When Do Markets Get It Wrong? Cases the Argument Does Not Cover

The case that a competitive market produces an efficient allocation depends on conditions that are stated explicitly in the underlying theory. Buyers and sellers must be numerous enough that none can influence the price. Everyone must know what they are buying. All costs and benefits of a transaction must fall on the parties to it. The goods must be things that can be owned and withheld. Nobody must be able to free ride on what others pay for. Where those hold, the standard conclusion follows, and where any fails it does not, which means the interesting question is never whether markets work but which condition is broken in a particular case and what follows.

Economists name a small number of distinct failures:

• •Externalities, where a cost or benefit falls on somebody outside the transaction, as with pollution
• •Public goods, which nobody can be excluded from and which one person's use does not diminish
• •Market power, where a seller or buyer can influence the price
• •Information problems, where one side knows something the other does not
• •Missing markets, where no way exists to trade something at all
• •Coordination failures, where an outcome everybody prefers is not reachable individually

Asymmetric information produces failures with a characteristic structure that is worth following. Where a seller knows the quality and the buyer does not, buyers offer only what an average item is worth, which drives good items out of the market and lowers the average further, potentially until the market collapses, which is the argument George Akerlof made about used cars in 1970 and for which he shared a Nobel Prize. The same logic operates in insurance, where people who know they are high risk buy more cover, raising prices and driving out low risk customers. The remedies are also structural, including warranties and certification that credibly signal quality, mandatory disclosure, professional licensing, and in insurance either compulsory participation or risk classification.

One case deserves separate treatment because the standard account of it has been substantially revised. A resource that nobody owns and everybody can use, including a fishery, a pasture or an aquifer, is predicted to be overexploited because each user gains the full benefit of taking more while bearing only a share of the cost, which is the argument popularised in 1968. The proposed remedies were private ownership or state control. Elinor Ostrom documented many real communities that managed such resources successfully for centuries through their own rules, and identified the conditions under which that works, including clearly defined boundaries, rules matched to local conditions, participation by users in making them, monitoring and graduated sanctions. She received a Nobel Prize in 2009, and the finding is that a third option exists.

Identifying a failure establishes that the market outcome is not efficient and does not by itself establish that intervention will improve matters. Governments have their own information problems, since a regulator generally knows less about an industry than the industry does. Interventions create incentives of their own and are captured by the interests they regulate with some regularity. The costs of administration are real. The relevant comparison is therefore between an imperfect market and an imperfect intervention rather than between a real market and an ideal government, which is a point made forcefully by economists across the political range. What the framework does supply is a diagnosis, since knowing which condition failed indicates which kind of remedy might address it and which will not.

The efficiency argument depends on stated conditions including numerous participants, full information and all costs falling on the parties, so the useful question is which condition breaks in a given case. Externalities, public goods, market power and information asymmetry are the recognised categories. Diagnosing a failure does not establish that an intervention will do better, since governments have their own information problems.