A negative externality is a spill-over cost imposed on third parties who are not involved in the transaction between producer and consumer. It is therefore not reflected in the market price.

For example, a coal-fired power station emits carbon dioxide and sulphur dioxide as by-products of electricity production. Local residents and future generations suffer increased rates of respiratory illness and climate-related damage — costs they bear without compensation. Because these social costs are external to the firm’s calculations, the market over-produces electricity relative to the socially optimal output.

The vertical axis is price, cost and benefit; the horizontal axis is the quantity of vaccinations. Marginal private benefit (MPB) slopes down and marginal private cost, which equals marginal social cost (MPC = MSC), slopes up; the free market settles where they cross, at quantity Q1. Marginal social benefit (MSB) lies above MPB: the vertical gap is the external benefit to people protected by someone else’s vaccination, such as babies too young to be vaccinated. The social optimum is where MSB = MSC, at Q*, to the right of Q1: left to the free market, too few vaccinations are bought. The welfare gain the free market misses is the triangle between Q1 and Q*, bounded above by MSB and below by MSC: the vaccinations whose social benefit exceeds their cost but which the market does not provide.

Social benefit = private benefit + external benefit. The graduate’s higher earnings are the private benefit; the productivity gain received by employers and co-workers is the external benefit.

Social benefit = 12,000 + 4,500 = USD 16,500 per graduate per year. External benefit as a percentage of social benefit = 4,500 ÷ 16,500 × 100 = 27.3% (to one decimal place).

A public good is non-rival — one person’s use does not reduce what is left for others — and non-excludable — people who do not pay cannot be stopped from benefiting. A private good is the opposite: rival and excludable.

The sea wall is both: protecting one more house behind it takes no protection away from the houses next to it, and a household that refuses to pay cannot be left unprotected while the houses on either side are kept dry. A sandbag is rival — a bag piled against one door cannot also protect the house next door — and excludable, because the shop sells it only to people who pay for it, so it is a private good.

Asymmetric information exists when one party to a transaction knows more than the other; here the pension provider knows far more about its fees, risks and likely returns than the saver does. Resources are misallocated when they go to uses that people would not choose if they were fully informed.

Because the fees are set out in long documents and performance is revealed only after decades, a saver cannot compare funds properly and may choose one with high fees and poor returns, or one that takes more risk than they realise. Savings then flow to funds that a fully informed saver would avoid, and some workers, unsure whom to trust, save too little for retirement — so too few resources go into pensions overall, and too many into poorly run funds.

A negative externality of consumption is a spillover cost imposed on a third party when an individual consumes a good, which is not reflected in the market price.

In the extract, a shopper who takes a single-use plastic bag pays nothing for the litter it becomes once it is thrown away: that cost falls on other people and on the environment, which is why Abu Dhabi reports the more than 172 million bags its ban kept out of the environment.

The private cost of consuming a good is the cost borne by the consumer who chooses to consume it: for a smoker, the price of the tobacco and the harm to their own health. The social cost is the private cost plus the external cost, the cost that falls on third parties who did not choose to consume, so social cost is greater than private cost by exactly the external cost.

The extract separates the two: of the more than 7 million deaths a year the World Health Organization attributes to tobacco, over 1.6 million are non-smokers exposed to second-hand smoke. The deaths of smokers are mostly part of the private cost, but the 1.6 million fall on people who did not choose to smoke, so they are an external cost, and they are what makes social cost greater than private cost. Those third parties are the smoker’s family, co-workers and anyone else who breathes the smoke; they are outside the transaction, and the smoker does not count the harm to them when deciding how much tobacco to buy.

A negative externality of consumption occurs when consuming a good imposes a cost on third parties, so its marginal social benefit (MSB) lies below its marginal private benefit (MPB), leading to over-consumption and allocative inefficiency at the free-market equilibrium. The AED 0.25 charge acts as an indirect (Pigouvian) tax on the consumer: it puts a price on a bag that used to be free, so the shopper who takes one now pays part of the external cost it causes.

