Working Paper: NBER ID: w13887
Authors: George J. Borjas; Jeffrey Grogger; Gordon H. Hanson
Abstract: In a recent paper, Ottaviano and Peri (2007a) report evidence that immigrant and native workers are not perfect substitutes within narrowly defined skill groups. The resulting complementarities have important policy implications because immigration may then raise the wage of many native-born workers. We examine the Ottaviano-Peri empirical exercise and show that their finding of imperfect substitution is fragile and depends on the way the sample of working persons is constructed. There is a great deal of heterogeneity in labor market attachment among workers and the finding of imperfect substitution disappears once the analysis adjusts for such heterogeneity. As an example, the finding of immigrant-native complementarity evaporates simply by removing high school students from the data (under the Ottaviano and Peri classification, currently enrolled high school juniors and seniors are included among high school dropouts, which substantially increases the counts of young low-skilled workers ). More generally, we cannot reject the hypothesis that comparably skilled immigrant and native workers are perfect substitutes once the empirical exercise uses standard methods to carefully construct the variables representing factor prices and factor supplies.
Keywords: Immigration; Labor Economics; Wages; Substitutability
JEL Codes: J01; J61
Edges that are evidenced by causal inference methods are in orange, and the rest are in light blue.
| Cause | Effect |
|---|---|
| High school students included in the data sample (I21) | upward bias in estimates of elasticity of substitution (C51) |
| Removing high school students from the analysis (I21) | downward-biased estimate of the elasticity of substitution (C51) |
| Misclassification of students as dropouts (I21) | inflated counts of low-skilled native workers (J69) |
| inflated counts of low-skilled native workers (J69) | skewed results to suggest complementarity (D10) |
| Using adjusted measures of wages and employment counts (J39) | consistently small and statistically insignificant coefficients (C20) |
| Previous findings of imperfect substitution (D43) | artifact of data mismanagement and sample selection biases (C83) |