Revisiting the Relationship between Ethnic Diversity and Preferences for Redistribution: Comment

Date01 April 2017
DOIhttp://doi.org/10.1111/sjoe.12209
Published date01 April 2017
Scand. J. of Economics 119(2), 268–287, 2017
DOI: 10.1111/sjoe.12209
Revisiting the Relationship between
Ethnic Diversity and Preferences
for Redistribution: Comment
Lena Nekby
Per Pettersson-Lidbom
Stockholm University, SE-106 91 Stockholm, Sweden
pp@ne.su.se
Abstract
In this paper, we revisit the question raised in Dahlberg et al. (2012, Journal of Polit-
ical Economy 120, 41–76) concerning a causal relationship between ethnic diversity and
preferences for redistribution. We find that their results are based on (i) an unreliable and
potentially invalid measure of preferences for redistribution, (ii) an endogenously selected
sample, and (iii) a mismeasurement of the refugee placement program. Correcting for any
of these three problems reveals that there is no evidence of any relationship between ethnic
diversity and preferences for redistribution. We also discuss what is currently known about
the refugee placement program, and to what extent it can be used for estimating causal
effects more generally.
Keywords: Ethnic diversity; income redistribution
JEL classification:D64; I30; J15; J18; Z18
I. Introduction
In this paper, we revisit the question raised in Dahlberg et al. (2012) –
henceforth DEL – concerning a causal relationship between ethnic diver-
sity and preferences for redistribution.1The question of whether ethnic
heterogeneity affects individual behavior, such as preferences, has received
considerable attention in the literature (e.g., Alesina and La Ferrara, 2005).
However, there are few studies that use a credible exogenous source of
This paper is a heavily revised version of Nekby and Pettersson-Lidbom (2012). We thank
Thina Carlsson at the Archives of the Swedish Board of Immigration for generous help with
data collection. We also thank Torsten Persson, David Str¨
omberg, M˚
arten Palme, Peter Skog-
man Thoursie, Peter Fredriksson, Mikael Lundholm, Mahmood Arai, and two anonymous
referees for useful comments.
Lena passed away in July 2014. Her friendship and her energy will be greatly missed.
1In a previous version of this paper (Nekby and Pettersson-Lidbom, 2012), we also discuss
other important issues concerning the replication of DEL. We also comment on their reply
to our critique (Dahlberg et al., 2013).
CThe editors of The Scandinavian Journal of Economics 2016.
L. Nekby and P. Pettersson-Lidbom 269
variation in ethnic diversity. In contrast, DEL exploit a plausible exogenous
variation in ethnic heterogeneity arising from a nationwide refugee
placement program, which placed refugees in municipalities throughout
Sweden during 1985–1994. DEL also match data on refugee placement to
individual survey data on the preferences for redistribution, and they use
the placement policy as an instrument for ethnic diversity as measured
by the share of immigrants in the municipality. The instrumental variable
(IV) results suggest a large and statistically negative relationship between
ethnic diversity and preferences for redistribution. However, the empirical
design is compromised by three problems: (i) the measurement of attitudes
towards redistribution lacks reliability and validity; (ii) endogenous sample
selection; (iii) the mismeasurement of the refugee placement program.
Evaluating the validity and reliability of a measurement item in a survey
(i.e., whether the response to the survey question measures what it is
purported to measure and whether the survey response is consistent) is
standard procedure in survey research. However, DEL only report results
from one measure of attitudes toward redistribution, even though there
are at least two other survey items in the Swedish election survey that
a priori should be equally good, or even better, measures of preferences for
redistribution, at least when it concerns redistribution towards immigrants.2
Importantly, the results from the two alternative outcome variables do not
corroborate the finding in DEL.
The endogenous sample selection is essentially due to the fact that DEL
only base their empirical analysis on individuals who answered the survey
in two consecutive elections, in order to create a panel.3Thus, the sampling
scheme is endogenous as it is based on the outcome (i.e., only individuals
who responded to the survey more than once are included in their sample).
As a result, there will be a strong sample selection bias as the attrition rate
in DEL’s sample is 73 percent relative to the original nationally represen-
tative sample of 14,297 observations. In other words, DEL’s result is based
on a highly selected panel of 1,917 individuals.4
The mismeasurement of the refugee placement program is due to the fact
that DEL measure the program by the number of grants paid out by the
Swedish Board of Immigration (SIV) to local governments.5However, it
2It is worth noting that the results from these two other outcomes are reported in a previous
working paper by Dahlberg and Edmark (2009).
3We pointed out the sample selection problem, and discussed how it could be solved with
DEL, at a seminar at Stockholm University in November 2011, and in several e-mails sent
in November and December 2011. To our knowledge, their paper had not been resubmitted
to the Journal of Political Economy at that time.
4DEL also make eight unreported sample restrictions on the panel data, which reduced the
number of individuals from 2,702 to 1,917.
5The rules governing the payments to municipalities are laid out in Swedish law (see links for
more information): F¨
orordning (1984:683) om statlig ers¨
attning f¨
or mottagande av flyktingar
CThe editors of The Scandinavian Journal of Economics 2016.

To continue reading

Request your trial

VLEX uses login cookies to provide you with a better browsing experience. If you click on 'Accept' or continue browsing this site we consider that you accept our cookie policy. ACCEPT