Abstract
We use OLS cross-country regressions and a competitive neural network approach to evaluate the performance of macroeconomic and political-economy data in assessing economic freedom. Our results point to a reasonable performance of the cross-sectional approach in explaining economic freedom scores but to a poor performance of the neural network approach in replicating country rankings provided by the Index of Economic Freedom. Our results thus justify the use of qualitative elements but at the same time call in question the relevance of macroeconomic data in the construction of economic freedom rankings
Notes
1 Doucouliagos (Citation2005) provides a recent review of the issues associated with the construction of the economic freedom index.
2 The results using PPP values for the GDP give similar results.
3 The results for the other specifications are available upon request.
Table 2. Mobility matrix between a neural network four-class grouping and IEF quartiles
4 Chile, Estonia, Lithuania, Bolivia, Guatemala, Jordan, Peru, Sri Lanka.