Parental Unemployment and Young People’s Right-Wing Extremism: Evidence from Germany

Thomas Siedler DIW Berlin and Institute for Social and Economic

Research (ISER)

University of Essex

31 January 2006 Recent years have witnessed a rise in right-wing extremism among German youth and

young adults. This paper investigates the extent to which the experience of parental

unemployment during childhood affects young people’s far right-wing attitudes and

xenophobia. Estimates from three different German data sets and multiple identification

strategies suggest that young people (aged 16-29) who grow up with unemployed

parents are more susceptible to right-wing extremism, and to xenophobia in particular.

This is consistent with classical theories of economic interest and voting behaviour.

They predict that persons who develop feelings of economic insecurity are more likely

susceptible to right-wing extremism and to show anti-foreign sentiments.

JEL codes: C23, D13, J62, Keywords: right-wing extremism, unemployment, intergenerational links, sibling differences, instrumental variables, propensity score matching

1. Introduction

Extremist right-wing ideas, parties and movements are a problem in contemporary

Germany. In a speech on April 10th 2005, Paul Spiegel, President of the Central Council

of Jews in Germany on the Occasion of the National Commemoration of the 60th

Anniversary of the Liberation of the Concentration Camps stated: “Since right-wing

extremist parties have gained or regained seats in the parliaments of Saxony and

Brandenburg, not a week goes by without the right-wing extremist managing to become

the focus of political discussion in Germany. What they are saying is nothing new: open

racism and anti-Semitism are complemented by firing up people’s weariness of the

political discussion in Germany”.1 In the above-mentioned federal state election in

Saxony in September 2004, the extremist right-wing party Nationaldemokratische Partei

Deutschlands (NPD) attracted 9.2 percent of the votes.2 Support among young male

voters was particularly strong with 1 in 5 men aged 25 and younger casting their vote

for the NPD (Statistisches Landesamt des Freistaates Sachsen).3 Germany also

witnesses right-wing violence by youth and young adults, with attacks and assaults

mainly against foreigners, asylum seekers, Jewish synagogues and cemeteries, political

opponents, disabled and homeless (Federal Ministry of the Interior, 2003). Between

1996 and 1999, more than 44,000 extremist right-wing crimes were officially registered

in Germany. Among those, 6.8 percent were violent crimes (Falk and Zweimüller,

2005). By September 15th 2000, the number of right-wing murders in Germany since the

reunification in 1990 is at least 93 and more probably 114 (Edelstein, 2002). The

majority of murders were committed by young people.

What are the driving forces for youth’s right-wing extremism4 in Germany? To

which extent do family background characteristics, such as parental unemployment

1 See <http://www.thueringen.de/de/politisch/veranstaltungen/17531/>. 2 Die Republikaner and the Deutsche Volks Union (DVU), two other extremist right-wing parties in Germany, also attracted many votes in regional elections in both East and West Germany. The Republikaner received 10.9 and 9.1 percent of the votes in Baden-Württemberg in the years 1992 and 1996, respectively. The DVU scored 12.9 percent of the votes in Sachsen-Anhalt in 1998 and 4.9 percent in Hamburg in 1997 (see, for example, http://www.bundeswahlleiter.de). 3 Similarly, nearly 1 in 4 first-time voters cast their vote for the DVU in the 1998 federal state election in Sachsen-Anhalt (Winkler and Falter, 2002). 4 To date, there exists no agreed definition of the term right-wing extremism. Right-wing extremism in this study is defined following the 2003 Annual Report of the Office for the Protection of the Constitution in Germany which states: “Right-wing extremist ideology is characterised by nationalist and racist beliefs and attitudes. It is governed by the idea that ethnic affiliation with a nation or race determines the value of human being. Because, according to right-wing thinking, human and civil rights are subordinated to this criterion, right-wing extremists fundamentally contradict the Basic Law, which accords these rights

1

experience during childhood, impact young people’s xenophobia and right-wing

behaviour?

This paper seeks to shed new light on these questions for young people in

Germany. It examines the extent to which maternal and paternal unemployment during

childhood impact on political far right-wing views, the propensity of feeling close to

extremist right-wing parties, and of belonging to skinhead neo-Nazi groups. Also

examined are young Germans’ prejudices and hostile attitudes toward foreigners and

asylum seekers. In addition, the study examines the role played by, and linkage with,

young people’s feelings of marginalization and economic insecurity, their

dissatisfaction with the political system, and the influence of parents’ political attitudes

and economic expectations.

One of the difficulties in estimating the true impact of parental unemployment

on children’s political outcomes is that the relationship of interest could be driven by an

intergenerational transmission of attitudes towards work and politics, or tastes. For

instance, parents who have a higher risk of being unemployed may also be more prone

to antidemocratic and xenophobic sentiments, and these may affect their children’s

political attitudes, irrespective of parental employment histories. Thus, a positive

relationship between right-wing political views and parental unemployment may be

driven by family characteristics which are unobservable to the researcher, rather than

representing a causal impact.

This study accounts for these potential omitted factors influencing both parental

unemployment and young people’s right-wing behaviour using several econometric

methods and thereby checking robustness. A baseline reference point is produced by

estimates from linear probability models. In a second step, both child random-effect

models and family fixed effect models are estimated by using longitudinal data from the

German Socio-Economic Panel Study. Third, instrumental variable models are applied

exploiting the differences in unemployment levels in the former German Democratic

Republic (GDR) and the Federal Republic of Germany (FRD) both before and after

German reunification in 1990 as an exogenous source of variation in parental

special priority and protection.” As such, the present definition broadly encompasses the five features which are most commonly used in defining right-wing extremism in the academic literature: nationalism, racism, xenophobia, anti-democracy and the strong state (Cas Mudde, 2000).

2

unemployment experience during childhood years. Finally, I apply propensity score

matching methods.

Investigating whether parental unemployment has an impact on young people’s

right-wing behaviour is of substantial interest since the majority of extremist right-wing

crimes in Germany are conducted by young men aged between 15 and 25 years

(Neubacher, 1999). In addition, attitudes and orientations in general are most

susceptible to influences and events during childhood and early adolescence (Krosnick

and Alwyn, 1989). For instance, Wahl (2003) argues that many individuals who were

either suspected or found guilty of extremist right-wing crimes in Germany have

mental-health problems stemming from destructive events during their childhood years.

Furthermore, since attitudes and political values are relatively stable after early

adulthood (Krosnick and Alwyn, 1989), young people’s right-wing extremism today

might have a lasting impact on their political values and behaviour in the future. Finally,

the German economy witnessed a major downturn during the 1990s, accompanied by a

large rise in unemployment rates. As a result, an increasing number of young people

have been growing up with unemployed parents (Statistical Office, 2005).

I find a strong and significant correlation between parental unemployment and

various extremist right-wing outcomes for young people in simple cross-sectional

estimations. Compared to individuals who did not grow up with unemployed parents,

young Germans who experienced parental unemployment during their childhood are

found to have a 5 percentage point higher chance of feeling close to an extremist right-

wing party and are 5-7 percentage point more likely to state that “it would be best if all

foreigners leave Germany”. I document that the relationship between parental

unemployment experience during childhood and young people’s right-wing extremism

is robust to changes in explanatory variables. Furthermore, using individual fixed-effect

and family fixed-effects models, instrumental variable regressions and propensity score

matching methods, I show that there is convincing evidence in favour of a causal effect

from parental unemployment during childhood on children’s right-wing extremism in

Germany.

