden information makes trades collapse. In each case there is a loss, a deadweight loss, that a well-designed intervention could reduce.
But every intervention is designed and run by people. They have limited information, their own incentives, and limited budgets. So an intervention can also go wrong.
Government failure happens when a government intervention makes the allocation of resources worse than it would have been without it. This includes an intervention whose costs are larger than the gains it brings.
Read the definition carefully, because the comparison is the whole point. Government failure does not mean "the policy was imperfect". Almost every policy is imperfect. It means the outcome with the policy is worse than the outcome without it. A tax that removes half of a pollution problem is a partial success, not a failure. A tax that leaves things worse than no tax at all is a failure.
Economists group the reasons for government failure into five causes. Each one is a mechanism, a chain of cause and effect. Learn the chain, not just the name.
Five ways an intervention can go wrong
1. Unintended consequences
A policy changes incentives, and people respond to the incentives, not to the policy's intentions. Some responses are hard to predict. The rat-tail reward is the clearest example.
You have met a quieter example already. In the lecture on price controls, San Francisco extended rent control in 1994. Covered tenants gained, as intended. But landlords of covered buildings reduced their supply of rental housing by 15%, and rents across the city rose by 5.1% (Diamond, McQuade and Qian, American Economic Review, 2019). A policy meant to make renting cheaper made it dearer for future tenants.
A third example is in the next lecture. The United States began requiring a minimum volume of biofuel in its transport fuel in 2005, and expanded the requirement in 2007. Most of it was ethanol made from corn. By 2009, roughly a third of the US corn crop was used for ethanol (US Congressional Research Service, 2010). Corn is also food and animal feed, so the effects reached far beyond fuel. How large those effects were is still debated, and the next lecture looks at them closely.
2. Information problems
To correct a market failure exactly, the government must know things that are very hard to know. To set the right pollution tax, it needs the true external cost of each extra tonne. To set a subsidy, it needs the true external benefit. These numbers cannot be read off a price tag. They must be estimated, and estimates can be wrong.
The lecture on negative externalities showed that a tax equal to the external cost moves the market to the best quantity. Here is the same idea in an invented fertiliser market. Runoff from fertiliser damages rivers, an external cost of $20 per tonne.
A tax of exactly $20 per tonne would remove the whole triangle. But suppose the government does not know the true cost is $20. It must guess. What happens when the guess is wrong? Try it below.
You are the government. You do not know the true external cost, so you choose the tax. Move it from $0 upwards. Watch the grey deadweight loss and compare it with the $100 000 a week lost with no tax at all. Find the tax that makes things worse than doing nothing.
Try:
Solid lines: demand (MPB = MSB) and supply (MPC). Dashed brown: marginal social cost (MSC). Dotted: supply with your tax. Grey hatched: the deadweight loss.
The table shows the fertiliser market under different taxes. The true external cost is $20 per tonne, and with no tax the deadweight loss is $100 000 a week.
| Tax per tonne | Quantity (thousand tonnes) | Deadweight loss per week | Compared with no tax |
|---|---|---|---|
| $0 | 40 | $100 000 | the market failure |
| $10 | 35 | $25 000 | better |
| $20 | 30 | $0 | best |
| $30 | 25 | $25 000 | better |
| $40 | 20 | $100 000 | the same |
| $50 | 15 | $225 000 | worse |
| $60 | 10 | $400 000 | worse |
What did you see? A tax that is too low still helps. A tax that is too high also helps, up to a point. At $40, twice the true damage, the tax cuts production to 20 thousand tonnes, which is 10 below the best quantity. That is exactly as far below as the free market was above it, so the loss is the same $100 000. Above $40, the intervention is worse than no intervention at all.
That is it. That is the whole idea of government failure, in one diagram. The government did not need bad intentions to fail. It only needed to be badly wrong about one number.
3. Regulatory capture
A regulator is set up to control an industry in the public interest. Regulatory capture happens when the regulator ends up serving the industry instead. The economist George Stigler stated the idea strongly in 1971 (Bell Journal of Economics and Management Science). He wrote that, as a rule, "regulation is acquired by the industry and is designed and operated primarily for its benefit". He later received the Nobel Memorial Prize in Economic Sciences, in 1982.
Why would this happen without any corruption? Look at who cares. A rule that protects an industry brings a large gain to a small number of firms, so each firm has a strong reason to lobby for it. The same rule costs each consumer a little, so no consumer has a strong reason to fight it. The organised side usually wins. Regulators also need expert knowledge, and the experts often come from the industry they regulate.
The last lecture ended with a warning: a licence can be used by a profession to keep competitors out. Here is how. A licence for electricians protects buyers, who cannot judge wiring. But suppose the licensing board is run by existing electricians. Each extra year of required training protects them from new competition and raises their prices. The rule began as a solution to information failure. It can become a barrier that mainly serves the people it was meant to control.
