Abstract
We present longitudinal survey data suggesting that the terrorist attacks in the USA on September 11, 2001, changed attitudes toward certain minorities in Sweden. This finding is consistent with results in previous studies. To investigate whether this change in attitudes also affected the labor market situation of these minorities, we study unemployment exit around 9-11 using detailed data on the entire Swedish working-age population. Contrary to what may be expected from many theories of labor market discrimination, the time pattern of exits and entries for different ethnic groups, as well as difference-in-differences analyses, shows no sign of increased discrimination toward these minorities. A possible explanation for this result is that employers act rationally in their hiring decisions and do not respond to changes in attitudes toward immigrants as a group.
Similar content being viewed by others
Notes
Such theories were initially formalized in Becker (1957).
The FBI reported a 17-fold increase in anti-Muslim crimes in 2001, which is in line with figures from Los Angeles county and Chicago (Human Rights Watch 2002). In the first year after the attacks, the US Equal Employment Opportunity Commission (EEOC) received more than twice as many charges based on “Religion—Muslim” compared to the year preceding 9-11. The EEOC has also found 654 charges of employment discrimination with an alleged connection to 9-11. The figure includes “charges alleging discrimination related to the events of September 11, 2001, by individuals who are—or are perceived to be—Muslim, Arab, Afghani, Middle Eastern or South Asian, or individuals alleging retaliation related to the events of September 11, 2001” (EEOC 2003). It is worth noting that out of the 449 cases that had been resolved by October 1, 2002, 288 were “closed with no cause findings.”
See for instance the International Social Survey Program at http://www.issp.org.
The respondents were asked “What do you think about immigrants?” and had three positive, three negative, and one neutral alternative to choose from when answering the question. The positive and negative categories are then further aggregated into only one category.
Unfortunately, there is no connection to demographic data of the respondents, thereby preventing a separate analysis of attitudes. The data we received from FSI were on a more aggregate level than in Fig. 1, which restricts us from expanding the figure in time.
As Larsson (2003) discusses, it is not certain that the answers give a representative picture of the general experiences of Swedish Muslims after 9-11. First, the questionnaire was distributed via Islamic organizations mostly in Gothenburg. Thus, the survey had regional boundaries and reached only a particular part of the population of interest. Second, only 176 out of 450 distributed surveys were answered. Still, considering how widespread the feelings of increased exposure are, they are probably not fully limited to the responding group.
See the Appendix B for a more formal presentation. How discrimination affects an unemployed person engaged in job search has been formalized in an increasing number of articles during the last 20 years. This literature focuses on equilibrium effects rather than on the (partial) effect on individual job search strategies (see for instance Altonji and Blank 1999). The purposes of these articles have been to extend the Becker (1957) model of discrimination and to show that wage differentials due to discrimination exist in equilibrium when frictions are included into the Becker model.
We believe that discrimination due to changing preferences is the most likely mechanism in this context. It is, though, also possible that 9-11 increased the so-called statistical discrimination (see Altonji and Blank 1999). Statistical discrimination arises when employers have imperfect information about individuals and therefore use measures of group productivity. It is of course unknown to us which productivity characteristics employers consider to be important but hard to observe. One could argue that 9-11 did not provide information about many of the factors commonly connected to this type of discrimination, such as language skills or institutional knowledge. In the 9-11 context, changes in statistical discrimination could occur, e.g., if employers consider the risk that an individual is planning terrorist activities. Indeed, such an explanation can be regarded as far-fetched and requires that the terrorist activities would incur a cost to the employer. Statistical discrimination may, however, also operate through increased risk aversion. If hiring certain immigrant groups is considered risky and the events made employers more risk-averse, we would expect a decrease in job offers given to certain groups.
This could reflect the cost of contacting potentially discriminatory employers. It may also be, for example, the discomfort of leaving one’s home.
If one believes that employers set out to discriminate Muslims only, the empirical analysis could be said to capture statistical discrimination of a group where many—but not all—are Muslim. We do not know how preferences are formed but note that the studies presented in Sect. 2 suggest some stereotyping in the attitudes.
According to Statistics Sweden (1993), about 90% of those who are unemployed according to the labor force surveys are also registered at the employment offices.
Roughly 10% of those included in the data have one or more dates (for transition) changed by our procedure.
