The migration response to local labour market shocks: Evidence from EU regions during the global economic crisis
| Published date | 01 April 2019 |
| Date | 01 April 2019 |
| DOI | http://doi.org/10.1111/obes.12271 |
| Author | Timo Mitze |
271
©2018 The Department of Economics, University of Oxford and JohnWiley & Sons Ltd.
OXFORD BULLETIN OF ECONOMICSAND STATISTICS, 81, 2 (2019) 0305–9049
doi: 10.1111/obes.12271
The migration response to local labour market shocks:
Evidence from EU regions during the global economic
crisis*
Timo Mitze†,‡,§
†Department of Business and Economics, University of Southern Denmark (SDU), Alsion 2,
DK-6400 Sønderborg, Denmark (e-mail: tmitze@sam.sdu.dk)
‡Rheinisch-Westf¨alisches Institut f¨ur Wirtschaftsforschung (RWI), Hohenzollernstr. 1-3, 45128
Essen, Germany
§Rimini Centre for Economic Analysis (RCEA), Wilfrid Laurier University, 75 University
Ave West, Waterloo, ON, Canada
Abstract
The global economic crisis has strongly affected Europe’s economic geography.This study
investigates the role of local labour market disparities in determining regional net in-
migration rates. While only a weak link is detected in the precrisis period, the local labour
market context of migration grows significantly stronger during the crisis. Decompositions
of the estimation results show that changes in migration rates are firstly a result of widened
disparities across European regions throughout the crisis. However, also behavioural ad-
justment processes occur, e.g. an orientation of migrants towards urban areas and away
from regions with persistently high long-run unemployment rates.
I. Introduction
A central question in labour and regional economics is whether and how local labour
market disparities affect internal migration (e.g. Kennan and Walker, 2011).1In this pa-
per, I present new evidence on this question by investigating the case of EU regions
from 2000 to 2012, which importantly includes the years of the global economic cri-
sis of 2008. As local labour market disparities have significantly widened during the
crisis (Crescenzi, Luca and Milio, 2016), I take this economic shock as a source of
JEL Classification numbers: C23, C26, F22, J61, R23.
*The author thanks Timothy Neal and Matthieu Gomez for helpful suggestions on the estimation of static and
dynamic common factor models and their implementation in STATA.Theauthor also acknowledges helpful comments
on earlier manuscript versions from Johannes Br¨ocker, Torben Dall Schmidt and Peter Sandholt Jensen as well as
editorial assistance from Lene Holbæk.
1See also Moretti (2011) for a more general treatment of the causes and the consequences of differences in labour
market outcomes across local labour markets within a country. Related to this strand of the literature, recent studies
in international economics have also shown a reinforced interest in studying local labour market dynamics, i.e. the
response to exogenous demand changes caused by international import competition to local manufacturers (Autor,
Dorn and Hanson, 2013) and trade liberalization (Dix-Carneiro and Kovak, 2017).
272 Bulletin
exogenous variation in local labour markets to test which factors have caused regional
migration rates to rise or fall. Surprisingly, empirical evidence on this matter is still largely
missing for the EU (with the exception of Jauer et al., 2014) and the growing litera-
ture on US internal migration is yet inconclusive (Mian and Sufi, 2014; Yagan, 2014,
2017; Monras, 2015; Cadena and Kovak, 2016). Particularly EU policy makers have
traditionally been concerned about internal migration because Europeans appear to be
generally less mobile than their US counterparts (Faini, 1999; Bonin et al., 2008; Molloy,
Smith and Wozniak, 2011; OECD, 2016). This makes the present study highly policy
relevant.
To carry out the analysis, I estimate a dynamic migration model utilizing data on
net in-migration rates, labour market indicators, and other regional economic variables
measured at different spatial scales (NUTS2 and NUTS3). Splitting the overall sample
into a pre- and postcrisis period, I estimate the relative contribution of local labour market
disparities in explaining the variations in regional in-migration rates over time. I also
implement a decomposition analysis to disentangle the estimated labour market effects
into two components. The first component relates to the degree to which changes in net
in-migration rates are driven by widened labour market disparities.The second component
relates to the extent to which changes in the migrants’ behavioural response take place
when holding labour market disparities constant at their precrisis values.
A fundamental challenge in the conduct of this empirical analysis is that labour market
conditions, such as income levels and (un-)employment rates, are determined simultane-
ously with the evolution of net migration rates at the local level. For this reason, I apply
an instrumental variable (IV) approach using Bartik-type shift-share instruments which
seeks to isolate local labour demand shocks as source of exogenous variation in driving
interregional migration. In addition to shift-share instruments, information on the regional
importance of the construction sector prior to the crisis is used as a complementary measure
for the unexpectedness of the crisis (Monras, 2015).
The empirical results from baseline fixed effects (FE) and structural IV estimation
show a significant link between regional net in-migration rates and local labour mar-
ket conditions during the global economic crisis. In comparison, I find that the local
labour market context is less important for determining regional migration rates prior
to the crisis. That is, while variations in local labour market conditions only account
for roughly 11% of the total variation in regional migration rates across NUTS3 re-
gions prior to the crisis, their explanatory power increases to 32% in the postcrisis pe-
riod. At the NUTS2 level, the local labour market context even accounts for roughly
40–50% of the observed overall variationsin regional net in-migration rates throughout the
crisis.
The decomposition of the estimation results further shows that changes in the migration
response over time are first and foremost attributable to widening local labour market
disparities. However, for some variables, such as the population density and components
of the unemployment rate, significant behavioural adjustment processes are observed as
well.That is, for instance, structurally weak regions with high long-run unemployment rates
experience significant lower net in-migration rates in the post- compared to the precrisis
period. Moreover, I observe an increased orientation of migrants towards urban areas with
a high population density.This indicates that migrants not only respond to amplified labour
©2018 The Department of Economics, University of Oxford and JohnWiley & Sons Ltd
To continue reading
Request your trial