4. Short-termism
Governments face elections, typically every four or five years. A policy with visible benefits now and costs later is easy to choose. A policy with costs now and benefits later is hard to choose, even when it is better overall. Serious action on climate change is the standard example: higher energy prices arrive at once, while most of the benefits arrive decades later, long after the next election. So the incentive is to delay, or to choose a weaker policy that looks strong.
5. Costs that exceed the benefits
An intervention uses real resources: officials, inspectors, forms, and the time firms spend complying. Even a perfectly designed policy can cost more to run than the market failure it removes. Worked Example 1 puts numbers on this.
A city's pollution tax removes about half of the damage from a factory. A newspaper calls it "government failure". Is that the right name?
Show the answer
Not from that fact alone. Government failure means the outcome is worse than it would have been with no intervention. Removing half the damage is an improvement, so it is a partial success. It would become government failure only if the tax's costs, such as running it or new problems it causes, were larger than the half of the damage it removed. To decide, compare the outcome with the policy against the outcome without it, not against perfection.
A real case: one day a week without your car
The policy. In 1989, Mexico City, which had severe air pollution, introduced Hoy No Circula, "Today it does not drive". Most drivers were banned from using their car on one weekday each week, chosen by the last digit of the number plate. Two digits are banned each weekday, so about a fifth of cars should stay off the road each day.
The prediction. Fewer cars on the road should mean less exhaust, so the air should become cleaner. Before reading on, make your own prediction below.
Lucas Davis studied the air in Mexico City before and after the ban. What do you think he found? Choose one.
The obvious prediction is that pollution fell by about a fifth. Davis (Journal of Political Economy, 2008) found no evidence that the restrictions improved air quality, for any pollutant he studied. He also found evidence that the ban increased the number of vehicles in use and shifted the fleet towards high-emission vehicles.
What happened. Lucas Davis used air measurements taken every hour at monitoring stations across the city (Journal of Political Economy, 2008). He found no evidence that the restrictions improved air quality, for any pollutant or method he used. He also found evidence that the ban increased the total number of vehicles in use and shifted the city's cars towards high-emission vehicles.
The verdict. This is government failure through unintended consequences. The designers predicted the first response, "drive less on one day". They missed the second, "buy another car". The policy also had real costs: lost trips, inconvenience, and the cost of the extra cars. With no measured gain in air quality, the outcome was worse than doing nothing. The deeper lesson is about information. The government knew how many cars there were, but not how households would respond. The ban targeted the visible thing, cars on a given day, and not the thing that mattered, total exhaust. That is the same mistake as paying for rat tails instead of fewer rats.
Weighing one failure against another
So markets fail, and governments fail. Does that leave us with nothing to say? No. It gives us a better question.
In 1969 the economist Harold Demsetz criticised what he called the "nirvana" approach (Journal of Law and Economics). This approach compares a real, imperfect market with an ideal, perfect government, and treats any gap as proof that the government should act. The fair comparison is real against real. An imperfect market must be compared with an imperfect government.
Intervention improves the outcome only when the part of the market failure that it actually removes is larger than the costs it adds: the cost of running it, plus the new problems it creates.
This rule explains two common patterns. When a market failure is large and well measured, even a clumsy policy is likely to help. When a failure is small or uncertain, and the policy is costly or easy to capture, leaving the market alone may give the better outcome. Good answers on this topic do not say "markets always know best" or "government should always fix market failures". They weigh the two failures, case by case, using evidence.
In the fertiliser market, the deadweight loss with no intervention is $100 000 a week. A regulator sets the tax at exactly the true cost, $20, and removes the whole loss. But the regulator's office, its inspections and the firms' paperwork cost $120 000 a week.
Weigh it. Gain: $100 000 a week. Cost: $120 000 a week. Net effect: , a loss of $20 000 a week. The policy is perfectly designed and still makes society worse off. This is government failure of type 5: the cost exceeds the benefit.
Check. If running the policy cost $60 000 a week instead, the net effect would be . The design is the same; only the running cost decides the verdict.
The government overestimates the damage and sets the tax at $30. Running the tax costs $10 000 a week. Is this better or worse than doing nothing?
The quantity. Firms produce where , so and thousand tonnes, 5 below the best quantity of 30.
The loss. At , MSB is and MSC is . The gap is 10, so the deadweight loss is , which is $25 000 a week.
Weigh it. With the policy, society loses a week. Without it, society loses $100 000 a week. The tax is wrong by 50%, and it still cuts the loss by $65 000 a week. Imperfect intervention is not the same as government failure.
Three classic mistakes
Government failure means the outcome is worse than with no intervention. A policy that improves things, but not completely, is not a government failure. Always state what the policy is being compared with.
Government failure is a risk to weigh, not a verdict. The same course that shows governments can fail also showed that markets can fail. The question is always which failure is larger in this case, with this policy.
Writing "regulatory capture" or "unintended consequences" earns little on its own. Show the mechanism: who responded, to which incentive, and how that made the outcome worse. In Mexico City: the ban, then (the evidence suggests) households buying extra cars, often older ones, then no improvement in the air.
Work each one on paper first. Open a hint only when you are stuck. All situations are invented unless a source is given.