For Natives, we use a 10% random sample.
Another contributing explanation may be that the figures include information on some spells starting before 1991. However, this is only the case for a very small fraction of the sample. Among Nordic immigrants included in HÄNDEL (who have the longest average stay in Sweden), less than 0.1% of the individuals have a spell that started before 1991.
This result can actually be seen in Fig. A2, since the confidence intervals for all estimates overlap the estimated effect in the “Sweden” group. The DD specification is a linear version of Eq. 1 where a DD dummy (group*post-9-11) is added. The coefficients for all individual variables are allowed to vary across groups (i.e., we include interactions in the model). Note also in Fig. 2 that the “Africa” group has a high peak around the month “−12”, i.e. the start of the pre-9-11 period in the DD analysis.
It should be noted that the people in the sample are unemployed and, in the strict sense, do not belong to any particular occupation. Since searching for other types of jobs may very well be a response to the 9-11 changes, we have excluded this variable from the baseline specification.
Of course, this conclusion assumes that the reservation wage effect does in fact not exactly cancel out the direct effect. If it does, we would not expect any differential impact on exit rates due to changed discrimination. As discussed in Sect. 4.1, several arguments suggest that this is unlikely to be the case. Furthermore, if opposing effects canceling each other out were the true story, the results would probably not be so robust across subgroups.
A more extensive presentation of this analysis is included in a previous version that can be retrieved from the authors.
For instance, the family of lognormal distributions and the family of Pareto distributions fulfill these requirements.
If we assume that \({\left| {\frac{{dw*}}{{dk}}k} \right|} = {\left| {\frac{{dw*}}{{d\lambda {\left( {s*} \right)}}}\lambda {\left( {s*} \right)}} \right|}\) then it is required that \({\left| {\frac{{ds*}}{{dk}}\frac{k}{s}} \right|} > 1\) for the results in Van den Berg (1994) to hold. We thank Peter Fredriksson for pointing this out.
References
Allen C, Nielsen JS (2002) Summary report on islamophobia in the EU after 11 September 2001. European Monitoring Centre on Racism and Xenophobia, Vienna
Altonji J, Blank R (1999) Race and gender in the labor market. In: Ashenfelter O, Card D (eds) Handbook of labor economics. Elsevier, Amsterdam, pp 3143–3259
Ams (2002) Definitioner inom AMV:s statistik. Historik över ett urval av våra begrepp. Uin 2002:2, Norrköping
Arab American Institute (2002) Healing the nation—the Arab American experience after September 11. Arab American Institute, Washington, DC
Becker G (1957) The economics of discrimination. University of Chicago Press, Chicago
de los Reyes P, Wingborg M (2002) Vardagsdiskriminering och rasism i Sverige. Integrationsverkets rapportserie 2002:13, Norrköping
EEOC (2003) Muslim/Arab employment discrimination charges since 9/11. U.S. Equal Employment Opportunity Commission http://www.eeoc.gov/origin/z-stats.html. Cited 17 Sept 2003
FSI (2001) Attityden till invandrare och flyktingar efter den 11 september. Forskningsgruppen för Samhälls—och Informationsstudier, Stockholm
FSI (2004) Några frågor om invandringen och invandrare/flyktingar. Forskningsgruppen för Samhälls—och Informationsstudier, Stockholm
Ghersetti M, Levin A (2002) Muslimer och Islam i Svenska Nyhetsmedier: om Rapporteringen av terrorattackerna i USA den 11 September 2001: en Förstudie. Integrationsverkets skriftserie III, Norrköping
Hertzberg F (2003) Gräsrotsbyråkrati och normativ svenskhet: Hur arbetsförmedlare förstår en etniskt segregerad arbetsmarknad. Ph.D. thesis, Stockholm University
Höglund S (1998) Svensk forskning om diskriminering av invandrare i arbetslivet. Swedish Council for Work Life Research, Sweden
Holm HJ (2001) What’s in a name? An ethnical discrimination experiment. Working paper 2000:3. Department of Economics, Lund University
Human Rights Watch (2002) We are not the enemy—hate crimes against Arabs, Muslims, and those perceived to be Arab or Muslim after September 11. United States Country Report 14(6) November 2002, USA
