Should Smokers Pay Higher Health Premiums, Part III

This is a follow up to my post on this topic earlier (and James’ post as well). I mentioned two problems with insurance markets, moral hazard and adverse selection, in the last post. While these are indeed real and serious problems, they are not the only problems. Another problem is nonobservability that is related to adverse selection.

The problem of nonobservability is basically where one party knows some information and the other party cannot observe it (or at least observe it easily). A simple example will help highlight the problem. Suppose a seller knows the quality of his good, but the buyer does not. In some cases (i.e. I am omitting some boring mathematical details here) the buyer will learn nothing about the quality from the price and will have to rely on the “average quality” for all such goods sold. In other cases (more mathemtically borind details omitted), I may learn something about the quality of the good, but still I wont want to buy the good (the average quality is still too low). Thus, the nonobservability can cause the market to break down and not function properly. George Akerlof1 called this problem the lemon problem. In general this problem is not as bad as it looks. For example, if the quality is related to reliability then a warranty can act as a signal for quality. Another possible signal an indivduals investment in education.2 A worker might invest in education (which is costly) to indicate their “type”, where “type” is high or low productivity.

However, signals from smokers are not likely to be forthcoming for the obvious reasons. However, there might be ways for a non-smoker to signal that they are not a smoker and obtain a lower premium. One way to offer such a signal is to submit to a test that detects tobacco (say in the blood or urine). Is this a good thing? The answer here is maybe. Typically in economics the presence of information asymmetries can cause markets to either not form, or work inefficiently (i.e., there are possible trades that would be “socially beneficial” that don’t take place). However, merely removing one of the problems associated with this kind of private information does not mean that things have to “better” (in the sense that some people are made better off and nobody is made worse off). For example, suppose first we have a world where smoking is not observable and there is no way to for non-smokers to signal their non-smoking. In this case, smokers health insurance maybe partially subsidized by non-smokers. If a reliable method of detecting smoking becomes available then the non-smokers can signal their “type”and the non-smokers will see their rates increase.

There is also another issue that has been raised with regards to privacy. I admit I don’t see much of a problem with privacy here. The smoker is voluntarily giving up a portion of his privacy that results in a lower rate. One thing about market transactions is that they are voluntary and those who engage in such transactions expect to benefit from the exchange. The privacy issue strikes me a red herring and might very well be a result of some form of: Health Care is a Right. Smokers could raise the privacy issue as a red herring in an attempt to keep their rates from going up. Those who really do believe that health care is a right could also oppose such things as signalling since signalling could raise premiums for various sub-groups of the population and thus result in more people opting to not have health insurance.

Another issue that some have raised is the fairness. One argument is that is not fair, for example, if people who have a genetic profile indicating a higher likelihood of a certain type of disease and thus have higher insurance permiums or mayby no insurance at all. I’m not sure I see much of a problem here. This strikes me as the same problem as a pre-existing condition. If the probablility of a certain outcome is 1 then there isn’t really any possibility for insurance, only subsidization. For example, if society decides (say via the democratic process) that subsidizing such people is the right thing to do, then some sort of subsidy would be put in place. However, the following points are not clear to me:

  1. That using genetic information in these situations is bad.
  2. That insurance companies/markets are the best mechanism for addressing these concerns.

The first one should be obvious. Suppose that there is a disease and there is also a way to measure how likely a person is to suffer from the disease based on genetic information. Should we remain ignorant of the problem and its potential size or should we seek to use that information? Here is one possibility, suppose there is a disease that is debilitating in the final years of life. If you know you have a higher probability of suffering from that disease you might be able to change your retirement plans to better address such a problem. There might be early treatments that could mitigate future costs or lower the probabilities. By limiting the uses of genetic information simply because insurance companies might want to use it to reduce their exposure to adverse selection could actually have larger problems unrelated to the adverse selection issue. As for the second point above, consider that by forbidding an insurance company from seperating out individuals based on their risks we could inadvertently be promoting risky behavior and also driving up insurance premiums for those least likely to afford such increases. Driving up insurance permiums might very well induce those who are on the margin in terms of being able to afford health insurance to opt for no health insurance at all. It might be better that a tax is used to subsidize those who have “bad” genetic profiles.

The problems with insurance is that it is a pretty complicated issue. First, it involves probability theory and my own experience tells me that most people really don’t understand probability theory (see for example the Monty Hall problem). The other problem is that moral hazard, adverse selection and nonverifiability really make an efficient insurance makret extremely difficult if not impossible. Moreover, with such problems in the insurance markets there are opportunities from some groups to improve their situation via signalling. Some object to signalling in one instance, but not in others which strikes me as inconsistent and strange. Further, the fact that insurance markets are not efficient (i.e. are second best outcomes) almost any change in policy is going to result in winners and losers. The bottomline is that anybody who says there is an easy answer is probably talking out of their posterior.
1Akerlof, G. 1970. The market for “lemons”: Qualitative uncertainty and the market mechanism. Quaterly Journal of Economics 84: 488-500.
2Spence, A Michael, 1973. “Job Market Signaling,” Quarterly Journal of Economics 87, pages 355-74.

FILED UNDER: Economics and Business, Health
Steve Verdon
About Steve Verdon
Steve has a B.A. in Economics from the University of California, Los Angeles and attended graduate school at The George Washington University, leaving school shortly before staring work on his dissertation when his first child was born. He works in the energy industry and prior to that worked at the Bureau of Labor Statistics in the Division of Price Index and Number Research. He joined the staff at OTB in November 2004.


  1. ken says:

    steve, your on the titanic.

    You are wasting your time trying to figure out why and how some people chose one life boat over another and how a life boat captain decides to let them on or not.

    Just like the passengers on the titanic, we as a nation are paying for something that there is just not enough of to go around for all of us. And just like the passengers on the titanic, each of whom needed a lifeboat, each of us is guaranteed to need medical care before the end of our voyage as well.

    So a far more interesting discussion would be how do we gaurantee that each of us gets medical care when that time comes? Any ideas?