How Does Insurance Work? Pooling Risk So That Nobody Is Ruined

Nobody knows whose house will burn, but the number that will burn in a population of a million is predictable to within a few percent, because random events average out as the sample grows. That is the law of large numbers, proved by Jacob Bernoulli around 1700, and it is what makes insurance possible. An insurer collecting premiums from a million households can estimate its total claims closely enough to set a price that covers them, adds a margin for expenses and profit, and is still a small fraction of any one house's value. The individual's uncertainty has been converted into the insurer's near-certainty, and the difference is what the insurer sells.

The premium each person pays is, in principle, the expected loss: the probability of the event multiplied by its cost, plus a loading. A house worth 300,000 with a one in five hundred chance of burning has an expected annual fire loss of 600, and a premium near that figure. The work of an insurer is largely the work of estimating those probabilities, which is the profession of the actuary, and of pricing each customer close to their own risk so that the careful are not made to subsidise the careless.

The arithmetic depends on conditions that do not always hold, and each failure has a name:

• •Adverse selection: if customers know their own risk better than the insurer, the high-risk ones buy and the low-risk ones do not, premiums rise to match, more low-risk customers leave, and the pool unravels; this is why insurers ask so many questions and why some cover is compulsory
• •Moral hazard: a person who is insured takes less care, since the loss falls on the pool; excesses, no-claims bonuses and exclusions exist to keep some of the loss with the insured
• •Correlated risk: the law of large numbers needs losses to be independent, and a hurricane, an earthquake or a pandemic hits everyone in the pool at once, which is why such risks are shared with reinsurers and, in the last resort, governments
• •Fraud: a claim for a loss that did not happen, which the industry estimates at several percent of all payouts and prices into everyone's premium

Merchants in Babylon four thousand years ago paid lenders an extra sum on a loan in exchange for cancelling the debt if the cargo was lost at sea, which is insurance under another name, and Roman burial clubs collected dues to pay for members' funerals. The modern industry has two birthplaces in London. Marine insurance grew up in the 1680s in Edward Lloyd's coffee house, where shipowners met the wealthy individuals who would underwrite a voyage by signing their names beneath the risk on a slip of paper, and Lloyd's still works on that principle. Fire insurance followed the Great Fire of 1666, which destroyed 13,000 houses and made the risk impossible to ignore, and life insurance became a science when Edmond Halley, of the comet, published the first proper mortality table in 1693.

Almost any risk can be insured if it is uncertain, measurable and not under the control of the insured. Life insurance pays a sum on death and is priced from mortality tables; health insurance pays for treatment and, where it is private, faces the sharpest adverse selection of all, since the sick most want to buy it; motor insurance is compulsory nearly everywhere because the loss falls on third parties; and property, liability, travel, pet, crop and professional insurance follow the same logic. Governments run the largest pools of all: state pensions, unemployment benefit and public health systems are insurance funded by compulsory contributions, which solves adverse selection by making everyone join. Reinsurance, insurance for insurers, spreads the largest risks across the world, and a hurricane in Florida is paid for partly in Zurich and Munich.

The purpose is not to make money on average, and a policyholder who never claims has not wasted the premium any more than someone who buys a lock and is never burgled. The purpose is to remove the possibility of ruin, and the rule for what to insure follows: cover the losses that would be catastrophic, such as a house, a liability or the loss of an income, and carry the small ones yourself, since the loading on a premium means insurance always costs more than the expected loss. A society in which risk can be pooled is one in which people can take out mortgages, start businesses, ship goods and drive cars, and the quiet, mathematical trade in probabilities is a larger part of why modern economies work than its reputation suggests.

Insurance pools the risks of many people so that the predictable total of their losses, guaranteed by the law of large numbers, can be paid from premiums that are each a small fraction of any one loss. Premiums are priced from the probability and size of the loss; the pool fails through adverse selection, moral hazard, correlated disasters and fraud, which the industry's rules exist to control; and its purpose is not profit for the insured but the removal of ruin.