The paper is organized as follows. The next section briefly discusses relevant

economic and political theories. Section 3 presents the various data sources that are used

in the empirical work. Eight different outcome measures of right-wing extremism are

3

described in section 4, and summarized in section 5. Regression estimates from multiple

identifying strategies are presented in sections 6-9. Sensitivity analyses are discussed in

section 10. The final section concludes.

2. Theoretical background

Theories of economic interest (Lipset, 1960; Downs, 1957) suggest reasons why

parental unemployment may help explain right-wing sentiments and xenophobia.

Individuals whose parents experienced unemployment may perceive employment

opportunities as scarce and develop feelings of socio-economic insecurity. As a

consequence, they may believe that foreigners take jobs away from them. Hence, there

exists a risk that these individuals are more receptive to extremist right-wing

propaganda and feel closer to right-wing parties which claim to protect the interest of

the natives first. For example, one of the electorate slogans of the extremist right-wing

German party Die Republikaner was “Jobs for Germans first!” (“Arbeit zuerst für

Deutsche!”).5

A related argument can be derived from the economic theory of voting

behaviour. This assumes that individuals are self-interested, rational and vote for the

party from which they expect to receive the highest utility. If individuals perceive a loss

in socio-economic status, they might be more prone to sympathises with and vote for

extremist right-wing parties, believing in the parties’ promises to prioritize jobs creation

for natives first.

This paper also borders on work in the political science literature (see, for

example Jaschke, 1994, Roth and Schäfer, 1994, Stöss, 1993). Two political theories

help to explain why family hardship may affect children’s extremist right-wing

sentiments. The first one concerns the hypothesis of a rational protest vote. The idea is

that individuals cast an extremist right-wing vote as a protest against current socio-

economic conditions, dissatisfaction with the political system or disillusionment with

the work of democratic parties. Hence, a person’s statement of support for a right-wing

party could channel for social and economic discontent, triggered by parents’

unemployment, rather than representing a right-wing ideological conviction. A second

theory argues that far right-wing voting is an expression of ideological conviction. The 5 See, for example, < http://www.im.nrw.de/pe/pm2001/pm2001/news_140.htm >.

4

hypothesis is that the support of extreme right-wing parties and organisations is based

on genuine political beliefs and right-wing convictions, rather than the result of

opportunistic protest voting and behaviour. Having these theories in mind is useful in

motivating and interpreting the empirical analysis below.

3. Data Sources

There are three German data sources used in this study. The first is the German Socio-

Economic Panel (SOEP), a panel survey that has interviewed approximately 17,000

individuals in about 6,000 households each year since 1984. In June 1990, following

German reunification, the SOEP was expanded to the territory of the former German

Democratic Republic, including nearly 4,500 individuals in 2,000 new households.

Ongoing representativeness of the population has been maintained by using a following

rule typical of household panel surveys. This study uses the years following the German

reunification from 1990 to 2004 (waves 7-21).6

The second data source is the Youth and Young Adult Longitudinal Study

(YYAL) which interviewed a national sample of youth and young adults in 1993 and

1995/1996. The YYAL collected data from nearly 5400 young people aged 16-30 in

both East and West Germany, including detailed questions about individual’s political

and social experiences and orientations.

The third data source is the DJI Youth Survey, a repeated cross-section survey

which was carried out in 1992 and 1997 on young adults aged 16-29 in East and West

Germany. The survey collected data from approximately 7,000 individuals in both

years. The survey was designed to be representative of the total German population in

that age range and contains extensive information on respondents’ political orientation

and behaviour. The survey also sought responses to questions about prejudices against

immigrants and asylum seekers.

I restrict analysis to native Germans aged 16-29 years. In addition, because

previous studies report significant differences in extremist right-wing crimes against

foreigners between East and West Germany (Krueger and Pischke, 1997), and because

East Germany was oversampled in all three surveys, separate regressions are estimated

for East and West Germany.

6 See Haisken-DeNew and Frick (2005) and Burkhauser et al. (1997) for further details about the SOEP.

5

4. Measures of Right-wing Outcomes

This subsection explains the definitions of the measures of right-wing behaviour used in

this study.7 Using several measures allows me to investigate young people’s right-wing

attitudes and xenophobia from different angles.

The first two variables measure affinity to right-wing parties in Germany, and

participation in extremist right-wing organizations such as skinhead neo-Nazi groups.

They are closely watched by the German intelligence service and are regularly

mentioned in the Annual Report of the Office for the Protection of the Constitution

(Verfassungsschutzberich) in its sections about political extremism.8 Furthermore,

members of skinhead neo-Nazi groups are found to be responsible for many xenophobic

crimes in Germany (Willems et al. 1993).

Leaning towards an extremist right-wing party. Respondents to the SOEP

and the YYAL are asked to which political party they are closest to. The variable takes

the value one if respondents name a right-wing extremist party, and is zero otherwise.9

Similarly, if respondents to the DJY Youth survey indicated closeness to the party Die

Republikaner, the variable was set equal to one, and zero otherwise.

Participation in skinhead neo-Nazi groups. The DJI Youth Survey and the

YYAL Study collect information about participation in and approval of extremist right-

wing groups. In the former, respondents were asked about their attitudes towards

“Faschos\Neonazi\right-wing skinheads”, and other “right-wing groups”. I defined

respondents as participants if they said either that they take an active part in or attended

at times an extremist right-wing group. The YYAL Study asks individuals about their

perception to “skinheads” on a six-point scale. I define participants to be the individuals

that said that they were part of a skinhead group.

Far right-wing political views. In the DJI Youth Survey, respondents were

asked about where they placed themselves on a Left-Right political spectrum. I define

7 The exact wording of the questions is provided in the Appendix. 8 See, <http://www.bmi.bund.de> for further information>. 9 Individuals who said they felt close to “other parties” were excluded from the analysis since they could sympathesis with other extremist right-wing parties in Germany.

6

respondents as having far right-wing political views if they placed themselves in the two

outmost right boxes on the political Left-Right scale.10

Recent studies point to a strong interconnection between right-wing political

ideology and anti-foreign sentiments within the German society (Alba and Johnson,

2003). The remaining outcome variables therefore measure young people’s attitudes and

sentiments toward foreigners and asylum seekers in Germany.

Anti-foreign sentiments. The DJI Youth Survey includes a module of items aiming

to measure respondents’ xenophobia. The following questions, which are worded in

terms of individual’s perceived negative sentiments about foreigners, are examined:

• “If jobs become scarce, foreigners living here should be sent home again”

• “It would be best, if all foreigners would leave Germany”

If respondents state the two strongest expression of agreement on a six-point scale, I set

the dependent variables to one and zero otherwise. Respondents are also asked about

whether they approve or disapprove if: “one shows asylum seekers in a palpable way to

return to their home country”. If young people approve of using violence against

asylum seekers, I define the variable to take the value one and zero otherwise. Attitudes

towards asylum seekers are of relevance in Germany since many extremist right-wing

riots during the early 1990s, such as the pogrom-like assaults in Hoyerswerda and

Rostock, were directed towards asylum seekers. In particular, some of the attacks were

supported by neighbouring Germans who cheered and clapped while the attacks took

place (Krueger and Pischke, 1997).

