Measuring state capacity

This past week I finished up some revisions to a paper that compares 15 different ways that the concept of state capacity has been operationalized in studies of civil conflict. These 15 measures fit into three broad theoretical categories: military capacity, bureaucratic/administrative capacity, and finally political insitutional coherence and quality. Basically, is state capacity the ability to put boots into the field? The ability to keep records and thus keep track of your citizens? Or is it the degree to which your political institutions reinforce one another (i.e., you’re either a consolidated, participatory democracy or a draconian, North Korea-style police state–but not in between)?

The paper will be posted soon, but here are the quick hits–some of which won’t be surprising, some of which might be:

1. Principal factor analysis (and oldie but a goodie) demonstrates the underlying dimensionality of state capacity is low, with three latent factors explaining over 90 percent of the variance in the 15 measures.

2. The dimensions do not map neatly on to the theoretical groupings: while the first factor, rational legality, captures bureaucratic and administrative capacity, the second, rentier-autocraticness, and third, neopatrimoniality, capture aspects of state capacity that cut across the theoretical categories. Rentier-autocraticness captures reliance on primary commodity exports, high state capture of economic resources (as proxied by taxation and total revenue), and low levels of democracy: this is your classic oil-rich, authoritarian state. Neopatrimoniality combines low extractive capacity with reliance on primary commodities and higher military expenditures–oil rich principalities and resource-rich tinpot dictators.

3. The first two dimensions do a better job of “sorting” the world into groups of states that have experienced civil conflict during the period of study (1984-1999) and those that haven’t. Here is a table from the paper in which I list the top five and bottom five countries, using their average scores for the panel. The countries in italics experienced a civil war/conflict in those years, the others did not:


4. The first dimension gives us a grouping we are pretty familiar with: the typical low-development, low-bureaucratic quality countries we all recognize as weak states. The second dimension gives us a slightly less intuitive (though consistent with Paul Collier’s recent book), nevertheless important grouping: low-revenue democracies.