Migration
More General Approach

Why do people migrate? Economic motives are clearly important, and a great deal of empirical work in labour economics supports the hypothesis that the migration rate is sensitive to the difference in income between sending and receiving countries. Most research in this area involves estimating a single regression equation, using time series data on migration and such explanatory variables as the real wage or income difference and a measure of relative unemployment levels. The problem with this 'single equation' approach is its assumption that the migration itself has no effect on the wage rate in either the sending or receiving country. Such an assumption might seem appropriate when migration is small relative to the labour forces in the two countries, but evidently this is not so with most important migrations. A large influx of migrants into the receiving country may increase the supply of labour and this may lower the real wage.

A decade ago Jeffrey Williamson argued for a general equilibrium approach to international migration. He proposed a simple model which explained labour supply and demand in the sending and receiving countries, wage levels and migration. The model could then be solved to obtain an equation explaining migration in terms of variables which are exogenous or determined outside the model. This 'reduced form' migration equation was estimated by Williamson using historical data on migration between four European economies and the United States between 1870 and 1914. His 'reduced form' results were in broad agreement with the underlying theory. When the complete model was estimated, however, these findings were not supported.

In Discussion Paper No. 72, Patrick Geary and CEPR Research Fellow Cormac O Grada undertake a closer examination of Williamson's model. They argue that it is less general equilibrium in nature than might first appear. The analysis and estimation involves migration into the US from a single sending country, and migration from other sending countries is not fully incorporated into the model. To allow for this would require far more data than are available. Geary and O Grada suggest ways of extending Williamson's model. These extensions yield more interesting results for the migration equation but distinctly mixed results overall. The authors conclude that Williamson's framework is potentially better suited to the study of modern migrations, for which there are better data.


Modelling Emigration to the New World: Some Problems
P T Geary and C O Grada

Discussion Paper No.72, August 1985 (HR)