Columns (8) to (11) include the ratio of capital intensities (K*/L*) / (K/L) = k* /k
and the interaction term u* × k*/k.39 This is motivated by the prediction of the Heckscher-
Ohlin-framework, where the spill-over depends on the relative capital-to-labor ratios form
the home country k and the foreign country k* . More precisely, suppose that a bad shock
on foreign institutions drives up foreign unemployment. If the foreign economy is relatively
capital-abundant relative to the domestic economy, after the shock its effective capital to
labor ratio is even higher than before and it produces more of the capital-intensive good.
The domestic economy, in turn, produces more of the labor-intensive good so that labor
demand goes up, and ultimately unemployment falls. This is, however, not what we
observe in the data, where an increase in u* drives up domestic unemployment by more
when the domestic economy appears relatively capital poor. However, the effect is weak,
and inference is possibly problematic since k and hence k*/k may be endogenous. Also,
results rely on a fairly small country sample. While not offering a conclusive test, our
results at least suggest that empirical support for the comparative advantage view is
probably weak. Given the well-known poor empirical performance of the Heckscher-Ohlin
model, this is not a surprise. Also note that Dutt, Mitra and Ranjan (2009) have not
found any effect of comparative advantage motives in the determination of unemployment
rates in a large cross-section of countries. The standard deviation of k*/k relative to its
mean (the coefficient of variation) is 1.91, while that of u is 0.54. Hence, our results do
not hinge on the absence of variance in k*/k in our sample.
Interestingly, the direct effect of k*/k on the equilibrium unemployment rate is negative
and statistically significant. However, that effect is hard to interpret. The overall extent
of comparative advantage of the domestic economy could be measured by |1 - k*/k| or
(1 - k* /k)2 . Both variables have no clear empirical effect on the unemployment rate.
4.4 Domestic unemployment and foreign institutions
In the next step, we analyze the direct effect of foreign labor market institutions on the
domestic rate of unemployment. We estimate an equation of the form:
Uit = λ ∙ LMRit + λ*∙LMR*,t + π ∙pmrit + γɪ ∙gapit + γ2 ∙gap* + Fi + Tt + Sit + εit, (31)
where LMRi*t collects foreign labor market variables.
Column (1) in Table 2 shows the most parsimonious specification, where we include
only the domestic and the foreign tax wedges (bit, bi*t) as well as the controls for the
domestic and the foreign business cycles and the complete set of fixed effects. We find
that the own and the foreign tax wedges help explain the domestic unemployment rate.
Both have coefficients with the signs predicted by our theoretical model and are accurately
estimated. Column (2) adds PMR as an additional control. PMR reflects different types
of entry regulations that limit competition on goods markets. Hence, PMR should be
39Capital intensities are proxied by the stock of capital computed as described above relative to the
total population instead of employment in order to mitigate the potential endogeneity of k*/k. The
variable u* × k*/k is instrumented by the interactions of k*-1∕kt-1 with exogenous foreign variables
LMRi*,t-1 and pmri*,t-1. See Dutt, Mitra, and Ranjan (2009) for a related empirical strategy.
37