Adverse selection is when one party to a contract or negotiation have a perfect information relates to the contract or negotiation, while the responding party is lacking the same information. This leads to asymmetric information, which becomes a hassle for responding party to make the best decision, thus suffer adverse effects. The adverse selection is closely related to the insurance industry. This is due to a group of people with private information about their conditions know they will get benefits and be better off under the insurance policy. The insurance company suffers adverse effects when the cost of coverage is higher than the actual risk exposure.
For example, an insurance company provides the individual health insurance which also covered the medical costs related to pregnancy and delivery, this insurance will attract mostly women that planning to have children in the future. The buyers are in advantage as they might have childbearing plans which are one of the examples of unobserved characteristics that have a huge effect on insurance costs.
Another example of adverse selection is the warranties on automobiles. The manufacturers were to issue car warranty for new cars which mean any problems (other than routine maintenance) that happened with the car within a certain time period will be covered by the manufacturer as in the agreement. Some automobile makers sell the extended warranties on automobiles that covering for a longer period. This policy would mostly be purchased by drivers who expect to put their cars to hard use with typically have high probabilities to be damaged. Those drivers who expect to use their cars for basis purposes such as go to work or carry passengers only would be less likely to purchase the extended warranty. As we can see, a driver who buys extended warranty has private information on how the car will be used.
Through the model of adverse selection illustrated in “The Market for Lemons: Quality Uncertainty and the Market Mechanism” written by George Akerlof, he examines the interaction between sellers and buyers in the presence of asymmetric information and in what way this situation affect the market of goods. Akerlof shows that the presence of asymmetric information is able to derive out the high-quality goods and leaving only bad quality goods. By using Akerlof 1970 as an example, a “peach” refers to the high-quality car while a “lemon” refers to the low-quality car. We assume that in the second-hand market, 50% of the cars are for peaches and another 50% are for lemons.
Suppose that buyers have imperfect information on which one is a high-quality car “a peach” and which one is low-quality car “a lemon”. The lack of information will affect the buyers’ willingness to pay a high price. Thus buyers will be only willing to pay the average price for two goods ( ). . The £1800 is the expected value of car those buyers willing to pay. However, in the presence of asymmetric information, the sellers have private information and they know whether the product is a peach or a lemon.
Buyer willingness to pay
Seller willingness to sell
Sells the lemon
Not sells the peach
According to the table above, the sellers know the fixed price that buyers will pay which is £1800. As we can see when buyers willingness to pay (£1800) are higher than the sellers willingness to sell (£1000), sellers are gladly accepting the price and sell the lemon. However, when buyers willingness to pay (£1800) are lower than the sellers willingness to sell (£2000), sellers will disagree to sell the peach. This situation shows that sellers will sell the lemons (since ) and they will leave the market without selling the peaches (since ). When the amount of sellers leaves the market increase, the buyers average willingness to pay will deteriorate (because buyers believe that only low quality of cars exist in the market), thus causing more sellers of high quality cars to leave the market. What left behind is lemons only, so the buyers will not pay £1800 for a lemon when they can get it at price £1200, this leads to the market equilibrium as now the lemon is sell between £1200 and £1000.
Akerlof shows how the price can determine the quality of goods and how the low quality of goods is able to chase away the high quality of goods. In conclusion, the adverse selection can be one of the causes of market collapse.
One of the solutions to overcome the adverse selection in the job market is through signalling which is introduced by Michael Spence in 1973. It is believed that asymmetric information leads to unfairness in the market for the exchange of goods and services. According to Michael Spence, he said that “two parties could get around the problem of asymmetric information by having one party send a signal that would reveal some piece of relevant information to the other party” (Michael Spence, 1973, page 355-374).
In a job market, they employers are willing to pay high wages for workers that have higher productivity. However, only the workers know their own abilities whether high or low. The potential employees should make the first move by giving a signal to the employers through the level of education they achieved. Employers will pay higher wages for educated employees because of they likely to have higher abilities and more productive. To support the signalling model, there are two constraints must be held. The first one is participation constraint, this means that the employees must be better off if they are working compared to their outside option such as being unemployed and receive unemployment benefits. The second one is incentive compatibility constraint, the individuals should be better off after revealing their types. There is no point for someone who has high abilities to pretend to have low abilities and the other way around.
To illustrate the signalling model, Spence applies the simplicity by using one firm and two types of workers; consist of the high type and low type. The low type of workers with a fraction of q and high type of workers with a fraction of 1-q in the population.
Cost of education Wage
According to the diagram 1, the workers with low type L have higher cost of education which is while the workers with high type H have lower education costs which is . The education costs can be both monetary and psychic. The low type has higher costs can be due to the longer period needed to learn and master something new while the high type can do it in short time period. Thus we know that the relationship between signalling cost (cost of education) and productivity is negatively correlated.
In diagram 2, it shows that low type workers’ have the productivity of 1 and high type workers’ have the productivity of 2. The employer believes that the higher productivity (above e*) represents by high type while the lower productivity (below e*) is represented by a low type, thus the higher wage 2 is received by high type and wage 1 is received by low type. However, when all the employees have the same level of education, the employer will not be able to distinguish the type thus pay wage equals to the average of productivity which is 2 – q.
Education = 0
Education = e*
Low type, L
W = 1 , C = 0 , Net = 1
W = 2 , C = e*, Net = 2-e*
High type, H
W = 1 , C = 0 , Net = 1
W = 2 , C = e*/2 , Net = 2- e*/2
The separating equilibrium is illustrated in diagram 2. The first step is by choosing the level of education equals to zero for low type and education at e* for the high type. In mathematical way if L choose e = 0, 1>2-e* thus e*>1. If by having no education make the low type better off this shows that the education cost is relatively cheap and it is possible for a low type to fake as high type thus receive high payment. In other cases, if H choose e = e*, 1<2-e*/2 thus e*<2. If the education cost is too expensive, the high wage will not be able to compensate cost. The separating equilibrium arises at any e* that lies between 1 and 2, this means that lower type will choose no education and get paid for his services at wage 1 while high type chooses the education e equal to e* and get paid at wage 2. The table 2 can also be used to shows how the pooling equilibrium occurs. This situation happens when a firm is unable to differentiate workers types as they have the same education, . The employer will take action to pay the average wage for everyone which at 2 – q. From zero to the expected cost of education (e^), it shows that ) = w = ), this mean that low type workers get low wages if they have no education and get the same wages as high type workers if they have e^ education. In conclusion, when the education give no effect or only small impacts on income, it is better to not have education. In conclusion, there is Pareto efficiency for both types of equilibrium. In separating equilibrium, the most effective is when the low type workers have the net wage of 1 at education equal to zero and the high type workers have net wage 2-e*/2 at education equal to e*. For the pooling equilibrium, the highest efficiency is when all the employees get paid with the same net wage which at 2 – q. The other solution to overcome adverse selection is through the screening. The screening model is introduced by Joseph E.Stigliz in 1975. By using this method, the underinformed party offers a variety of choices that informed parties can self-select. When the firm makes first move, for example by setting up the wage scheme at ( , and ( , then let the workers to proceed with the next step by self-select the effort level. As the conclusion, the adverse selection happens when there is asymmetric information between two parties thus one party have to bear adverse effect. The deviation of information can prevent the uninformed party from relevant decision making. As we can see above, the severe adverse effect can cause the market collapse and unfairness in market transactions. There are a few solutions for adverse selection such as through signalling and screening. A few assumptions have to be stated in order to make the framework of those models works.However, these solutions also have the constraints such as we only measure a certain size of the population and the actual result might differ for the all population.