In a recent exchange on Twitter, I responded to a comment by Harvard researcher, Amitabh Chandra, who advocated for an insurance system that used community rating (a common price for everyone) for disease like cancer that were genetic and risk rating (individuals charged different prices based upon their risk of illness) for diseases like obesity that were related to lifestyle.

In the simplest case, such a system makes sense. An efficient insurance system pools risk for things that are typically beyond individual control and separates, classifies, and charges for risks that individuals can control. All of us take some risk in driving.  Riskier drivers, as demonstrated by their past experience, pay more.

There are at least two possible problems that deserve consideration.  First, both genetic and lifestyle factors may result in prices that are beyond the means of some people. Efficiency isn’t our only societal goal. Do we allow some to go uninsured or be unable to pay for medical care? Most people probably agree that there needs to be some provision, perhaps through income transfers, for this situation.

Of course, if you do that, those transfers result in economic inefficiency, too. Transfers require taxing some economic activity, and that distorts prices–for labor, for property, and so on.  So, we have to consider whether a system that rates risks and has transfers is more inefficient than a system that pools risks and, perhaps, has fewer transfers.

A second issue is the simple division of diseases into those that are genetic and those that are lifestyle.  This ignores the presence of a third cause–environmental factors–that also influence health.  More importantly, it assumes that diseases can be easily classified.  Cancer certainly has genetic causes.  But, almost every type of cancer has important links to behavior and environment. Some cancers, like lung cancer or skin cancer, have significant behavioral links.  Obesity, which Chandra identifies as a lifestyle disease, has significant genetic links.

Trying to identify what risks are genetic and should be pooled and what risks are behavioral and should be rated is not simple; there are few diseases that do not involve genes, behavior and environment.  Appropriately classifying and pricing risks is also an activity with costs and inefficiencies.  We are a long, long way from a detailed apportionment of causes to diseases and diseases to truly needed and appropriate health care spending. Such a system would also require those doing the rating (insurers? government? employers?) to gather, store and use sufficient data on each individual’s or group’s genetics, behavior, and environment. More than a few people are uncomfortable with sharing this information, and the costs of securing it are not small.

In conclusion, these complexities question whether the approach of rating behavioral risks for disease is more efficient than a simpler community rated pool and doing such rating raises important questions beyond simple efficiency.