opportunities for private investment, nominal wage data are of little use. Retail sales, RS, per
worker is therefore a measure of consumption expenditure per wage earner. RS should have a
negative effect on the number of activists. Three indicators of public consumption -
enrollment in higher education institutions, ST, new public housing construction in square
meters, NH, and physicians per capita, PH - should also have a non-positive effect (negative
if provision of public goods is correlated with wages and zero otherwise). In addition,
enrollment in the institutions of higher education serves as a proxy for the availability of the
“outside option” for potential activists: non-party professional careers.30 Therefore, enrollment
is likely to have a strictly negative effect on the number of activists. Finally, investment
measures, investment per worker and the change in investment rate, used here as correlates of
the changes in the capital-labor ratio should have positive effect of less-than-unitary
magnitude.
The model is estimated using first-order log differences in order to exclude the time
trend and fixed effects. The complete empirical specification is given by:
(32) ∆ln(Na it) = β0 + β1 ∆ ln(Nb it) + β2 ∆ ln(R it) + β3 ∆ ln(RS it) + β4 ∆ ln(L it)
+ β5 I it + β6 ∆ ln(ST it) + β7 ∆ ln(PH it) + β8 ∆ ln(NHit) + εit ,
where I it replaces K/L in the prototype specification (31) and is alternatively log of
investment per worker, ln(I/L ), or the change in investment rate, ∆ ln(I/ Y).
30 Because all education was free of charge, demand for college education (number of applicants)
typically exceeded admission. Therefore, total enrollment is equivalent to the number of “promotion
tickets” available to the population outside of the party promotion machine. Although party
membership was a plus for an applicant, it was not a prerequisite for admission and higher education is
therefore a distinct alternative to activist service.
34