Examples of government failure — Economics

Before this

This leads to

Fiscal policy — government spending and taxation
Non-tariff barriers — quotas, subsidies, regulations · coming soon

Mountains of butter nobody wanted

In November 1986, the stores of the European Community held more than 280 000 tonnes of butter that was already at least two years old (UK Parliament, Hansard, 3 December 1986). Nobody had bought it. Nobody was going to. And the Community had paid farmers to produce it.

This was not an accident or a mistake in one office. It was the predictable result of a policy working exactly as it was written. In 1980, that policy took 73.2% of everything the Community spent (European Commission, CAP expenditure data).

The last lecture gave you five causes of government failure: unintended consequences, information problems, regulatory capture, short-termism, and costs that exceed the benefits. This lecture works through four real cases: farm price support in Europe, rent control, corn ethanol in the United States, and the "welfare cliff". For each one, ask two questions. Which cause, or causes, are at work? And what is the chain of cause and effect?

Case 1: a guaranteed price for farmers

The Common Agricultural Policy (CAP) began in the early 1960s. Western Europe remembered wartime food shortages. So the goals made sense: stable food supplies and a decent income for farmers. The main tool was a guaranteed minimum price. If the market price fell below it, public agencies bought the produce at the guaranteed price.

A guaranteed minimum price is a price floor. You met floors in the lecture on price controls. Here is the same idea in an invented wheat market, with numbers you can check.

An invented wheat market. Without the floor, the market clears at $100 a tonne and 100 thousand tonnes. At a guaranteed $150, farmers grow 150 thousand tonnes but buyers want only 50 thousand. The surplus is 100 thousand tonnes a year.

At $150, farmers move up their supply curve and grow more. Buyers move up their demand curve and buy less. The gap is a surplus of $150 - 50 = 100$ thousand tonnes. To keep the floor in place, the government must buy that surplus. Here that costs $100000 \times $150 = $15$ million a year, for wheat nobody wanted at that price.

The surplus does not disappear when it is bought. Follow it, one step at a time.

Step through — from a guaranteed price to farmers abroad

Press Next to add one link in the chain. Read the reason before you go on.

The chain from a guaranteed price to its effects, one link at a time:

The government guarantees $150 a tonne when the market price would be $100. The aim is higher and steadier farm incomes.
Farmers grow more and buyers buy less. A higher price calls out more output and reduces the quantity demanded.
A surplus appears: 150 thousand tonnes grown, 50 thousand bought.
The government buys and stores the surplus. Otherwise sellers would cut prices to clear their stocks, and the floor would collapse.
Storage is costly and food spoils. So the surplus is sold abroad at the lower world price, with the government paying the difference.
Extra supply on world markets pushes world prices down.
Farmers in poorer countries, who have no such support, receive lower prices and lose sales.

One policy creates three groups who pay: taxpayers, consumers, and farmers abroad.

No step in this chain needs anyone to be foolish. Every step is a sensible response to the step before.

Real case — The Common Agricultural Policy, Europe (1960s onwards)

The policy. Guaranteed minimum prices for farm products such as grain, milk and butter, backed by public buying of anything the market would not take.

The prediction. Supporters expected stable supplies and secure farm incomes. Economic theory predicted something more: a binding floor creates a surplus, and the surplus must be bought, stored or sold off.

What happened. Both predictions came true. Food supplies were secure. But surpluses grew. By the 1980s, newspapers spoke of "butter mountains" and "wine lakes". In November 1986, more than 280 000 tonnes of butter in the Community's stores was at least two years old (Hansard, 1986). Surpluses were sold abroad with export subsidies. In 1980, the CAP took 73.2% of the Community's spending (European Commission). Critics also linked high guaranteed prices to more intensive farming, with heavier use of fertiliser and pesticides and damage to wildlife and water.

