econstor
Make Your Publications Visible.

A Service of

zbw
Leibniz-Informationszentrum
Wirtschaft
Leibniz Information Centre
for Economics

Endrich, Marek

Working Paper

A Window to the World: The long-term effect of
Television on Hate Crime

ILE Working Paper Series, No. 33

Provided in Cooperation with:
University of Hamburg, Institute of Law and Economics (ILE)

Suggested Citation: Endrich, Marek (2020) : A Window to the World: The long-term effect of
Television on Hate Crime, ILE Working Paper Series, No. 33, University of Hamburg, Institute of
Law and Economics (ILE), Hamburg

This Version is available at:
http://hdl.handle.net/10419/214659

Standard-Nutzungsbedingungen:

Die Dokumente auf EconStor dürfen zu eigenen wissenschaftlichen
Zwecken und zum Privatgebrauch gespeichert und kopiert werden.

Sie dürfen die Dokumente nicht für öffentliche oder kommerzielle
Zwecke vervielfältigen, öffentlich ausstellen, öffentlich zugänglich
machen, vertreiben oder anderweitig nutzen.

Sofern die Verfasser die Dokumente unter Open-Content-Lizenzen
(insbesondere CC-Lizenzen) zur Verfügung gestellt haben sollten,
gelten abweichend von diesen Nutzungsbedingungen die in der dort
genannten Lizenz gewährten Nutzungsrechte.

Terms of use:

Documents in EconStor may be saved and copied for your
personal and scholarly purposes.

You are not to copy documents for public or commercial
purposes, to exhibit the documents publicly, to make them
publicly available on the internet, or to distribute or otherwise
use the documents in public.

If the documents have been made available under an Open
Content Licence (especially Creative Commons Licences), you
may exercise further usage rights as specified in the indicated
licence.

www.econstor.eu



 

 

 

 

 

  

                                                                                                                                                                                        

 

 
 

 

INSTITUTE OF LAW AND ECONOMICS 

WORKING PAPER SERIES 

 

 

A Window to the World: The long-
term effect of Television on Hate 
Crime 
 

Marek Endrich 

 
 

Working Paper 2020 No. 33 
 

 

NOTE: ILE working papers are circulated for discussion and comment purposes. 

They have not been peer-reviewed. 

© 2020 by the authors. All rights reserved. 

February 2020 



A Window to the World:
The long-term effect of Television on

Hate Crime

Marek Endrich1

Institute of Law and Economics, University of Hamburg

07.02.2020

Abstract

This paper analyzes the long-term impact of television on hate crimes in Germany.
In the German Democratic Republic (GDR) foreign television served as a window to the
world and exposed viewers to foreign influences. But certain parts of the GDR were
excluded from receiving Western television due to geographical features. I argue that this
resulted in long-lasting differences in the attitude towards foreigners. Using the spatial
variation in signal strength as a natural experiment, the paper tests the effect of Western
broadcasts on the rate of hate crimes. Municipalities with no access to foreign broadcasts
exhibit a higher degree of xenophobic violence in the period of the migration crisis in
Germany between 2014 to 2017. It shows that media can lead to preference changes that
persist for a long time after the exposure.
Keywords: hate crimes, refugees, natural experiment, media
JEL Classification: J15, K42

1 marek.endrich@uni-hamburg.de. I am grateful to Davide Cantoni for sharing his data on the signal strength
of Western television with me and would like to thank Stephan Michel, Konstantinos Pilpilidis and Stefan
Voigt for helpful comments. I also gratefully acknowledge funding by the German Research Foundation.



1 Introduction

Migration has become one of the main features of modern societies. Nevertheless, there is a

cultural backlash against migrants and they are facing harassment and violence. In Europe, 24

percent of persons with a migrant background state that they were a victim of hate-motivated

harassment in the previous year and three percent experienced an attack due to their foreign

origin.2 These xenophobic attacks entail social and economic repercussions. They decrease

the integration effort of foreigners (Gould and Klor 2015, Steinhardt 2018), worsen the image

of affected regions and can lead to deteriorating economic conditions (Endrich and Michel

2018). Despite these costs, little systematic evidence on the determinants of hate crimes exists.

Peculiar political events such as the terrorist attacks of 9/11 or political upheavals such as

Trump’s presidential campaign in the US seem to trigger surges in violence against foreigners

(Gould and Klor 2015, Mueller and Schwarz 2019). The question remains why some regions

are more prone to outbursts of xenophobic violence than others.

This paper contributes to the debate on the determinants of hate crimes. It demonstrates

how the exposure to foreign television can lead to a lower inclination to commit acts of xenopho-

bic violence. For the analysis, I look at one specific episode of surges in hate crimes, Germany

in the years 2014 to 2017. A large influx of refugees since the beginning of the Syrian civil war

in 2011 was accompanied by the rise of a new right-wing party and xenophobic demonstrations

and attacks.3 Violence against refugees occurred country-wide, but the Eastern regions in Ger-

many, synonymous with the area of the former GDR, were particularly affected (Figure A1). It

continues a pattern of xenophobic violence against foreigners in the Eastern part of Germany

after reunification (Krueger and Pischke 1997).

Economic and cultural reasons are proposed to explain the susceptibility of the Eastern

part of Germany to xenophobic violence. Some argue that the legacy of communism has led to

different preferences in the former GDR compared to former West Germany (FRG) (Alesina

and Fuchs-Schuendeln 2007).4 While political attitudes are predicted to catch up within 20

to 40 years after the reunification, social norms appear more resistant to change (Brosig-Koch

et al. 2011). Economic explanations relate to the turmoil after reunification. Inhabitants in

2 The results come from the Second European Union Minorities and Discrimination Survey (2017). It was
conducted by the European Union Agency for Fundamental Rights.

3 Other European countries such as Austria and Italy experienced similar surges in hate crimes and increases
in the support for far-right parties.

4 Research finds a legacy of the socialist state after reunification on preferences about the welfare state (Alesina
and Fuchs-Schuendeln 2007), solidarity behavior (Brosig-Koch et al. 2011), social trust (Heineck and Süssmuth
2013), personality traits (Pannenberg et al. 2015), education and labor market outcomes (Fuchs-Schündeln
and Masella 2016) and gender roles (Campa and Serafinelli 2019).

1



the area of the former GDR had to endure a period of economic hardship and degradation and

such an experience can change preferences and incite xenophobic behavior (Falk et al. 2011).

How does foreign television play a role? Watching Western TV was a popular leisure activ-

ity in the GDR. The broadcasts covered foreign cultures and foreigners much more prominently

than the national TV stations in the GDR. But geographical features only allowed some parts

of the former GDR to receive the Western TV signal and excluded others from viewing West-

ern broadcasts. I argue that the exposure had a positive impact on the sentiment towards

foreigners. It manifested into long-lasting differences in xenophobia. These subsequently be-

come relevant when anti-foreigner topics gain salience and result in diverging intensities of

anti-refugee violence.

I exploit the exogeneity in the quality of the transmission signal as a natural experiment and

analyze how the exposure to foreign television can lead to differences in xenophobia. Looking at

a cross section of German municipalities in the area of the former GDR, I compile a dataset on

violence against refugees from an online data repository for the years of 2014 to 2017. I find that

regions that were exposed to Western TV show a significantly lower level of violence against

refugees than municipalities without access to Western TV. The result is consistent across the

different kinds of hate crimes, namely assaults, arson attacks against refugee homes and other

attacks against refugee housing. The effect of Western TV is diminished in regions with a

history of xenophobic attitudes. It suggests that a more positive predisposition to foreigners

made viewers more receptive to the media exposure. Further, I find that it’s the degradation

of a municipality but not the economic situation per se that is associated with a higher level

of hate crimes.

The findings complement a literature on the causes of social and political backlashes against

minorities and immigration. Large scale immigration can activate anti-foreigner sentiments

and lead to hostile reactions of the local population (Hangartner et al. 2019). But if there

is frequent contact between locals and refugees, it can also increase tolerance (Gamalerio et

al. 2018, Steinmayr 2016). Xenophobic violence becomes more likely when the cultural distance

between the migrants and the local population is large (Tabellini 2018) and when the influx

leads to a rapid compositional change in the population (Entorf and Lange 2019). An increase

in the visibility of minorities from different ethnicities can lead to political extremism and

assaults (Colussi et al. 2016). But violence against migrants is mitigated when government

efficiency is high and integration is well organised (Ziller and Goodman, forthcoming). The

influx of refugees to Germany between 2014 and 2017 provides one recent event that triggered

anti-foreigner sentiments. I elaborate why this has led to different degrees in the outbursts of

hate crimes across the Eastern part of Germany.

Cultural traits such as xenophobia exhibit remarkable persistence. Cantoni et al. (2019)

2



show how voting results for the NSDAP at the end of the Weimar Republic predict the vote

share of the xenophobic German party AFD 80 years later. Anti-semitic norms have prevailed

over centuries and fuel violence against jews (Voigtländer and Voth 2012). A history of right-

wing voting is a good predictor for the propensity to cast a vote for parties with an anti-

foreigner manifesto (Schwander and Manow 2017). One explanation for the persistence is the

transmission of xenophobic traits across generations (Avdeenko and Siedler 2017). Past events

such as the exposure to Western TV therefore can have long-lasting effects on attitudes.

Economic conditions are the alternative explanation for xenophobic outbursts. Economic

crises can perpetuate racial animus, leading to an increase in hate crimes (Anderson et al. 2018).

Right-wing violence occurs more frequently in times of high unemployment (Falk et al. 2011),

and a perception of relative deprivation can motivate citizens to vote for right-wing extremist

parties (Dorn et al. 2018). Dippel et al. (2015) find that an increase in trade exposure benefits

extreme right-wing parties and that labor market adjustments are the driver behind the effect.

The economic hardship in the eastern part of Germany after reunification and a perception

of refugees as competitors on the labor market and for resources can make them targets of

scapegoating. In the empirics I consider the possible confounding influence of economic factors.

