Why does an insurer not give everyone the same price: what is adverse selection and how does it push the premium up?

Sanobar opa runs the "Kaftar" teahouse. Before the New Year she asked the insurer for the price of a property policy and was upset by the answer: "In our teahouse the electrical wiring was completely replaced at the start of the year, and nobody has the habit of leaving a lamp on. The neighbouring teahouse has 20-year-old wiring, so why do we pay the same?" The question is precise: why does an insurer not set the same price for everyone, and what happens if it does?

In economics the answer is called adverse selection. Below we calculate it step by step on a hypothetical example. The example was composed by the author, and the figures belong neither to any tariff nor to real statistics.

What is adverse selection?

When an insurance contract is made, the two sides do not know the same things. The teahouse owner knows more than the insurer about their wiring, hearth and staff. The insurer cannot see the true risk of every customer who comes in. With information this unequal, the people who need a policy most are exactly those with high risk. A low-risk customer considers the price too high and leaves, and the average risk of those who remain rises.

This term does not mean fraud. Even if a person does not lie, they know their own risk and decide by the price. The idea was analysed in the well-known 1970 paper by the economist George Akerlof; insurance is mentioned in it as one of the examples (see the list of sources).

Example: four groups, one price

Suppose there are 100 teahouses in each of four neighbourhoods, 400 in total. The event (a fire) happens at most once a year, and each time the loss is 80 mln soum. The groups differ in the probability of the event during the year:

Expected loss = probability × 80 mln soum. The willingness-to-pay limit is a hypothetical assumption: each owner agrees to pay up to 1.5 times their expected loss, and if it costs more, they do not take the policy. In reality this limit differs for each person (loan conditions, caution, financial capacity). The figure was chosen for simplicity.

The insurer cannot tell the groups apart and sets the same price for all customers. The price is such that payouts equal losses: the average expected loss. Additional costs and a profit margin are not taken into account here.

How does the chain work?

At each step the insurer uses the average expected loss of the remaining customers as the price, and customers stay or leave according to their limit.

Calculation: at step 1, (1.6 + 3.2 + 4.8 + 8.0) / 4 = 4.40. At step 2, (3.2 + 4.8 + 8.0) / 3 ≈ 5.33. At step 3, (4.8 + 8.0) / 2 = 6.40 mln soum. For the first group the price is 2.75 times more expensive than its own risk, so it left.

Result: the single price eventually rose from 4.40 to 6.40 mln soum, and of 400 policies 200 remained. The best customers left, and those who remained became high-risk. The insurer lost, and so did those who left. The group 1 teahouses, with a probability of only 2%, were left unprotected, and group 2 also left after one step.

Why does "let everyone pay the same" not work, even though it seems fair?

A single price looks simple and fair at first. But 4.40 mln soum in the example is nearly three times too much for group 1, and almost half price for group 4 (4.40 compared with 8.00). Group 4 is exactly the one that takes the cheap policy first: it is advantageous for them. Group 1, seeing the price, gives up the policy or goes uninsured and bears the risk itself. As a result policies pile up where the risk is strong and are absent altogether where the risk is low.

So it is more accurate to regard the price difference across risk groups not as the insurer's whim but as a mechanism that keeps the market alive. If the groups in the example are separated, everyone takes a policy: not 200 instead of 400, but all 400. In return, the high-risk group pays the price of its own risk.

How does the insurer respond?

In most cases the insurer takes one route: it divides the price by risk group. For this the customer is asked questions: age of the building, condition of the wiring, type of hearth and so on. If the groups in the example are identified correctly, each pays its own price: group 1 pays 1.60 mln, group 4 pays 8.00 mln soum (without a margin). Then nobody leaves because of the price: this is a logical assumption, while in reality the insurer adds a margin and never separates the groups perfectly.

Other tools are also aimed at the same problem; they are covered in separate articles: - A deductible (franchise) leaves part of the loss with the customer. - Bonus-malus gives a discount for loss-free years.

These are methods used by the insurer: which insurer applies which one depends on the contract and tariff conditions, so read the policy.

What does this give you?

Two practical conclusions follow.

1. Answering correctly is in your interest. Answer the questionnaire fully and accurately. What is recorded and how incorrect information affects the contract is stated in the policy and the documents applied: it is not confirmed in this article, so read and clarify.
2. Reducing your risk has a price. In the example Sanobar opa replaced the wiring. If the insurer takes such information into account when assigning the group, her price may move from group 2 to group 1: 1.60 mln soum instead of 4.80. This is not a guarantee: you need to ask the insurer.

Practical advice

• Prepare the information before comparing prices. Age of the building, wiring, fire-fighting equipment, inspection documents: if everything is written down, the insurer will put you in the right group.
• Ask "why this price?" Ask precisely which factors raised the price and which might lower it. Get the answer in writing.
• Do not assume the cheapest policy is safe. A low price sometimes means less cover or a large deductible. Use the table in the deductible article for comparison.
• Expense accounting is separate. The accounting of the premium paid for a policy is in the article "How is property insurance recorded in the accounts". This article gives no journal entries, because it is about the logic of price, not accounting.
• The example shows only the logic. The real tariff, probability and margin depend on the insurer's documents, so do not take any figure as a real price.