You are about to draw a diagram that says something quite startling. It says that when a government subsidises a good, it spends more money than consumers and producers gain between them. Some of the money simply disappears. On the diagram, a subsidy destroys value.
And yet there is not a government on earth that does not subsidise something. Food, fuel, buses, farms, vaccines, solar panels, childcare, trains, university places, electric cars.
So who is wrong — the diagram, or every government in the world?
Five quickfire questions from earlier topics. Thirty seconds on the clock for each, then the answer. Every one of them is a tool you are about to need in the next ninety minutes — this is not a general warm-up, it is the loading of exactly what Part B onwards depends on.
Start with the obvious objection. If people wanted more of a thing, they would buy more of it, and firms would make more of it. So why would a government step in and hand producers money?
Because there are situations where the market gets the quantity wrong — and gets it wrong in a particular direction. A subsidyA payment from the government to producers (or sometimes consumers) which reduces the cost of supplying a good, increasing supply and lowering the price paid by buyers. is the tool for when the market produces too little. Hover or tap any underlined word for its definition.
Four reasons come up again and again, and they are worth holding as four separate arguments rather than one vague one, because an exam question is usually about only one of them.
Somebody other than the buyer benefits. A person who gets vaccinated protects the people around them; a household that insulates its roof burns less gas for everyone. Firms are not paid for those extra benefits, so the market under-produces. This is the merit good case and it is by far the most examined.
The good is a necessity and some households cannot afford enough of it — food, heating, bus fares, childcare. Here the aim is not efficiency at all. It is who gets what, and it is judged on different grounds.
A new domestic industry cannot yet compete with established foreign firms, but might if given time to grow and bring its costs down. Support now, in exchange for an industry later.
The country does not want to depend on anyone else for it. Food and energy are the usual examples, and the argument gets much louder whenever trade looks unreliable.
You already have the memory aid for the whole toolkit. LETS intervene in the market — Legislation, Education, Taxation, Subsidies. You met it on the previous page when the tool was Taxation. Today it is the S. Taxation and Subsidies are the two price-based tools and they point in opposite directions, which is why a question asking you to choose between them is really asking which way the market is failing.
Eight goods. For each one, decide which way the market is failing — is society getting too much of this, or too little?
A subsidy paid to producers lowers what it effectively costs them to supply each unit, so it moves the supply curve. And the acronym you already know for what shifts supply has the answer sitting in it.
Exam answers lose marks here in a very specific way, and it is worth naming. Students draw the diagram for a cash payment per unit produced — which is the right diagram — and then explain it by saying the subsidy “reduces the firm’s costs of production”, which describes a different kind of subsidy. Both are real. They are not the same thing and they do not give the same diagram.
Did you know? The International Monetary Fund puts total global fossil-fuel subsidies at $7.4 trillion in 2024 — about 6.4% of world GDP. But only $725 billion of that is money governments actually hand over or prices they hold down. The other $6.7 trillion is what the IMF calls implicit subsidy: the environmental and health costs that fuel users never pay for.
Which raises a question worth carrying through the rest of this page. If not charging someone for the damage they cause counts as subsidising them, then almost everything is subsidised, and the word begins to lose its edges.
Tap every one that is a subsidy to producers. Several are other kinds of intervention entirely, and telling them apart is the point — a question comparing methods of intervention needs you to keep them separate.
A subsidy diagram has two prices, two quantities and four areas on it, and dropping the whole thing on a class at once is why most people never really learn it. So it is built here one piece at a time. Click through the six steps and read each one before moving on — the last step asks a question that the numbers answer.
The market below is deliberately the same one used for the specific-tax walkthrough in the previous subtopic. Same demand, same supply, same $5 equilibrium. That is so you can see at the end that a $0.90 subsidy and a $0.90 tax on the identical market do the identical amount of damage — in opposite directions.
Six steps. Read each one, then press Next. You can click back to any step you have already seen.
Same market, same $0.90 subsidy. Work each cell out and type it in. Correct cells lock green and you can keep fixing the rest.
The government wrote the cheque to producers. So producers got the money — obviously?