-
Name the main cause of government failure in each case, and give the chain in one sentence. (a) A fishing licence board, run by existing fishing firms, issues almost no new licences. (b) A fuel subsidy introduced before an election costs more every year, but no government removes it. (c) A town pays $1 for every old tyre handed in for recycling, and people start bringing tyres from neighbouring towns. (d) A tax is set at $40 when the true damage is $15.
Hint 1
The five causes are: unintended consequences, information problems, regulatory capture, short-termism, and costs exceeding benefits.
Hint 2
For each, ask who responds and why. Is it the regulated industry, voters and politicians, ordinary people reacting to a reward, or a government that guessed a number?
Full answer
(a) Regulatory capture: the board is run by firms that gain from fewer competitors, so it protects them rather than the public. (b) Short-termism: removing the subsidy would bring a visible cost before the next election, while the benefit of removing it arrives later. (c) Unintended consequences: the reward pays per tyre handed in, not for fewer tyres dumped in this town, so people respond to the reward as written, like the rat tails. (d) Information problems: the government did not know the true external cost, and set the tax more than twice too high (). In a market with straight-line curves like the fertiliser example, that leaves a bigger loss than no tax.
-
In the fertiliser market of this lecture, the government sets the tax at $45 per tonne. Find the quantity produced and the deadweight loss. Is this government failure? Then find the range of taxes that leaves society better off than no tax.
Hint 1
Firms produce where , so .
Hint 2
The loss is . The gap is twice the distance from 30.
Full answer
thousand tonnes, which is 12.5 below the best quantity. At , MSB is 82.5 and MSC is 57.5, a gap of 25. The loss is , which is $156 250 a week. That is more than the $100 000 lost with no tax, so yes, this is government failure. The tax beats doing nothing when the loss is below $100 000, that is, when the quantity is within 10 of 30: from 20 to 40 thousand tonnes, not counting the ends. That means any tax above $0 and below $40. A tax of exactly $40 does no better and no worse than no tax.
-
A new rule removes 80% of a market failure that costs society $500 000 a year. Running the rule costs the government $300 000 a year, and firms spend $150 000 a year complying with it. Is the rule worth having? What is the most the two costs together could be for the rule to be worth it?
Hint 1
The gain is the part of the failure that the rule actually removes, not the whole failure.
Hint 2
Gain: . Costs: add both, because firms' time is a real cost to society too.
Full answer
Gain: a year. Costs: a year. Net: , a loss of $50 000 a year. The rule is not worth having in its current form: cost exceeds benefit. It would be worth having only if the total cost were below $400 000 a year. A common mistake is to leave out the firms' compliance cost because the government does not pay it. Society still pays it.
-
Evaluate this statement: "Because government failure exists, governments should leave markets alone." Use this lecture's examples and at least one calculation.
Hint 1
Use Demsetz's point: what should an imperfect government be compared with?
Hint 2
Put Worked Example 2 (a wrong tax that still helps) beside Mexico City (a policy with no measured benefit). What makes the difference between them?
Full answer
For the statement. Interventions can make things worse. Mexico City's driving ban had real costs and, in Davis's study, no measured improvement in air quality. In the fertiliser market, a tax above $40 leaves a bigger loss than no tax. A perfect tax with running costs of $120 000 a week still loses $20 000 a week. Against it. The statement compares a real government with a market that is assumed to work. But the market here also fails: it loses $100 000 a week. A tax of $30, wrong by 50%, cuts the total loss to $35 000 a week. So an imperfect government can still beat an imperfect market. Judgement. The statement is too strong. Government failure is a reason to design policies carefully, not a reason never to act. Intervention makes most sense when the market failure is large and well measured, and running costs are small. It also helps when the policy targets the harm itself: the exhaust, not the car; fewer rats, not the tails. When the failure is small or the policy is easy to capture or to get around, leaving the market alone may be better.
Notice what the rat tails, the driving ban and the wrong tax have in common. In each case a rule targeted something easy to see or count: tails, cars on a given day, one estimated number. The real goal was something harder to see: fewer rats, cleaner air, the true harm. People respond to what a rule actually rewards, not to what its designers hoped for.
This is why economists study incentives so closely, and why they compare real choices with real choices. The question is never "Is the market perfect?" or "Is the government perfect?". Neither is. The question is which imperfect arrangement gives the better outcome here, and how we would know. That question, more than any diagram, is the habit of mind this course has been building.
- Government failure happens when an intervention leaves the allocation of resources worse than it would have been without the intervention.
- The five main causes are unintended consequences, information problems, regulatory capture, short-termism, and costs that exceed the benefits.
- A corrective tax set too high can do more harm than no tax at all; with straight-line curves, this happens when it is more than twice the true external cost.
- Mexico City's 1989 driving ban showed how people's responses can defeat a policy: Davis (2008) found no evidence that air quality improved.
- The fair test compares an imperfect market with an imperfect government, and intervention is worth it only when the gain it delivers exceeds the costs it adds.