Thursday, January 06, 2005

Economic freedom more than just adding up 

I'm on my way to the AEA meeting I mentioned below, and am stuck waiting for a plane. Maybe I can get to this post I meant to do last night.

There has been ample discussion of the new Index of Economic Freedom that was written up earlier this week in the Wall Street Journal. (See this discussion at the Fire Ant Gazette, including a correction noted by Brett Schaefer at Heritage in the comments infra. My questions about this index have long dealt with methodology; this post might be dry for a paragraph or two, and if you want to skip the gory details just slide down to the end para.

The methodology this index uses is a methodology that many such studies use. They take a set of factors that they think represent economic freedom. In Heritage's case, that's fifty. They get sorted into ten categories:

The ten factors are scored individually for each country using some of those fifty factors, then the scores in each of the ten are averaged.

One of my problems is that there's no reason for each of those factors to be equally weighted. The authors call this a common-sense approach. But it appears to me that some of these factors are prior to others, and property rights would appear to be prior to all of them. You can have on paper the best regulatory system on the planet, but with insecure property rights what does it matter? Most of the others fall in that category as well. My colleagues and I are working on a paper now which shows that if you take the things that economic freedom is supposed to help -- like per capita GDP -- and check correlations between each individual component, the one that dominates is property rights. It does not appear that most of the others contribute any explanation to why standards of living differ across countries.

The other problem is a problem about how you rank within each category. These are not cardinal measures we are using in each category but ordinal rankings. (In some cases, like inflation or tariff rates, you are using a cardinal number to create an ordinal rank.) The problem there is that you are assuming you've got the ranks right, and that the contribution of each step to economic freedom is equal. That's not necessarily true. In some work I did on central bank independence measures -- done in a similar way to the Heritage measure -- we found that many times the steps were not equidistant.

Why does this matter? In the central bank case, we found that when we had measures that had too much simplification you got the anomalous result that central banks that were not subject to political pressure had no greater effectiveness in reducing inflation that those that were subject to political pressure. That result would contradict a good bit of macroeconomic policy literature. We took our finding to support the traditional view that inflation is often the result of political bias towards paying for government goods with newly printed money. But it now dawns on me that the problem we found is a broader problem for political economy, oversimplification as a result of wanting more quantitative results.

I am not meaning to denigrate the broad work done by Heritage in gathering information on economic freedom. It is excellent data. The problem comes when one tries to reduce the dimensions of economic freedom into a single number that can be put in a table or a map. Not everything adds up the same way, and not every dimension matters in the same amount. We need much more information about these measurements than we have now. If you want to work with that series, map just the property rights sub-index and see what difference it might make compared to the map in the Heritage report.