Optimal Design of an Immigration Points System

Abstract

There is growing interest in the United States and elsewhere in the use of a points-based system for selecting immigrants on the basis of their observed human capital. This paper explores the design of an optimal skills-based immigrant selection system based on two basic elements: a predicted-earnings threshold for determining whom to accept and reject, and a human-capital-based earnings regression for making error-minimizing predictions of immigrant success in the host labor market. We first show how to design a points system based on what are assumed to be the optimal predicted-earnings threshold and the optimal prediction regression. We next develop a method for identifying the optimal threshold given the prediction regression. The method produces a ¿selection frontier¿ that describes the options facing policy makers. The frontier shows the tradeoff between the average quality of admitted immigrants and the number of immigrants admitted. The frontier shifts out with improved accuracy in predicting earnings as well as with increases in the variation and average quality of the applicant pool. Finally, we show how the policy maker chooses the optimal selection system given the selection frontier.