effect is about two orders of magnitude smaller than the own effect.
Result 1c is our central result about the interaction of labor market institutions. If
labor market institutions in country 1 deteriorate, unemployment in that country goes up,
real wages increase, but income falls (due to lower production). These effects also reduce
firms’ incentives to create jobs in countries 2 and 3. There are three channels through
which a change in b1 has an impact on Φ2 and Φ3 and therefore on unemployment: An
income effect, a competitiveness effect and a selection effect.
The income effect results from the reduced income in country 1. Country 1 spends
part of its income on foreign varieties. Hence, increased unemployment in that coun-
try reduces demand for goods from countries 2 and 3, thereby lowering those countries’
exports and export prices, which tends to increase unemployment. The second effect is
a competitiveness effect. As the increase in the workers’ outside option pushes up the
real wage in country 1, the prices of country 1’s varieties go up relative to varieties from
countries 2 or 3. Moreover, as employment contracts, the number of firms in country 1
and hence the number of varieties produced falls. This endows firms in countries 2 and 3
with a better competitive stance: Residual demand for each firm is higher, which tends to
decrease unemployment in countries 2 and 3. The income and the competitiveness effect
give rise to an additional effect that relies on the selection of firms. The reduced demand
for exports of firms from countries 2 and 3 (the income effect) harms the most productive
firms, as those are the only ones that export. As those firms lose weight, average produc-
tivity goes down and thus Φ2 and Φ3 fall. Similarly, decreased competitiveness of firms in
country 1 implies that domestic firms in countries 2 and 3 earn higher profits. It follows
that the entry-threshold falls: Some firms that were too unproductive to survive before
the reform, can now stay in the market and drive down average productivity. This again
drives down Φ2 and Φ3 and thus lowers incentives to post vacancies.
Overall, we have three channels through which a worsening of labor market institu-
tions in one country affects unemployment in its trading partners. The competitiveness
effect tends to decrease unemployment, while the income and the selection effects tend
to increase it. It turns out that the latter are dominating. This finding is very robust
to alternative calibrations. However, spill-over effects are fairly small quantitatively. The
own effect of inefficient labor market institutions in country 1 is by about two magnitudes
stronger than the effect on the unemployment rates in countries 2 and 3. This finding is
independent from varying other labor market parameters such as ci or mi. We will show
in section 3.6 that the magnitude of spill-overs is much higher when real wages are rigid.
3.3 How geography and country size shape the spill-over
The only channel of transmission of institutional changes in country 1 to labor market
outcomes in countries 2 and 3 is trade in intermediary goods. Since our model implies a
straight-forward gravity-type link between trade costs, country sizes, and bilateral trade
volumes, it is natural to study the implications of these variables on the strength of
institutional spill-overs.
Starting with the role of geography, we change the centrality of the “bad” country,
21