Imagine you’re in the market for a new car. The used car you purchased last year is still fresh in your mind – it broke down every other day – so you decide to buy new. You sign some papers and drive-off in your new car. An hour later, for whatever reason (maybe you remember you’re allergic to automobiles), you decide to sell your car. To your dismay, you find that you can only get about 75% of what you originally paid for it.
Part of that 25% reduction is simply due to the dealership mark-up. Best case, a dealership would repurchase your car at wholesale. But the dealership may also wonder what happened in that hour which you owned the car. Maybe you discovered something that they don’t know about. Maybe the car is haunted (or more boringly a lemon).
This gap in knowledge – the same gap that stopped you from buying used in the first place – is known as information asymmetry and it’s the subject of today’s post. In our example, information asymmetry caused us to lose a couple thousand dollars. Taken to the extreme, information asymmetry can cause complete market failures.
The Market for Lemons
This may sound obvious so far but before 1970, economists didn’t think very hard about the role information played in markets. Consider the marginal revenue productivity theory of wages, which states that employees are paid a wage equal to the revenue they generate for a firm. But how do firms know the productivity of a potential employee? Economists just assumed perfect information and moved on.
In the late 1960’s, George Akerlof wrote “The Market for Lemons”, which upended this notion. His idea was both simple and groundbreaking. Consider the market for used cars. If perfect information is available, buyers are willing to pay $2,000 for a well-functioning used car (peach) and $500 for a poorly functioning used car (lemon).
However, buyers don’t have perfect information. They don’t know if the car will be a peach or a lemon. Maybe they split the difference and offer $1,250 to cover the risk. Now, consider the seller who knows for sure whether the car is a peach or a lemon. They will never give away their peach for less than $2,000, but they will gladly give away their lemon for $1,250.
So only lemons are offered at $1,250. Buyers realize this and reduce their bids to $500. The market for a well-functioning used car completely disappears. Through this example, Akerlof’s paper showed that asymmetric information can lead to a complete market failure.
Recall that idea of paying an employee the value of their generated revenue. In 1973, Michael Spence wrote “Job Market Signalling“, which updated the idea to include the asymmetry problem. Spence argued that since employers can’t tell which employees will be the most productive, they rely on signals (college degree, credit score, experience, etc) to filter out applicants. The harder it is for low-productive employees to obtain the signal, the better it works.
Signals aren’t perfect though. The best candidate might still be overlooked or underpaid. A bad candidate might be hired and overpaid. People might be forced to pay for a college education they don’t need just for a seat at the table. But when everyone understands the problem, they can work towards a better solution.
In the used car example, the salesman might have been able to signal a car’s quality through testimonials, online reviews, or an online car history.
Adverse selection occurs when either a buyer or a seller has information about a product that the other does not, and that information incentivizes them to act in a damaging way. We touched on this idea with the used car seller who was only willing to sell lemons at $1,250. In that example, lemons were adversely selected.
In the insurance industry, people who are most at risk are the ones that get insured. A healthy teenager might avoid health insurance all together, but someone with a history of medical problems is going to make sure they are covered. In the same way, a risky driver will seek out car insurance more frequently than a safe driver (ignoring mandatory laws in both cases). In these examples, risk prone customers are adversely selected.
To get around this, insurance companies offer different packages. They assume that only risky customers will pay up for a full coverage and low deductible package. Safe drivers will be fine with the high deductible. Once again, the solution isn’t perfect but understanding the problem helps us improve it.
Similarly to adverse selection, moral hazard occurs when protection makes people more likely to take risks. If we know we are protected from the consequences of a risky behavior, we tend to do more of it. For example, we drive rougher when we pay for rental insurance than we would without.
Consider the financial crisis of 2008. The government would have prefered if banks did not take on huge risky bets, but they were still committed to bailing them out when things went bad. The banks, knowing that they will be bailed out, can just take on more risky bets.
Consider the agent-principle problem as it relates to investors and management. Management may be incentivized to make large bets in the hope of hitting their bonus targets, at the cost of long term viability. If the bets go bad, management can just take their paychecks and move on. Investors are left holding the bag.
Signalling, adverse selection, and moral hazard are all byproducts of asymmetric information. They lead to sub-optimal outcomes and market failures. The solution in every case is to either increase the amount of information symmetry or to get the incentives right. We are less likely to lie when our reputations are on the line. We are less likely to take unneeded risk when the price of said risk is incorporated. And we are more likely to buy a used car when we are given detailed information about its past.