Book review: Loss Coverage by Guy Thomas

Author: Guy Thomas
Publisher: Cambridge University Press
Review: Mark Farrell FIA

Actuaries are generally known for their rational dispositions. Yet they are, on occasion, still prone to the human flaws of bias, making faulty assumptions and following prior knowledge without fully questioning the underlying logic.

Loss Coverage attempts to dispel one such misconception regarding the value of adverse selection in insurance pricing markets.

It is defined as a process whereby low-risk individuals drop out of the insurance pool, leaving a higher proportion of high-risk individuals. This happens because individuals know more than insurers about their own risks.

The book disputes the orthodox view that adverse selection is something to be avoided at all costs. Conventional wisdom tells us that restrictions on insurance risk classifications will increase adverse selection, causing a drop in demand (from low-risk individuals) and hence an increase in prices and ultimately a decrease in the number of insured individuals. This is known as the adverse selection spiral (death spiral) which is essentially considered to be an undesirable outcome.

Thomas uses succinct prose and clear, logical arguments (with helpful figures) to argue that actuaries are overlooking an important point. The impact of adverse selection, he contends, should be measured, not in terms of the number of insured, but rather in terms of loss coverage. He defines this as the expected losses compensated (not numbers insured) under any particular risk classification scheme. As loss coverage encompasses the individual risk probabilities, the public policy implications the book focuses on begin to look very different. While actuaries may naturally tend to think of insurance premiums from an 'actuarialy fair' perspective there are, of course, wider considerations which policymakers should take into account.

Losing low-risk individuals to the pool can be more than compensated for via an increase in high-risk individuals. Some adverse selection may lead to a beneficial aggregate position for society in terms of increasing the overall level of loss coverage. Thomas builds the argument with a rigorous mathematical foundation in three chapters that examine the demand elasticities for both higher and lower-risk groups in different markets.

One assumption made throughout the book is that it's okay and even desirable for the low risks that remain in the insured pool, to subsidize the higher risks for the greater good of society. It assumes the societal benefit of insuring low versus high risks is a function solely of their respective probabilities of loss. To use the book's own terms, insurance is viewed as a 'probabilistic' rather than 'reassurance' good. It would have been interesting, and more balanced, to have seen more focus placed on discussion of this aspect.

The book finishes with a chapter on big data and the potential perils of using these unstructured large datasets for risk classification. It would have been fascinating to have examined counter arguments on this particular chapter. For example, big data
(wearable technology) has the potential to also help make previously uninsurable individuals more insurable (diabetics).

Overall, this is a very well written, researched and thought-provoking book that intelligently proposes that some limits on some risk classifications can have significant merit, especially from a policy perspective.