gives the responses of unemployment to every shock in the sample at time t:
Rtj = cρje,, j =0,1,2, ...t,
(22)
where Rtj denotes the t period response of unemployment to the jth shock.
unemployment (fɔ
We can thus define the combined response of
at time t as the sum of all the responses in the tth row:
to all actual shocks
t
fl, = ∑ Rtj. (23)
j=1
Essentially, the combined response funcion is obtained by the superimposition of the
impulse response functions (IRF’s) generated by the e’s.
Finally, the sum of all combined responses measures the total effect of the evolution
of the exogenous variable on the unemployment rate:
T Tt
Σ⅛ = ∑ ∑¼ (24)
t=1 t=1 j=1
Observe that the above total effect is just the sum of all the elements in matrix (21).
The total unemployment effect of the evolution of the exogenous variable and the con-
tribution of the exogenous variable to the unemployment rate differ in one main respect.
The former measures the impact of an exogenous variable in the absence of all other
shocks, whereas the latter measures its impact in the presence of all other shocks.
Table 8 below gives the total effect of the evolution of each exogenous variable over
the estimation period (1982-1995), the boom period (1985-1991), and the recession years
(1992-1995). The results confirm our findings in the previous section: the variables whose
evolution is most important for the unemployment rate are oil prices and investment.
Although taxes, benefits, and import prices put upward pressure on the unemployment
rate, their effects are much weaker.
Oil prices had the biggest role in the reduction of unemployment during the booming
years - the decrease in the high unemployment regions was 7 pp, almost two thirds of that
in the low unemployment regions.
During the recession years, the growth rate of capital stock (investment) was the
main factor behind the rise in unemployment - the increase in the high unemployment
regions was more than double of that in the low unemployment regions. This should
be contrasted to the economic upturn of 1985-1991, where the 5 pp decrease in the high
unemployment rate regions was only 70% of the decrease in the low unemployment regions.
Once again, this is what we observed in the previous section: while in good times the
high unemployment group does not benefit as much as the low unemployment group, in
bad times the high unemployment group is hit more severely than the low unemployment
group.
These results are also supported by Bande et al. (2005) who show that in Spain,
24