Right-wing behaviour and xenophobia. Finally, I construct two variables

which are aimed at providing a rigid and strict proxy for (a) right-wing behaviour and

(b) xenophobia. My measure of right-wing behaviour equals one if respondents in the

DJI Youth Survey state leaning towards an extremist right-wing party, participate in a

Skinhead neo-Nazi group and show right-wing political views, and is zero otherwise.11

Accordingly, I define the xenophobia variable as being one if respondents agree with all

three anti-foreign statements described above, and is zero otherwise.

10 Qualitatively similar results are obtained when using a more restrictive measure of right-wing views, i.e. a dummy variable that equals one if the respondents labeled themselves on the outmost right box on the political Left-Right continuum. 11 In the YYAS, the variable right-wing behaviour equals one if respondents lean towards an extremist right-wing party and state participating in a Skinhead Neo-Nazi group, and is zero otherwise.

7

5. Data and Descriptive Statistics

The definition of the main explanatory variable ‘parental unemployment during

childhood’ varies across the three data sets. First, parental unemployment is reported

retrospectively by respondents (aged 16-29) in both the DJI Youth Survey and the

YYAS. Contrary, in the SOEP, parents’ unemployment is self-reported. Second, the

YYAS and the SOEP cover young people’s entire childhood (ages 0-16), whereas the

DJI Youth Survey provides parental unemployment information only in the year

respondents were aged 16. Finally, the period over which parental unemployment is

measured varies too. In the DJI Youth Survey and the YYAS the time period spans the

years 1976-1997 and 1977-1996, respectively. The SOEP also covers more recent years

up until 2003. In case young people grew up in a single parent family, parental

unemployment covers unemployment of the single parent only.12 Differences in both

the definition of the key explanatory variable and the time period over which it is

measured explain the variation in the proportion of young people who grew up with

unemployed parents across the three data sets.

A common set of explanatory variables (in addition to parental unemployment)

can be derived from each survey, including age, sex, year of birth, highest general

schooling degree of the children and parents, a maximum set of year and federal state

dummies and the current local unemployment rate (measured at the federal state level).

The measure of educational qualification has three categories: general secondary school

qualification or less, intermediate school qualification and higher school qualification.

In addition, in the DJI Survey, I also control for father’s occupation (five categories).

Summary statistics are provided in the Appendix, Table A1.

Do the data show any differences in political right-wing outcomes according to

parental unemployment experience during childhood? Table 1 presents summary

statistics for all outcome variables by sample, region of residence and parental

unemployment experience during childhood. The table shows that the proportion of

young people with right-wing outcomes is considerably higher among those whose

parents were unemployed in all but one case. These differences are always statistically

significant (not shown in table). This is equally true for young people living in East and

West Germany. For example, in both the YYAS and the DJI Youth Survey, the

8

proportion of young persons participating in skinhead neo-Nazi groups is considerably

higher among those who experienced parental unemployment in their childhood.

Similarly, more than 40 percent of young people who experienced parental

unemployment agrees that foreigners should leave Germany if jobs become scare

compared to 25 percent of young people whose parents were not unemployed during

their childhood. Overall, the proportion of young people with right-wing attitudes and

xenophobic sentiments is higher in East Germany than in West Germany.

6.1 Parental unemployment and Young People’s Right-Wing Extremism

I start by estimating OLS linear probability regressions of the form13:

r = f +t + uβ + Xγ + e, (1)

where r represents a right-wing extremist outcome variable for each individual, f are

federal state dummies, t are year dummies, u is a dummy variable equal to one if a

person experienced parental unemployment during childhood and zero otherwise and X

is a vector of child- and family-specific variables. The parameter of most interest is β,

which captures an association between growing up with an unemployed parent and

young people’s propensity to show right-wing extremist attitudes later in life. The year

dummies t capture common shocks to young people’s right-wing behaviour, the state

dummies f control for regional differences and the error term e captures all other

omitted factors.

Table 2 presents the baseline results. For brevity, only the estimates of β are

reported. Panel A shows a positive significant association between experience of

parental unemployment and children’s right-wing extremism for 11 out of 12 outcome

measures. In the majority of regressions, the point estimates are larger in magnitude and

significance for young people living in East Germany compared to those living in West

Germany. In contrast, the impact of parental unemployment on individuals’ xenophobia

is larger for young persons living in West Germany (Panel B). Note, however, that there

exists a positive and significant intergenerational link in all six regressions. The largest

point estimates are obtained for the outcome variable “if jobs become scarce, foreigners

12 Note that both the DJI Youth Survey and the YYAS comprise both mother and father-only families, whereas young people who grew up in father-only families were excluded from the SOEP sample. 13 I present linear probability models because of the difficulties in conducting instrumental variable estimates for probit models.

9

living here should be sent home”. Individuals who grew up with an unemployed parent

show a 6-11 percent higher likelihood to agree with this statement. Similarly,

individuals are 5-7 percent more likely to state that it “would be best if all foreigners

leave Germany” and show a significant higher propensity to approve of using violence

against asylum seekers if they experienced parental unemployment during childhood.

Finally, Panel C reports the point estimates for the two outcome variables Right-wing

behaviour and Xenophobia. There exists a statistically significant correlation between

parental unemployment experience during childhood and Right-wing behaviour for

young people living in East Germany, but not in West Germany. Contrary, the

association with people’s xenophobia is positive significant for young people living in

East and West Germany.

These estimates capture both the direct and indirect effect of parental

unemployment during childhood on children’s right-wing extremism. For example,

parental unemployment is likely to correspond to lower household income, which might

also influence young adults’ political behaviour. Overall, the results suggest a

significant positive association between parental unemployment experience during

childhood and various extremist right-wing outcomes for young people in Germany.

6.2 Alternative Explanations why Parental Unemployment affect Young People’s

Right-wing Extremism

Having determined that parental unemployment is positively associated with young

people’s right-wing extremism I now examine reasons why this might be the case. I

consider four hypotheses. The first is that young people who grow up with an

unemployed mother or father might be more likely to be unemployed themselves later

in life (O’Neill and Sweetman, 1998) and their political believes might be influenced by

their own labour market prospects. Previous empirical research points to a positive

correlation between being unemployed and the propensity to cast an extremist right-

wing vote or to commit right-wing crimes (Lubbers and Scheepers, 2001; Bacher,

2001). To explore the robustness of the baseline results, I examine how young people’s

own unemployment affects their right-wing attitudes. The explanatory variables in the

10

baseline regressions are now supplemented by a dummy variable that equals one if the

young person was unemployed when interviewed, and is zero otherwise.

Second, I examine the potential direct influence of young people’s feelings of

socio-economic deprivation and insecurity. If parental unemployment is a measure of

general economic uncertainty rather than of economic hardship during childhood, the

positive correlation between parental unemployment and right-wing outcomes should

weaken. To test this hypothesis, I control for young people’s financial worries, and

whether or not they feel being disadvantaged by the German reunification. In addition,

in the regressions based on the SOEP, I can control for whether parents’ are currently

worried about their financial situation.