Farmers' groups in poorer countries complained about the effects abroad. In 2008, Ghana's national tomato producers' federation said that about 27 000 tonnes of processed tomatoes arrived from Europe each year (The New Humanitarian, 2008). Campaigners warned that subsidised European products were hurting local farmers. How much of such harm came from EU subsidies, rather than from other suppliers, is hard to measure.

What changed. From 1992, a series of reforms moved support away from guaranteed prices and towards direct payments to farmers. In 2015, members of the World Trade Organization agreed in Nairobi to eliminate agricultural export subsidies, with developed countries acting immediately for almost all products (WTO). By 2023, the CAP's share of EU spending had fallen to about 24.6% (European Commission).

The verdict. The main cause is unintended consequences: the price floor created surpluses, storage costs and harm abroad. Many economists also see regulatory capture: farmers are a small, well-organised group with a large gain from support, while the cost is spread thinly across millions of consumers and taxpayers. That is exactly the pattern Stigler described in the last lecture.

Case 2: rent control, the full picture

The lecture on price controls showed what a rent ceiling does. Fewer flats are offered, a shortage appears, and illegal "key fees" and poorer repairs follow. It also gave the evidence from San Francisco after rent control was extended in 1994. Landlords of covered buildings cut their rental supply by 15%, and rents across the city rose by 5.1% (Diamond, McQuade and Qian, American Economic Review, 2019).

There is one more layer, and it is about who gets the help. Rent control is meant to help people who cannot afford market rents. But a controlled flat is not given to the person who needs it most. It goes to whoever already holds it, or whoever found it first. People who hold one have a strong reason never to move, even when their income rises or their family changes. A large family may squeeze into a small controlled flat, while a single person stays in a large one.

Edward Glaeser and Erzo Luttmer studied this in New York City, which has had rent regulation in some form since the Second World War (American Economic Review, 2003). They compared who lives in which kinds of flats in New York with the pattern in cities without rent control. They found that an economically and statistically significant share of New York's flats appeared to be held by the "wrong" households, compared with a free market. They argued that this misallocation could cost society more than the shortage itself.

The verdict. Rent control shows two causes at once. Unintended consequences: landlords respond by supplying less. Information problems: the policy cannot see who most needs a flat, so it helps whoever holds one.

Case 3: fuel from corn

The last lecture promised a closer look at ethanol. Here it is. In 2005, the United States began to require a minimum volume of renewable fuel in its transport fuel, and in 2007 it raised that requirement sharply. Most of the renewable fuel was ethanol made from corn (maize). The aims were less dependence on imported oil, lower greenhouse gas emissions, and new demand for farmers.

Real case — The US Renewable Fuel Standard (2005 and 2007)

The policy. A legal minimum volume of renewable fuel, rising each year, to be blended into petrol. Corn ethanol supplied most of it.

The prediction. More demand for ethanol means more demand for corn. Corn is also food for people and feed for animals. So theory predicts higher corn prices, and higher prices for the foods that depend on corn.

What happened. The share of the US corn crop used for ethanol rose from about 7% in 2000 to about 40% in 2011 (US Department of Agriculture, Economic Research Service). Part of that corn returns to the food chain as animal feed, a by-product of making ethanol, so the net share is smaller. The average price US farmers received for corn rose from $2.00 a bushel in the 2005/06 season to $4.20 in 2007/08, and to $6.22 in 2011/12 (USDA ERS). In 2007 and 2008, world food prices rose sharply, and many poorer countries suffered.

How much was ethanol to blame? Here economists disagree, and you should say so. Other forces pushed prices up at the same time: high oil prices, droughts, a weaker dollar, low stocks and export bans. An IFPRI study estimated that biofuel demand explained about 30% of the rise in average grain prices from 2000 to 2007, and 39% for maize (Rosegrant, US Senate testimony, 2008). A World Bank working paper put the share much higher (Mitchell, 2008). It attributed about 70–75% of the rise in food commodity prices from 2002 to mid-2008 to biofuels and their related effects. Its author noted that this estimate is hard to compare with others. The range is wide. But no serious estimate puts the effect at zero.