No. In this market the subsidy was $0.90 a unit, and of that, 50 cents reached consumers as a lower price and 40 cents stayed with producers as a higher price received. The government paid producers, and consumers walked away with the larger share of it.
That is what incidenceWho actually ends up better or worse off as a result of a tax or subsidy — which is not necessarily the person who hands over or receives the money. means, and it is the same idea you met with taxes, running the other way. Who the cheque is written to does not decide who benefits. The market decides that, by moving the price.
And what it uses to decide is elasticityHow responsive quantity is to a change in price. Demand is inelastic when buyers barely change how much they buy; elastic when they change a lot.. The rule is short and worth memorising in this form:
The share of a subsidy that reaches consumers as a lower price is
In plain words: the side of the market that responds less to price captures more of the subsidy. Whoever cannot easily walk away is the one the price has to move to persuade — and moving the price in their favour is how the benefit reaches them.
Four instructions, four regions of the subsidy diagram. Click the area that matches each one — and note that one of the four instructions needs more than one region.
Here is where this topic stops being mechanical and starts being examinable.
Below are four markets. Everything is held constant except one thing: how elastic demand is. Same supply curve, same starting equilibrium of $5 and 10 units, same $0.90 subsidy. Only the steepness of the demand curve changes.
Before you work anything out, predict: in which market will the subsidy raise the quantity traded by the most?
Work out, for each market, what share of the subsidy consumers keep as a lower price. Then read what happens to quantity in each one.
Look at what the two columns do in opposite directions. Where demand is very inelastic, consumers keep 86% of the subsidy — but quantity rises by only 0.31 units. Where demand is elastic, consumers keep only 29%, but quantity rises by 1.59 units, five times as much.
So consider a government that subsidises a merit good in order to increase consumption of it. If demand for that good is price inelastic — and for many merit goods it is — then almost all the money becomes a price cut for people who were buying it anyway, and consumption barely moves.
That single observation answers a whole family of evaluation questions, and it is one of the most reliable routes to the top band on this topic. A subsidy is not a good tool for changing behaviour when behaviour is not price-sensitive. It is a good tool for changing prices.
Six sentences making up one complete chain of analysis, shuffled. Tap two rows to swap them until the logic runs from top to bottom.
“Subsidise it” is not a policy. How much is the policy — and the answer is not “as much as possible”.
Take the same market again and vary only the size of the subsidy. Work out the deadweight loss at each size and a pattern appears that settles the argument.
The quantity and the government’s spending are given at each rate. Work out the deadweight loss — then look at what the column is doing.
Double the subsidy from $0.90 to $1.80 and spending rises from $9.90 to $21.60 — a bit more than double. But the deadweight loss goes from $0.45 to $1.80: it quadruples. Double it again and the loss quadruples again, to $7.20.
The waste rises with the square of the subsidy while the spending rises roughly in proportion to it. So a small subsidy is cheap in waste per dollar spent and a large one is not. This is why an economist will usually argue for a modest, targeted intervention rather than a generous one — and it is a point you have now derived rather than been told.
Now back to the question at the top of this page. The diagram says a subsidy destroys value. Every government subsidises things anyway. Time to work out who is right — and the answer is more interesting than “both a bit”.
Go back to the deadweight loss you calculated. It appeared because the last few units cost more to produce than buyers valued them at. But “valued them at” means valued by the buyer. The demand curve records private benefit and nothing else.
So if the good has genuine external benefits — if a vaccinated person protects their neighbours, if an insulated house lowers everyone’s emissions — then the demand curve sits below the true value of those units to society. The extra output the subsidy encourages may be worth more than the diagram can show, and the “deadweight loss” may be nothing of the kind.
This resolves the contradiction, and it is worth stating carefully in an answer. A subsidy on an ordinary good with no externality is straightforwardly inefficient — the diagram is complete and the loss is real. A subsidy on a good with a large positive externality can raise total welfare, because the thing the diagram fails to draw is exactly the thing that justifies the policy. Saying which case you are in is the analysis. Examiners see a great many answers that praise a merit-good subsidy and then, two lines later, say it creates a deadweight loss, without ever noticing that those two claims are in tension.