There are only few studies that look at the association of media and hate crimes.5 Ivandic

et al. (2019) find that media reports on terrorist attacks trigger anti-migration sentiments and

motivate local hate crimes against Muslims. The penetration of social networks can instigate

ethnic hate crimes and lead to an increase in xenophobic attitudes (Bursztyn et al. 2019).

Propagations of anti-foreigner sentiments through channels such as twitter can have real world

repercussions and bring about attacks on the targeted minorities (Mueller and Schwarz 2019).

The closest to our study is the work by Ralsmark (2017) who deals with the visibility of

minorities on television and how this can affect social tolerance. She observes the visibility

of people of color in US television and finds that a higher visibility leads to less hate crimes

in the subsequent year. Her argument that the exposure to minority cultures increases social

tolerance is similar to ours, but I measure the long-term effects of such exposure on the attitude

towards a minority group.

5 The literature on media and political preferences is larger and shows how biased broadcasting can affect
voting decisions (DellaVigna and Kaplan 2007) or how entertainment TV leads to larger support for populist
movements (Durante et al. 2019). On media and criminal behavior, violent movies induce a short-term
increase in violent crime (Dahl and DellaVigna 2009). The portrayal of different ways of life on television
can affect fertility (La Ferrara et al. 2012) and the attitude and behavior related to women (Jensen and
Oster 2009). Radio transmissions with nationalistic content can fuel ethnic animosities and, if used as
a propaganda channel, can be instrumental in ensuring public support for dictatorial regimes (Adena et
al. (2015), Vigna et al. (2014)). A government can use mass media as a coordination device in state-sponsored
violence (Yanagizawa-Drott 2014).

3



A few others have used the same natural experiment to analyze the influence of Western

TV on preferences and behavior after reunification.6 Bursztyn and Cantoni (2016) exam-

ine consumption patterns of former GDR citizens and detect that the exposure to Western

broadcasts has moved preferences towards products that were heavily advertised. Hyll and

Schneider (2013) provide evidence that viewing Western television has led to different hedonis-

tic, consumptive and wealth aspirations. The decades of communist experience coupled with

no reception of Western TV can explain persistent differences in financial risk taking and in

the rate of participation in the stock exchange (Laudenbach et al. 2018). Friehe et al. (2018)

are interested in the long-lasting effect on criminal behavior and find that regions with access

to Western TV exhibit less violent crime, sex crime and theft, but more fraud. Our project

deals with a different kind of crime, namely xenophobic violence. The empirical framework will

account for potential dependencies between crimes and xenophobia and for the other channels

of influence.

Two studies argue that Western broadcasts have helped to create a democratic culture and

to develop a political attitude closer to the Western system. Friehe et al. (2019) look at voting

results after reunification and observe that regions with no Western TV reception come with

a higher vote share for extremist parties. The effect is attenuated for parties on the political

right. Hornuf and Rieger (2017) interpret the association of Western television and a larger

vote share for right-wing parties as an indication that Western TV has led to less xenophobia.

Their explanation is in line with my argument. However, this paper offers a more direct test of

the argument. Hate crimes are an act of resentment against foreigners and represent a better

proxy for anti-foreigner sentiment. The motivation for voting for a right-wing populist party

can stem from economic and cultural reasons unrelated to xenophobia. One example is the

expression of discontent with the reigning parties during economic crises (Funke et al. 2016).

Hate crimes are hence more suited for answering the research question.

The paper is organised as follows. Section 2 elaborates on the historical and institutional

background and develops the theoretical conjectures. Section 3 describes the data and sets up

the empirical framework. Section 4 presents the results and section 5 concludes.

6 Others have used the framework to elaborate on the effect of Western TV on behavior in the GDR and
find little evidence for an unstabilizing role during the breakdown of the regime or for triggering effects on
anti-government protests (Kern and Hainmueller 2009, Crabtree et al. 2015)

4



2 Western TV and Hate Crimes - Institutional setting

This part elaborates on the design of the broadcasting system in the former GRD and the

programming with regard to the visibility of foreigners and foreign cultures. I then describe

the development of xenophobia in the former East Germany, offer theories that describe how

television and xenophobia can be related and introduce the setting of the natural experiment.

2.1 Television in the GDR

After World War II, Germany was divided in two blocs that became the Federal Republic

of Germany (FRG) and the German Democratic Republic (GDR). Travelling and personal

interactions between the two entities were frequent in the beginning, but the border closure and

construction of the Berlin Wall in 1961 set up a spatial and mental separation. Communication

across borders became prohibited and the GDR governed the restrictions with an iron fist. As

a result, the trade with the West in cultural goods such as books or music almost completely

stopped. FRG broadcasting was the only exposure of GDR citizens to Western influence.

Television became the window to the western world.

Regular broadcasting in the GDR started after a brief trial period in 1956 under the name

of Deutscher Fernsehfunk (DFF). It initially sent out one program, but got expanded into two

stations in 1969. Television quickly became a mass medium and in the beginning of the 1960s

one million TV sets had been sold. In later periods over 95 percent of households reported

having a television set and for a majority of citizens watching TV constituted their favourite

leisure time activity (Hesse 1988). After 1969 most viewers could choose between at least five

channels, including three from the FRG.

The DFF was a state controlled broadcasting system under the auspices of the governing

party, the Socialist Unity Party (SED). The department of agitation in the central committee

of the SED commanded the political control and often used television as an instrument of pro-

paganda.7 The programming of the television reflected the strong influence of the Soviet Union

on the GDR. The second channel, for instance, often played Soviet movies and documentaries

such as ’For friends of the Russian language’. The state tried to shield its citizens from watch-

ing Western broadcasts and to promote its own stations but remained largely unsuccessful. In

7 The constitution of the GDR contained no right to freedom of information. The DFF’s mission was to
strengthen the fraternal friendship with the Soviet Union in the greater struggle against the West and to
build identification with the socialist state. No independent newspaper or radio broadcasts were available in
the GDR.

5



the regions that received Western TV, a majority preferred the broadcasts from the FRG. The

viewers also had low trust in the credibility of the news broadcast of the DFF in comparison

with Western news (Hesse 1988).

The programming of the DFF covered mainly domestic issues and focused on accomplish-

ments of the GDR and other socialist countries. Coverage of foreign cultures was minor. This

remained the constant principle, even though the competition with Western stations led to a

step wise internationalisation of the program in the 1980s and to the screening of some Amer-

ican and West European movies (Dittmar 2005). Still, even in the 1980s, West German TV

stations exhibited a much larger share of broadcasts with foreign content than the Eastern

counterparts (Hornuf and Rieger 2017).

Reunification led to the dissolution of the DFF in 1991 and the FRG stations became the

official public broadcasting system. The regional stations MDR and RBB that are part of the

Association of Public Broadcasting Corporations provide specific media coverage on the area

of the former GDR.

2.2 Xenophobia in the GDR

Citizens in the GDR had little interactions with foreigners in comparison to the FRG. Travel-

ling to countries outside of the Eastern Bloc was restricted with the exemption for an elite. In

the GDR, foreigners consisted mostly of guest workers or personnel of the Soviet Army. Foreign

workers got recruited on fixed-term contracts from countries such as Vietnam, Cuba and Alge-

ria.8 But personal contacts between locals and foreigners remained sparse. The guest workers

had to live in separate housing, were restricted in their freedom of movement and required to

move back to their home country after the contract ended (Behrends et al. 2003).

After reunification, a surge in hate crimes against foreigners was particularly endemic in the

former part of the GDR (Krueger and Pischke 1997). East German cities such as Hoyerswerda

and Rostock-Lichtenhagen gained prominence for xenophobic riots. Surveys confirm that areas

in the former GDR showed a greater level of anti-foreigner sentiment (Krüger-Potratz 1991).

But the xenophobia had precursors. Despite the self proclaimed image of an anti-fascist state,

hostility against foreigners had been a recurring phenomenon since the beginning of the 1980s.

Evidence is rare as the regime in the GDR suppressed information on xenophobic violence, but

surveys reveal the negative opinion of citizens towards some foreigners. Indeed, the image of

8 At the high point at the end of the 1980s there were 60.000 guest workers from Vietnam and 15.000 from
Mosambik in the GDR.

6



different nationalities corresponded with the portrayal of their nation in the East German state

propaganda of the time (Friedrich and Schubarth 1991).9

2.3 TV and Xenophobia

How can the exposure to Western television affect the attitude of the viewers towards other

nationalities? I rely on the cultivation theory and the intergroup contact theory from a literature

in psychology to explain how television can lead to a dissemination of values in tolerance and

instigate learning processes about foreigners.

According to the cultivation theory, the cumulative exposure to television can change moral

values and the beliefs about the world. Television is said to create an artificial world that

exposes its viewers to different realities (see, e.g. Gerbner et al. (2002)). Viewers distil infor-

mation about different ways of life, use the portrayed characters as role models and absorb the

expressed values into their own moral compass.

If the depiction of foreigners departs from previous experiences, the exposure can affect the

attitude towards foreigners and the perception of their role in a society. The experience of

GDR citizens and the depicted reality on GDR television was void of foreign influence. The

society was homogeneous, the presence of foreigners was deliberately muted and foreign influ-

ence related if at all to life experiences in partner countries of the Eastern Bloc. In contrast,

the popular entertainment and soap operas from the West represented the more diverse society

of the FRG. FRG television frequently showed international imports.10 The broadcasts trans-

mitted a world view that assigned foreigners a prominent role in a society. Viewers of Western

television became more open to foreign influence and the exposure led to an increase in social

tolerance towards foreigners.

Complementary, exposure to others on television can lead to learning processes and mit-

igate biases. The intergroup contact theory tells us that the exposition to other groups can

reduce intergroup prejudice (Williams 1947). Prejudice decreases when anxiety about future

interactions with out-groups is lowered. Just as the personal contact between locals and for-

eigners can undo wrong beliefs and animosities, so can the exposition to other ethnicities on

television decrease anxieties. The unbiased reporting on the minority, free expression and a fair

representation are crucial factors for a tolerance-increasing exposure. Most often such unbiased

9 Information that could harm the GDR regime was commonly kept secret. The survey data by the Leipzig
Institute for Youth Research on xenophobia fell under such a classification.

