Social scientists use both controlled and natural experiments to estimate the impact of various sorts of public policies. One problem with this approach is that it is very difficult to ascertain the long run effects of policy, which might differ dramatically from the short run effects.

Consider how rates of alcoholism vary in different parts of the world:


1. Alcoholism tends to be more of a problem in northern Europe than in the Mediterranean countries of southern Europe.

2. Alcoholism rates are especially high among indigenous peoples in the Americas, Australia and the Pacific islands.

How can we make sense of this pattern?

One possibility is that alcoholism is least severe in areas with the longest history of drinking alcohol.  Wine has been a popular beverage in the Mediterranean area for millennia.  In contrast, indigenous peoples in the New World had little exposure to alcoholic beverages until recently.  

This might seem an odd example to use when thinking about the effects of public policy, as these differences were due to cultural factors, not differing legal systems.  But this example does illustrate that cultures evolve gradually over very long periods of time, in such a way that one group of people may react quite differently to the availability of alcohol than another group, even if the two groups currently face the same legal and economic constraints.  

This fact is important when considering experiments on public policies such as “Universal Basic Income”.  You can certainly imagine an experiment where you give a group of people a lump sum of money.  But what does that tell us about the long run impact of such a program?  Suppose the program continued over many generations, until there was no longer a stigma associated with living your entire life on UBI and putting no effort into work? 

Any time we institute a permanent new public policy such as Social Security or federal deposit insurance, we are running a very long run experiment with our society.  We don’t know how these programs will shape attitudes and behavior over the ultra-long run, as we don’t have the patience to run such experiments before the program is instituted.  

In these cases, I rely more on my intuition about how society is likely to be affected in the very long run, and put little weight on social science experiments (many of which fail to replicate, even in the short run.)