From January 2023 the US federal government offered buyers of a qualifying new electric vehicle a tax credit of up to $7,500. The credit ended on 30 September 2025.
It moved the market. Battery-electric vehicles reached 12% of US light-vehicle sales in September 2025 — a record, and the last month the credit was available. In the quarter that followed they fell to under 6%, and were still at about 6% in mid-2026. Withdraw the subsidy and the quantity halves: the diagram, running backwards, in public.
But an evaluation found something less comfortable. Around 75% of the credits went to people who would have bought an electric vehicle anyway. For those buyers the subsidy changed nothing except their bank balance. Counting only the cars that genuinely would not otherwise have been sold, the cost to the taxpayer worked out at roughly $36,500 per additional vehicle — nearly five times the headline credit.
And yet the same study concluded the policy still passed a cost–benefit test, returning about $1.11 of US benefit for every $1 of government spending once the emissions saved and the production shifted to American plants were counted.
75% of the money went to people who would have bought anyway. Explain, in your own words, why that single figure can be used by both sides of the argument about whether this subsidy was a good idea.
Every dollar of subsidy is a dollar not spent on something else, and the alternative is rarely nothing — it is a hospital, a school, or a lower tax. A subsidy returning $1.11 per dollar is only a good idea if nothing available returns more.
The additionality problem, and the one the case above makes concrete. If most recipients would have acted anyway, the cost per additional unit is far above the headline rate.
A firm that can rely on support has weaker reason to cut its costs. The infant-industry argument only works if the industry eventually grows up — and subsidies are famously hard to withdraw.
Setting the right rate needs a number nobody can measure: the size of the external benefit. And the decision is political, so lobbying, votes and the electoral cycle all push on it.
A subsidy on a good that richer households buy more of hands more money to richer households. Fuel subsidies are the classic case: in Indonesia the richest fifth of households received 46% of the benefit and the poorest two-fifths just 18%.
You were asked whether the diagram or the governments were wrong. The honest answer is that the diagram is right about what it measures, and it does not measure everything.
If your instinct was that the diagram must be missing something — you were right, and the thing it is missing is external benefit. If your instinct was that governments subsidise for political rather than economic reasons — you were also right, and that is the government-failure argument, worth just as many marks.
Both instincts are load-bearing in the 12-mark question at the bottom of this page. An answer that uses only one of them cannot reach the top band, because a one-sided response scores zero for evaluation however good the analysis is.
Fourteen tiles, four groups — but the groups are not all the same size, so you cannot count your way to the answer. And finding the right tiles is only half of it: you then have to say precisely what they share.
Examiner tip. AO2 marks are for using the diagram, not for having one. A beautifully drawn, correctly labelled subsidy diagram that your writing never mentions earns the mark for drawing it and nothing more. Write sentences that begin “the diagram shows…” and “reading the new equilibrium off the diagram…”.
Always name both prices. “The price falls” is ambiguous once a subsidy exists, because one price falls and the other rises. Say the price consumers pay and the price producers receive, every time.
Say which kind of subsidy you are drawing. One sentence — “a specific subsidy of $X per unit, so supply shifts down by exactly $X” — prevents the commonest mismatch between diagram and explanation.
On an evaluation question, decide something. A mark is routinely reserved for a justified conclusion, and “it has advantages and disadvantages” decides nothing. Name the condition your answer turns on.
Ten questions written in the format of the multiple-choice paper. One attempt each, then the explanation — and the explanations say why each wrong answer is wrong, which is usually the more useful half. Several are built around the errors this page has been warning you about.
Five questions in the shape of the current paper. Write first, then reveal the mark scheme — the reveal will not open until you have made a real attempt, because reading the mark scheme first defeats the point. Your first submission is the one that counts; later attempts are stored as revisions and never overwrite it.
A specific subsidy is a payment made by the government to producers of a fixed amount of money for each unit of the good they produce or sell. Because the payment is the same on every unit, it reduces the price a firm needs to receive from buyers by exactly that amount, and so shifts the supply curve vertically downwards by the value of the subsidy.
