We incorporate the now standard knowledge-capital model of multinational firms in a new economic geography setting. The theoretical predictions of our model suggest that unskilled labor mobility leads to less concentration of production than skilled labor mobility does. This is in line with empirical evidence that agglomeration of production among European nations is less pronounced than among US regions. Our model shows that the different patterns in labor mobility can explain actual differences in the spreading of industries. According to our welfare analysis, trade liberalization is likely Pareto-improving for a larger (smaller) country with mobile unskilled (skilled) labor.<br>In the supplement, we investigate the sensitivity of our results in several respects. In the first section, we provide the figures of real factor rewards for the trade liberalization scenarios discussed in and underlying Figures 7 and 8 of the paper. Second, in Figures 3(n) - 5(v) (6(n) - 6b(v)) we infer the existence, or non-existence, of each firm type separately in the τ - λ<i><sub>L</i></sub>-space (τ - λ<i><sub>S</i></sub>-space) for country <i>i</i> firms and all four scenarios of firm regimes. Third, we illustrate how changes in the parameters μ, ρ and σ affect the outcome. Finally, we analyze how the asymmetric endowment with the immobile factor influences the core-periphery patterns.
