What Is a Model, Anyway?

In yesterday's post, I briefly outlined the events of a small altercation involving NHS-related comments made by Professor Steve Levitt. Professor Noah Smith asked for a model, but was not happy with what Professor Levitt provided (see yesterday's post for links).

But Professor Levitt did provide a model! Granted, he starts out by discussing the policy proposal, so the small model is a bit buried in his blog post, but it is there, and it goes like this:

If it turns out that consumers are sensitive to prices (i.e. that the most basic principle of economics holds, and demand curves slope downwards), total spending on health care will decrease.

Professor Levitt goes on to highlight two important sources of the decrease in expenditure, namely that low-valued services consumed only because they come at a zero price will be cut out by consumers who have to pay a positive price for them, and that competition for consumers who are now price sensitive will lead to increased efficiency.

 

The model is not very fleshed out, but I am not sure it really needs to be (Professor Levitt refers to simulations of the model also, but does not go into details). The essence of many hugely successful models in economics can be described in just a few simple words and that is a most welcome thing.

There are details for which Professor Levitt's model does not account, but maybe there are not too many of them, since Professor Smith embraces the policy proposal which is based upon the model. This strengthens the case for viewing health care as a market fairly much like other markets.

What would a model of Weetabix consumption look like? Well, one should start by positing that marginal willingness to pay is decreasing and that consumption stops once marginal willingness to pay is below the price. Welfare losses can result if prices are nor permitted to equate marginal cost and marginal willingness to pay; services are consumed which cost more to produce than the value consumers place upon them. An analogous market for health care will have similarly important policy implications.

This is not too advanced and will impress few people, but the criticism which Professor Levitt's model has come across, coupled with the widely agreed-upon sense of the resultant policy proposal, is evidence, both that the simple lesson needs to be said, and that remarkably simple models can be remarkably useful.

How Different a Market is Health Care?

Professor Steve Levitt has come under a bit of attack lately because of some comments he made, in part to UK Prime Minister David Cameron, and in full to Cameron's team when the PM had walked out. He asked, essentially, if the PM would be happy with a system under which Britons could come to a car dealership and pick up a car, financed - just like the National Health Service - by the state. Among those upset with these comments is Professor Noah Smith, who has argued (here and here) that health care is a different kind of market than that for cars.

 

I think Professor Levitt is essentially right on this. Now he never actually claimed that there are no differences between health care and other markets, but it is clear from context that he considers health care and other markets to have rather a lot in common. Certainly you would expect huge quantities to be demanded in the market for health care with services available at a zero price, just like you would expect huge quantities to be demanded under similar conditions in basically any other market.

 

So what is different about health care? Among the contenders are adverse selection and moral hazard. Adverse selection means that insurance buyers know more about their health than do sellers, but this is true in very many other markets as well (by the way, the health insurance market is characterized by propitious, not adverse, selection, according to very many studies - just one terrific paper with data on the issue is here). Moral hazard means that insured folks have an incentive not to take care of themselves as well, but this, too, is present in many other markets. (This issue is not primarily about insurance, but it is a common way of financing health expenses and Professor Smith mentioned these specifics in one of the posts linked to above.)

 

A related problem of asymmetric information is that those seeking to purchase health care do not know a lot about the product. But this, again, is true for a wide variety of other markets, such as cars, computers, plumbing services and dentistry. For most food I buy I have no idea what has been added to it. Nor do I care much, because sellers compete and have reputations to maintain.

Another issue pertaining to (lack of) information is how to determine which hospital is best while in acute need of medical attention. Will stroke sufferers "shop around"? Of course they will not, but that holds under any system. Would hospitals charge "exorbitant" fees for those coming in for heart attacks? They clearly have an incentive to do so, but competing hospitals have the same incentive. Distance may give some monopoly power, but not too much because of the presence, a bit farther away, of other hospitals.

One issue on which health care really does differ is that of (partial) unpredictability tied to possibly lethal consequences. I think this may be a reason behind the widespread impression that health care should be the business of the government because it is so "important". Bread and Weetabix are also "important" but they lack the other aspects. I think this also explains the presence of the subsidy in Professor Levitt's policy proposal for the British politicos. Given these widespread beliefs, it would be very hard to argue for no subsidy at all, so Professor Levitt's proposal seems a very good one to me.

Notice, lastly, that the aforementioned difference is not really a difference in the mechanics of the market for health care. Rather, it is simply a difference in widespread perceptions of the services offered in this market. So if free markets tend to work better than regulated ones, it really should show in performance. I am in the deep end of the pool attempting to assess the evidence on this, but I believe it is noteworthy that the Singaporean model - which as far as I can gather comes close to what Professor Levitt tried to propose to Cameron - typically receives a great deal of praise.