Heaven's Swing Door: Endogenous Skills, Migration Networks, and the Effectiveness of Quality‐Selective Immigration Policies

• Published date: 01 April 2015
• Author: Simone Bertoli,Hillel Rapoport
• DOI: http://doi.org/10.1111/sjoe.12095
• Date: 01 April 2015

Scand. J. of Economics 117(2), 565–591, 2015

DOI: 10.1111/sjoe.12095

Heaven’s Swing Door: Endogenous Skills,

Migration Networks, and the Effectiveness

of Quality-Selective Immigration Policies∗

Simone Bertoli

University of Auvergne, CERDI, FR-63000 Clermont-Ferrand, France

simone.bertoli@udamail.fr

Hillel Rapoport

Paris School of Economics, CES, FR-75647 Paris, France

hillel.rapoport@univ-paris1.fr

Abstract

A growing number of OECD countries are leaning toward the adoption of selective immi-

gration policies, which are expected to raise the quality (or education level) of migrants.

This view neglects two important dynamic effects: the role of migration networks, which

could reduce the quality of migrants, and the responsiveness of education decisions to the

prospect of migration. We propose a model of self-selection into migration with endogenous

education choices, which predicts that migration networks and the quality of migrants can

be positively associated when destination countries adopt sufficiently selective immigration

policies. Empirical evidence, presented as background motivation, suggests that this is indeed

the case.

Keywords: Brain drain; discrete choice models; migration; self-selection

JEL classification:F22; J61; O15

I. Introduction

A growing number of OECD countries are leaning toward adopting more

restrictive and increasingly quality-selective immigration policies, which

confer better chances of admission at destination to those applicants with

a higher level of education. This tendency is apparent from the gradual

introduction of points-based immigration systems, first in Canada in 1967,1

∗The authors are grateful to Slobodan Djajic, Aitor Lacuesta, and the participants at various

conferences and seminars for their comments. This research was supported by the Agence

Nationale de la Recherche of the French government through the program Investissements

d’avenir (ANR-10-LABX-14-01) and by the FERDI. The usual disclaimers apply.

1Education alone can provide an applicant with up to 25 out of the 67 points that are

currently necessary for admission into Canada (Bertoli et al., 2012), and its pivotal role in

shaping the chances of admission is magnified by its positive correlation with labor-market

experience (21 points) and language proficiency (24 points).

CThe editors of The Scandinavian Journal of Economics 2014.

566 Endogenous skills, migration networks, and immigration policies

followed by Australia in 1989, New Zealand in 1991, and more recently

the UK in 2008. Elsewhere, immigration policies have also evolved towards

becoming more restrictive quantitatively and more selective qualitatively,

be it through the introduction of specific visa categories for highly skilled

professionals (e.g., the H1-B visa category in the US, or the European

“Blue Card” project currently in its infancy) or through the introduction

of biased selection criteria, making low-skill immigration more difficult

while at the same time encouraging permanent high-skill immigration (e.g.,

France’s short-lived “chosen immigration” reform of 2007).2

Observed migration flows are the result of the combined effect of self-

selection (i.e., size and skill composition of a given pool of candidate

immigrants) and out-selection (i.e., external selection among existing can-

didates) mechanisms. The underlying assumption behind quality-selective

immigration policies is that more selection will raise the average education

level of migrants. This makes perfect sense from a static standpoint, when

we regard the pool of applicants as given. Still, this pool evolves over

time as the migration process unfolds: migration networks tend to reduce

the moving costs (Massey et al., 1994; Carrington et al., 1996; Munshi,

2003; Kanbur and Rapoport, 2005), thus influencing the self-selection into

migration (McKenzie and Rapoport, 2010; Beine et al., 2011a; Bertoli,

2010a).

The cost-reducing effect of migration networks is best illustrated by

the “swing door” metaphor in the title of our paper: while the first mi-

grants to push the door will encounter the most resistance, their followers

will be able to enjoy the lower resistance of a swing door in movement.

The literature has established that this swing door effect is not neutral

with respect to the quality of migrants. McKenzie and Rapoport (2010)

find that the probability of (f irst-time) migration from Mexico to the US

increases with education up to relatively high education levels in commu-

nities with small networks (high migration costs). This is consistent with

positive self-selection and, conversely, with the fact that migration propen-

sities decrease with education in communities with large networks (low

migration costs), which is consistent with negative self-selection. A similar

result is obtained by Bertoli (2010a) from the analysis of Ecuadorian mi-

gration, while Beine et al. (2011a) provide evidence that larger networks

translate into more negative self-selection patterns using bilateral data from

2The decline in the level of education of immigrants (Borjas, 1999), with its possible

contribution to rising inequality and increased pressure on underprivileged segments of the

native population (Borjas et al., 2010), has prompted proposals to increase the degree of

selectivity of immigration policies (see, for instance, the specific proposals advanced by

Borjas, 1999); the immigration reform bill that was introduced in the US Senate in April

2013 contains a provision for the admission of 120,000 immigrants per fiscal year through

a merit-based system.

CThe editors of The Scandinavian Journal of Economics 2014.