24
Michael Fertig and Christoph M. Schmidt
of ageing preferences for specific goods and services change, it is likely that the
structure of employment across different sectors of the economy is affected by
this phenomenon as well. This is certainly an even more predominant factor in
rather closed economies. However, due to the similarities in the demographic
change in almost all industrialized economies one would expect that open
economies will be affected by changes in product demand as well.
From the above discussion it should have become transparent that the conse-
quences of population ageing for European labor markets are not fully under-
stood. Theoretical models deliver conflicting hypotheses on this phenomenon,
especially regarding the impact ofan ageing population on (un-) employment.
The next section, therefore, presents some empirical evidence on the relation-
ship between population ageing and the probability of being employed - i.e.
for employment rates - for individuals living in EU-countries.
4. Cohort Size and Unemployment - European Cross-Country Evidence
In our empirical application, we utilize the 1999 wave of the European Com-
munity Household Panel (ECHP) for all EU-15 countries except Luxem-
bourg3. We restrict the analysis to the economically active population, i.e. all
15-64 old individuals not being in school any longer. The dependent variable
in our analysis is the individual employment status, taking the value of 1 if the
individual is regularly employed (works 15 hours or more per week) and zero
otherwise. We analyze this variable in a discrete choice framework (Probit
model). In this endeavor, we jointly employ three different sets of explanatory
variables: (i) individual characteristics, (ii) variables measuring the demo-
graphic change, and (iii) a full set of country indicators. Table 4 in the Appen-
dix provides a brief description of all variables and Table 5 reports some de-
scriptive statistics.
The first set of explanatory variables, i.e. individual characteristics, comprises
the individual’s gender, his/her level of educational attainment, and a variable
indicating whether or not an individual has any chronic physical or mental
health problem, illness or disability. This first subset of variables also includes
the individual’s age (in years). In our Probit regressions, we allow age and age
squared to exert different effects for three different age groups. The first age
group comprises young adults aged 15-29 years, the second group comprises
the “prime earning years” 30-55, and the third age group contains all individu-
als older than 55 (i.e. 55-64). For each of these employment-age profiles we
would expect to find a concave, i.e. inversely u-shaped, relationship. This pat-
tern is bound to reflect a large number of influences, among them the initially
3
Luxembourg had to be dropped due to missing observations for one of our explanatory varia-
bles.