Re-Evaluating the Returns to Language Skills Using Latent Trait Estimates

35 Pages Posted: 17 Nov 2017

Date Written: October 15, 2017

Abstract

Past studies have established a sizable wage premium for immigrants who learn the language of their destination country. However, these measurements are usually based on self-reported ordinal measures of spoken language fluency, which are problematic and inconsistent in several ways. Using detailed survey data from United States and France, this paper constructs more robust latent trait measures of language fluency using well-established psychometric methods, and re-evaluates the evidence for the wage premium. The results show that measures of spoken fluency alone conflate those with and without non-verbal skills (reading, writing, and comprehending) and therefore overestimate the wage premium for speaking. I find that the additional wage premium attributable to full verbal/nonverbal fluency is as large as that for verbal fluency alone. In addition, I provide evidence that the skills separating verbal-only from full fluency are generally related to education and training before immigration rather than to skills acquired after. Finally, I show that systematic differences in survey response between demographic groups are not a major source of measurement bias.

Suggested Citation

Marrone, James V, Re-Evaluating the Returns to Language Skills Using Latent Trait Estimates (October 15, 2017). RAND Working Paper Series WR- 1212. Available at SSRN: https://ssrn.com/abstract=3072060 or http://dx.doi.org/10.2139/ssrn.3072060

James V Marrone (Contact Author)

University of Chicago ( email )

1126 E 59th St
Chicago, IL 60637
United States

Here is the Coronavirus
related research on SSRN

Paper statistics

Downloads
22
Abstract Views
561
PlumX Metrics