Right-wing extremism cannot narrowly be interpreted as a ‘youth problem or

phenomenon’ rather it should be seen in wider, general population context. Koopmans

and Olzak (2004) point out that public reactions to, and opinions in the population

about, right-wing extremism play a crucial role in encouraging or discouraging right-

wing violence. The third hypothesis is that parents’ political beliefs play a key role in

determining children’s political attitudes. To the extent that the influence of parental

unemployment on right-wing outcomes is diminished by the inclusion of variables that

capture parents’ political beliefs, one could conclude that parental political orientations

– rather than their unemployment – are the driving force behind young people’s right-

wing extremism. This is tested using SOEP data. Two proxy variables are used for

parents’ political views: parents’ affinity to an extremist right-wing party and whether

parents indicate of being in favour of Democracy.

It is commonly believed and argued that right-wing extremism is closely

associated with disillusionment with the political system (Backer, 2000). To investigate

the effects of young people’s dissatisfaction with established parties, politicians and

political institutions, covariates capturing respondents’ disaffection with politicians and

the government are included in the regressions. Political dissatisfaction is proxied by

whether respondents think that politicians trick the people (DJI Youth Survey) and

whether they report having no or low levels of trust into the government (YYAS).

Table 3 investigates the robustness of the baseline results by progressively

adding in further covariates that might lie behind the positive association between

parental unemployment and right-wing extremism. For the sake of brevity, I report

11

results for one outcome only. The results for the other right-wing measures were

broadly similar to the results reported in Table 3 and are available from the author upon

request. Panels A and B show the estimates obtained for young people living in East and

West Germany, respectively. I start by investigating how young people’s own

unemployment affects their right-wing party affinity. The results are mixed. Overall,

there exists a positive significant association for young people in East Germany, which

varies considerably in magnitude among the three data sets. The link is less clear cut for

those living in West Germany. The key finding is that the correlation between parental

unemployment during childhood and young people’s right-wing extremism remains

robust. In the majority of regressions, the point estimates do not change in significance

or magnitude.14

Column (2) also controls for young people’s current worries about their financial

situation. Columns (3) further add proxies for parents’ financial worries (SOEP) and

young people’s feelings of being disadvantaged by the German reunification (DJI Youth

Survey). The results shown in Table 3 paint a picture consistent with the economic

interest theories. Indeed, being worried about or dissatisfied with the current financial

situation, and having feelings of being disadvantaged by the German reunification are

positively correlated with individuals’ propensity to lean towards an extremist right-

wing party. This association is statistically significant in the majority of regressions.

Including these additional variables into the regressions does not alter the association

between parental unemployment and the right-wing outcome measure under study,

however.

Using data from the SOEP, columns (4) and (5) present results for an

intergenerational link in political attitudes. The estimates suggest that having a parent

who ever responded to feel close to an extremist right-wing party is positive associated

with young people’s closeness to these extremist parties, and this effect is very strong.

On the other hand, there is no significant link between parents’ expressed democratic

values and their children’s affinity to an extremist right-wing party. Again, the

association between parental unemployment during childhood and young people’s

outcome under study remains robust and the point estimates decrease only slightly in

14 I have also controlled for individuals’ unemployment duration (measured in months). The results were insensitive to the inclusion of this additional variable.

12

magnitude in the West German sample and remains unchanged in the East German

sample.

Finally, columns (6) add controls for young people’s current disaffection with

the political system. In all regressions, there is a positive and statistically significant

association between having no trust into politicians or the government and young

people’s closeness to extremist right-wing parties in Germany.15 This result is

consistent with Backer (2000). Again, the correlation between parental unemployment

and the young people’s closeness to a right-wing party remains unchanged. Taken

together, the results suggest that the association between parental unemployment during

childhood and young people’s right-wing extremism is unlikely to be due to

individual’s own unemployment, their current financial worries, disaffection with the

political system or their parents’ financial worries and political beliefs.

To further examine the potential influence played by socio-economic

deprivation, I distinguish between young people who experienced unemployment of

both parents during childhood and those experienced only mother’s or only father’s

unemployment.16 One would expect a stronger intergenerational association with

extremist right-wing outcomes if both parents were unemployed as this is likely to result

in higher levels of material deprivation and feelings of socio-economic insecurity.

Furthermore, since fathers have higher labour earnings than mothers on average, one

might hypothesise that paternal rather than maternal unemployment had a larger impact

on children’s right-wing extremism. Table 4 presents results for two outcome measures:

leaning towards an extremist right-wing party and participation in skinhead neo-Nazi

groups. There is a strong correlation between family hardship and the right-wing

outcomes under study when both parents were unemployed during childhood in

regressions based on the YYAL and DJI Youth Survey. In the majority of regressions in

Table 4, the association is larger in magnitude compared to those who experienced

unemployment by one parent only. No clear cut evidence emerges with respect to the

second hypothesis. In two regressions (SOEP and YYAS), it is father’s unemployment

which seems to play a stronger role than mother’s unemployment. In regressions based

on the DJI Youth Survey, it is maternal rather than paternal unemployment during

15 I also controlled for individuals’ distrust into the parliament in regressions based on the DJI Youth Survey. This did not change the results. 16 Due to small sample sizes, regressions are pooled for respondents living in East and West Germany.

13

childhood which has a stronger impact in magnitude and significance on right-wing

attitudes.

In sum, the results so far give a broad picture of the association between parental

unemployment experience and various right-wing outcomes for youth and young adults

in Germany. However, if parental unemployment and child right-wing extremism are

correlated with some unobservable characteristics, then the linear probability estimates

will be biased. In the following sections, I therefore present several econometric

approaches which help to unravel a causal intergenerational effect – random effect panel

and sibling fixed effect estimates, instrumental variable estimates and propensity score

matching – and discuss the results on the basis of the different identifying assumptions.

7. Panel data estimates

Using panel data from the SOEP has several important advantages over cross-sectional

surveys. First, the panel data structure allows one to control for unobserved family-

specific and child-specific effects. Second, since parental unemployment is self-

reported, this is likely to result in lower levels of measurement error than if reported by

their children retrospectively. Third, using parents’ retrospective employment histories

enables me to measure parental unemployment over young people’s entire childhood as

opposed to a single point (or year) in time (Wolfe et al. 1996). Fourth, the SOEP data

allows me to control for other family background characteristics, such as family

structure (growing up in a non-intact family; whether or not person is an only child and

birth order). Finally, mother’s and father’s employment status is observed not only

during individuals’ childhood years (ages 0-16), but also at later ages, so one can

measure parental unemployment over a longer period. This allows me to examine the

impact of parents’ unemployment during adolescence on right-wing extremism.

According to the impressionable years hypothesis (Krosnick and Alwin, 1989), parental

unemployment during adolescence should have a smaller financial and psychological

impact on young people as compared to parents’ unemployment during childhood.

Table 5 presents the effect of experience of parental unemployment during

childhood and ever experienced parents’ unemployment on the propensity to lean

towards an extremist right-wing party by using data from the SOEP. Panel A shows the

impact of parental unemployment during childhood (ages 0-16), Panel B reports the

14

impact of parental unemployment both during childhood and at later ages on right-wing

party affinity. Again, results are reported separately for young people living in East and

West Germany. The first two columns present results from linear probability models

which include additional control variables such as birth order and mother’s marital

history during childhood. The remaining columns report estimates from the following

two estimation methods which account for the potential endogeneity of parents’

unemployment: individual random-effect models and family fixed effect linear

probability models.

