Temporary and permanent migrant selection: Theory and evidence of ability‐search cost dynamics

• Author: Joyce J. Chen,Katrina Kosec,Valerie Mueller
• Date: 01 November 2019
• Published date: 01 November 2019
• DOI: http://doi.org/10.1111/rode.12617

Rev Dev Econ. 2019;23:1477–1519. wileyonlinelibrary.com/journal/rode

|

1477

© 2019 John Wiley & Sons Ltd

DOI: 10.1111/rode.12617

REGULAR ARTICLE

Temporary and permanent migrant selection:

theory and evidence of ability‐search cost dynamics*

Joyce J.Chen1

|

KatrinaKosec2

|

ValerieMueller3

*We thank Innovative Development Strategies, Limited, Hak Lim Lee, Mekamu Kedir, Arshad Khurshid, Hina Nazli, and

Sohail Malik for their assistance in the data collection and analysis, and Audrey Light, Robert E. B. Lucas, Pia Orrenius, and

Caroline Theoharides for comments on previous versions of the paper.

1University, Department of

Agricultural,Environmental, and

Development Economics, The Ohio State

University, Columbus, Ohio

2Development Strategy and Governance

Division,International Food Policy

Research Institute, Washington, DC

3Arizona State University School of

Politics and Global Studies, Tempe,

Arizona

Correspondence

Valerie Mueller, Arizona State University

School of Politics and Global Studies,

Tempe, AZ

Email: vmuelle1@asu.edu

Abstract

We integrate two workhorses of the labor literature, the

Roy and search models, to illustrate the implications of mi-

gration duration—specifically, whether it is temporary or

permanent—for patterns of selection. Consistent with our

stylized model, we show that temporary migrants are inter-

mediately selected on education, with weaker selection on

cognitive ability. In contrast, permanent migration is asso-

ciated with strong positive selection on both education and

ability, as it involves finer employee–employer matching

and offers greater returns to experience. Networks are also

more valuable for permanent migration, where search costs

are higher. Labor market frictions explain observed net-

work–skill interactions.

KEYWORDS

migration, networks, Pakistan, self‐selection

JEL CLASSIFICATION

J61; O15

1478

|

CHEN etal.

1

|

INTRODUCTION

As the geographic mobility of workers continues to rise, migration—the movement of individuals and

families—will be as central to economic development as trade. Remittances now far exceed official

aid flows (Ratha etal., 2014), and the skill composition of migrants has increasingly important con-

sequences for destination markets (Card, 1990, 2005; Borjas, 2005, 2006; Ottaviano and Peri, 2012).

But, unlike trade, individuals must move with their “goods,” raising complex questions about both the

motives and the scope for migration. To better understand the impact of migration on both spatial and

economic inequality, we must consider more deeply the migration decision itself (Henderson, 2010;

Young, 2014; Gollin etal., 2017).

The literature has extensively explored the profile of immigrants to inform policy‐makers of po-

tential shifts in labor force composition (Chiquiar and Hanson, 2005; Orrenius and Zavodny, 2005;

McKenzie and Rapoport, 2010; Moraga, 2011, 2013; Bertoli etal., 2013; Kaestner and Malamud,

2014). Previous studies have focused heavily on immigration to the USA, but internal migration

is an increasingly common form of migration worldwide.1

Historically, internal migration has

also been central to economic growth by reallocating rural labor to support the expansion of the

urban manufacturing sector (Lewis, 1954). Yet the extant literature still largely fails to differentiate

among migrants, despite drastically varied motives and skills (Rogerson, 1990; Newbold, 2001,

2012). The profile of internal migrants is of particular concern to the Pakistani government, given

objectives to shift production away from resource‐based exports to manufacturing‐based exports

(News, 2017).

