The sample: For most equations, the number of observations = 6 years times 21 countries = 126. The benchmark years covered are 1880, 1890, 1900, 1910, 1920, and 1930. In some cases data from adjacent years had to be substituted. The widest departures from benchmark dates are these: The 1920 dependent variables for Sweden are actually spending ratios from 1917, while those for Austria and Belgium are ratios from 1922.
The twenty-one countries of the main 1880–1930 panel are Argentina, Australia, Austria (without the rest of the Austro-Hungarian Empire), Belgium, Brazil, Canada, Denmark, Finland, France, Greece, Italy, Japan, Mexico, Netherlands, New Zealand, Norway, Portugal, Spain, Sweden, United Kingdom, and United States. All are viewed as sovereign nations, despite limitations on the sovereignty of Australia, Finland, and New Zealand before the turn of the century.
For the equation explaining the number of public primary-school teachers per 1000 children of ages five to fourteen, only eleven of these countries supplied usable teacher counts on all six dates: Austria (without the rest of the Austro-Hungarian Empire), Belgium, Canada, Finland, France, Italy, Netherlands, Norway, Sweden, United Kingdom, and United States.
For the equations explaining the revenues from income tax and inheritance tax as percentages of GNP, the nineteen-country sample consists of the twenty-one-country sample minus Spain and Argentina.
Regression techniques: The regressions were run on SHAZAM for the Macintosh. The regressions used the TOBIT and the generalized-least-squares POOL commands without special restrictions.