The first method uses all person-year observations for which young people’s

party affinity is observed more than once. This method controls for any unobserved

individual-fixed effects which are assumed to be uncorrelated with the explanatory

variables. In this case, the random-effect estimates are consistent if parental

unemployment is the only endogenous variable. The second method uses the sub-

sample of young people in which at least two siblings of the same mother were

interviewed. This econometric method controls for unobserved environmental or genetic

characteristics which do not change over time. Although family fixed effect models are

not a panacea against unbiasness, they are very useful in controlling for the effects of

family characteristics which may have a strong impact on children’s right-wing

behaviour later in life (Ermisch and Francesconi, 2001). For example, young adults who

grow up with xenophobic parents (which is unobserved in the SOEP) might be more

likely to lean towards an extremist right-wing party, irrespective of parental

employment history. Assuming that parents’ anti-foreign sentiments remain stable over

time, family fixed effect models help to control for family specific influences which

might have an impact on both parents’ unemployment propensity and their children’s

political attitudes.

I start by discussing the results in the first row in Panel A, Table 5. The first two

columns show that the association between parental unemployment and each of the

right-wing outcomes is robust to including further covariates and are in line with those

reported in Table 3. The random effect models also suggest a positive and significant

impact of parental unemployment during childhood on the right-wing outcome measure

under study. In contrast, sibling difference estimates are positive and statistically

significant for young people living in East Germany only. The impact is large,

15

suggesting that East Germans who grow up with an unemployed parent have a 11

percentage higher risk to lean towards an extremist right-wing party compared to those

in East Germany whose parents were not unemployed.

The remaining columns in Table 5, Panel A report whether the impact of

parental unemployment varies with the age at which it was experienced. To this means,

I break down the most recent parental unemployment into two mutually exclusive

childhood stages: early childhood (birth to age 12) and late childhood (ages 13-16).

From the theoretical considerations in section 2, it became apparent that right-wing

extremism might be triggered by parental unemployment, channelled through feelings

of deprivation and socio-economic insecurity. This implicitly implies that young people

were aware of their parents’ unemployment, which is more likely at older childhood

age. Hence, one would expect an impact of parental unemployment at late childhood

rather than at early childhood. Consistent with this hypothesis, Table 5 shows a positive

significant impact of parental unemployment during late childhood on right-wing party

affinity in the majority of regressions. No significant effect appears for parental

unemployment during early childhood.

The results for parental unemployment experience during childhood and

adolescence are shown in Panel B. I start by discussing the results in the first row of

Panel B. With the exception of the sibling difference estimates for individuals living in

West Germany, the point estimates from the various estimation methods are smaller in

magnitude compared to the corresponding ones in Panel A. This suggests that parental

unemployment during late childhood – rather than at ages 17 and above – plays a

crucial role on young people’s right-wing extremism.

Taken as a whole, the results in Table 5 provide first evidence for Germany that

the positive link between parental unemployment and young people’s closeness to an

extremist right-wing party might be causal. Furthermore, they indicate that parental

unemployment during late childhood (ages 13-16) plays a crucial role on children’s

affinity to an extremist right-wing party.

8. Instrumental variable estimates

To further investigate a causal effect of parental unemployment during childhood on

young people’s extremist right-wing outcomes, instrumental variable regressions are

16

estimated. As a source of exogenous variation in parents’ unemployment, I exploit the

stark differences in unemployment rates between the former German Democratic

Republic and the Federal Republic of Germany both in the years before and after the

German reunification in 1990. The GDR had a centrally planned economy: prices,

production and allocation of resources were all planned by the state and measured

unemployment was virtually non-existent. In the years following the German

reunification, unemployment levels surged in the former GDR, reaching 19.5 percent in

2002 (ZUMA, 2005). In contrast, the FRG had a free market economy, with

unemployment levels in the range of 3.8 to 9.3 percent during the 1980s, and ‘more

moderate’ increases in unemployment levels in the years following the fall of the Berlin

Wall compared to East Germany.17

To clarify this, Figure 1 shows the proportion of young people growing up with

an unemployed parent over time by using data from the YYAS, separately for

individuals living in East and West Germany. The vertical line represents the year of

German reunification in 1990. The figure shows (1) that the proportion of young people

with unemployed parents during childhood was lower in the GDR before 1989

compared to the FRG, and (2) that unemployment rates increased more dramatically

within 1-2 years after reunification in East Germany than in West Germany. These huge

differences over time and across both parts of Germany provide an exogenous source of

variation for parental unemployment experience. The IV estimation strategy uses an

identification strategy which is similar to a difference-in-difference approach. The

regression model consists of equation (1) as well as:

up = βo + β1childhoodGDR + xitα3 + ν (2)

The equations are estimated by Two Stage Least Square (2SLS), whereas equation (2)

represents the first-stage equation and the variable childhoodGDR serves as an instrument

for parental unemployment during childhood.

The first measure of childhoodGDR is the number of years an individual lived in

the GDR during childhood (ages 0-16). The idea is that the longer a young person spent

her childhood in the former GDR, the less likely it is that she experienced parental

17 The unemployment rate in West Germany in 2002 was 8.7 percent.

17

unemployment during childhood years. Results of the first-stage regressions are shown

in Table 6. As expected, there exists a negative significant association between the

number of years a person spent in the former GDR and her chances of growing up with

an unemployed parent across all three data sets. The first-stage relationships are very

strong, with t-statistics in the range of 7-16, and a lowest F-statistic of 56.5, suggesting

that the instrumental variable two-stage least squares estimates are unlikely to be biased

toward ordinary least squares (Staiger and Stock, 1997). Similarly, the partial R2‘s are

large, also implying that the instruments are strong (Bound, Jaeger, and Baker, 1995).

The estimates suggest that having lived an additional childhood year in the former

socialist system decreases the chances of experiencing parental unemployment during

childhood by around 5-7 percentage points.

Pooled OLS results for young people living in Germany and 2SLS estimates are

presented in Table 7, separately for the three data sets. For the DJI Youth Survey,

results from two different 2SLS estimations are presented. As with earlier OLS

estimates (Table 2), there is evidence that parental unemployment has a positive and

statistically significant association with the majority of right-wing outcomes.

Most of the first 2SLS results (IVI) are positive and statistically significant. With

two exceptions, the 2SLS point estimates are significantly higher than the corresponding

OLS results. Assuming randomness in parental unemployment experienced during

childhood, this indicates a causal intergenerational impact of parental unemployment on

children’s right-wing extremism. The last column in Table 7 shows results from the

second instrumental variable strategy (IVII). Throughout, this improves the precision of

the estimates slightly in comparison to the first IV strategy. The coefficients of the first

three outcomes decrease and, as a result, the coefficient on the outcome variable Right-

wing political views becomes insignificant. The second IV approach allows me cast

some light on whether the instruments are orthogonal to the error term. Results of the

test are presented in the last column in Table 7. No stark differences emerge between

the two IV approaches with respect to the outcome variables capturing individuals’ anti-

foreign sentiments. Taken together, the results in Table 7 provide further evidence of a

significant causal relationship running from parental unemployment experience during

childhood on young people’s right-wing extremism.

