no employees 1 employee more than 1 employees
Figure 7 Parameter estimates for age by number of employees, according to
model 11
Based on the arguments mentioned above, we decided that model 9 is the optimal
model, which includes a regional variable, a size-variable and economic activity
variable, as well as the three time-dimensions.
Parameter estimates indicate whether mortality for certain characteristics is higher than
average or lower. If a parameter value is higher than zero this means that mortality is
higher, values lower than zero indicate the opposite. The more a value differs from zero,
the stronger the effect is.
According to model 9 the first three explanatory variables behave exactly as
hypothesized. Inside the EMS mortality rates are higher (0.24) than outside the EMS
(0.0). Firms with no employees have the highest mortality rates (0.0), and with more
than one employee the lowest (-0.89), firms with one employee fall in between (-0.66).
For the sector variable the parameters also behave as expected: from the lowest to the
highest mortality rates we find respectively industry (0.0), services (0.06), other (0.19)
and trade (0.47).
In figure 8 the parameters and standard errors for age are plotted. Apart from the first
age group, mortality rates clearly decrease with age. The impact of age on mortality is
the strongest on the highest ages. With an increasing age also the standard errors
increase.
Figure 9 shows parameter estimates and their standard errors for the period dimension.
Indeed there seems to be a relation between the economic business cycle and survival of
firms. Especially in the most recent years mortality chances are considerably higher,
than in the beginning of the period. The period 1989-1993 shows relatively low
mortality chances.
Parameter estimates for the cohort variable have the smallest, though still significant
values (figure 10). The pattern is clear. In terms of mortality chances the periods 1988-
15