And the climate? When the policy began, corn ethanol was expected to cut emissions compared with petrol. In 2008, Searchinger and colleagues argued in Science that, once the clearing of new farmland around the world is counted, corn ethanol nearly doubles greenhouse gas emissions over 30 years. Other studies give smaller land-use effects, and this question is still debated.

The verdict. The main causes are unintended consequences, through the link between fuel and food markets, and information problems. The true effect on emissions depends on land-use changes across the world, which were hard to predict and are still hard to measure. Many economists add regulatory capture, because corn farmers and ethanol producers gained a great deal and lobbied to keep the requirement. A fair verdict is "a policy whose costs were larger and whose benefits were smaller than its designers expected", not "the policy that caused the food crisis".

Stop and think

A student writes: "Corn prices rose after the ethanol mandate, so the mandate caused the rise." What is wrong with this argument, and how would a careful economist check it?

Show the answer

"After" does not prove "because". Oil prices, droughts, the dollar and low stocks all changed at the same time, and each of them can raise corn prices too. A careful economist tries to separate the causes. For example, they build a model of the corn market and ask how prices would have moved with everything else the same but no extra ethanol demand. That is why the estimates differ: different models separate the causes in different ways. The honest conclusion is that ethanol was one cause among several, and its share is uncertain.

Case 4: when working more pays almost nothing

The last case is not about prices in a market. It is about how a rule is designed. Benefits support people who cannot earn enough to live on. But a benefit must stop somewhere, or it would be paid to everyone. How it stops changes people's incentives.

Imagine an invented benefit of $800 a month for a person with no job. Now consider three ways to withdraw it as their earnings rise.

A cliff. The full $800 is paid while earnings are $600 a month or less. Above $600, it stops completely. Someone earning $600 has $1 400 in total. If they earn $650, they have only $650. Earning $50 more costs them $750.
100% withdrawal. Every dollar earned reduces the benefit by a dollar. A job paying $600 leaves total income at exactly $800. Taking the job gains nothing.
A taper. Every dollar earned reduces the benefit by only a part of a dollar. With a 55% taper, a job paying $600 reduces the benefit by $0.55 \times 600 = 330$ , to $470. Total income is $600 + 470 = $1 070$ , a gain of $270.

The first two designs create a poverty trap: a situation where working more, or taking a job at all, leaves a person no better off, or even worse off. The person is not lazy. Like the people with the rat tails, they respond sensibly to the rule. Try the three designs yourself.

Graph builder — design the withdrawal of a benefit

You design how an $800 benefit is withdrawn. Choose a design. For a taper, choose how many cents of benefit are withdrawn for each extra dollar earned. Watch what taking a $600 job is worth, and where the benefit stops.

Taper rate: 55%

Try:

The graph shows total income (earnings plus benefit) for three designs of an invented $800 benefit.

With a cliff, income drops from $1 400 to $600 at the $600 line. With 100% withdrawal, income stays flat at $800 until earnings pass $800. With a 55% taper, income always rises, by 45 cents for each extra dollar, until the benefit ends at about $1 455.

Design	Income with a $600 job	Gain from taking it	Gain from $100 more	Benefit stops at
Cliff at $600	$1 400	$600	−$700	$600
100% withdrawal	$800	$0	$0	$800
63% taper	$1 022	$222	$37	≈ $1 270
55% taper	$1 070	$270	$45	≈ $1 455

What did you see? A cliff punishes the step across it: earning a little more can cost hundreds of dollars. With 100% withdrawal, the line is flat, so work up to $800 pays nothing. Lowering the taper rate makes work pay at every step. But look at where the benefit stops. With a 55% taper, someone earning $1 400 a month still receives a small payment. This is the real design problem: there is no rate that is best on every count.