18

Several explanations are suggested in the literature why IV estimates might yield

in larger point estimates than corresponding OLS results: (1) measurement error in the

explanatory variable parental unemployment and (2) heterogeneity in the effect of

parental unemployment experience across the population of young people.18 The first

issue occurs if measurement error, which results in downward bias of OLS estimates,

outweighs potential upward bias due to omitted variables.19 As discussed in section 5,

the definition of the main explanatory variable varies among the three data sets. In

particular, since unemployment of parents is reported by their children in both the DJI

Youth Survey and the YYAS, measurement error in form of recall errors is more likely

in both surveys compared to the SOEP. Moreover, in the DJI Youth Survey, parental

unemployment is only observed in the year young people were aged 16. This might

further increase the problem of measurement error in the OLS estimates based on the

DJI Youth Survey. Indeed, the difference between the OLS estimates and the

corresponding IV results are larger in the DJI Youth Survey than in the other two

samples for the outcome variables leaning towards extremist-right wing party and

participation in skinhead neo-Nazi group. Finally, retrospective employment histories

from the SOEP not only indicate that unemployment levels were lower among parents

in the former GDR compared to parents living in the FRG, but also that unemployment

duration was shorter on average. Hence, the instrumental variables might not only

explain differences in the incidence of parental unemployment during childhood but

also for differences in parental unemployment duration between East and West

Germany. Another reason for larger IV results might be due to the fact that these

estimates represent the average effect of a particular and selected group of young people

only. The impact of parental unemployment experience on young people’s right-wing

extremism is likely to be larger among children from low educated, low income

families. In other words, the IV estimates capture the marginal effect for the young

people who are affected by the instrument, the so called local average treatment effect

(Angrist, Imbens and Rubin, 1996), whereas unbiased OLS estimates (i.e. in the absence

of measurement error and omitted variable bias) correspond to the population average

treatment effect (ATE). Hence, the present IV estimates capture the impact of parental

18 See, for example, Card (2001) for a recent survey of the returns to schooling literature. 19 In particular, this might be of relevance for the results using DJI Youth Survey since parental unemployment is measured during late childhood (age 16) only.

19

unemployment on young people’s right-wing extremism among a sub-group of the

population of young people who would have grown up with unemployed parents had

they not lived in the former GDR. I argue that these are primarily individual’s with low

income, working class parents who would have had a higher risk to experience parental

unemployment during childhood when growing up in a free market economy rather than

in a socialist, planned economy. In fact, in the years following German reunification,

East Germans with no or low levels of education faced the highest risk of becoming

unemployed.

9. Propensity score estimates

The matching method followed here has two advantages over the previous IV approach

(Blundell and Costa Dias, 2000). First, in the IV estimations randomness of parental

unemployment is assumed to guarantee that the instrument is not related to the

outcomes other than through parental unemployment experience. As such, the IV

method extracts from potential selection problems. Second, the above 2SLS procedure

is only valid under the assumption that young people with given observable

characteristics respond identically to parental unemployment experience during

childhood (Heckman, 1997), assuming homogenous treatment effects. The propensity

matching methods allows for heterogeneous reactions to treatment across individuals by

accounting for potential selection process. The propensity score method compares

treated individuals (who experienced parental unemployment during childhood) with a

comparison group of young people who did not grow up with unemployed parents and

have observable characteristics which are similar to those of the treated.

The key parameter of interest is the so-called average treatment effect on the

treated (ATT). The ATT here measures the impact of parental unemployment on right-

wing extremism among those who actually experienced mother’s or father’s

unemployment.

Table 8 presents ATT estimated using propensity score matching. Columns (1)

show estimates from local linear regressions, columns (2) contain results from bivariate

Kernel estimations. Independently of the data set or matching method, all the estimates

point to a positive and statistically significant impact of parental unemployment on

young people’s right-wing extremism. In line with the previous IV results, the

20

propensity score estimates are larger in absolute value than the corresponding OLS

estimates (albeit to a lower magnitude), also suggesting a causal impact from parental

unemployment experience during childhood on young people’s right-wing extremism.

For example, OLS point estimates suggest that parental unemployment during

childhood increases the probability of young people having right-wing political views

by 4 percentage points. The marginal effect in the propensity score estimates is 5

percentage points.

11. Summary and Discussion

This paper attempts to examine whether parental unemployment experience during

childhood (ages 0-16) increases the risk that young people in Germany become right-

wing extremist. Estimates from cross-sectional regressions suggest that the experience

of parents’ unemployment during childhood is significantly positive associated with the

majority of extremist right-wing attitudes and anti-foreign sentiments for young people

aged 16-29. The intergenerational relationship appears to be stronger in significance and

magnitude for young people living in East Germany. Furthermore, the association

remains robust to controlling for parents’ political believes and economic worries,

young people’s disaffection with politics, their own financial worries and

unemployment status.

By exploiting large differences in unemployment levels between East and West

Germany, both before and after the German reunification as an exogenous variation for

parental unemployment, this paper provides evidence of a causal relationship running

from parental unemployment during childhood on young people’s right-wing

extremism. The majority of 2SLS estimates are larger in magnitude than the

corresponding OLS results. For example, the 2SLS estimates suggest that young

Germans who experienced parental unemployment during childhood have a 20

percentage point higher likelihood to agree that it “would be best if all foreigners leave

Germany” and are 5-7 percentage points more likely to participate in a skinhead neo-

Nazi group compared to those who did not experience parental unemployment during

childhood. Similar conclusions emerge from propensity score matching methods, albeit

slightly lower in magnitude compared to the IV estimates. Overall, using three different

data sets produce similar conclusions: growing up with unemployed parents

21

significantly increases young people’s right-wing extremism in Germany. The results

are robust to different identifying strategies and across ten different measures of right-

wing outcomes.

Finally, some cautionary notes are in order here. In this paper, the impact of

neighbourhood and peer group effects has not been investigated. Interaction of youths in

their neighbourhood and with their peers may result in adoption of their political

believes. For instance, growing up in an area with a right-wing culture may increase

young people’s susceptibility to right-wing extremism. In contrast, local concentration

of ethnic minorities and contact with immigrants might decrease young people’s risk to

be xenophobic. These important topics are left for future research.

22

5. Reference

Federal Ministry of the Interior (2003) Annual Report of the Office for the Protection of the Constitution. Available at http://www.bmi.bund.de. Blundell, R. and Costa Dias, Monica (2000), ‘Evaluation methods for non-experimental data’, Fiscal Studies, 21(4), 427-468. Burkhauser, R. V., Kreyenfeld, M. and Wagner, G. G. (1997), ‘The German Socio- Economic Panel – A representative sample of reunited Germany and its parts’, DIW-Vierteljahresbericht, vol. 66, pp. 7-16. Backer, Susann (2000), “Right-wing extremism in unified Germany”. In Hainsworth, Paul (ed.), The Politics of the Extreme Right. London and New York: Pinter. Bauer, Thomas, Magnus Loftstrom and Klaus F. Zimmermann (2000 or 2001),“Immigration policy, assimilation of immigrants, and natives’ sentiments towards immigrants: Evidence from 12 OECD countries“, Swedish Economic Policy Review, 7(2), 11-53. Card, D. (2001), ‘Estimating the return to schooling: progress on some persistent econometric problems’, Econometrica, 69(5), 1127-1160. Downs, A. (1957), An Economic Theory of Democracy. New York: Harper & Row. Ermisch, John and Marco Francesconi (2001), “Family structure and children’s achievements”, Journal of Population Economics, 14(2), 682-715. Falk, Armin and Josef Zweimüller (2005), “Unemployment and right-wing extremist crime”, IZA discussion paper no. 1540, Bonn. Haisken-DeNew, J. and Frick, J. (2005), Desktop companion to the German Socio-Economic Panel Study (SOEP), DIW Berlin. Heckman, James (1997), “Instrumental variables – A study of implicit behavioral assumptions used in making program evaluations”, Journal of Human Resources, 32(3), 441-462. Koopmans, Ruud and Olzak, Susan (2004), “Discursive opportunities and the evolution of right-wing violence in Germany”, American Journal of Sociology, Krosnick, J. A. and D. F. Alwyn (1989), “Aging and susceptibility to attitude change”, Journal of Personality and Social Psychology, 57, 416-425. Krueger, Alan and Pischke, Jörg-Steffen (1997), „A statistical analysis of crime against foreigners in unified Germany“, Journal of Human Resources, 32(1), 182-209.

