Economic Effects of Labor Migration: A Romanian Case Study

The thesis aims to analyze the impacts of labor migration on the Romanian economy. It is motivated by the large-scale exodus of people that has taken place from the country over the past 20-25 years. It places itself within the larger body of migration literature, most of which assesses the impacts of immigration on receiving countries. Defining the independent variable of the analysis as net migration stocks as shares of the overall population (a variable which takes on negative values in cases of net out-migration), it is possible to utilize the standard theoretical framework applied in the literature. Specifically, three models are considered; one with roots in labor economics ( the labor model ), a Heckscher-Ohlin model and a Ricardo-Viner model. All models use labor and capital as factor inputs, and labor is assumed to be homogenous. The Ricardo-Viner (R-V) and the Heckscher-Ohlin (HO) models assume that two goods are produced; one labor intensive and one capital intensive. The labor and Ricardo-Viner models are suitable for short-term analysis, whereas the Heckscher-Ohlin model - given its assumption of perfectly elastic capital transfers between the two production sectors - yields better predictions for the long term. Different implications of the three models are discussed at some length, and for the labor and Heckscher-Ohlin models we solve for equilibrium outcomes and migration s expected impact on these. The predicted effect of labor migration differs somewhat between the models. The labor model focuses on the short-term effect on wages, and finds that these are expected to decrease with positive migration changes. The Ricardo-Viner model emphasizes the adjustments of structural variables (such as the relative sizes of the two production sectors) in the short term. It finds that both sectors will benefit from inward migration. Over the long term, the Heckscher-Ohlin model predicts that factor prices will be unaffected by endowment changes, and that it is the labor intensive sector that will benefit from immigration, whereas the capital intensive sector will go into decline. The thesis also discusses some mechanisms that could reasonably be expected to influence the findings of the empirical analysis that are not incorporated in the formal models. Examples of this include endogenous labor market participation choices of natives and factor intensity changes in the production sectors. A general problem in analyzing Romanian labor migration is lack of quality data. Ideally, our empirical analysis would have included the education and work experience levels of migrants as well as non-migrants, in order to see e.g. if Romania is experiencing brain drain (as claimed by some), and how the skill attainment of migrants is predicted to affect the economic variables of the models. To my knowledge, the thesis utilizes all relevant county-level data that are freely available. The empirical analysis considers net migration s effect on five dependent variables (four of which are separated by production sector), specifically earnings, employment, net investments, turnover revenue and gross value added (GVA). Since there could be some simultaneous causality between these variables and net migration shares, three instruments are introduced. These are historical migration shares, the size of minority populations and foreign wage levels. The instruments are tested and mostly found to be exogenous and relevant, with the exception of foreign wages. The empirical analysis also considers the possibility that observations could be correlated across panels; a concern which is tested and generally found to be justified. Cross-panel correlation robust (CPCR) regressions are therefore included in the analysis. Overall, we find that the estimated effects are broadly in line with HO theory, indicating that one year is sufficient time for the structural variables to adapt (at least partly) to migration shocks. This finding appears to be most robust in the case of investments, where the relevant regression specifications all yield positive migration effects in the labor intensive sector and negative effects in the capital intensive sector. Also for the employment and turnover variables are the CPCR findings convincingly in line with the HO model. Results are less clear-cut for gross value added, something which could be due to the exclusion of several industries in the GVA observations. The effect on earnings might be zero (as predicted by HO theory), but estimates vary and I argue that trustworthy instruments for this variable are lacking