A real example of this choice comes from the United Kingdom. Universal Credit replaced several older benefits and tax credits with one payment that is withdrawn gradually as earnings rise. In November 2021, the government cut its taper rate from 63% to 55% (The Universal Credit (Work Allowance and Taper) (Amendment) Regulations 2021). A claimant then kept 45 pence of each extra pound earned above any work allowance, instead of 37 pence. Universal Credit has also faced criticism of its own, including waits for first payments, so a better taper did not end the debate.

The verdict. A badly designed withdrawal rule is government failure through unintended consequences and poor design: a safety net that also discourages work. The fix is not to remove the safety net. It is to design the withdrawal more carefully, and to accept the cost of doing so.

Three questions before you judge a policy

Before you conclude that a government should, or should not, intervene, ask three questions.

Weighing an intervention

1. How large is the market failure? A large, well-measured failure makes a strong case for action, even with an imperfect policy.

2. How likely and how severe is government failure? Does the government have the information it needs? Is there an organised group likely to capture the policy? Will short-term politics shape it?

3. Is there a better design? Often the real choice is not "intervene or not" but "this design or a smarter one".

Question 3 is the one most answers miss. Europe kept supporting farmers, but changed the form of support. The United Kingdom kept its benefit, but lowered its taper. In both cases, the improvement came from a better design.

Three classic mistakes

Trap 1 — Listing the harms without weighing them

Saying "the CAP caused surpluses" is only half an answer. The question is whether the harms were larger than the gains, compared with doing nothing or with a better policy. Secure food supplies were a real gain. Weigh both sides, then conclude.

Trap 2 — Treating a contested number as settled

Estimates of ethanol's effect on food prices range from about 30% to about 75% of the rise. Writing "ethanol caused the food crisis" claims more than the evidence shows. Say what the evidence shows, say that it is debated, and explain why.

Trap 3 — Blaming the people instead of the incentive

The farmer who grows unwanted wheat, the tenant who never moves, and the worker who turns down extra hours all respond sensibly to the rules. An answer that calls them greedy or lazy has missed the mechanism. Name the incentive the rule created, and show how people responded to it.

Try it yourself

Work each one on paper first. Open a hint only when you are stuck. All situations are invented unless a source is given.

In the invented wheat market of this lecture (demand $P = 200 - Q$ , supply $P = Q$ , with $Q$ in thousand tonnes), the guaranteed price is raised from $150 to $160. Find the new surplus and the cost of buying it. By how much does the cost rise?

Hint 1
Put $P = 160$ into each equation to find what buyers buy and what farmers grow.

Hint 2
Buyers: $160 = 200 - Q$ . Farmers: $160 = Q$ . The cost is surplus × price.

Full answer
Buyers buy $200 - 160 = 40$ thousand tonnes. Farmers grow 160 thousand tonnes. The surplus is $160 - 40 = 120$ thousand tonnes. Buying it costs $120000 \times $160 = $19.2$ million a year, compared with $15 million at $150. So the cost rises by $4.2 million. A $10 rise in the floor (about 7%) raised the cost by 28%, because the surplus grows from both ends and each tonne costs more.
An invented benefit of $500 a month is withdrawn at 80% of earnings. Asha earns $400 a month. She is offered 10 more hours of work at $10 an hour. How much better off would she be each month? What is her real hourly gain? At what earnings does the benefit stop?

Hint 1
Work out her total income (earnings plus benefit) at $400 and at $500.

Hint 2
Benefit $= 500 - 0.8 \times earnings$ . At $400 of earnings it is $500 - 320$ .

Full answer
At $400: benefit $500 - 320 = 180$ , so total income is $400 + 180 = $580$ . At $500: benefit $500 - 400 = 100$ , so total income is $500 + 100 = $600$ . She is $20 better off for 10 hours of work, a real gain of $2 an hour, although the job pays $10 an hour. The benefit stops when $0.8 \times earnings = 500$ , so at earnings of $500 \div 0.8 = $625$ . Check: $625 + (500 - 0.8 \times 625) = 625 + 0 = 625$ ✓. A high taper rate is a weaker form of the poverty trap.
A newspaper says: "The US ethanol mandate caused the 2007–08 world food crisis." Using the evidence in this lecture, explain what is right and what is wrong with this claim. Name the causes of government failure involved.

