unexplained by the firm-specific shock. Given that Shimer (2005) estimates the monthly
rate of job separation to be 0.034, it follows that the rate of arrival of match-specific
shocks χ should be equal to 0.025 per month.
Cost parameters. We set the interest rate to 4% per year. In order to calibrate the
value of non-market activity, we follow Shimer (2005) and set bi = 0.4 for all i in the
benchmark to match an earnings replacement ratio close to 40%. The cost of posting a
vacancy, ci , is set 50% above the vacancy filing rate for all three countries. Given that
the equilibrium wage is around wi = 1.137, this value yields an average recruitment cost
of around 5.7 weeks of workers’ earnings, as suggested by empirical estimates.
Variable and fixed costs of trade and entry. We choose variable trade costs τij
equal to 1.3 for all country-pairs ij in the benchmark equilibrium, following Ghironi and
Melitz (2005). Given the Pareto distribution for firm productivities, the share of firms
that export is
'j = τ j ( Pj )’ ( Rf ) 1-σ. (27)
i ifij
That number is put at about 21% by Bernard, Eaton, Jensen, and Kortum (2003). To-
gether with τij = 1.3 for all country-pairs ij and assuming a symmetric benchmark equi-
librium, this pins down the ratio fij /fii at about 1.7. We use the values of entry costs,
fe , and the flow fixed costs, fij , to match the following two moments. First, we ensure
that the equilibrium tightness θi = 0.5 for all countries in the benchmark equilibrium.
Second, we target an average firm size equal to 21.8 employees, as estimated by Axtell
(2001). The calibrated entry costs are equivalent to 2.82 years of income per capita. This
figure can be compared to the assessment by Ebell and Haefke (2009) that regulatory
barriers to entry in the US amount to 0.6 month of yearly income. The parametrization
therefore suggests that technological innovation costs outweigh entry fees by an order of
magnitude. The Appendix contains a summary table of all chosen parameter values.
3.2 How domestic institutions impact outcomes world-wide
We now deviate from the symmetric benchmark equilibrium and allow for differences in
unemployment benefits, trade frictions, and country sizes. We pay particular attention
to cross-country differences in unemployment benefits as they are easily observable in
the data, exhibit substantial variation across countries, and are shown to consistently
explain unemployment rates in empirical research.22 Moreover, we know that the model
reacts similarly to changes in search costs ci or the search technology mi (see Lemma
1). Specifically, we study what effects an increase in unemployment benefits of country
1 (the “bad” country) has on unemployment in that country and, more importantly, in
its trading partners (countries 2 and 3). We vary b1 in the interval [0.4, 0.8] and hold
unemployment benefits for countries 2 and 3 constant at the benchmark value of 0.4.
22See, e.g., Bassanini and Duval, 2006.
18