Wednesday, March 18, 2009

Attribution Error, Elections and Macroeconomics

Attribution error is the name given to psychologists that we tend to attribute blame (or thanks) to people or things that are close by: it’s very similar to guilt by association.

In macroeconomics, this comes up with business cycles and elections. Reagan, Clinton and Bush II get reelected because the economy is doing OK, while Bush I doesn’t because the economy is doing badly. This is in spite of the weak connection – if there is one at all – between presidents and economics performance.

An excellent example of how strong this attribution error can be is contained in this post from Core Economics (the most important Australian economics blog) where they show that state elections in Australia can be predicted by the U.S. unemployment rate.

This doesn’t mean that unemployment in the U.S. causes Australian elections. That’s silly.

But, there is a strong correlation.

What’s going on here is that U.S. unemployment rates are strongly positively correlated with the U.S. economy, which is weakly positively correlated with the Australian economy, which is strongly positively correlated with Australian state economies, whose performance voters attribute to the prominent nearby targets: local politicians.

This is a big problem for the U.S. in the future. The presence of attribution errors suggests that Bush will always be blamed for the recession that started on his watch. Obama will get the credit for ending the recession, since they all end relatively quickly (on the scale of elections).

This will extend to the stimulus package as well. It is very likely to be declared a success if the economy recovers. However, if the economy starts to recover before very late in 2009, it’s very likely that it is the Bush stimulus of 2008 that made the difference, not the Obama stimulus of 2009.

No comments:

Post a Comment