In this paper we take an initial step toward modeling heterogeneity within internal migration by

focusing on variation in the duration of migration episodes among individuals originating from the

same areas and in the same time period. To formalize differences between permanent and temporary

migration, we embed a classic Roy (1951) model with heterogeneous moving costs, as in previous

work (Chiquiar and Hanson, 2005; McKenzie and Rapoport, 2010; Orrenius and Zavodny, 2005),

into a standard search model (Lippman and McCall, 1976). Variation in the duration of migration

opportunities implies important distinctions not only in the decision process but also in patterns of

self‐selection, resulting in differing effects on both origin and destination labor markets. For example,

temporary migration can dampen business cycles by increasing labor supply when wages are high,

while permanent migration can have the opposite effect if labor supply is slow to respond to changes

in market conditions.2

We use data from a unique panel survey of rural households in Pakistan spanning 22 years (1991–

2013) to study the drivers of selection into temporary and permanent internal migration. The data

offer an array of detailed information on worker attributes including education, cognitive ability

(digit span and Raven's test scores), and visible physical ability (height). Our empirical findings

confirm stronger positive selection for permanent migrants than for temporary migrants with re-

spect to formal schooling. However, networks play a nuanced role, both drawing high‐ability types

and dissuading the highly educated. This suggests that, while networks can reduce search costs,

particularly for high‐ability workers, there are limits to what they can do. Using an alternative data

source (Pakistan Bureau of Statistics, 1991, 2002, 2013), we show that our findings on education are

consistent with networks alerting highly educated workers of a low elasticity of demand for skilled

labor at destinations.

The remainder of the paper is organized as follows. Section 2 presents our theoretical framework,

describing the choice between migrating permanently, temporarily, or remaining at the origin. Section

3 describes our dataset, while Section 4 outlines our empirical approach. Sections 5 and 6 present our

main results and robustness checks. Section 7 concludes.

|

1479

CHEN etal.

2

|

THEORETICAL FRAMEWORK

In this section we lay out a stylized search model, as in Lippman and McCall (1976), to illustrate key

differences between permanent and temporary migration. We then embed features of a classic Roy

(1951) model to consider how patterns of self‐selection may differ with the type of migration.3

2.1

|

Search with permanent and temporary opportunities

Consider two different types of migration which can be, and often are, employed by different individ-

uals in the same period or by the same individual in different periods.4

The first involves a permanent

employment opportunity and, with it, a permanent change of residence. This type of migration often

occurs over longer distances and requires greater upfront investment (Kleemans, 2015). The second

involves a temporary employment opportunity, necessitating a temporary change of residence and an

eventual return to the origin. This type of migration often occurs over shorter distances, and is more

frequently used as a short‐term diversification or risk‐coping mechanism, requiring relatively little

upfront investment (Bryan etal., 2014). Migration is no longer a binary choice; in each period, the

individual considers both opportunities and selects the one that maximizes expected life‐time utility.

Given differences between the two types of opportunities, different patterns of selection will emerge

as well.

We formalize this choice process using a standard search model. We assume an individual faces a

fixed working life, at the beginning of which he receives a lifetime wage offer in the tehsil (i.e. sub‐dis-

trict) of origin.5

This operates much like unemployment insurance in labor search models, though the

“benefits” in this case do not expire, and ensures that not everyone will choose to engage in permanent

migration during their working life. Each individual also has employment opportunities outside of the

home tehsil, some temporary and some permanent. For ease of exposition, we consider an extreme

case where temporary employment episodes last only a single period, while permanent employment

offers last for the remainder of the working life. However, the implications of the model do not hinge

on the exact duration of temporary and permanent migration episodes. The key distinction is simply

that permanent migration provides employment over a substantially longer period of time than does

temporary migration.

A temporary employment opportunity is available in each period but for only a single period, akin

to entering a spot market for labor at the destination. In this case, the individual observes a wage distri-

bution for temporary employment opportunities that may vary from period to period.6

The individual

must then incur a search cost (i.e. move to the destination) in order to accept the temporary migration

opportunity and obtain a specific wage draw. A permanent offer from outside the home tehsil can also

be obtained in each period. Permanent employment provides a lifetime wage drawn from a known

distribution. In this case, search costs must be incurred before the wage draw is received, consistent

with the notion that permanent migrants often identify a (set of) specific employment opportunities

before migrating.7

We assume that the individual maximizes expected discounted lifetime income, net of search costs,

given the following choices in each period:8

(1) work at home and do not search for another job; (2)

work at home and search; (3) accept a temporary wage offer and do not search; (4) accept a temporary

wage offer and search; or (5) accept a permanent wage offer. The individual may also recall (i.e. return

to) any previous lifetime employment opportunity. Intuitively, a temporary offer will be accepted if

the wage gains exceed the search cost (see the Appendix for details). A permanent wage offer will be

accepted when the wage gains exceeds the continuation value.