The case for geographically concentrated vaccine doses

Here goes:

A central yet neglected point is that vaccines should not be sent to each and every part of the U.S. Instead, it would be better to concentrate distribution in a small number of places where the vaccines can have a greater impact.

Say, for the purposes of argument, that you had 20,000 vaccine doses to distribute. There are about 20,000 cities and towns in America. Would you send one dose to each location? That might sound fair, but such a distribution would limit the overall effect. Many of those 20,000 recipients would be safer, but your plan would not meaningfully reduce community transmission in any of those places, nor would it allow any public events to restart or schools to reopen.

Alternatively, say you chose one town or well-defined area and distributed all 20,000 doses there. Not only would you protect 20,000 people with the vaccine, but the surrounding area would be much safer, too. Children could go to school, for instance, knowing that most of the other people in the building had been vaccinated. Shopping and dining would boom as well.

Here is one qualifier, but in fact it pushes one further along the road to geographic concentration:

Over time, mobility, migration and mixing would undo some of the initial benefits of the geographically concentrated dose of vaccines. That’s why the second round of vaccine distribution should go exactly to those people who are most likely to mix with the first targeted area. This plan reaps two benefits: protecting the people in the newly chosen second area, and limiting the ability of those people to disrupt the benefits already gained in the first area.

In other words, if the first doses went (to choose a random example) to Wilmington, Delaware, the next batch of doses should go to the suburbs of Wilmington. In economics language [behind this link is a highly useful Michael Kremer paper], one can say that Covid-19 infections (and protections) have externalities, and there are increasing returns to those externalities. That implies a geographically concentrated approach to vaccine distribution, whether at the federal or state level.

Here is another qualifier:

…there will be practical limits on a fully concentrated geographic distribution of vaccines. Too many vaccines sent to too few places will result in long waits and trouble with storage. Nonetheless, at the margin the U.S. should still consider a more geographically concentrated distribution than what it is likely to do.

Do you think that travel restrictions have stopped the spread of the coronavirus? (Doesn’t mean you have to favor them, all things considered.)  Probably yes.  If so, you probably ought to favor a geographically concentrated initial distribution of the vaccine as well — can you see why it is the same logic?  Just imagine it spreading out like stones on a Go board.

Of course we are not likely to do any of this.  Here is my full Bloomberg column.