Applied to Dubai, the response was large for so small a charge: within a month the Al Maya Group reported its shoppers taking approximately 40% fewer plastic bags, and Al Adil more than 20% fewer. Once a bag carried a price, many shoppers found it was worth less to them than AED 0.25 and brought their own. Some shoppers carried on paying, so the charge did not stop consumption: it removed the bags worth less than AED 0.25 to the person taking them and kept those worth more.

The chain of reasoning is: the charge raises the private cost of taking a bag closer to its full social cost, which moves shoppers along their demand curve to fewer bags, which reduces the quantity consumed towards the socially optimal level, which in turn reduces the welfare loss triangle associated with over-consumption. The result is an improvement in allocative efficiency in Dubai’s retail market.

The market failure is over-consumption: when bags are free, shoppers take them until the last one is worth almost nothing to them, beyond the socially optimal level where marginal social benefit equals marginal social cost, because the cost of the litter falls on others. A charge prices part of that external cost and leaves the choice to the shopper, who still takes any bag worth more than the charge; a ban sets the quantity directly, which is the optimum only if almost no bag is worth its full social cost to anyone.

On the extract’s figures the ban cut use much further: retailers in Abu Dhabi saw plastic bag use fall by 90% to 95% in the ban’s first year, while Dubai’s AED 0.25 charge cut plastic bags by approximately 40% at the Al Maya Group and by more than 20% at Al Adil in its first month. A charge removes only the bags worth less than it to the shopper, so if the external cost of a bag is larger than AED 0.25, the charge is too low and some over-consumption remains. The smaller fall in Dubai shows the charge left more bags in use than the ban did, though not by itself whether too many.

Abu Dhabi’s ban did not leave shoppers a free alternative either: retailers charged at least AED 0.50 for each reusable bag, twice Dubai’s charge, so part of the ban’s effect is a higher price on the bag shoppers switched to. A charge set at the right level needs the government to know the external cost of a bag, which it cannot measure precisely; a ban avoids that and removes the external cost with certainty, but it also removes the benefit of bags that some shoppers valued at more than their full social cost, which is a welfare loss of its own.

However, the figures are not like for like: Dubai’s are one month of reports from two retail groups, Abu Dhabi’s a year across its retailers. And Dubai planned its charge as a step towards a ban from the start, so its ban in 2024 is not evidence that the charge had failed. The ban is more effective at cutting bag use; whether it corrects the market failure better depends on how many of the bags it removed were worth their full social cost to the shoppers who took them, which the extract cannot show.

A negative externality of consumption is a cost that falls on third parties when a good is consumed and that is not reflected in its market price, so the social cost of consuming the good is greater than the private cost the consumer pays. Sugary drinks carry such a cost, ill health whose treatment falls partly on others, alongside an information failure: consumers under-value the long-run harm they will bear themselves. The International Diabetes Federation’s estimate that 1.3 million adults in the UAE had diabetes in 2024 shows the scale of the health problem the tax is aimed at.

The 50% excise did change behaviour where it was measured. In Saudi Arabia prices rose by more than the tax itself (a pass-through rate of 110%), and carbonated drink sales volumes fell by 35% relative to other Arab Gulf states. A price rise of at least 50% producing a 35% fall means demand is price inelastic: the price elasticity of demand is about -0.7 or smaller in size (35 ÷ 50), so consumption fell, but most of it remained. The chain is: the excise raises the price the consumer pays, so the private cost of each drink moves closer to its full social cost; consumers move along their demand curve and buy fewer drinks; and consumption moves from the free-market quantity towards the social optimum, shrinking the welfare loss that over-consumption creates.

But a tax on price was blind to sugar: until the end of 2025 a drink with half the sugar paid the same 50%, so producers had no reason to reformulate. Under Table 1 a drink with 8 g or more of sugar per 100 ml pays AED 1.09 a litre and one with 5 g to less than 8 g pays AED 0.79, but one below 5 g pays AED 0, so a producer that cuts the sugar below 5 g removes the tax entirely. The UK’s levy, which uses the same 5 g and 8 g thresholds, suggests the incentive works: the average sugar content of the drinks it covers fell by 46% between 2015 and 2020. A tax that leads producers to cut the sugar in each drink reduces the external cost without consumers having to drink less; a tax on price could reduce it only by cutting the number of drinks.