Hint 1
What does the evidence say about the size of ethanol's effect? Is the range from zero to large, or from moderate to large?

Hint 2
List the other forces that raised food prices in the same years.

Full answer
What is right. The mandate greatly raised demand for corn: the share of the US crop used for ethanol rose from about 7% in 2000 to about 40% in 2011, and corn prices roughly doubled between 2005/06 and 2007/08. Studies find a real effect on food prices, from about 30% of the grain price rise (IFPRI) to about 70–75% of the food commodity price rise (World Bank working paper). What is wrong. "Caused" suggests it was the only cause. Oil prices, droughts, a weaker dollar, low stocks and export bans all pushed prices up at the same time, and estimates of ethanol's share differ widely. Causes. Unintended consequences (the link between fuel and food markets), information problems (uncertain effects on prices and on emissions), and possibly regulatory capture (organised farm and ethanol interests). A better sentence: "The ethanol mandate was one significant cause of the rise in food prices, though economists disagree about how large its share was."
A government wants to support farmers' incomes. Option A is the guaranteed price of $150 a tonne from this lecture. Option B is a fixed payment to each farm, the same whatever it grows. Use the invented wheat market to compare the two options for quantity, price paid by consumers, surplus, world markets, and cost to taxpayers. Which is the better design, and what risk remains?

Hint 1
Does Option B change the price that farmers receive for one more tonne of wheat? If not, does it change how much they grow?

Hint 2
With Option B, the market still clears at its equilibrium. Find it, then compare consumers, the surplus and the budget.

Full answer
Option A. Price $150. Farmers grow 150 thousand tonnes and buyers buy 50, a surplus of 100 that costs taxpayers $15 million a year to buy, plus storage. Consumers pay $50 a tonne more than the market price. The surplus is sold abroad and pushes world prices down. Option B. The payment does not depend on output, so it does not change the reward for one more tonne. The market clears at $100 and 100 thousand tonnes: no surplus, no storage, no dumping, and consumers pay the market price. Taxpayers pay only the fixed payments, which can be set to give farmers whatever support the government chooses. Verdict. Option B removes the surplus mechanism, so it is the better design. This is the direction CAP reform took after 1992. The remaining risk. Fixed payments can still be captured: if they are paid by farm size, most of the money goes to the largest landowners, who need it least. A better design changes the incentive problem, but the capture problem still needs watching.

The bigger idea

Look back at the four cases. A guaranteed price paid for output, so farmers produced output. A rent ceiling protected whoever held a flat, so people held on to flats. A fuel rule rewarded ethanol volume, so corn went into tanks. A benefit rule took away most of each extra dollar, so extra work stopped paying. In every case, people did exactly what the rule rewarded.

This is the deepest lesson of this topic, and it reaches far beyond economics. A rule does not get the behaviour its designers hoped for. It gets the behaviour it rewards. Good policy design starts with one question: "If I were the farmer, the tenant or the worker, what would this rule make me do?" Asked before a policy begins, it makes many government failures predictable, and so avoidable.

Key concepts in this lesson

Europe's guaranteed farm prices acted as a price floor, creating surpluses that harmed taxpayers, consumers and farmers in poorer countries.
Rent control reduces supply and also misallocates housing, because help goes to whoever holds a controlled flat rather than to whoever needs it most.
The US corn ethanol mandate raised corn demand and prices, but economists disagree about how much it contributed to the 2007–08 food price rise.
A benefit withdrawn too steeply creates a poverty trap, and a lower taper makes work pay at the cost of a more expensive benefit.
Before judging an intervention, ask how large the market failure is, how likely government failure is, and whether a better design exists.