Who should get swine flu vaccines?

David Karp comments on the Hamilton city government's policy for swine flu vaccines (also here):

I think part of the problem trying to figure out how to allocate vaccines is figuring out what our policy goal is. Is it to cut down on externalities? Giving preference for emergency workers and child care workers makes sense in that regard, because most people in society benefit from these people being healthy enough to work during a pandemic. Is it about fairness? If so, is it really fair that some people are considered more worthy of a vaccine than other people simply because of their age, how they caught the flu, or the type of work they do? Is it about using the vaccine available to cure as many people as possible? In that case, the fifth criteria -- giving the vaccine to those most likely to survive a particular strain -- seems like the best to use. Or should the criteria be economic efficiency: those who have the highest maximum willingness to pay for the vaccine get it?

Actually I suspect the health authority has another criterion in mind: to vaccinate people who are likely to be highly-connected nodes in the swine flu network, i.e. those who may pass it to others. The economic interpretation is to consider the vaccine in this case not as a consumption but an investment good, and direct it to where it makes the highest return.

My guess is that the "textbook" economic answer to this problem would be to hold an auction for the vaccine, which would mean people who want the vaccine most (at least in terms of their willingness to pay) would get it. But it's an unappealing option because it means that rich people have an upper hand over poor people...

He suggests a lottery instead, then allowing recipients to sell the vaccine if they way. But in classical economic terms, that has all the same disadvantages as an auction - and, instead of the money going to society as a whole, it goes privately to a small number of people randomly selected - many of whom will be quite well-off.

But - though I'm not sure if David was thinking this way - the psychology of the auction approach is probably weaker, and the lottery may be more appealing to the non-economist public.

First, the auction more obviously reduces the transaction to a financial one; because the sale is conducted in public instead of as a series of private transactions, it makes it obvious that it is going with the money, and this weakens legitimacy among the public.

Second, the money in an auction is perceived as going to "the government" instead of to real people. Presumably the government will (to a first approximation) use the money for the benefit of citizens, but the impersonality of the state weakens people's belief in that outcome.

Third, because it will actually be economically efficient. While this seems like it would be a good thing, many people will feel that efficiency makes the process somehow soulless and cynical. The fact that, under a lottery, some middle-class people would end up holding onto their vaccine instead of selling it (endowment effect, anyone?) and thus some lucky person would be saved when they aren't willing to pay for it, may make the required compromises more acceptable in the public's eyes. Often, a bit less rationality is what people relate to.

Some would say these considerations are not the job of an economist. But it simply depends on where we draw the boundaries around our models. If our goal is to achieve maximum public welfare, then understanding psychology and how it affects the public's utility is a requirement. If our goal is to create stable institutions which result in a successful society (economically and otherwise) then legitimacy is a valid attribute to explore.

Economists have done a reasonable job in recent decades of convincing some parts of the public that considerations of economic efficiency are good criteria on which to set policy. But in certain fields - public health being one - sentiment has a big influence. And instead of pretending to live in a rational world, we are better off accepting that factor and modelling it.