Country 1, ∆ Unemployment
Country 2, ∆ Unemployment
1
0.8
0.6
0.4
0.2
0
90
2.2
2
1.8
1.6
1.4
1.2
1
0.8
90
60
0
0
τ1 (%)
τ1 (%)
S1=L1/Lw
S1=L1/Lw
Figure 9: Change in unemployment [on the vertical axis] as a function of centrality and
size of the “bad” country 1 for a given increase of country 1’s search costs from 1 to 1.3
under perfect real wage rigidity.
where wages are bargained individually, with the opposite extreme assumption of perfect
real wages rigidity. The lack of adjustment in prices naturally increases the scope of
adjustment in quantities. Hence, real wage rigidity should increase spill-overs in terms
of unemployment rates. Comparing the cases of individual bargaining with rigid wages
spans the interval in which the ‘true’ size of spill-overs lies.
When real wages are rigid, the wage curve (18) is replaced by the requirement that
wi/Pi = ωi. We recalibrate the model such that our choice ωi reproduces the unemploy-
ment rates, firm and job turnover rates, export penetration rates, and the average firms
sizes as shown in Table A1. All external parameters are the same as in the earlier cali-
bration.29 Since the replacement rate bi appears only in the now redundant wage curve,
we vary the cost of vacancy creation ci over the interval [1, 1.3].
Figure 9 reproduces Figure 3 for the new scenario (c1 rather than b1 is changed) and
under the assumption of rigid real wages. When c1 grows from 1 to 1.3, the unemployment
rate in country 1 moves up by about 0.8 to 2.2 percentage points, depending, as before,
on the relative size of country 1 and on its geographical location relative to its trading
partners. The spill-overs to countries 2 and 3 (again treated symmetric) are now much
more sizeable than before and vary between 0 and 1 percentage points. The model predicts
that the strength of spill-overs is up to 45% of the effect in the reforming country. This is
in strong contrast to our earlier results for flexible wages (Figure 3). To make sure that
this difference does not come from our change in the experiment (changing c1 instead of
b1), we repeat the same exercise with flexible wages. Figure 10 confirms that the change
29The structure of the model implies that, in the baseline equilibrium, the value of ωi will be identical
to the real wage that results under individual bargaining.
30