10 The foreign shows had many viewers in the GDR, and US soap operas such as Denver were among the most
popular broadcasts (Hesse 1988).

7



portrayal are achieved in a political context with a free press (Ayoub and Garretson 2017),

something that stood in stark contrast to the institutional environment in the GDR.

The mission of the GDR television was to strengthen the bond to the communist regime.

It portrayed Western cultures in a bad light and the coverage focused on failures of countries

aligned with the ’class enemy’. The result was a nurturing of hostile attitudes towards foreign-

ers. Western television on the other hand offered a culturally diverse programme with news

and documentaries that often dealt with foreign cultures. At the same time it focused on the

advantages of a globalised western world. One example is the coverage of the experience with

guest workers in the FRG in the 1970s. It was covered in a better light in the Western media.

While the FRG media emphasised their contribution to the economic miracle and painted a

more positive picture of migrants in general, GDR television focused on the problems in the

integration process.

Not all citizens express the same responsiveness to television. The success of propaganda

depends on the predisposition of recipients. If the viewer is negatively predisposed, they are

unlikely to believe the message and can be even dissuaded off the cause. Adena et al. (2015)

show how broadcasting by the Nazi party NSDAP in the 1930s was persuasive in regions with

historic anti-semitism, but led to the opposite effect in regions that were more friendly to jews.

For our case, if a viewer carries a negative predisposition towards foreigners, she will be less

responsive to images on Western TV and disregard new information on foreign cultures. A

positive predisposition induces an openness to new influences and the exposure can persuade

her of the advantages of a diverse society.

The evidence tells that the visibility of foreigners and the reporting on foreign cultures was

much more pronounced and positive in the Western media. I expect, first, that citizens in

regions that received Western TV show a lower inclination to commit hate crimes. Secondly,

the effect of television depends on the predisposition towards foreigners. Regions that are

historically less hostile to foreigners should experience a larger change in attitude.

2.4 Natural Experiment

The partial reception of Western TV in the GDR provides for a natural experiment to test the

effect of television on xenophobic attitudes. Antennas along the FRG and GDR border as well

as from the Western enclave in West Berlin sent a terrestrial signal that transmitted Western

television. The quality of the transmission depended on the distance to the next transmitter

combined with the topological features of the viccinity. Regions surrounded by mountains or

located in a valley were vulnerable to distractions and break offs. Bursztyn and Cantoni (2016)

apply an irregular terrain model that uses information on all antennas that transmitted the

8



main public TV stations in the FRG. They calculate the average strength of the TV signal for

each municipality in the former GDR based on the municipality borders of 1997. I rely on their

measurement and transform it by the population weight to the municipality borders of 2016.

Figure 1 depicts the signal strength on the municipality level. The region around Dresden in

the South East and around Greifswald in the North East received the worst signal. As noted in

many anecdotes, citizens in these vicinities could not view Western TV. I follow Bursztyn and

Cantoni (2016) and use the signal strength of Dresden as the cut-off value for the assignment

into a treatment and control group. TV signal quality falls discontinuously and the quality of

reception does not react substantially to any variation in the signal if it is far from the cut-off

level. Dresden was assumed to be close to the signal strength discontinuity. The Municipalities

with a better signal strength than Dresden are assigned into the treatment group of reception of

Western TV and with a worse or equal signal strength are in the control group of no reception.

Insert Figure 1 about here

The quality of the signal can serve as a natural experiment if the following preconditions

are fulfilled:

(1) GDR citizens viewed Western TV

(2) The treatment and control region are comparable

(3) There was no spatial sorting

(1) GDR citizens viewed Western TV

The first necessary condition for identification is that citizens who received Western TV did

view it. Authorities in the GDR failed to prevent the transmission of the Western broadcast

and after 1973 FRG television even lost its official taboo.11 Watching TV constituted the most

preferred leisure activity in the GDR (Hesse 1988). On average households spent 28.6 hours

per week on viewing television and this was consistent across different types of occupations

(Hanke 1990).

Hesse (1988) conducted a survey with refugees from the GDR on their consumption of

Western television (see Table 1). Among the subjects who lived in areas with Western TV

reception, 82 percent viewed it every day. In regions where the signal was too distorted,

nobody in the sample used the television daily and 53 percent did never watch it.

Insert Table 1 about here

11 In the beginning of the 1960s the GDR authorities planned to jam transmitters or remove parts of existing
receivers to disturb the signal, but the envisaged protest of the public stopped the execution

9



A survey by the Zentralinstitut fuer Jugendforschung among GDR residents supports the

popularity of Western TV. In the GDR districts with Western TV reception such as Magdeburg

and Cottbus, the average respondent watched the broadcasts almost daily. In the district of

Dresden where for most households the signal was too weak, 63.5 percent stated that they never

watched Western television (Hennighausen 2015). The lack of foreign television did not lead

to fewer purchases of TV sets, indicating that the citizens instead consumed GDR television

(Bursztyn and Cantoni 2016).

The treatment and control region are comparable

The second assumption is that under a scenario of no Western broadcasts citizens in the treat-

ment region would be no different to citizens in the control region. In the GDR, the uniform

educational system and little parental influence on children’s development resulted in homoge-

neous individual preferences (Hyll and Schneider 2013). The economic structure varied little

between the treatment and the control region. Both had industrial centers with either Leipzig

or Dresden and included parts characterised by agriculture such as the area around Schwerin

for the control or around Greifswald for the treatment group. A comparison of indicators on the

district level at the beginning of the GDR in 1950 and at the end in 1990 in Table 2 reveals that

the two groups were almost indistinguishable from each other (Bursztyn and Cantoni 2016).

The difference in the employment structure, population density and extent of retail sales does

not reach a significant level and neither does the change over time in the indicators between

the two groups.

Insert Table 2 about here

Socioeconomic characteristics in 1988 such as housing, education or the level of crime and

the divorce rate show a high similarity and support the homogeneity assumption (Kern and

Hainmueller 2009). Political attitudes before the introduction of Western TV did not differ, as

measured by the voting results in 1946.

The homogeneity across the two groups until the end of the GDR suggests an analysis

with data from right after reunification. However, the sample of hate crimes covers the years

2014 to 2017. Even though the similarity until 1990 suits the analysis, the economic turmoil

after reunification hampers the empirical set-up. Mass layoffs and plant closures during the

transition from the socialist to a market economy had massive effects and could have led to

different regional developments. Table 3 presents regional characteristics for the two groups in

the year 2013. The control group fared considerably worse since reunification. It performed

significantly worse in Gdp per capita and in the unemployment rate, its share of high school

drop outs was larger and the loss of inhabitants more pronounced. If these divergences influence

xenophobia, our results on the long-term effect of Western TV could be biased. In the empirical

analysis I mitigate the concern by including a battery of variables that cover for the different

10



economic, political and social developments.

Insert Table 3 about here

Contact theory predicts that interactions with foreigners decrease hostility. Inflows of mi-

grants after reunification could have led to a change in the degrees of xenophobia. But the

effect depends on the favorability of the meeting conditions. While facilitated contact and little

competition for local resources can alleviate negative sentiments, little contact and infrequent

interactions do not lead to changes in attitudes (Gamalerio et al. 2018, Steinmayr 2016). Sur-

veys report that regular contact with foreigners has picked up from a low level but remains far

behind the level in the former Western part (Schmidt and Weick 2016). In the years after re-

unification, many foreigners in the Eastern part were refugees who were allocated to the region.

Language and cultural barriers inhibited communication and interactions remained scarce. Our

sample shows a proportion of only 1.17 percent foreigners for 2011 in the area of the former

GDR compared with 8.6 percent in the Western part. Potential contacts with foreign tourists

remain less likely in the East where in 2013 foreign tourists make up 7.9 percent of all tourists,

or 0.15 foreign tourists per capita, compared with a proportion of 20.7 percent of all tourists or

0.38 foreign tourism per capita in the former West. In the empirics I account for the varying

exposure in the regions to foreign citizens, refugees and tourism.

Spatial sorting

Another confounding factor is the possibility of spatial sorting. In the times of the GDR, viewers

who preferred to watch Western TV could have migrated into regions with good reception.

If they exhibited different xenophobic attitudes, this would bias the results of the natural

experiment. But citizens in the GDR seldom moved. The centrally planned economy was

based on low labor mobility. Regional specialisation limited the mobility of workers and a

shortage in the housing market kept residential mobility at a low level. During the period of

1970 and 1990, only 2.5 percent of inhabitants moved places in a year, and in 1988 only 0.9

percent relocated across county borders (Bursztyn and Cantoni 2016, Hyll and Schneider 2013).

These numbers are three times lower than for citizens of the FRG.

While spatial sorting in the GDR does not hamper with the empirical setting, the strong

surge in migration after reunification can do. Borders opened in 1989, and in both 1989 and

1990 2.5 percent of the East German population migrated to the West (Hunt 2006). They

consisted mostly of younger, well-educated citizens that were to a larger share female (Fuchs-

Schündeln and Schündeln 2009). After the initial surge, the improvement in relative wages in

the former East (Hunt 2006) quickly reduced internal migration (Chevalier and Marie 2017),

though the out-flow increased again in the late 1990s. Importantly, at least for the beginning

years migration rates were statistically similar across the treatment and control regions, a

result that is consistent across the different age groups (Bursztyn and Cantoni 2016). The

11



fact that East Germans continue to show a low inclination to geographical mobility (Boenisch

and Schneider 2013) further mitigates the risk of biases from spatial sorting. In the empirical

framework I take into account the potential effect of differences in population losses. It does

not offer a complete relief from the problem of spatial sorting though as the type of people

that migrated is not observable. A bias from compositional changes could arise if regions with

Western TV experienced less out-migration of people with little xenophobic attitudes or became

attractive destinations for in-migration of such individuals.

3 Empirical specification and Data

3.1 Data

The best data to test the theories presented would be individual information on xenophobic

sentiments across the area of the former GDR. I am not aware of such data that allows a con-

vincing measurement of anti-foreigner sentiments. The project therefore relies on data on the

number of hate crimes collected at the lowest possible administrative level. The municipality

is the lowest administrative unit in Germany, situated below the State and the county level.

