THE ECONOMIC PROGRESS OF IMMIGRANTS: Immigrant Economic Progress and Source Country Characteristics 2

2. The Gini coefficient of the source country’s income distribution. Boijas (1987) has argued that immigrants originating in countries that offer a high rate of return to skills are more likely to be negatively selected, will have a smaller effective human capital stock at the time of migration, and will earn less in the United States. A higher rate of return to skills implies a more disperse distribution of income. The Gini coefficient of the source country’s income distribution should then have a negative impact on the rate of wage growth. Deininger and Squire (1996) have constructed various measures of income inequality, including the Gini coefficient, for most countries since 1960. I used these data to obtain measures of the Gini coefficients in four years: 1960, 1965, 1970, and 1975.

3. A measure of “openness” of the source country’s economy. The openness index is defined as the ratio of exports plus imports to GDP (in percentage terms). I used the Penn World Tables to get this index for the calendar years 1960,1965, 1970, and 1975. Immigrants originating in countries with open economies are more likely to have some contact with foreign industries and economic institutions prior to migration, are more likely to have the types of skills that other countries value, and would be expected to have a higher level of effective capital when they enter the United States. Weak relative complementarity in the production function implies that the openness index should be positively correlated with both the log entry wage and the rate of wage growth of immigrant cohorts.
4. A Herfindahl index measuring how immigrant cohorts cluster geographically once they enter the United States. It has long been suspected (without much evidence) that the
where Eljr gives the fraction of immigrants from the (i,j) cell who live in state r. The Herfindahl index takes on a maximum value of one if all immigrants live in a single state, and becomes smaller the more randomly the immigrants are distributed over the United States. I use data on states, rather than on metropolitan areas, to calculate the clustering index. The Herfindahl index is sensitive to the number of geographic units, and the number of metropolitan areas identified by the Census has grown significantly over time (particularly between 1970 and 1980). The state-based calculation, therefore, makes the Herfindahl index comparable over time. The measures of the Herfindahl index for the immigrant cohorts that arrived in either 1960-64 or 1965-69 are obtained from the 1970 Census, while the measures of the index for the cohorts that arrived in 1970-74 or 1975-79 are obtained from the 1980 Census.