The benchmark of this paper is the Fujita and Thisse (2002) core-periphery model, which adds a R&D sector with skilled labor to create new varieties for the modern sector. The number of R&D firms increases not only with the number of existing patents and knowledge spillovers but also with the number of skilled workers who can migrate and choose theregion offering the better lifetime salary.The main objective of the present work is to analyse the long-term consequences of the choice of the migration law in Fujita and Thisse(2002)and in other comparable models. After describing throughoutly our benchmark,we introduce a different migration law à la Krugman (1991).Although the change in the migration law implies that individuals do not foresee price changes and hence their choice is somehow less optimal, the steady state outcome does not vary qualitatively:the unique steady state is a symmetric distribution of labor across regions. Later we change the benchmark model to avoid the so called monotonic convergence hypothesis, about which we discuss at large in the paper. When we model the economy using Romer (1990) two sector model applied to two regions allowing for skilled migration, then there exists a solution path that converges to a steady state which exhibits a distribution of skilled workers amongst regions which is no longer symmetric. In effect, the new steady state depends on technology, fixed costs, knowledge spill-overs and transportation costs.