In 2016, the area of the former GDR was divided into six States, 80 counties and 2663 munic-

ipalities. It has a population of 12.5 million, excluding Berlin, and the municipalities vary in

size between 34 and about half a million inhabitants. Before reunification, the GDR consisted

of 14 districts and 216 counties and had, excluding East-Berlin, 14.8 million inhabitants.

Germany between the years 2014 and 2017 offers a nice case study on xenophobic violence.

A large influx of refugees since the beginning of the Syrian civil war in 2011 has led to a drastic

increase in xenophobic demonstrations and attacks and the rise of a new right-wing party. The

NGOs Amadeu Antonio Stiftung and Pro Asyl keep track of violence and social unrest aimed

at refugees. They provide an online data repository on all incidents that is freely accessible.

Each incident is manually coded and includes information on the timing, the location of the

event, the source of the information and if applicable the report and the number of victims.

The source of the data are publicly available reports from the media, the police and from state

institutions. The data collection started in 2014.

The xenophobic activities are categorized into assaults, arson attacks, miscellaneous attacks

against refugee housing and xenophobic demonstrations. I aggregate assaults, arson attacks and

attacks against housing into a hate crime indicator. Anti-refugee incidents do not appear in

the database when they cannot be attributed to one of the four categories or when there is

no public information. This means that the data is not a complete representation of all hate

12



crimes against refugee and accounts only for a small part of all right-wing crime.12 There is no

reason to expect differences in the spatial coverage of hate crimes across Germany. Henceforth,

the indicator of hate crimes against refugees represents a good proxy for the overall extent of

xenophobic violence.

Hate crimes against refugees surged in the autumn of 2014, reached a peak in 2016 and

declined afterwards (Figure 2). I collect the data on hate crimes from the data repository

between October 2014 until the end of 2017 and aggregate it on the municipality level. The

sample includes 4,715 cases of miscellaneous attacks, 968 assaults and 272 arson attacks. Figure

3 shows the spatial distribution of hate crimes by municipality for the region of the former GDR,

normalized per 100.000 inhabitant.

Insert Figure 2 about here

Insert Figure 3 about here

The largest rate of hate crimes happened in the municipality of Bad Gottleuba-Berggießhübel

in Saxonia, followed by Guben and Freital. The majority of municipalities did not experience

any hate crimes. One explanation for the zero counts is the fact that the allocation quota of

refugees by the state authorities depends on the number of inhabitants. Municipalities with

low population numbers did not get assigned any refugees. As the presence of refugees and

their housing is a necessary condition for the crimes to take place, I restrict the sample to

municipalities that have more than 5000 inhabitants. This leaves us with 510 municipalities of

which 175 have zero hate crimes as shown in Figure A2. In the empirics I will apply models

that consider the excess number of zeroes.

A municipality is considered as treated if the signal strength of Western TV was better

than the signal strength in Dresden. 460 municipalities belong to the treated group and 50

municipalities to the control group. Regions without access to Western TV experienced on

average 9.53 hate crimes per 100.000 inhabitants. Regions with Western TV exposure had a

lower level of 5.68 criminal incidents, a difference in the rate that is statistically significant. The

different degrees indicate a long-term effect of Western TV on xenophobia. But even though

the regions were homogeneous during the times of the GDR, the different development since

reunification can bias the results and I control for it by using a battery of variables.

12 According to the German Federal Criminal Police, Germany experienced on average 22,345 cases of right-
wing crime per year in the period of 2015 to 2017. The average in the sample from the online repository for
Germany is at 1948 incidents per year over this period.

13



Covariates

The control variables belong to the characteristics of the municipality, the economic and social

circumstances, the presence of foreigners and cultural factors. If not mentioned differently, the

data comes from the Federal Statistical Office of Germany (destatis ) and is on the municipality

level for the year 2013, one year before the upsurge in hate crimes. All data is adjusted by

population weights to the municipality borders of 2016.

The first set of control variables captures the characteristics of the population and the

municipality. I use the density of the geographical unit to estimate if more densely populated

areas exhibit a larger rate of hate crimes. Young people are more prone to commit violent

crimes and I include the proportion of the population below 25 years. The effect of religious

affiliations is ambiguous. On the one hand, the doctrines of the two main religions Catholicism

and Protestantism refer to the necessity to help others in need. It predicts a more welcoming

reception of refugees and less xenophobic violence in religious areas. But intergroup bias theory

tells us that identification with a local religious group can lead to derogations of people who are

part of an out-group (LaBouff et al. 2012). Most refugees came from muslim majority countries.

I control for religiosity by the proportion of the population that are Catholics or Protestants

and the data comes from census data in 2011. It is not possible to account for the intensity of

religious belief, which could be stronger for the minority group of Catholics.

An economic explanation for hate crimes relates to competition on the labor market and

for resources. Locals can regard refugees as competitors especially for jobs with low human

capital requirements. I expect a harsher reaction in regions with less educated inhabitants.

The proportion of high school drop outs covers for the educational attainment and the data is

from the 2011 census and available at the county level. A worse economic situation dampens

the labor market and can lead to a stronger competition for resources. One recurse is violence

against out-groups such as refugees (Krueger and Pischke 1997, Falk et al. 2011). To control

for conditions on the labor market, I include the unemployment rate and the unemployment

rate of inhabitants under 25 which come from the Federal Employment Agency of Germany.

GDP per capita and the share of welfare recipients cover for the general economic conditions

and are provided on the county level.

The well-being depends not only on the personal economic situation but on the economic

and social development of the surroundings. The area of the former GDR faced a large out-

migration to the West. Population loss leads to the depletion of public and private services

in the affected regions and can nurture a feeling of exclusion. I look at the population change

between 1997 and 2013 as a proxy for the relative deprivation of a municipality. There is a large

variation in the change in population, ranging from a loss of 41 percent for the municipality

of Weißwasser to an increase of 123 percent in Glienicke. The occurrence of hate crime can as

well depend on the general level of crime, and crime data on the county level comes from the

14



German Federal Criminal Police. Friehe et al. (2018) show that crime is significantly higher

in regions without former reception of Western TV, and including general crime as a control

allows to distinguish the results from this confounding factor.

Intergroup contact theory predicts that the exposure to foreigners and the quality of in-

teractions can reduce ethnic intergroup prejudice and xenophobia. I use information on the

proportion of foreigners who lived in a municipality before the beginning of the refugee crisis

from the census data 2011. Data on the number of foreign tourists in 2013 is on the county

level and normalized by the population. The unemployment rate of foreigners proxies for the

integration success of foreigners and comes from the Federal Employment Agency of Germany.

A better integration of foreigners leads to a lower probability of negative sentiments. The in-

flux of refugees during the migration crisis led to a rapid compositional change for some regions

which can fuel animosity among the residential population (Entorf and Lange 2019). I use the

change per capita in the number of refugees between 2015 and 2017 on the county level.

Last, cultural traits such as xenophobic attitudes exhibit a remarkable persistence and are

prone to be triggered by events such as the refugee crisis (Voigtländer and Voth 2012, Cantoni

et al. 2019). I control for such legacies by measuring the vote share of the party NSDAP in

the year 1932 (Falter and Haenisch 1990).13 Another historical event that can reflect political

preferences of former GDR citizens are the protests before the break-down of the GDR regime.

Crabtree et al. (2015) provide information on protest events for the year 1989 on the county

level for the former GDR.

Table A1 shows the correlation matrix for the outcome variable and all controls. Our

outcome rate of hate crimes has a strong negative correlation with Western TV. Additionally,

the rate of hate crimes is negatively correlated with population change, while there is a positive

correlation with the unemployment rate and its sub-indicators for the young and foreigners.

The evidence points in the direction of a negative effect of Western TV on xenophobic violence.

But we observe that regions that had no Western TV reception have a higher unemployment

rate and more welfare recipients but also more foreign tourists. If these variables affect the

rate of hate crime, it will distort the evidence on the effect of Western TV. A regression model

helps to take account of the confounding factors.

13 The election of 1932 is regarded as the last election that was conducted under free circumstances in contrast
to the election in 1933 that brought the NSDAP into power.

15



3.2 Empirical specification

The theory predicts that Western TV has a negative long-term effect on the number of hate

crimes. Hate crimes are count data and I normalize them by 100.000 inhabitants to make

the incidents comparable across municipalities. I use a Poisson regression to estimate the

determinants of the rate of xenophobic violence. The Poisson regression is typically used for

count variables, but can account for the rate of incidents. The number of incidents of hate

crime Hm in municipality m between the years 2015 and 2017, depending on the size of the

population nm, is expressed as

Hm(nm) = Poisson(λ̄mnm) (1)

14

The conditional mean of the rate of violence against refugees is given by λ̄ and is specified

as

log(λ̄m) = log(mn) + β0 + β1WesternTVm + X
′
mψ + φb + εm (2)

The main explanatory variable of interest is WesternTV . The dummy variable is equal

to one for a municipality that received Western TV during the times of the GDR and zero

otherwise. The coefficient β1 tells by how much the rate of xenophobic violence differs for

municipalities with Western TV and is the percentage difference in the rate of hate crime. Xm

is a vector of covariates. I include State fixed effects φb to take account of any state specific

factors that could affect the occurence of hate crimes. As the interest is in the rate of hate

crimes, I add the log of population size of a municipality log(mn) as an offsetting factor to the

estimation.

The Poisson model assumes that the conditional variance and the conditional mean of the

dependent variable are equal,

E(H) = V ar(H) = λ (3)

The hate crime indicator shows a dispersion of 3.77. As it is significantly different from

the equi-dispersion at a value of zero, the restrictive condition is not fulfilled. Under over-

dispersion, the Poisson quasi-maximum likelihood estimation provides for a consistent and

14 The Poisson probability specification is given by P(λ) =
e−λmλhmm

hm!
where λm = Xmβ

16



efficient estimator of the conditional mean as long as the conditional mean is correctly specified.