23

Lipset, Seymour Martin (1964), “The sources of the radical right”. In: Daniel Bell, The radical right, Garden City, pp. 75-95. Lubbers, M. and Scheepers, P. (2001), “Explaining the trend in extreme right-wing voting: Germany 1989-1998”, European Sociological Review, 17(4), 431-449. Mudde, Cas (2000), The ideology of the extreme right. Manchester and New York: Manchester University Press. O’Neill, Donal and Sweetman, Olive (1998), “Intergenerational Mobility in Britain: Evidence from Unemployment Patterns”, Oxford Bulletin of Economics and Statistics, 60(4), 431-447. Wahl, Klaus (2003), Skinheads, Neonazis, Mitläufer. Täterstudien und Prävention, Opladen. Willems, H. (1993), “Gewalt und Fremdenfeindlichkeit: Anmerkungen zum gegenwärtigen Gewaltdiskurs”, in Otto, H.-W. and Roland Merten (eds.) Rechtsradikale Gewalt im vereinigten Deutschland. Opladen: Leske & Budrich, 88-108. Wolfe, Barbara, Robert, Haveman, Donna K. Ginther, and C.B. An (1996), “The ‘window’ problem in studies of children’s attainment: A methodological exploration.” Journal of the American Statistical Association, 91, 970-82. Center for Survey Research and Methodology (ZUMA). (2005). System of social indicators for the Federal Republic of Germany: Key indicators 1950-2005. Social indicators department, December.

24

Figure 1: Proportion of young people growing up with an unemployed parent by region of residence (East vs. West Germany), YYAS

0.0

2.0

4.0

6.0

8P

ropo

rtion

with

Firs

t par

enta

l une

mpl

oym

ent s

pell

1983 1985 1987 1989 1991 1993 1995 1997year_parue2

East Germany West Germany

Table 1: Means of the Outcome Variables by Sample, Region of Residence and Parental Unemployment German Socio-Economic

Panel (SOEP) Youth and Young Adult

Survey DJI Youth Survey

East Germany West Germany East Germany West Germany East Germany West Germany [1] [2] [1] [2] [1] [2] [1] [2] [1] [2] [1] [2] Leaning towards extremist right-wing party 0.054 0.098 0.020 0.047 0.030 0.091 0.026 0.014 0.029 0.090 0.023 0.101 N 795 427 3,422 1,066 1,270 121 2,181 145

Participation in Skinhead Neo-nazi group 0.015 0.048 0.015 0.036 0.014 0.070 0.009 0.058 N 2,021 249 3,243 223 Right-wing political views 0.030 0.092 0.034 0.079 Totally\strongly agree with: “If jobs become scarce, foreigners living here should be sent home again”

0.359 0.476 0.200 0.353

“It would be best if all foreigners leave Germany”

0.180 0.280 0.107 0.201

“It is alright if one shows asylum seekers in 0.054 0.138 0.029 0.111 a palpable way to return to their home country” a) Right-wing behaviour 0.007 0.042 0.004 0.014 Xenophobia 0.018 0.067 0.006 0.050 N 2,922 357 5,858 139 Notes: N is person-year observations in the SOEP and the number of individuals in the Youth and Young Adult Survey and the DJI Youth Survey. [1] No parental unemployment during childhood [2] Parental unemployment during childhood a) Number of individuals for this outcome variable are: 1,398; 246; 2,687 and 90.

1

Table A1: Summary Statistics by Sample German Socio-

Economic Panel

Youth and Young Adult

Survey

DJI Youth Survey

Age 23.43 (4.00)

22.07 (3.87)

22.34 (4.28)

Year of birth 1975.91 1970.98 1972.05 Female 0.429 0.506 0.480 East Germany 0.244 0.396 0.353 Highest school degree No degree or secondary general school degree 0.193 0.203 0.162 Intermediate school degree 0.371 0.396 0.356 High school degree 0.307 0.247 0.421 Still in schooling 0.128 0.154 0.061 Mother’s highest school degree No degree or secondary general school degree 0.460 0.465 0.465 Intermediate school degree 0.416 0.375 0.383 High school degree 0.124 0.160 0.152 Father’s highest school degree No degree or secondary general school degree 0.451 0.426 0.396 Intermediate school degree 0.342 0.321 0.338 High school degree 0.206 0.254 0.266 Parental unemployment during childhood 0.261 0.082 0.053 Local unemployment rate 11.17

(4.71) 11.47 11.87

(5.28) (5.25)

N 1,773 5,736 9,276 Note: N is the number of individuals. In the SOEP, sample means are measured in the last year individuals are observed in the panel. Figures are sample means with standard deviations in parentheses.

2

Table 2: Parental Unemployment and Young People’s Right-Wing Extremism German Socio-

Economic Panel Youth and Young

Adult Survey DJI Youth

Survey East

GermanyWest

GermanyEast

GermanyWest

Germany East

GermanyWest

Germany

Panel A Leaning towards an extremist right- wing party

0.050* (0.027)

0.021* (0.012)

0.050** (0.019)

-0.019 (0.013)

0.045** (0.011)

0.067** (0.013)

Participation in Skinhead neo-Nazi group 0.031** (0.010)

0.020** (0.009)

0.044** (0.008)

0.043** (0.009)

Right-wing political views 0.045** (0.011)

0.034* (0.016)

Panel B Totally\strongly agree with:

“If jobs become scarce, foreigners living here should be sent home”

0.056** (0.027)

0.114** (0.033)

“Would be best if all foreingers leave Germany”

0.049** (0.023)

0.068** (0.026)

“It is alright if one shows assylum seekers in a palpable way to return to their home country”

0.054** (0.017)

0.070** (0.019)

Panel C Right-wing behaviour 0.019** -0.001 0.031** 0.007 (0.006) (0.002) (0.006) (0.005) Xenophobia 0.027** 0.038** (0.009) (0.007) Notes: N denotes number of person-year observations in the German Socio-Economic Panel and number of individuals in the Youth and Young Adult Survey and the DJI Youth Survey. OLS regressions with standard errors in parentheses. In regressions based on the SOEP, standard errors are corrected for person-year clustering. Each estimate represents the coefficient from a different regression. * significant at the 10%, ** significant at the

3

Table 3: Parental Unemployment, Feelings of Economic Insecurity, Political Disaffection and Propensity to feel close to extremist right-wing party German Socio-Economic

Panel Youth and Young

Adult Survey DJI Youth

Survey (1) (2) (3) (4) (5) (1) (6) (1) (2) (3) (6)

