Adverse selection

Adverse selection was first described for life insurance. It creates a demand for insurance which is positively correlated with the insured's risk of loss.[1]

For example, non-smokers typically live longer than smokers. If the price of insurance does not vary according to smoking status, then it will be more valuable for smokers than for non-smokers. Thus smokers will have a greater incentive to buy insurance and will purchase more insurance than non-smokers. This increases the average mortality rate of the insured pool, causing the insurer to pay more claims.

In response, the company may increase premiums to correspond to the higher average risk. However, higher prices cause rational non-smokers to cancel their insurance as insurance becomes uneconomic for them, exacerbating the adverse selection problem. Eventually, higher prices will push out all non-smokers in search of better options, and the only people left who will be willing to purchase insurance are smokers.

To counter the effects of adverse selection, insurers may offer premiums that are proportional to a customer's risk by distinguishing high-risk individuals from low-risk individuals. For instance, medical insurance companies ask a range of questions and may request medical or other reports on individuals who apply to buy insurance. The premium can be varied accordingly and any unreasonably high-risk individuals are rejected (cf. pre-existing condition). This risk selection process is part of underwriting. In many countries, insurance law incorporates an "utmost good faith" or uberrima fides doctrine, which requires potential customers to answer any questions asked by the insurer fully and honestly. Dishonesty may be met with refusals to pay claims.

Empirical evidence of adverse selection is mixed. Several studies investigating correlations between risk and insurance purchase have failed to show the predicted positive correlation for life insurance,[5] auto insurance,[6][7] and health insurance.[8] On the other hand, "positive" test results for adverse selection have been reported in health insurance,[9] long-term care insurance,[10] and annuity markets.[11]

Weak evidence of adverse selection in certain markets suggests that the underwriting process is effective at screening high-risk individuals. Another possible reason is the negative correlation between risk aversion (such as the willingness to purchase insurance) and risk level (estimated beforehand based on hindsight observation of the occurrence rate for other observed claims) in the population. If risk aversion is higher among lower-risk customers, adverse selection can be reduced or even reversed, leading to "advantageous" selection.[12][13] This occurs when a person is both less likely to engage in risk-increasing behavior are more likely to engage in risk-decreasing behavior, such as taking affirmative steps to reduce risk.

For example, there is evidence that smokers are more willing to do risky jobs than non-smokers.[14] This greater willingness to accept risk may reduce insurance policy purchases by smokers.

From a public policy viewpoint, some adverse selection can also be advantageous. Adverse selection may lead to a higher fraction of total losses for the whole population being covered by insurance than if there were no adverse selection.[15]

When raising capital, some types of securities are more prone to adverse selection than others. An equity offering for a company that reliably generates earnings at a good price will be bought up before an unknown company's offering, leaving the market filled with less desirable offerings that were unwanted by other investors. Assuming that managers have inside information about the firm, outsiders are most prone to adverse selection in equity offers. This is because managers may offer stock when they know the offer price exceeds their private assessments of the company's value. Outside investors, therefore, require a high rate of return on equity to compensate them for the risk of buying a "lemon".

Adverse selection costs are lower for debt offerings. When debt is offered, this acts as a signal to outside investors that the firm's management believes the current stock price is undervalued, as the firm would otherwise be keen on offering equity.

Thus the required returns on debt and equity are related to perceived adverse selection costs, implying that debt should be cheaper than equity as a source of external capital, forming a "pecking order".[16]

The example described assumes that the market does not know managers are selling stock. The market could gain access to this information, perhaps by finding it in company reports. In this case, the market will capitalize on the information found. If the market has access to the company's information, the presence of information asymmetry is removed, and as such there is no longer a state of adverse selection.

The presence of adverse selection in capital markets results in excessive private investment. Projects that otherwise would not have received investments due to having a lower expected return than the opportunity cost of capital, received funding as a result of information asymmetry in the market. As such, governments must account for the presence of adverse selection in the implementation of public policies.[17]

In modern contract theory, "adverse selection" characterizes principal-agent models in which an agent has private information before a contract is written.[18][19] For example, a worker may know his effort costs (or a buyer may know his willingness-to-pay) before an employer (or a seller) makes a contract offer. In contrast, "moral hazard" characterizes principal-agent models where there is symmetric information at the time of contracting. The agent may become privately informed after the contract is written. According to Hart and Holmström (1987), moral hazard models are further subdivided into hidden action and hidden information models, depending on whether the agent becomes privately informed due to an unobservable action that he himself chooses or due to a random move by nature.[20] Hence, the difference between an adverse selection model and a hidden information (sometimes called hidden knowledge) model is simply the timing. In the former case, the agent is informed at the outset. In the latter case, he becomes privately informed after the contract has been signed.

Adverse selection models can be further categorized into models with private values and models with interdependent or common values. In models with private values, the agent's type has a direct influence on his own preferences. For example, he has knowledge over his effort costs or his willingness-to-pay. Alternatively, models with interdependent or common values occur when the agent's type has a direct influence on the principal's preferences. For instance, the agent may be a seller who privately knows the quality of a car.

Seminal contributions to private value models have been made by Roger Myerson and Eric Maskin, while interdependent or common value models have first been studied by George Akerlof. Adverse selection models with private values can also be further categorized by distinguishing between models with one-sided private information and two-sided private information. The most prominent result in the latter case is the Myerson-Satterthwaite theorem. More recently, contract-theoretic adverse selection models have been tested both in laboratory experiments and in the field.[21][22]

In markets where the seller has private information about the product they wish to sell, reputation mechanisms help to reduce adverse selection by acting as a signal of quality.[23] An example would be the online marketplace, Ebay. A seller known for selling high-quality goods can further enhance its reputation by utilising Ebay's reputation system. There is an incentive for the seller to do so, as buyers who derive utility from purchasing the product are naturally inclined to source their purchase from high-quality sellers. As such, buyers are able to rely on the reputation system as a signal to filter high-quality sellers from low-quality sellers.[24]

Lemon laws act as a form of consumer protection in the event the buyer purchase a defective product. While usually applied to automobiles, lemon laws are also used for most consumer goods. Such regulations were enacted to reduce cases where manufacturers knowingly sold defective products. Lemon laws vary by countries, but generally require the seller to repurchase the product or replace it. For example, the Texas Deceptive Trade Practices allows for consumers to sue for triple damages in the event of sustaining harm as a result of purchasing a defective product as a result of the seller withholding information at the time of the transaction. As such, such government regulations act as a deterrent against sellers exploiting the asymmetric information between the parties involved. This, in turn, reduces the problem of adverse selection, as buyers who are knowingly protected by lemon laws are more inclined to engage in transactions they previously would not have done so due to the lack of viable information available to them.

By offering a warranty for the product the seller intends to sell, they are able to indirectly communicate private information about the product to the buyer. Warranties assist in conveying information about the seller's confidence in the product for its quality, by acting as a guarantee on the product.[25] A common example is in the used car market, where apart from warranties offered by the seller itself, the buyer may purchase additional warranties in the form of insurance from third-party companies.