This is true even when the data-generating process is not following a Poisson distribution

(Wooldridge 1999). It estimates a dispersion parameter δ2 in V ar(h|x) = δ2E(h|x) that for
δ2 = 1 is just the assumed variance in the Poisson model. The quasi-maximum likelihood

estimator includes robust standard errors clustered at the municipality level.

The outcome variable hate crime per capita exhibits an excess number of zeroes that can

be larger than the Poisson model predicts (see Figure A2). Zero-inflated models supplement

the basic count models with a binary process that helps to model the excessive number of

zeros. I apply the zero-inflated Negative Binomial model as it takes into account the issue of

over-dispersion.

Large economic and political changes after the breakdown of the Wall can affect the level of

xenophobia. The battery of covariates aims to mitigate the concern. In one extension I include

county fixed effects which provides for a conservative test of our theory.

In order to understand who was most affected by the exposure to Western TV, the following

estimation includes the conditioning effect of xenophobic predispositions. According to the

prediction, the effect should be weaker for regions who exhibited stronger xenophobic sentiments

before the introduction of Western TV. The vote share of the NSDAP per municipality in the

1932 elections proxies for long-term cultural sentiments towards foreigners and is interacted

with the WesternTV dummy.

log(λ̄m) = log(mn) +β0 +β1WesternTVm +β2nsdapm +β3WestTVm∗nsdapm +X′mψ +φb +εm
(4)

The coefficient β3 tells if regions that had a weaker predisposition to xenophobia were more

strongly affected by the appealing effect of Western TV.

4 Results and discussion

The theory predicts that hate crimes occur less often in the treated municipalities that were

able to receive Western TV. Table 5 presents the result of the regressions. Regression (1) starts

with variables for the characteristics of the municipality and step by step I add the controls for

the characteristics of the population and the economic situation (2), the presence of foreigners

and their integration success (3) and long-term cultural factors (4). Across all specifications the

treated municipalities exhibit a significantly lower rate of hate crimes than the control regions.

17



In the fully specified model (4), a municipality that received Western TV experienced a 52

percent lower rate of hate crimes.

Insert Table 5 about here

Municipalities that have a larger proportion of Catholics experience less hate crimes against

refugees. It seems that in the context of the refugee crisis in Germany, a larger role for Catholics

leads to more social tolerance towards refugees. I find limited support for an economic explana-

tion of hate crimes. Economic prosperity measured by gdp per capita, the proportion of welfare

recipients and the general unemployment rate do not explain the variation in xenophobic ac-

tivity. Only the unemployment rate of inhabitants under the age of 25 is positive and weakly

significant, suggesting that some of the young are most strongly involved in hate crimes.

The picture is different for the degradation of a municipality over time. For a municipality

that had to endure a higher loss of inhabitants, the rate of hate crimes is significantly larger. It

suggests that it is not the economic situation per se that leads to repercussions against refugees

but the relative feeling of deprivation and exclusion. The finding that the crime rate per capita

is negatively related to the number of hate crimes underlines the different nature of xenophobic

violence compared to other crimes.

The long-term number of foreigners living in a vicinity, the change in the influx of refugees

and the number of foreign tourism appear to have no relation to xenophobia. Interestingly, the

higher the unemployment rate among foreigners, the stronger the hostility. Tolerance seems

not to depend on the frequency in contacts with foreigners but on the quality of the interaction

and the perception of their integration effort.

The historical voting record for the NSDAP as a proxy for long-term cultural traits has

no predictive power to explain violence against foreigners today. Neither does the number of

protests in 1989 predict the level of hostility towards refugees.

In Table 6 I distinguish between the different categories of hate crimes. The result for the

long-term effect of Western TV is consistent for minor attacks against refugee housing (1),

assaults (2) and arson attacks against refugee homes (3). The treatment regions with Western

TV perform much better with regard to xenophobia than the control regions. The magnitude

of the effect varies between 48 per cent for miscellaneous attacks and 68 per cent for arson

attacks. The effects of the confounding factors vary a bit across the different categories, but

they are mostly in line with the results of the aggregated indicator. Assaults occur more often

in municipalities with a lower share of high school drop outs. Arson attacks happen more often

in less densely populated areas, are committed by populations characterizied by a larger share

of young inhabitants and less welfare recipients and occur in municipalities with lower levels of

foreign tourism.

18



Insert Table 6 about here

Xenophobic demonstrations provide one additional category of xenophobic expressions. Re-

gions that received Western TV exhibit a significantly smaller rate of demonstrations aimed at

refugees, supporting the argument of a lower level of xenophobia. Further, in municipalities

with higher educational attainment xenophobic demonstrations happen more often.

In the next step, Figure 4 presents the marginal effect of Western TV on hate crimes depend-

ing on the level of pre-existing xenophobia. For a municipality with a relatively low percentage

of votes for the NSDAP in the election of 1932, the effect of Western TV is significant and

negative, while for municipalities with a large vote share for the NSDAP Western TV had no or

even a positive effect on hate crimes. Apparently Western TV was only effective in diminishing

the hostility against foreigners in regions with little history of xenophobia. Recipients who have

a stronger xenophobic predisposition got less convinced by the imagery of the West.

Insert Figure 4 about here

As of now the regressions use fixed effects for the States. A more conservative approach

applies county fixed effects for the 76 counties in the region of the former GDR. It allows

to see if municipalities within a county differ in xenophobia along the dimension of Western

TV reception, but comes at the costs of less variation in the outcome variable. Table A2

shows that even under these strict restrictions, the long-term effect of Western TV remains

significant. While the negative effect of a population loss remains significant, the influx of

refugees and the number of foreign tourists now lead to more social tolerance towards refugees.

For the population indicators, areas with better educated inhabitants but plagued by higher

unemployment are associated with a larger number of xenophobic incidents.

The descriptives show that we are dealing with an excess number of zeroes in the ob-

servations on hate crimes. The zero inflated Negative Binomial model explicitly models the

probability that any hate crime is committed. Using the same determinants for both stages,

in Table 7 we can observe that Western TV is a strong negative predictor for the number of

incidents but has no effect on the binary stage of hate crimes. Population loss is again a strong

predictor for the likelihood of hate crimes and for a larger number as is the unemployment rate

of foreigners.

Insert Table 7 about here

19



5 Conclusion

Media can change the attitude towards minorities, affect voting decisions and induce criminal

behavior. This study shows the long-term effect of television on xenophobic violence. Using

the upsurge in hate crimes in Germany during the migration crisis as a case study, we detect

that municipalities that received Western TV in the GDR show a lower level of violence against

refugees than regions that were excluded from the reception. Even though the times of the

GDR are far gone, the long-term exposure to foreigners and foreign cultures on Western TV

has led to long-lasting differences in the attitude towards foreigners. Events such as the large

influx of migrants can act as a trigger to bring out such predispositions.

The observation of such long-term trends merits consideration. One suggestion for improve-

ment is the usage of data on hate crimes from the beginning of the 1990s from right after

reunification. Official information from state authorities or the collection of news reports on

xenophobic violence through a media analysis in line with Krueger and Pischke (1997) and

Endrich and Michel (2018) are the potential sources for such data. An alternative is to work

with data on the individual level from surveys. The analysis provides for a new explanation

for the large differences in the degrees of hate crimes in the area of the former GDR. It can

inform policy makers about locations that are vulnerable to xenophobic outbursts and help in

targeting policies that combat xenophobia.

20



References

Adena, Maja, Ruben Enikolopov, Maria Petrova, Veronica Santarosa, and Ekaterina Zhu-

ravskaya (2015). “ Radio and the Rise of The Nazis in Prewar Germany.” The Quarterly

Journal of Economics 130 (4): 1885–1939.

Alesina, Alberto, and Nicola Fuchs-Schuendeln (2007). “Good-Bye Lenin (or Not?): The Effect

of Communism on People’s Preferences.” The American Economic Review 97 (4): 1507–

1528.

Anderson, Mark, Benjamin Crost, and Daniel Rees (2018). “Do Economic Downturns Fuel

Racial Animus?” mimeo.

Avdeenko, Alexandra, and Thomas Siedler (2017). “Intergenerational Correlations of Extreme

Right-Wing Party Preferences and Attitudes toward Immigration.” The Scandinavian

Journal of Economics 119 (3): 768–800.

Ayoub, Phillip M., and Jeremiah Garretson (2017). “Getting the Message Out: Media Context

and Global Changes in Attitudes Toward Homosexuality.” Comparative Political Studies

50 (8): 1055–1085.

Behrends, Jan, Thomas Lindenberger, and Patrice G. Poutrus (2003). Fremde und Fremd-

Sein in der DDR: zu historischen Ursachen der Fremdenfeindlichkeit in Ostdeutschland.

Metropol Verlag.

Boenisch, Peter, and Lutz Schneider (2013). “The social capital legacy of communism-results

from the Berlin Wall experiment.” European Journal of Political Economy 32:391–411.

Brosig-Koch, Jeannette, Christoph Helbach, Axel Ockenfels, and Joachim Weimann (2011).

“Still different after all these years: Solidarity behavior in East and West Germany.” Jour-

nal of Public Economics 95 (11): 1373–1376.

Bursztyn, Leonardo, and Davide Cantoni (2016). “A Tear in the Iron Curtain.” Review of

Economics and Statistics, no. 1: 25–41.

Bursztyn, Leonardo, Georgy Egorov, Ruben Enikolopov, and Maria Petrova (2019). “Social

Media and Xenophobia: Evidence from Russia.”

Campa, Pamela, and Michel Serafinelli (2019). “Politico-Economic Regimes and Attitudes:

Female Workers under State Socialism.” The Review of Economics and Statistics 101 (2):

233–248.

Cantoni, Davide, Felix Hagemeister, and Mark Westcott (2019). “Persistence and Activation

of Right-Wing Political Ideology.” mimeo.

21



Chevalier, Arnaud, and Olivier Marie (2017). “Economic Uncertainty, Parental Selection, and

Children’s Educational Outcomes.” Journal of Political Economy 125 (2): 393–430.

Colussi, Tommaso, Ingo E. Isphording, and Nico Pestel (2016). “Minority Salience and Political

Extremism.” IZA Discussion Papers, no. 10417.