A subsidy paid per passenger journey lowers the cost to operators of providing each journey, so the supply curve shifts vertically downwards by the value of the subsidy. The diagram shows the new supply curve crossing demand at a lower fare and a larger number of journeys.
However, the fall in the fare is smaller than the subsidy. As the fare falls, more people want to travel, and operators supply the extra journeys only if the amount they receive per journey rises. The subsidy is therefore split: passengers gain a lower fare and operators gain higher revenue per journey. How the split falls depends on relative elasticities — if demand for bus travel is price inelastic, because many passengers have no alternative way of getting to work, passengers capture the larger share.
Evaluation. Whether this weakens the case depends entirely on what the subsidy was for. If the objective was purely to make travel affordable, money retained by operators is money that missed its target. But bus subsidies are commonly intended to keep loss-making routes running at all, in which case the share retained by operators is not leakage — it is the policy working as designed.
From January 2023 the government of a large economy offered buyers of a new electric vehicle a tax credit worth up to $7,500. The credit was withdrawn at the end of September 2025.
Battery-electric vehicles reached 12% of new light-vehicle sales in September 2025, the highest monthly share ever recorded and the final month the credit was available. In the following quarter the share fell to below 6%, and it remained at about 6% a year later.
An independent evaluation estimated that around 75% of the credits were claimed by buyers who would have purchased an electric vehicle without them. Taking that into account, the cost to the taxpayer of each additional vehicle sold was put at approximately $36,500. The same evaluation concluded that the scheme nonetheless generated about $1.11 of domestic benefit for every $1 of government spending, once reduced emissions and increased domestic production were valued.
What the data shows. The subsidy clearly moved the market. Battery-electric vehicles reached a record 12% of sales in September 2025 and fell to under 6% in the quarter after the credit ended — a halving. That is the standard effect of a subsidy being introduced and then withdrawn: supply shifts down and quantity rises, then the shift reverses.
But most of the money changed nothing. If 75% of credits went to buyers who would have bought regardless, three-quarters of the spending was a pure transfer to those households and bought no extra adoption at all. This is why the cost per additional vehicle, at $36,500, is roughly five times the $7,500 credit: the whole cost of the scheme is carried by the quarter of purchases that were genuinely additional.
Against that, the scheme still passed. At $1.11 of benefit per $1 spent, the emissions avoided and the domestic production gained were together worth more than the policy cost. A subsidy can be simultaneously an inefficient way to buy adoption and a net gain for the country.
Evaluation. The verdict therefore depends on the criterion, and the data is not ambiguous about that — it simply answers two different questions. Judged as a way of accelerating adoption at least cost, this was poor value: a policy paying $36,500 per additional car would lose to almost any alternative use of the money, including a smaller credit better targeted at marginal buyers. Judged as a net welfare question, it succeeded, if narrowly. There is also a real difficulty in reading the September spike, which may be purchases brought forward by buyers racing a known deadline rather than purchases created; the sustained 6% afterwards suggests the underlying market was smaller than the record month implied.
Conclusion. It should be judged a qualified success that was badly designed rather than a failure. The benefit–cost ratio is the right test of whether to do it at all, and it was met; the 75% figure is the right test of how to do it, and it was failed. A means-tested or lower credit targeted at buyers close to the margin would have delivered most of the same benefit for far less of the cost.
A specific subsidy is a payment from the government to producers of a fixed sum for every unit sold. Because producers receive that sum on top of the market price, they are willing to supply any given quantity for a lower price from buyers, so the supply curve shifts vertically downwards by exactly the value of the subsidy.
The diagram shows the original equilibrium E₁, where demand crosses the original supply curve, and the new equilibrium E₂, where demand crosses the subsidised supply curve. Reading the two points off the diagram, the quantity traded has risen and the price on the vertical axis has fallen.
The diagram also shows something a pre-subsidy diagram cannot: two prices. The lower one is what consumers hand over; the higher one, found by going vertically up from the new quantity to the original supply curve, is what producers actually receive. The vertical distance between them on the diagram is the subsidy per unit, and the rectangle formed by that distance and the new quantity is the total cost to the government — which the diagram makes clear must be measured at the new, larger quantity, not the old one.
