Market Failure 5: Imperfect Information (When life hands you lemons, call your mechanic)

There are things we know that we know. There are known unknowns. That is to say there are things that we now know we don’t know. But there are also unknown unknowns. There are things we do not know we don’t know. So when we do the best we can and we pull all this information together, and we then say well that’s basically what we see as the situation, that is really only the known knowns and the known unknowns. And each year, we discover a few more of those unknown unknowns.

Donald Rumsfeld

For people to make good decisions, they need to have good information. A person who believes that bleach is a tasty treat will happily (and rationally) poison themselves to death. As I noted in Part One, perfect information is a requirement of perfectly efficient markets for this reason – even the most rational of beings will make bad decisions if they decide based on incomplete or inaccurate information.

However, this series is about more than just pointing out when markets aren’t perfect – the real question is what to do about it. In a lot of cases imperfect information is not a serious problem. For one thing, individual errors often cancel out in large samples (this is popularly referred to as the wisdom of crowds), which means that the system as a whole usually functions better than the ignorance of its participants would suggest. For another thing, market participants themselves recognise the problem and attempt to fix it. Almost all of you will personally participate in one of these fixes – insurance.

Insurance is a risk management tool – it converts an uncertain set of future expenditures into a certain (or at least more certain) expenditure, and risk itself is a product of imperfect information. Imagine if it was known for a certainty which houses would burn down and when. Insurance companies would happily offer very cheap rates for houses that would never burn down, but why would anyone buy insurance if they knew they would never have to make a claim? Conversely, insurance for houses that were going to burn down would be very expensive; insurers would charge the cost of rebuilding the house, plus a bit to account for their operating and capital costs. Since the insurance would have to cost more than replacing the house, no-one would buy it. Risk and ignorance are two sides of the same coin – you are uncertain about a thing precisely because you don’t know everything about it.

So does this mean the market has done all the work for us this time? I’m afraid not. For one thing, individual errors don’t always cancel out. If people have a cognitive bias that leads them to more often make an error in one direction than the other direction, that will impact markets at the aggregate. We’ll discuss this problem in the next part. But there’s another situation that can cause a problem insurance can’t solve, and that’s when the parties to a transaction have different amounts of information. It is the problem of asymmetric information that I will spend the rest of this post discussing.

There are two potential problems with asymmetric information. The first, adverse selection was first described in detail by George Ackerloff using an example called the market for lemons. Imagine a market for used cars where there were two kinds of cars: peaches, which are high quality, and lemons, which are low quality. The seller of each car knows whether it is a peach or a lemon, but the buyer can’t tell peaches and lemon apart – they want a peach more than a lemon, but they have no way of telling whether a given car is a peach or a lemon until after they have bought it.

A buyer would pay more for a peach than a lemon, but since they can’t tell them apart, their willingness to pay is going to be the average of their willingness to pay for a peach and a lemon, weighted by the relative frequency of peaches and lemons in the market ¹ (if a buyers values peaches at $50,000 and lemons at $10,000, they will be willing to pay $30,000, assuming there are as many peaches as lemons). But the owners of peaches are aware that their cars are worth more than that, which will make them reluctant to sell at that price. Sure, if you no longer have any use for your car, you may sell anyway, but if the car still has use value to you, then you won’t be as interested in selling at an average price than you would be at a peach price. Conversely, lemon owners will be thrilled to sell at an average price. This means that the car market will be flooded with lemons, and have few, if any, peaches. This will lead buyers to revise down their odds of getting a peach, thereby reducing their willingness to pay. This leads to a downward spiral in which nearly all used cars for sale are lemons.

Note who loses out because of adverse selection. People who wanted to buy or sell a lemon anyway aren’t much affected – buying and selling a lemon is not harder than it would be with full information. The people who are harmed are those trying to buy or sell a peach, as sellers have to put up with lower prices and buyers have trouble even finding a peach.

So why doesn’t insurance get us out of this mess? One of the reasons is that if prospective buyers can’t tell if a car is a lemon, the insurance company may not be able to either. In which case, how could they tell if your claim was legitimate? The other reason is that insurance markets are themselves affected by adverse selection.

Different people represent a different level of expected cost to insurers. An athlete in their 20s and a smoker in their 60s represent a different level of expected health care expenditure. Those two people may be easy to differentiate, but what about two seemingly identical car owners, one of whom drives rarely and carefully, the other often and recklessly? From a car insurer’s perspective, one of these customers is a peach and one is a lemon, and the lemon will be much more interested in getting insurance. This is the reason insurance companies ask so many questions of prospective customers, and why they reserve the right to reduce coverage if you lie to them.

In fact, these requirements of insurers show how governments can help solve adverse selection problems.

1. One common solution is to ban making misleading statements about a product you are selling. If there was a penalty for selling a lemon as a peach, then peach seller could credibly declare that their cars were peaches. This would permit the market to move from a pooling equilibrium (where all cars are treated the same) to a separating equilibrium (where different cars are treated differently).
2. Another solution is to mandate testing and certification of cars. If each car had to be tested by an independent expert and bore an official sticker saying if it was peach or a lemon, that would also fix the information problem.
   Another option would be to require minimum car standards – if you ban lemons, the only cars for sale will be peaches. The problem with this as a solution is that some people would prefer to buy a lemon at lemon prices than a peach at peach prices (for example, poor people who can’t afford peach prices). While it might seem noble to push up average car quality by forcing lemons off the road, unless you increase the supply of peaches all you are doing is driving price-sensitive consumers (who will be predominantly poor) out of the market entirely. The problem here is a lack of information, and therefore solutions that involve providing information are likely going to be the best solutions.
3. Another thing to bear in mind is that the market will often attempt its own fixes, though many of them rely on some government support. For example, you can take that car to a mechanic, but how do you solve the adverse selection problem in the market for mechanics? Similarly, reputation can give a business a reason to deal honestly with its customers, but that requires trademarks to protect its brand from imitators. Where possible, policy makers should look at what the market is already trying to do, and see if it can support private solutions rather than trying to reinvent the wheel.
4. Finally, there’s the standard advice of not making things worse. Letting people hide information from insurers may lead the insurers to respond in unfortunate ways. Worse still, making a business treat two observably distinct groups as if they were the same can lead that business to make decisions that are utterly perverse from the perspective of your policy goals.

There is a second problem that can result from asymmetric information. That problem goes by the insurance industry’s name for it – moral hazard. Moral hazard is where people behave differently knowing that their actions cannot be easily observed. For example, your insurance company can’t monitor your driving all the time, so your insurance policy might lead you to drive a little more recklessly than you might otherwise (not a lot more recklessly, nobody enjoys crashing their car, but recklessness is a little cheaper with insurance than it would be without). Similarly, if the car market can’t distinguish lemons and peaches you may feel less inclined to maintain your car quite as diligently. The effects are going to be marginal, but those small effects tend to accumulate across a whole economy. A subset of moral hazard is the principal-agent problem, whereby someone hired to perform a task does not act entirely in their client’s best interests.

Unfortunately there is not a lot government can do to avoid moral hazard, most government policies make moral hazards worse, not better. Monitoring people’s behaviour may help, but it won’t be cost-effective much of the time. Moral hazard is often just one of those things you have to live with.

But what if people are making bad decisions, even accounting for their lack of information? That will be the topic of the next part where I will be covering behavioural economics.

1. Assuming buyers are risk neutral.