Crabtree, Charles, David Darmofal, and Holger L Kern (2015). “A spatial analysis of the impact

of West German television on protest mobilization during the East German revolution.”

Journal of Peace Research 52 (3): 269–284.

Dahl, Gordon, and Stefano DellaVigna (2009). “Does Movie Violence Increase Violent Crime?”

The Quarterly Journal of Economics 124 (2): 677–734.

DellaVigna, Stefano, and Ethan Kaplan (2007). “The Fox News Effect: Media Bias and Voting.”

The Quarterly Journal of Economics 122 (3): 1187–1234.

Dippel, Christian, Robert Gold, and Stephan Heblich (2015). “Globalization and Its (Dis-)

Content: Trade Shocks and Voting Behavior.” NBER Working Papers, no. 21812.

Dittmar, Claudia (2005). “Television and Politics in the Former East Germany.” Comparative

Literature and Culture 7 (4).

Dorn, Florian, Clemens Fuest, Lea Immel, and Florian Neumeier (2018). “Inequality and Ex-

tremist Voting: Evidence from Germany.”

Durante, Ruben, Paolo Pinotti, and Andrea Tesei (2019). “The Political Legacy of Entertain-

ment TV.” American Economic Review 109 (7): 2497–2530.

Endrich, Marek, and Stephan Michel (2018). “The good tourist, the bad refugee and the ugly

German: Xenophobic activities and tourism.” ILE Working Paper Series, no. 16.

Entorf, Horst, and Martin Lange (2019). “Refugees Welcome? Understanding the Regional

Heterogeneity of Anti-Foreigner Hate Crimes in Germany.” IZA Discussion Papers, no.

12229.

Falk, Armin, Andreas Kuhn, and Josef Zweimüller (2011). “Unemployment and Right-wing

Extremist Crime*.” Scandinavian Journal of Economics 113 (2): 260–285.

Falter, Juergen W., and Dirk Haenisch (1990). Election and Social Data of the District and

Municipalities of the German Empire from 1920 to 1933. GESIS Data Archive, Cologne.

Friedrich, W., and W. Schubarth (1991). “Ausländerfeindliche und rechtsextreme Orientierun-

gen bei ostdeutschen Jugendlichen: eine empirische Studie.” Deutschland Archiv 24 (10):

1052–1065.

Friehe, Tim, Helge Müller, and Florian Neumeier (2018). “The effect of Western TV on crime:

Evidence from East Germany.” European Journal of Political Economy 55:346–372.

22



Friehe, Tim, Helge Müller, and Florian Neumeier (2019). “Media’s Role in the Making of a

Democrat: Evidence from East Germany.” CESifo Working Paper Series, no. 7485.

Fuchs-Schündeln, Nicola, and Paolo Masella (2016). “Long-Lasting Effects of Socialist Educa-

tion.” The Review of Economics and Statistics 98 (3): 428–441.

Fuchs-Schündeln, Nicola, and Matthias Schündeln (2009). “Who stays, who goes, who returns?”

Economics of Transition and Institutional Change 17 (4): 703–738.

Funke, Manuel, Moritz Schularick, and Christoph Trebesch (2016). “Going to extremes: Politics

after financial crises, 1870–2014.” European Economic Review 88:227–260.

Gamalerio, Matteo, Mario Luca, and Max Viskanic (2018). “Finding the Warmth of Other

Suns? Refugee Reception, Extreme Votes and Hate Crimes.” SSRN Electronic Journal.

Gerbner, G., L. Gross, M. Morgan, N. Signorielli, and J. Shanahan (2002). “Growing up with

television: Cultivation processes.” In LEA’s communication series. Media effects: Advances

in theory and research, edited by J. Bryant and D. Zillmann, 43–67. Mahwah, NJ, US:

Lawrence Erlbaum Associates Publishers.

Gould, Eric D., and Esteban F. Klor (2015). “The Long-run Effect of 9/11: Terrorism, Backlash,

and the Assimilation of Muslim Immigrants in the West.” The Economic Journal 126 (597):

2064–2114.

Hangartner, Dominik, Elias Dinas, Moritz Marbach, Konstantinos Matakos, and Dimitrios Xef-

teris (2019). “Does Exposure to the Refugee Crisis Make Natives More Hostile?” American

Political Science Review 113 (2): 442–455.

Hanke, Helmut (1990). “Media culture in the GDR: characteristics, processes and problems.”

In Media, Culture and Society, 12:175–193. SAGE.

Heineck, Guido, and Bernd Süssmuth (2013). “A different look at Lenin’s legacy: Social capital

and risk taking in the Two Germanies.” Journal of Comparative Economics 41 (3): 789–

803.

Hennighausen, Tanja (2015). “Exposure to television and individual beliefs: Evidence from a

natural experiment.” Journal of Comparative Economics 43 (4): 956–980.

Hesse, Kurt R. (1988). Westmedien in der DDR. Nutzung, Image und Auswirkungen bundesre-

publikanischen Hörfunks und Fernsehens. Verlag Wissenschaft und Politik.

Hornuf, Lars, and Marc Oliver Rieger (2017). “Can Television Reduce Xenophobia? - The Case

of East Germany.” CESifo Working Paper Series, no. 6632.

Hunt, Jennifer (2006). “Staunching emigration from East Germany: Age and the determinants

of migration.” Journal of the European Economic Association 4 (5): 1014–1037.

23



Hyll, Walter, and Lutz Schneider (2013). “The causal effect of watching TV on material aspi-

rations: Evidence from the “valley of the innocent”.” Journal of Economic Behavior and

Organization 86:37–51.

Ivandic, Ria, Tom Kirchmaier, and Stephen Machin (2019). “Jihadi Attacks, Media and Local

Hate Crime.” London, Centre for Economic Policy Research.

Jensen, Robert, and Emily Oster (2009). “The Power of TV: Cable Television and Women’s

Status in India.” The Quarterly Journal of Economics 124 (3): 1057–1094.

Kern, Holger Lutz, and Jens Hainmueller (2009). “Opium for the Masses: How Foreign Media

Can Stabilize Authoritarian Regimes.” Political Analysis 17 (4): 377–399.

Krueger, Alan B., and Joern-Steffen Pischke (1997). “A Statistical Analysis of Crime against

Foreigners in Unified Germany.” The Journal of Human Resources 32 (1): 182–209.

Krüger-Potratz, Marianne (1991). Anderssein gab es nicht. Ausländer und Minderheiten in der

DDR. Waxmann Verlag.

La Ferrara, Eliana, Alberto Chong, and Suzanne Duryea (2012). “Soap Operas and Fertility:

Evidence from Brazil.” American Economic Journal: Applied Economics 4 (4): 1–31.

LaBouff, Jordan P., Wade C. Rowatt, Megan K. Johnson, and Callie Finkle (2012). “Differences

in Attitudes Toward Outgroups in Religious and Nonreligious Contexts in a Multinational

Sample: A Situational Context Priming Study.” The International Journal for the Psy-

chology of Religion 22 (1): 1–9.

Laudenbach, Christine, Ulrike Malmendier, and Alexandra Niessen-Ruenzi (2018). “The long-

lasting effects of experiencing communism on financial risk-taking.”

Mueller, Karsten, and Carlo Schwarz (2019). “From Hashtag to Hate Crime: Twitter and Anti-

Minority Sentiment.” SSRN Working Paper.

Pannenberg, Markus, Tim Friehe, and Michael Wedow (2015). “Let Bygones be Bygones?

Political Regimes and Personalities in Germany.”

Ralsmark, Hilda (2017). “Media visibility and social tolerance: Evidence from USA.” Working

Papers in Economics. University of Gothenburg, Department of Economics, no. 703.

Schmidt, P., and S. Weick (2016). “Einstellungen der deutschen Bevölkerung zu Zuwanderung

von 1980 bis 2016.” Informationsdienst Soziale Indikatoren 19.

Schwander, Hanna, and Philip Manow (2017). “It’s not the economy, stupid! Explaining the

electoral success of the German right-wing populist AfD.”

Steinhardt, Max F. (2018). “The Impact of Xenophobic Violence on the Integration of Immi-

grants.” IZA Discussion Papers, no. 11781.

24



Steinmayr, Andreas (2016). “Exposure to Refugees and Voting for the Far-Right: (Unexpected)

Results from Austria.” IZA Discussion Papers, no. 9790.

Tabellini, Marco (2018). “Gifts of the Immigrants, Woes of the Natives: Lessons from the Age

of Mass Migration.” Harvard Business School Working Papers, nos. 19-005.

Vigna, Stefano Della, Ruben Enikolopov, Vera Mironova, Maria Petrova, and Ekaterina Zhu-

ravskaya (2014). “Cross-Border Media and Nationalism: Evidence from Serbian Radio in

Croatia.” American Economic Journal: Applied Economics 6 (3): 103–132.

Voigtländer, Nico, and Hans-Joachim Voth (2012). “Persecution Perpetuated: The Medieval

Origins of Anti-Semitic Violence in Nazi Germany.” The Quarterly Journal of Economics

127 (3): 1339–1392.

Williams, R.M. (1947). The reduction of intergroup tensions. New York: Social Science Research

Council.

Wooldridge, Jeffrey M. (1999). “Quasi-Likelihood Methods for Count Data.” Chap. 8 in Hand-

book of Applied Econometrics Volume 2: Microeconomics, 321–368. John Wiley & Sons,

Ltd.

Yanagizawa-Drott, David (2014). “Propaganda and Conflict: Evidence from the Rwandan Geno-

cide.” The Quarterly Journal of Economics 129 (4): 1947–1994.

Ziller, Conrad, and Sara Goodman (forthcoming). “Local Government Efficiency and Anti-

Immigrant Violence.” The Journal of Politics.

25



Figure 1: Signal Strength Western TV in the former GDR area

Signal Strength of Western TV reception in the area of the former GDR. Borders are at the municipality level
of 2016.

Table 1: Frequency of Watching Western TV

Frequency Reception of Western TV
possible (in % )

Reception of Western TV
Impossible (in %)

almost every day 82 -
often 12 -
sometimes 5 5
rarely 1 42
never - 53

Survey among 205 refugees from the GDR, 162 from regions with reception of West TV and 43 from
regions with no reception (Hesse 1988).