Hvitfelt H (1998) Den Muslimska Faran—Om Mediebilder av Islam. In: Brune Y (ed) Mörk Magi i Vita Medier. Svensk Nyhetsjournalistik om Invandrare, Flyktingar och Rasism, Carlsson
Lange A (2000) Diskriminering, integration och etniska relationer. Integrationsverket, Norrköping
Larsson G (2003) Muslimer i Sverige ett år efter 11 September 2001. Diskriminering, mediebilder, och alternativa informationskanaler in Samtal med svenska muslimer. Situationen för svenska muslimer efter terroristattacken i USA den 11 september 2001. Integrationsverkets rapportserie 2003:3, Norrköping
le Grand C, Szulkin R (2002) Permanent disadvantage or gradual integration: explaining the immigrant-native earnings gap in Sweden. Labour 16(1):37–64
Mortensen D (1986) Job search and labor market analysis. In: Ashenfelter O, Layard R (eds) Handbook of labor economics. Elsevier, Amsterdam, pp 849–920
Rooth D-O (2002) Adopted children in the labour market—discrimination or unobserved characteristics? Int Migr Q Rev 40 (1):71–98
Statistics Sweden (1993) Studie av arbetslösa enligt AMV och enligt SCB: en jämförelse av arbetslöshetsstatistik från SCBs arbetskraftsundersökningar och AMVs sökanderegister, Bakgrundsfakta till arbetsmarknads—och utbildningsstatistiken 1993:5. Statistics Sweden, Stockholm
Statistics Sweden (2002) En longitudinell databas kring utbildning, inkomst och sysselsättning 1990–1999: LOUISE, Bakgrundsfakta till arbetsmarknads—och utbildningsstatistiken 2002:3. Statistics Sweden, Stockholm
Statistics Sweden (2005) Tillsammans—integration i det svenska samhället. Statistics Sweden, Stockholm
Van den Berg G (1994) The effects of changes of the job offer arrival rate on the duration of unemployment. J Labor Econ 12(3):478–498
Vilhelmsson R (2002) Wages and unemployment of immigrants and natives in Sweden. Ph.D. thesis, Stockholm University
We thank the responsible editor, two anonymous referees, Anders Forslund, Peter Fredriksson, Per Johansson, Maria Melkersson, Eva Mörk, Oskar Nordström Skans, Björn Öckert, participants at seminars at IFAU and SOFI, the Conference on Ethnic Minorities, Integration and Marginalisation at the Danish National Institute, and the 2004 Nordic Conference on the Effects of Labour Market Policy and Education for valuable comments. Åslund acknowledges financial support from the Jan Wallander and Tom Hedelius Foundation. A research grant from the Swedish Council for Working Life and Social Research is gratefully acknowledged by Rooth.
Author information
Authors and Affiliations
Corresponding author
Additional information
Responsible editor: Olof Åslund
Appendices
Appendix A: Supplementary tables and figures
- Variable:
-
Definition
- Post-9-11:
-
1 in the period after 9-11, 0 in the preceding period
- Female:
-
1 for females, 0 for males
- Age:
-
Age on Dec. 31, 2000 for post-9-11, on Dec. 31, 1999 for pre-9-11
- Age squared:
-
- Cohabiting:
-
1 if married or cohabitant with children in common (with spouse)
- Kid:
-
Child <18 years old in the household
- Cohabiting*Female:
-
- Kid*Female:
-
- Education dummies:
-
Dummy for highest completed level; see Table 1 for further description
- Country of origin dummies:
-
Birth “countries” as available in the IFAU database; see below
- Days in current spell (interval dummies):
-
Dummies for the following intervals: 0–49, 50–99, 100–199, 200–299, 300–399, 400–
- Days in previous spells (interval dummies):
-
Dummies for the following intervals: 0, 1–99, 100–299, 300–499, 500–999, 1,000–
- Search category:
-
Search category according to HÄNDEL at the sampling date (see Ams 2002). Includes categories 11, 12, and 13 in investigations of the exit to employment. Specifications for entry to unemployment include all categories
Regions and countries of origin
Region | Countries included |
Sweden | Sweden |
Nordic | Finland, Denmark, Norway + Iceland |
Western | GB + Ireland, Germany, Southern Europe (Greece + Italy + Spain + the Vatican + Monaco + Malta + San Marino), Other European countries (Andorra + Belgium + France + Liechtenstein + Luxemburg + the Netherlands + Switzerland + Austria), USA + Canada |
Eastern Europe | Poland, The Baltic states (Estonia + Latvia + Lithuania), Eastern Europe 1 (Rumania + the former USSR + Bulgaria + Albania), Eastern Europe 2 (Hungary + the former Czechoslovakia) |