The new tax is a fixed amount per litre: on a 330 ml can of a high-sugar drink it is AED 1.09 × 0.33 ≈ AED 0.36, whatever the can costs. That is less than a 50% tax on any can priced above about AED 0.72 (0.36 ÷ 0.5), so unless producers reformulate, the price rise on a sugary drink may be smaller than under the old tax, and so may the fall in consumption. The change works through reformulation and through drinkers switching to untaxed low-sugar drinks, not through the price of the sugary one. The evidence is also weaker than it looks. The UAE study reported in 2024 found no statistically significant change in consumption since the tax was introduced, and the Saudi fall is one country’s carbonated drinks measured against its neighbours. The 1.3 million adults with diabetes cannot show either tax’s effect: diabetes builds up over decades and has many causes besides sugary drinks.

A fixed amount per litre is also regressive: it is a larger share of the price of a cheap drink than of an expensive one, and a larger share of a low income than of a high one. Against that, the untaxed low-sugar drinks give low-income drinkers a way to avoid the tax that the 50% rate did not. And neither tax corrects the information failure: a tax changes the price, not what consumers know about the harm to themselves, so labelling or education would still be needed alongside it.

The sugar-based tax is likely to correct the market failure more effectively, because it targets the sugar that causes the external cost and gives producers a reason to remove it, as the UK’s 46% fall suggests. But its case rests on reformulation rather than on price, the price rise on a sugary can may be smaller than under the 50% tax, and neither tax corrects consumers’ information failure.

Market failure occurs when the free market leads to a misallocation of resources — producing too much (negative externalities), too little (positive externalities, merit goods), or none at all (public goods). The question is whether government intervention is always required to correct these failures, or whether alternative mechanisms — including private bargaining and market-based solutions — can sometimes achieve efficient outcomes without state action.

In many cases, government intervention is essential because the market cannot self-correct. Public goods such as street lighting and national defence are non-excludable and non-rivalrous — the free rider problem means no private firm can profitably supply them. Only the state can fund provision through taxation. The UK’s sugar tax (Soft Drinks Industry Levy), announced in 2016 and introduced in 2018, was followed by a fall of about a third in the sugar sold in the drinks it covers, mainly because manufacturers reformulated their products to pay less of it. Without this intervention, the overconsumption of sugar — and its associated health-system costs — would have continued uncorrected.

The Coase theorem suggests that if property rights are well-defined and transaction costs are low, private parties can negotiate an efficient outcome without government involvement. For example, if a factory pollutes a farmer’s land and the farmer has clear legal ownership, they can negotiate compensation or a reduction in pollution — reaching the socially optimal output through bargaining. In practice, however, transaction costs are often high and property rights unclear, limiting the applicability of this approach.

Even when intervention is attempted, it can lead to government failure — where the intervention makes the allocation of resources worse rather than better. Governments face information failure — they may not know the exact size of an external cost, leading them to set taxes too high or too low. Regulatory capture means regulators may serve industry interests rather than the public. Unintended consequences are common — for example, minimum pricing on alcohol may harm low-income moderate drinkers without significantly reducing problem drinking. The cost of intervention itself (bureaucracy, enforcement, compliance) may exceed the welfare gain from correcting the failure.

On balance, government intervention is often necessary but not always effective or desirable. For pure public goods and large-scale externalities where private solutions are impractical, intervention is essential. However, the form and extent of intervention matters — well-designed market-based instruments (tradable permits, targeted taxes) tend to outperform heavy-handed regulation. The key judgement is whether the costs of government failure are likely to be greater or less than the costs of market failure left uncorrected. Where transaction costs are low and property rights clear, private solutions should be tried first.

Moral hazard occurs when an economic agent takes greater risks because it knows that some of the cost of things going wrong will be borne by someone else. Deposit protection is a promise, usually by a government-backed insurer, that savers will get their money back, up to a limit, if their bank fails. Banking already suffers from asymmetric information — depositors cannot see how their bank lends their money — so the question is whether protection makes the resulting risk-taking worse than the failures it prevents.