26



Table 2: Data comparison GDR districts

variable signal mean nonsignal mean dif st.error p value

year 1955
Population density 206 202 4 77 0.959

Share of employed in agriculture 23.7 27.8 -4.1 11.1 0.744
Share of employed in industry 34.1 28.7 5.4 10.0 0.635

Retail sales per capita 1691 1694 -3 102 0.979
Savings per capita 277 297 -20 28 0.544

year 1990
Population density 181 176 5 62 0.941

Share of employed in agriculture 13.5 11.3 2.2 5.1 0.706
Share of employed in industry 33.2 39.5 -6.3 7.5 0.479

Retail sales per capita 7577 7250 327 188 0.190
Savings per capita 9312 9381 -69 928 0.946

Cars per capita (1000) 237.4 237.6 -0.2 12.1 0.992

change 1955 - 1990
Population density -19 -26.2 8.2 15.4 0.626

Share of employed in agriculture -14.5 -12.6 -1.9 6.0 0.778
Share of employed in industry 5 5.5 −0.5 3.0 0.870

Retail sales per capita 5862 5557 305 157 0.142
Savings per capita 8946 8994 -48 770 0.954

The data is on the district level of the fomer GDR and comes from GDR statistical
yearbooks (Krueger and Pischke 1997).

Table 3: Regional Characteristics in 2016 by treatment status

variable signal mean nonsignal mean Difference st.error p value

Unemployment rate 0.12 0.15 -0.04 0.01 0.00
Unemployment rate foreigners 0.24 0.27 -0.03 0.02 0.13

Unemployment rate young 0.12 0.14 -0.03 0.01 0.01
Crime per capita 0.07 0.07 -0.004 0.002 0.03

Density 254.89 390.04 -135.15 59.15 0.03
4refugee per capita 0.01 0.01 0.001 0.0003 0.002

Foreign tourists per capita 0.25 0.48 -0.23 0.06 0.001
GDP per capita 23, 086.03 22, 344.74 741.29 384.59 0.06

Share inhabitants below 25 0.19 0.19 -0.001 0.003 0.65
Share high school drop outs 0.03 0.03 -0.003 0.001 0.02

Vote Share of NSDAP in 1932 0.38 0.36 0.02 0.01 0.01
Number of Protests 1989 3.08 3.12 -0.05 1.16“ 0.97

Population change -0.09 -0.15 0.06 0.02 0.001
Share Catholics 0.04 0.05 -0.02 0.01 0.17

Share Protestants 0.19 0.18 0.01 0.01 0.26
Share foreigners 0.01 0.01 -0.001 0.001 0.37

Share welfare recipients 0.01 0.02 -0.003 0.0004 0.00

Note: The sample consists of municipalities with more than 5000 inhabitants and excludes the area of the former
East Berlin. Treated regions were able to receive Western TV in the times of the GDR. Share of Catholics,
share of protestants and share of foreigners are from the year 2011. Population change covers the percentage
change in population between the years 1997 and 2016

27



0

200

400

600

2014 2015 2016 2017 2018

Time

H
a
te

C
ri

m
es

Figure 2: Development of Hate Crimes

Hate crimes against refugees over the years October 2014 until 2017 for the area of the former GDR. The data
is aggregated on the monthly basis.

28



Figure 3: Hate Crimes per 100.000 inhabitants on Municipality Level, 10/2014 - 2017

Hate crimes against refugees per 100.000 inhabitants at the municipality over the years October 2014 until 2017
for the area of the former GDR.

Table 4

Statistic Mean St. Dev. Min Max

Western TV 0.10 0.29 0.00 1.00
population change −0.10 0.18 −0.41 1.23
population 19,204 43,269 5,004 581,980
density 268.14 317.50 19.27 2,652.28
vote share NSDAP 0.38 0.07 0.12 0.55
unemployment rate 0.12 0.04 0.04 0.31
unemployment rate young 0.12 0.04 0.01 0.29
unemployment rate foreign 0.24 0.15 0.00 1.00
share catholics 0.04 0.06 0.01 0.80
share protestant 0.19 0.09 0.04 0.63
gdp per capita 23,013.35 3,329.50 18,056.00 36,592.00
share welfare recipients 0.004 0.002 0.002 0.01
foreign tourism per capita 0.27 0.25 0.03 1.46
4refugee per capita 0.01 0.003 −0.0005 0.03
crime per capita 0.07 0.02 0.04 0.14
share population <25 0.19 0.02 0.14 0.29
share drop outs 0.03 0.01 0.02 0.05
hatecases 3.99 9.89 0.00 135.00
miscellaneous cases 2.63 6.65 0.00 80.00
arson attacks 0.16 0.51 0.00 5.00
assaults 1.20 3.46 0.00 50.00
demonstrations 0.45 1.80 0.00 29.00

Note: Data on municipalities with more than 5000 inhabitants, 510 observations.

29



Table 5: Cross-Section Quasi Poisson

Dependent variable:

(hatecases)

(1) (2) (3) (4)

Western TV −0.540∗∗ −0.540∗∗∗ −0.698∗∗∗ −0.719∗∗∗
(0.223) (0.195) (0.211) (0.204)

density −0.0002 0.0001 −0.00005 −0.0002
(0.0001) (0.0002) (0.0002) (0.0002)

share population <25 3.117∗ 2.941 2.528
(1.770) (1.808) (1.778)

share catholics −2.467∗ −2.733∗ −3.200∗∗
(1.404) (1.517) (1.468)

share protestants −1.126 −1.129 −0.734
(0.768) (0.824) (0.765)

share drop outs −19.925 −18.477 −17.043
(12.524) (12.378) (12.053)

log(gdppc) 0.885∗ 0.969∗∗ 0.693
(0.455) (0.419) (0.453)

crime per capita −11.715∗∗∗ −11.691∗∗∗ −15.361∗∗∗
(3.462) (3.661) (4.611)

share welfare recipient 12.406 3.049 10.498
(18.277) (20.038) (21.221)

unemployment rate 3.089 1.217 0.988
(2.682) (2.943) (3.026)

unemp. rate young 3.969 4.244∗ 4.569∗

(2.623) (2.572) (2.616)
population change −1.876∗∗∗ −1.574∗∗ −1.702∗∗∗

(0.605) (0.633) (0.588)
share foreigner 3.693 7.021

(7.166) (6.949)
unemp. rate foreigner 1.137∗∗∗ 1.054∗∗

(0.441) (0.429)
foreign tourism per capita −0.253 −0.188

(0.182) (0.174)
difference refugee per capita 6.553 4.745

(12.464) (11.091)
vote share Nsdap −1.533

(1.147)
protests 1989 0.010

(0.008)
Constant −7.694∗∗∗ −17.128∗∗∗ −17.801∗∗∗ −14.244∗∗∗

(0.294) (4.681) (4.307) (4.906)

Observations 510 510 510 510

Note: Standard errors are heteroskedasticity robust and clustered at municipality level.
The unit of observation is the municipality. Significance at ∗p<0.1; ∗∗p<0.05; ∗∗∗p<0.01

30



Table 6: Cross-Section Quasi Poisson. Subcategories

Dependent variable:

(misc) (assault) (arson) (demonstration)

(1) (2) (3) (4)

Western TV −0.651∗∗∗ −0.828∗∗∗ −1.164∗∗∗ −0.802∗∗∗
(0.227) (0.240) (0.340) (0.306)

density −0.0002 0.0001 −0.002∗∗∗ −0.001
(0.0003) (0.0003) (0.001) (0.0004)

share population <25 2.005 2.690 8.511∗ 1.585
(2.190) (2.865) (4.684) (4.706)

share catholics −3.221∗∗ −3.151 −3.260 −1.688
(1.353) (2.699) (2.180) (2.365)

share protestants −0.558 −0.770 −2.007 0.095
(0.866) (1.062) (2.280) (1.487)

share drop outs −14.768 −27.343∗ 12.865 −72.390∗∗∗
(12.907) (15.246) (30.357) (26.222)

log(gdppc) 0.528 0.954 1.615 −1.513
(0.515) (0.588) (0.983) (0.942)

crime per capita −16.938∗∗∗ −13.505∗∗ −7.490 −0.398
(5.022) (6.482) (16.264) (10.139)

share welfare recipients 18.530 16.582 −201.653∗∗∗ −44.331
(23.468) (27.717) (64.090) (71.225)

unemployment rate −1.670 7.595∗ 3.900 −0.571
(3.193) (3.934) (7.975) (6.462)

unemployment rate young 6.884∗∗ −0.612 1.482 9.807
(2.841) (3.329) (6.939) (6.228)

population change −1.621∗∗ −1.997∗∗∗ −0.570 −1.351
(0.691) (0.753) (1.300) (1.128)

share foreigner 8.173 8.940 −19.481 22.589
(7.666) (8.504) (16.152) (19.512)

unemployment rate foreign 1.217∗∗∗ 0.645 1.259 0.946
(0.437) (0.678) (0.846) (0.770)

foreign tourism per capita −0.120 −0.210 −0.889∗ −0.011
(0.192) (0.253) (0.466) (0.328)

4refugee per capita 14.743 −20.851 5.104 9.253
(12.445) (15.975) (34.152) (27.496)

vote share NSDAP −2.016 −0.173 −2.386 −2.490
(1.264) (1.551) (2.668) (2.247)

protests 1989 0.009 0.013 0.013 0.008
(0.010) (0.011) (0.023) (0.014)

Constant −12.969∗∗ −18.344∗∗∗ −25.837∗∗ 6.246
(5.548) (6.264) (10.366) (9.805)

Observations 510 510 510 510

Note: Standard errors are heteroskedasticity robust and clustered at municipality level.
The unit of observation is the municipality. Significance at ∗p<0.1; ∗∗p<0.05; ∗∗∗p<0.01

31



Table 7: Zero Inflated Neg. Bin.: Count Model Coefficients

Dependent variable:

hatecases

(1) (2) (3)