Incidence means who ends up better off, and it is not settled by who receives the cheque. Both sides gain here: consumers through the lower price they pay, producers through the higher price they receive.
Evaluation. How the gain divides depends on relative elasticity. The share reaching consumers as a lower price is PES ÷ (PES + |PED|), so the more inelastic demand is, the larger the consumers’ share. Intuitively, the side that responds less to price is the side the price must move furthest to accommodate.
But it would be wrong to say price elasticity of demand determines the split, for two reasons. First, the expression contains PES as well, and a market with highly elastic supply gives consumers most of the benefit whatever PED is. Second — and this matters more for policy — a large consumer share is not the same as a successful subsidy. Where demand is very inelastic, consumers keep most of the money but quantity barely moves, so a government subsidising a merit good in order to increase consumption of it gets the worst possible combination: high cost, and little of the behaviour change it wanted. Elasticity determines who gains; it does not tell you whether the policy achieved anything.
A merit good is one whose consumption generates benefits for people other than the consumer, and which consumers also tend to undervalue. Because firms are not rewarded for those external benefits, a free market produces less than the socially desirable quantity. A subsidy is the government’s most direct answer: a payment per unit to producers which lowers the effective cost of supply.
The diagram shows the supply curve shifting vertically downwards by the value of the subsidy, and a new equilibrium E₂ at a lower price and a larger quantity than E₁. Reading both prices off the diagram, consumers pay less than before while producers receive more; the vertical distance between the two is the subsidy, and the rectangle formed with the new quantity is what the government spends. Where the good has positive externalities, moving output towards the socially optimal quantity is precisely the objective, so the increase in quantity is the measure of success.
The case for is stronger than the diagram alone suggests. A standard analysis identifies a deadweight loss, because the extra units cost more to produce than buyers valued them at. But the demand curve records only private benefit. If the external benefit is real, those units are worth more to society than the diagram can show, and the apparent loss may not be a loss at all. This is the central argument, and it is why a subsidy on a merit good is defended in a way that a subsidy on an ordinary good cannot be.
The case against rests on three different objections, and they are not equally strong. The weakest is cost alone, since cost is only decisive relative to what the money would otherwise do. Stronger is the measurement problem: setting the right subsidy requires valuing the external benefit, a number nobody can observe, so the rate is chosen politically and is vulnerable to lobbying and the electoral cycle. Strongest of all is additionality. Evidence from vehicle-purchase subsidies suggests around three-quarters of payments can go to people who would have acted anyway, which multiplies the cost per additional unit several times over while changing nothing for the majority of recipients.
But the decisive consideration is elasticity, and it is the one most answers miss. How much extra consumption a subsidy buys depends on how responsive demand is to price. Many merit goods have inelastic demand — people do not vaccinate twice because it became cheaper. Where demand is inelastic, the subsidy is captured overwhelmingly as a lower price by consumers who were already buying, and quantity barely moves. The government then spends a great deal to achieve a transfer rather than a behaviour change. Where demand is price elastic, the same money buys a much larger increase in consumption.
That also points to when a different tool is better. If people under-consume because they misjudge the benefit, the failure is one of information, and correcting it shifts the demand curve itself — something no price cut can do. If the failure is that some households cannot afford the good at all, direct provision or a targeted transfer reaches them more reliably than a subsidy whose benefit is shared with everyone else in the market.
Conclusion. A government should use subsidies to raise consumption of merit goods where two conditions hold together: the external benefit is large enough to be worth paying for and can be estimated with some confidence, and demand is responsive enough to price that the money actually changes behaviour rather than merely reducing the bill for existing buyers. Where either fails — and for many merit goods the second one does — the subsidy is not wrong so much as mismatched, and information provision or direct provision will achieve more per dollar. The question is less whether to subsidise than whether a price is the thing that needs changing.
The closing paragraph of a strong answer to the 12-mark question above, with six words removed. Drag them back into place — and note that the words taken out are not the technical terms but the ones carrying the argument. Three words in the bank are not needed at all.