Former Yugoslavia | Bosnia–Herzegovina, former Yugoslavia (Yugoslavia + Croatia + Macedonia + Slovenia) |
Latin America | Mexico and Central America, Chile, Other South American countries (Argentina + Bolivia + Peru + Colombia + Uruguay + Ecuador + Guyana + Paraguay + Surinam + Venezuela) |
Africa | African Horn (Ethiopia + Somalia + Sudan + Djibouti), Other African countries (all African countries not included in African Horn or North Africa) |
Middle East + North Africa | North Africa + Middle East (Lebanon + Syria + Morocco + Tunisia + Egypt + Algeria + Israel + Palestine + Jordan + South Yemen + Yemen + the United Arab Emirates + Kuwait + Bahrain + Qatar + Saudi Arabia + Cyprus), Iran, Iraq, Turkey |
Asia | East Asia (Japan + China + Korea + Hong Kong + Taiwan), Southeast Asia (Vietnam + Thailand + the Philippines + Malaysia + Laos + Burma + Indonesia + Singapore), Other Asian countries (Sri Lanka + Bangladesh + India + Afghanistan + Pakistan + Brunei + Bhutan + Kampuchea + the Maldives + Mongolia + Nepal + Oman + Sikkim) |
The left column contains the regional grouping used in our analysis. The bold-faced names in the right column indicate the aggregations in the original IFAU data. We exclude “Oceanic” (e.g., Australia and New Zealand) countries from our analysis, since this category mixes very different countries. People from this part of the world constitute a very small part of the Swedish immigrant population |
Appendix B: Discrimination in a standard search model
This appendix presents how discrimination following an attitude change affects the escape rates out of unemployment in a standard job search model (Mortensen 1986). We assume that the attitude change affects two factors: employers’ willingness to hire (the job offer arrival function) and the costs for job search.
1.1 Escape rates and discrimination
In this type of job search model, the escape rate out of unemployment is
where s* is the (endogenous) search intensity, λ(s*) is the function for wage offer arrivals, while [1−F(w*)] is the probability that a wage offer w from the cumulative wage distribution F(w*) is greater than the reservation wage (w*). First, changed preferences and attitudes could affect λ(s*), which means fewer job offers at a given search intensity. We assume that λ(s*)=λs*, where λ can be regarded as an efficiency parameter of search. Further, the cost (of search) function affecting the optimal level of search is assumed to be a simple convex function c(s)ks 2, where k is the parameter that shifts due to increased discrimination.
The next step is to analyze how a change in these two channels of discrimination affects the escape rate out of unemployment. Eq. 3 shows how the escape rate out of unemployment is affected by an increase in wage offer arrivals.
For plausible assumptions regarding the shape of the wage offer distribution (see Van den Berg 1994), Eq. 3 is positive.Footnote 22 This implies that we would expect the escape rate out of unemployment to decrease for discriminated job searchers when employers decrease the availability of jobs to these individuals (the efficiency of search will now be lower).
How the escape rate out of unemployment is affected by an increase in the search cost parameter k is shown in Eq. 4.
The first term in Eq. 4 is the negative direct effect of an increase in the costs of search (ds*/dk is negative), while the second term is an indirect effect that lowers the reservation wage (dw*/dk is negative since the job searcher becomes less “choosy” when costs increase). As in Eq. 3, there are two opposing effects on the escape rate out of unemployment. Even if the sign of Eq. 4 is indeterminate, it is likely that the negative direct effect is greater than the counteracting indirect effect for the same reason that makes Eq. 3 positive.Footnote 23 Then, the hazard rate would decrease when the costs of search increases.
Rights and permissions
About this article
Cite this article
Åslund, O., Rooth, DO. Shifts in attitudes and labor market discrimination: Swedish experiences after 9-11. J Popul Econ 18, 603–629 (2005). https://doi.org/10.1007/s00148-005-0036-9
Received:
Accepted:
Published:
Issue Date:
DOI: https://doi.org/10.1007/s00148-005-0036-9