Western TV −0.605∗∗ −0.522∗∗ −0.550∗∗
(0.235) (0.256) (0.235)

density −0.00002 0.0002 0.0001
(0.0002) (0.0002) (0.0002)

share population <25 4.836 5.203
(4.705) (3.469)

share catholics −0.475 −0.965
(0.870) (0.848)

share protestants −0.280 0.208
(0.993) (0.731)

vote share NSDAP 1.134 −1.365
(9.467) (9.409)

log(gdppc) 0.500 0.736∗

(0.450) (0.376)
crime per capita −6.103 −9.386∗∗

(4.691) (4.076)
share welfare recipients −3.817 2.872

(18.525) (19.470)
unemployment rate 4.682∗ 2.332

(2.691) (2.626)
unemployment rate young 2.374 3.307

(2.469) (2.348)
population change −1.954∗∗∗ −1.634∗∗∗

(0.600) (0.572)
unemployment rate foreign 1.618∗∗∗

(0.449)
foreign tourism per capita −0.213

(0.194)
4refugee per capita 10.381

(11.426)
Constant −7.794∗∗∗ −14.434∗∗∗ −16.907∗∗∗

(0.235) (4.803) (3.740)

Observations 510 510 510
Log Likelihood −1,037.883 −967.493 −953.669

Zero Inflation Coefficients

(1) (2) (3)

Western TV −6.461 −4.169 −5.115
(1.929) (8.324) (4.985)

density −0.018∗∗ −0.005 −0.063
(0.008) (0.014) (0.048)

share population <25 −296.121∗∗ −496.645∗∗∗
(149.96) (93.262)

share catholics 110.348∗∗∗ 125.323
(149.46) (87.581)

share protestants 40.976∗ 72.198∗∗∗

(23.203) (16.554)
share drop outs −0.845 −71.082

(527.07) (253.130)
log(gdppc) 6.613 26.718

(37.438 (22.528)
crime per capita 52.090 110.536

(65.974) (116.740)
share welfare recipients −127.854 −141.529

(1260.00) (220.650)
unemployment rate 49.944 −157.653∗∗

(46.560) (66.354)
unemployment rate young −44.365 112.268∗∗∗

(36.652) (31.919)
population change 47.573∗∗ 53.303∗∗∗

(34.153) (14.574)
unemployment rate foreign 25.745∗∗

(10.080)
foreign tourism per capita −18.115

(18.031)
4refugee per capita −85.159

(103.060)
Constant −14.347 −55.637 −209.880

(171.050) (367.630) (216.540)

Observations 510 510 510
Log Likelihood −1,037.883 −967.493 −953.669

32



-4

-2

0

2

0.1 0.2 0.3 0.4 0.5
NSDAP Vote in 1932

E
st

im
a
te

d
C

o
effi

ci
en

t
fo

r
W

es
te

rn
T

V

CI(Max - Min): [3.492, 7.272]

Figure 4: Effect of West TV on Hate crimes by NSDAP vote share

Marginal effects of Western TV on hate crimes against refugees depending on the vote share of the NSDAP in
the 1932 election.

33



6 Appendix

Figure A1: Hate Crimes per 100.000 inhabitants on District Level, Germany

Hate crimes against refugees per 100.000 inhabitants over the years October 2014 until 2017 for Germany. The
data is aggregated on the county level.

34



0

50

100

150

0 50 100 150 200

Hate Crimes per 100.000 p.

N
u

m
b

er
C

o
m

m
u

n
it

ie
s

Figure A2: Histogram of Hate Crimes

Aggregate of hate crimes against refugees over the years October 2014 until 2017 for the area of the former
GDR. The data is aggregated on the municipality level. 510 observations.

35



Table A1: Correlation Matrix 1

Rate Hate Crimes Western TV population change density vote share NSDAP protests 1989 alq alq young alq foreign

Rate Hate Crimes 1 -0.152 -0.205 0.050 0.009 0.038 0.334 0.292 0.220
Western TV -0.152 1 0.096 -0.127 0.086 -0.002 -0.249 -0.168 -0.058

population change -0.205 0.096 1 0.386 -0.348 -0.058 -0.518 -0.323 -0.304
density 0.050 -0.127 0.386 1 -0.330 0.422 -0.050 0.00003 0.033

vote share NSDAP 0.009 0.086 -0.348 -0.330 1 0.011 0.090 0.039 0.056
protests 1989 0.038 -0.002 -0.058 0.422 0.011 1 0.074 0.049 0.100

alq 0.334 -0.249 -0.518 -0.050 0.090 0.074 1 0.874 0.437
alq young 0.292 -0.168 -0.323 0.00003 0.039 0.049 0.874 1 0.358
alq foreign 0.220 -0.058 -0.304 0.033 0.056 0.100 0.437 0.358 1

share catholics -0.063 -0.088 0.069 0.026 -0.212 0.009 -0.080 -0.101 -0.053
share protestants -0.085 0.043 -0.060 -0.237 0.427 -0.069 -0.336 -0.407 -0.064

gdppc 0.005 0.066 0.042 0.219 -0.214 0.363 -0.079 -0.049 -0.082
share welfare recipients -0.015 -0.208 -0.228 -0.175 0.137 0.065 0.338 0.308 0.069

foreign tourism pc -0.052 -0.276 0.192 0.198 -0.151 0.152 -0.014 0.065 -0.088
4refugee per capita 0.041 0.090 -0.004 0.093 -0.055 0.322 0.116 0.158 0.040

crime pc -0.001 -0.081 0.126 0.190 -0.265 0.381 0.217 0.256 0.039
share population <25 0.001 -0.037 0.417 0.387 -0.127 0.364 -0.127 0.001 -0.070

share drop outs 0.039 -0.124 0.155 -0.091 -0.092 -0.107 0.371 0.467 0.0123
6



Table A1: Correlation Matrix (continued)

share catholics share protestants gdppc share welfare recipients foreign tourism pc 4refugee per capita crime pc share population <25 share drop outs
Rate Hate Crimes -0.063 -0.085 0.005 -0.015 -0.052 0.041 -0.001 0.001 0.039

Western TV -0.088 0.043 0.066 -0.208 -0.276 0.090 -0.081 -0.037 -0.124
population change 0.069 -0.060 0.042 -0.228 0.192 -0.004 0.126 0.417 0.155

density 0.026 -0.237 0.219 -0.175 0.198 0.093 0.190 0.387 -0.091
vote share NSDAP -0.212 0.427 -0.214 0.137 -0.151 -0.055 -0.265 -0.127 -0.092

protests 1989 0.009 -0.069 0.363 0.065 0.152 0.322 0.381 0.364 -0.107
alq -0.080 -0.336 -0.079 0.338 -0.014 0.116 0.217 -0.127 0.371

alq young -0.101 -0.407 -0.049 0.308 0.065 0.158 0.256 0.001 0.467
alq foreign -0.053 -0.064 -0.082 0.069 -0.088 0.040 0.039 -0.070 0.012

share catholics 1 -0.092 -0.003 -0.011 -0.038 0.013 -0.095 0.161 -0.064
share protestants -0.092 1 -0.139 -0.200 -0.194 -0.267 -0.466 -0.056 -0.372

gdppc -0.003 -0.139 1 -0.056 0.341 0.239 0.367 0.213 0.013
share welfare recipients -0.011 -0.200 -0.056 1 0.106 0.343 0.388 -0.001 0.502

foreign tourism pc -0.038 -0.194 0.341 0.106 1 0.125 0.351 0.262 0.148
4refugee per capita 0.013 -0.267 0.239 0.343 0.125 1 0.498 0.069 0.300

crime pc -0.095 -0.466 0.367 0.388 0.351 0.498 1 0.117 0.497
share population <25 0.161 -0.056 0.213 -0.001 0.262 0.069 0.117 1 0.053

share drop outs -0.064 -0.372 0.013 0.502 0.148 0.300 0.497 0.053 1

3
7



Table A2: Cross-Section with kreis fixed

Dependent variable:

(hatecases)

(1) (2) (3) (4)

Western TV −0.827∗∗∗ −0.700∗∗∗ −0.712∗∗∗ −0.700∗∗∗
(0.219) (0.226) (0.224) (0.224)

density 0.0003∗∗ 0.0002 0.0002 0.0001
(0.0002) (0.0002) (0.0002) (0.0002)

share population <25 3.369∗ 2.843 3.234
(1.721) (1.840) (1.998)

share catholics −1.083 −1.146 −1.086
(1.909) (1.922) (1.850)

share protestants −0.584 −0.378 0.074
(0.790) (0.820) (0.817)

share drop outs −65.252 −223.173∗∗∗ −226.528∗∗∗
(172.371) (82.041) (77.178)

log(gdppc) 4.870∗∗ 2.937 2.156
(2.156) (2.842) (2.593)

crime per capita −3.427 69.432∗ 58.256
(36.783) (36.507) (35.486)

share welfare recipients −103.138 295.917 381.010
(221.735) (258.298) (233.981)

unemployment rate 11.315∗∗∗ 10.657∗∗∗ 10.737∗∗∗

(2.410) (2.500) (2.608)
unemployment rate young −0.470 −0.857 −0.661

(2.453) (2.487) (2.423)
population change −1.186∗∗ −1.124∗∗ −1.424∗∗

(0.532) (0.533) (0.604)
share foreigner 3.670 4.877

(6.444) (6.451)
unemployment rate foreign 0.574 0.477

(0.395) (0.410)
foreign tourism per capita −9.835∗∗∗ −8.506∗∗

(3.755) (3.721)
4refugee per capita −220.429∗∗ −201.242∗∗

(94.664) (85.287)
vote share NSDAP −2.110∗∗

(0.943)
protests 1989 0.006

(0.007)
Constant −8.162∗∗∗ −56.324∗∗∗ −40.017 −32.355

(0.243) (20.510) (26.217) (23.977)

Observations 510 510 510 510

Note: Standard errors are heteroskedasticity robust and clustered at municipality level.
The unit of observation is the municipality. Significance at ∗p<0.1; ∗∗p<0.05; ∗∗∗p<0.01

38