Composition, similarity, and the measurement of economic homogeneity

The concept of economic homogeneity in regional science is based on two very different notions: composition and similarity. In the case of the states of the US, measures of homogeneity based on these different notions are negatively correlated with each other, suggesting that the results of studies that use measures of economic homogeneity are likely to be sensitive to the conceptual basis of the specific measure. In order to overcome this problem, a portfolio-theoretic measure of economic homogeneity with some attractive properties is presented. It is naturally decomposable into two components reflecting each of these notions and easy to construct using widely available data. Using this measure, patterns of homogeneity for the states of the US are illustrated and discussed, and hypotheses about the relationship between homogeneity and economic instability are tested.