Education is not positively and significantly related to economic growth for most samples

One of the challenges of a complex process is that many times you have to operate with assumptions in which you have great confidence but for which you have little evidence. Robin Grier raises this in A New Database on Educational Quality by Robin Grier.

I frequently have my undergrad and grad students read Bill Easterly’s excellent book, The Elusive Quest for Growth: Economists’ Adventures and Misadventures in the Tropics. They are often a bit depressed after reading the chapters on growth, noting how little we seem to know about growth in the real world. It is the education chapter, however, which really gets to them. In it, Easterly points out what has been known for ages but is rarely mentioned in print (see Lant Pritchett’s “Where has all the education gone?” for a refreshing exception), namely, that education is not positively and significantly related to economic growth for most samples. Sometimes you can find a weakly positive t-stat in a sample of OECD countries, but you are lucky if you merely find zero correlation (instead of a negative and significant) in the developing world.

We see that economic growth and increased education are weakly correlated but it is very difficult to determine the flow of causality much less to establish the mechanisms by which education might cause economic growth. There is lots of useful speculation as to how increased education could cause economic growth but very, very little evidence that it does cause growth.

There is a similar issue with reading. Book creation and book reading is positively correlated with economic growth and personal success. Just as with macro-economics, it is easy to speculate how enthusiastic reading could be causative of life success but very, very difficult to demonstrate that it does cause life success.

It is worth remembering as we become passionate about the details of how encourage reading, or how to enable greater education, just how weak the foundation is. Not to say that that the unsupported assumptions are wrong, simply that they are unsupported regardless how much we believe them to be true.