Protection weakens the discipline depositors would otherwise impose. A saver whose deposit is guaranteed has no reason to check how risky the bank’s lending is, or to move to a safer bank, so a risky bank can keep attracting deposits without paying a higher interest rate for them. Its managers and shareholders then keep the profits if risky loans pay off, while losses beyond their own capital fall on the insurer, and possibly on taxpayers, if they do not. The bank takes more risk than it would if it bore the whole cost. Credit flows to riskier projects than is efficient — a misallocation of resources, which is what makes moral hazard a market failure rather than only a risk to the bank. The 2023 case strengthens the argument. The US authorities protected all deposits at the two failed banks, including balances above the USD 250,000 limit that deposit insurance normally covers. Much of those banks’ funding came from large deposits held by firms — exactly the depositors best placed to monitor a bank — so guaranteeing them after the event signals that even uninsured deposits may be rescued, and future large depositors have less reason to check.

Against this, a bank lends most of its deposits out as long-term loans, so it can never repay every depositor at once. Without protection, a depositor who fears a loss has every reason to withdraw first, and the withdrawals themselves can force even a solvent bank to fail — a bank run. A run can spread as depositors wonder which bank is next, and the fall in lending that follows hits producers who cannot borrow to invest or pay suppliers, and the workers whose jobs depend on them. Protection also corrects an information failure: an ordinary saver cannot judge a bank’s balance sheet, so a limited guarantee protects the depositors least able to monitor — whose discipline over the bank was weak anyway.

How much harm moral hazard does depends on design. A limit keeps large, informed depositors exposed to losses, so they still have a reason to watch how the bank lends. Charging banks insurance premiums that rise with the risk they take makes riskier banks pay more for protection, and capital requirements make shareholders lose their own money before the insurer pays. Each narrows the gap between the risk a bank takes and the cost it bears. The harm is therefore concentrated in guarantees that go beyond the limit after the event, as in 2023. A one-off rescue may stop a run, but if depositors come to expect one every time, the limit stops meaning anything and the moral hazard it was meant to contain returns.

Who bears the cost matters too. In the 2023 case the US deposit insurer recovered its losses through a special charge on larger banks rather than from taxpayers, so the cost stayed in the banking system — but it fell on banks that had not taken the risks, which weakens the link between risk and cost again. The time frame changes the weighing. The benefit of protection is immediate and visible — a run stopped in days — while the cost of moral hazard builds up slowly, through years of riskier lending, and shows only when the next failure comes.

On balance, limited and properly priced deposit protection does more good than harm: runs can destroy solvent banks and spread to firms and workers, and the depositors it protects were never able to discipline banks anyway. The view in the question is right about unlimited, after-the-event guarantees like those of 2023, which protect the depositors best placed to monitor and teach every depositor to expect a rescue. The judgement therefore rests on design: protection does more harm than good only when it is unlimited, unpriced and not matched by regulation of the risks banks take.

A negative externality of production occurs when the production of a good imposes costs on third parties who are not involved in the transaction. The marginal social cost (MSC) exceeds the marginal private cost (MPC) — the gap between them is the external cost.

For example, a steel factory discharging chemical waste into a nearby river imposes costs on downstream fisheries, local residents, and the environment. The factory only considers its private costs (raw materials, energy, wages) when deciding how much to produce. It does not pay for the pollution damage — this external cost falls on third parties.

Because the market price reflects only private costs, not social costs, the price is too low and output is too high relative to the socially optimal level. On a diagram, the free market produces at Q₁ (where MPC = MPB) but the social optimum is at Q* (where MSC = MSB). The area between Q* and Q₁, bounded by MSC and MSB, is the deadweight welfare loss — the excess cost to society from overproduction.

This is a clear case of market failure because the price mechanism sends the wrong signal — it tells producers to produce more than is socially desirable. Resources are misallocated because the market does not account for the full cost of production. Government intervention (e.g. a Pigouvian tax equal to the external cost) could internalise the externality and move output toward the social optimum.