Gerontocracy in Motion?
25
increasing and later declining individual productivity over the life cycle, as
well as education, family formation, and retirement decisions. Overall, these
variables are supposed to control for individual characteristics impinging
upon individual employment chances in a pragmatic fashion.
The second subset of exploratory factors is measuring demographic change.
This set contains the relative own cohort size and its square, measured as the
share of individuals in each five-year age-brackets (i.e. 15-19,20-24, ..., 60-64)
relative to the total population. These two variables are intended to capture
the independent effect of belonging to a large or small cohort, given each indi-
vidual’s age. If wages do not (fully) respond to changes in cohort size, we will
expect a negative relationship between the individual probability of being em-
ployed and the size of one’s own cohort. If, on the other hand, political econ-
omy effects come into play, these relations might be more intricate.
Furthermore, this set of variables comprises the education structure of each
individual’s own cohort, measured as the share of highly educated individuals
in each cohort. With this variable we intend to capture the effect on individuals
of responses in a generation’s average human capital accumulation to changes
in its relative cohort size. Specifically, we expect human capital accumulation
to counteract generational crowding. If the secular trend towards increasing
education confounds the negative effects of cohort size, in particular for the
baby boomers and the subsequent smaller birth cohorts, the separate inclusion
of the educational stance of each cohort will allow the separate identification
of cohort size effects.
Finally, our explanatory variables comprise a full set of country indicators act-
ing as country fixed effects in our estimations. These country indicators are
supposed to control for unobservable country-specific peculiarities which, on
average, equally impinge upon the employment probability of all individuals
living in this country. Among these country-specific aspects, one might think of
wage bargaining systems or behavioral differences in labor market participa-
tion. Since these variables are merely country-specific constants, their inter-
pretation is difficult and we will abstain from placing any structural interpreta-
tion on their estimates. The only purpose of these variables is to prevent the
estimates for all other variables from being contaminated by such coun-
try-specific idiosyncrasies.
Table 3 reports the results of our Probit model for the full sample of 98,568 in-
dividuals and the results for a restricted sample of 48,172 men. This restricted
analysis explicitly takes the differences in the labor market participation be-
havior between men and women into account. By isolating questions of parti-
cipation from the impact of demographic change on individual employment
probabilities, the restricted analysis portrays the effects of generational
crowding for those workers with low labor supply elasticities.