East Germany Young people’s own unemployment 0.099** 0.087* 0.082* 0.079* 0.079* 0.021** 0.039** 0.041** 0.040** 0.036** 0.035** (0.046) (0.048) (0.048) (0.044) (0.044) (0.013) (0.018) (0.010) (0.011) (0.011) (0.011) Worried about financial situation 0.041** 0.034* 0.031* 0.032* 0.002 0.003 0.001 (0.017) (0.017) (0.017) (0.017) (0.006) (0.007) (0.007) Feels disadvantaged by reunification 0.013** 0.004 No trust into politicians/government 0.033** 0.023** (0.015) (0.007) Parents are worried about finances 0.038* 0.020 0.020 (0.017) (0.015) (0.016) Parents lean towards extremist right-wing 0.601** 0.593** party (0.104) (0.103) Parents are in favour of Democracy 0.016 (0.021) Parental unemployment 0.050* 0.051* 0.047* 0.042* 0.044* 0.020** 0.063** 0.046** 0.046** 0.051** 0.052** (0.027) (0.027) (0.027) (0.024) (0.023) (0.009) (0.022) (0.011) (0.011) (0.018) (0.011)

N 1,207 West Germany Young people’s own unemployment 0.014 0.006 0.005 0.006 0.006 0.019 0.049** 0.045** 0.044** 0.043** (0.024) (0.022) (0.022) (0.022) (0.023) (0.018) (0.011) (0.011) (0.011) (0.011) Worried about financial situation 0.035** 0.034** 0.034** 0.033** 0.015** 0.013** 0.012** (0.014) (0.014) (0.014) (0.014) (0.004) (0.004) (0.004) Feels disadvantaged by reunification 0.020** 0.018** (0.005) (0.004)

4

No trust into politicians/government (0.014) 0.033** 0.016** (0.012) (0.004) Parents are worried about finances 0.002 0.000 -0.001 (0.008) (0.007) (0.007) Parents lean towards extremist right-wing 0.341** -0.028** party (0.116) (0.013) Parental unemployment 0.021* 0.021* 0.020* 0.022* 0.022** -0.035 0.060** 0.059** 0.065** 0.066** (0.011) (0.011) (0.012) (0.011) (0.011) (0.024) (0.013) (0.013) (0.014) (0.014)

N 4,357

5

Table 4: Parental Unemployment and Young People’s Propensity to lean towards Extremist Right- Wing Party German Socio-

Economic Panel Youth and Young Adult

Survey DJI Youth

Survey Right-wing

party Right-wing

party Skinhead group Right-wing

party Skinhead

group Mother unemployed 0.024 0.027** 0.004 0.043** 0.048** (0.017) (0.012) (0.007) (0.010) (0.007) Father unemployed 0.036** 0.006 0.019** 0.033** 0.039** (0.016) (0.013) (0.008) (0.013) (0.009) Both parents unemployed 0.006 0.029** 0.031** 0.130** 0.035** (0.019) (0.014) (0.009) (0.023) (0.016)

6

Table 5: Parental Unemployment and Young People’s Propensity to lean towards Extremist Right-Wing Party (SOEP) OLS Random effect

estimates Sibling difference

estimates East

Germany West

Germany East

Germany West

Germany East

Germany West

Germany Panel A Parental unemployment 0.051** 0.020* 0.061** 0.021** 0.116** -0.002 during childhood (0-16) (0.027) (0.011) (0.028) (0.010) (0.042) (0.015) Most recent parental unemployment at ages: 0-12 -0.034 0.026 -0.021 0.014 0.118 -0.009 (0.030) (0.018) (0.049) (0.013) (0.081) (0.019) 13-16 0.063** 0.016 0.075** 0.024* 0.116** 0.005 (0.029) (0.013) (0.029) (0.013) 0.041 (0.017) Panel B Parental unemployment: 0.037** 0.013 0.038** 0.011 0.034* 0.016** (0.018) (0.008) (0.018) (0.008) (0.018) (0.007) Most recent parental unemployment at ages: 0-12 -0.035 0.026 -0.022 0.015 -0.009 0.005 (0.030) (0.018) (0.051) (0.013) (0.043) (0.011) 13-16 0.063** 0.014 0.082** 0.024* 0.047* 0.019 (0.030) (0.012) (0.031) (0.014) (0.025) (0.012) 17 and above 0.064 0.020 0.075* 0.013 0.064* -0.021 (0.040) (0.032) (0.038) (0.027) (0.033) (0.024)

7

Table 6: First-Step Regressions SOEP YYAL DJI Youth

Survey Panel A (IVI) Number of childhood years -0.048** -0.065** lived in GDR (0.006) (0.004)

Lived in GDR at age16a -0.142** (0.009) Panel B (IVII) Parental unemployment in East Germany in the years:

1990-1993 0.109** (0.009) 1994-1997 0.184** (0.011)

Partial R2 0.027 0.052 0.023 0.028 F-Statistic 56.46 342.03 240.70 277.57

a Includes the years 1990 and 1991. Table 7: OLS and Instrumental Variable Estimations (East and West Germany)

German Socio-Economic Panel

Youth and Young Adult Survey

DJI Youth Survey

OLS IV OLS IV OLS IVI IVII

Leaning towards an extremist right-wing party 0.025** -0.027 0.009 0.016 0.053** 0.129** 0.080* (0.011) (0.054) (0.011) (0.059) (0.008) (0.050) (0.045)

Participation in Skinhead neo-Nazi group 0.026** 0.074** 0.046** 0.069** 0.054* (0.007) (0.028) (0.005) (0.035) (0.032)

Right-wing political views 0.043** 0.112** 0.075 (0.009) (0.056) (0.050)

Totally\strongly agree with:

“If jobs become scarce, foreigners living 0.081** 0.110 0.162 here should be sent home” (0.019) (0.124) (0.113)

“Would be best if all foreigners 0.062** 0.210** 0.192** leave Germany” (0.016) (0.101) (0.091)

“It is alright if one shows asylum seekers in a palpable way to return to 0.065** 0.180* 0.191** their home country” (0.012) (0.095) (0.081) Right-wing behaviour 0.009** 0.026** 0.024** 0.047* 0.023 (0.002) (0.011) (0.004) (0.024) (0.022)

Xenophobia 0.037** 0.206** 0.171** (0.005) (0.036) (0.032)

8

Table 8: Propensity Score Matching (East and West Germany) German Socio-

Economic Panel Youth and Young

Adult Survey DJI Youth

Survey (1) (2) (1) (2) (1) (2)

Leaning towards an extremist right-wing 0.022** 0.023** 0.017 0.018 0.056** 0.060** party (0.008) (0.007) (0.013) (0.013) (0.014) (0.013)

Participation in Skinhead neo-Nazi group 0.025** 0.027** 0.048** 0.050** (0.009) (0.008) (0.012) (0.012)

Right-wing political views 0.051** 0.053** (0.013) (0.013)

Totally\strongly agree with:

“If jobs become scarce, foreigners living 0.101** 0.132** here should be sent home” (0.026) (0.023)

“Would be best if all foreigners 0.079** 0.095** leave Germany” (0.020) (0.020)

“It is alright if one shows asylum 0.073** 0.076** seekers in a palpable way to return to (0.020) (0.019) their home country” Right-wing behaviour 0.010** 0.010** 0.027** 0.027** (0.005) (0.005) (0.008) (0.007) Xenophobia 0.032** 0040** (0.012) (0.011) Notes: [1] Local Linear Regression [2] Kernel Estimations