Good Bye Lenin Revisited: East-West Preferences Three Decades after German Reunification

1 Introduction

Experiences shape preferences. This has meanwhile been shown for a wide range of preferences, from risk and inflation tolerance (Malmendier and Nagel 2011; Malmendier 2021) to support of democracy (Fuchs-Schündeln and Schündeln 2015). Alesina and Fuchs-Schündeln (2007) analyze the effect of having lived under Communism versus Capitalism on preferences for a strong welfare state. Based on the 1997 and 2002 releases of the German Socio-Economic Panel, Alesina and Fuchs-Schündeln (2007) find East Germans to be significantly more in favor of the government taking a strong role in caring for vulnerable groups than West Germans. At the same time, they document convergence of preferences between 1997 and 2002 and conclude that it would take 20–40 years for the views of East and West Germans to converge if convergence were to continue at the same (linear) speed. By documenting convergence of preferences on the individual level, and stronger East-West differences for older individuals who lived under the different systems for a longer time, they can establish a causal effect of living under different systems on preferences.

In this paper, we revisit the long-term effects of having lived under Communism versus Capitalism on preferences. Adding the 2017 release of the German Socio-Economic Panel, we update the results obtained by Alesina and Fuchs-Schündeln (2007). We show that East Germans continue to be significantly more pro-state compared to West Germans even 27 years post reunification. Moreover, we find that the speed of convergence slowed down significantly. The convergence of preferences between 2002 and 2017 is mainly driven by the change in the composition of the population, with younger cohorts replacing older cohorts: On the individual level, there was on average no further significant convergence of preferences between 2002 and 2017. Finally, we present evidence on the intergenerational transmission of preferences: we find that young individuals born after reunification to parents who used to live in the East are significantly more likely to exhibit pro-state preferences than young individuals born to parents who used to live in the West, independent of their own current residence. Linking children to their parents, we document a strong correlation between the preferences of parents and children. This intergenerational transmission of preferences implies that the effects of political regimes on preferences can be extremely persistent.

This paper is closely related to a growing strand of literature on the formation of economic and political preferences. Ravallion and Lokshin (2000) and Corneo and Grüner (2002), among others, show that preferences for redistribution depend on income and perceived social mobility. Di Tella et al. (2007) find that land titles increase pro-market materialist attitudes. Guiso et al. (2006) and Luttmer and Singhal (2011) show that ethnicity and culture of the birth country are significantly correlated with preferences for redistribution among second-generation migrants. Alesina and Giuliano (2011) provide an overview of the theoretical literature and conclude that both personal (age, gender, ethnicity, socioeconomic status) and societal (political ideology, cultural norms) characteristics have an impact on preferences for redistribution. Fong (2001), Alesina and La Ferrara (2005), and Durante et al. (2014) establish a connection between preferences for redistribution and other values and beliefs such as concern for fairness, risk attitude, preferences regarding efficiency and equality, and belief in luck.

Following Alesina and Fuchs-Schündeln (2007), several studies have looked at political and social preferences in the context of German reunification. Svallfors (2010) uses the International Social Survey Program data for 1990, 1996, and 2006. He finds that attitudes towards state intervention of East Germans converge quickly to those of West Germans, and that there is no East-West difference in attitudes among the younger cohorts. Brosig-Koch et al. (2011), on the other hand, see no convergence in attitudes towards solidarity, cooperation, and fairness between East and West Germans, relying on laboratory experiments among students. Finally, Avdeenko (2018) shows that East Germans living in the border regions are less likely to vote for left parties, which the author interprets as retrospective electoral punishment.

An increasing number of studies use German reunification as a natural experiment in other contexts than the effect on preferences, see Fuchs-Schündeln and Hassan (2016) for an overview. If separation and reunification of Germany were exogenous, and there were no East-West differences prior to the separation in 1949, then West Germans constitute a valid control group for East Germans, and the differences in outcome variables right after reunification can be attributed to having lived under different economic and political regimes. The literature evaluates the effect of Communism on a wide range of variables, including savings (Fuchs-Schündeln 2008; Fuchs-Schündeln and Schündeln 2005), income and economic development (Burchardi and Hassan 2013), productivity (Burda and Severgnini 2018), labor market outcomes (Fuchs-Schündeln and Masella 2016), inflation expectations (Goldfayn-Frank and Wohlfart 2020), stock market participation (Fuchs-Schündeln and Haliassos 2021; Laudenbach et al. 2020), financial literacy (Davoli and Hou 2021) and gender norms (Bauernschuster and Rainer 2012; Lippmann et al. 2020). Becker et al. (2020), however, question the validity of German reunification as a natural experiment. They document pre-existing differences between East and West before separation and argue that there was selective out-migration from East Germany between 1949 and 1961. Following the approach taken by Alesina and Fuchs-Schündeln (2007), we argue that we are immune to this criticism. First, we show that the length of experiences with the different regimes is correlated with differences in preferences. Second, we analyze convergence not only on the aggregate, but also on the individual level.

The paper is organized as follows. Section 2 describes the data source and the sample. Section 3 replicates and expands the main analyses of Alesina and Fuchs-Schündeln (2007), including the new data wave from 2017. Section 4 analyzes the preferences of cohorts born after reunification. Finally, Section 5 concludes.

2 Data and Methodology

2.1 The German Socio-Economic Panel

As in Alesina and Fuchs-Schündeln (2007), we use the German Socio-Economic Panel (SOEP) as our primary data source. SOEP is a large representative panel of German households.^[1] The survey started in 1984, first sampling only West German households, but adding an East German sample already in the spring of 1990. The respondents of the survey answer a wide array of questions, providing information about their socio-economic background, family ties, labor market experiences, attitudes, and political views. The question concerning preferences about the role of the government in providing financial security was asked only in the 1997, 2002, and 2017 waves. This question reads: “In addition to the state, private individuals such as free-market companies, organizations, associations or individual citizens are responsible for a large number of social tasks in our society today. What is your opinion on who should be responsible for the following areas: financial protection of families/financial security for unemployment/financial protection in case of illness/financial security in old age/financial security for people in need of care”. The answers are given on a scale of 1–5, which correspond to “only the state”, “mostly the state”, “both the state and private forces”, “especially private forces”, and “only private forces”. We use the answers to this question to generate five independent dummy variables. Specifically, the dummy variables take the value of 1 when the respondent states that financial security is the responsibility of “only the state” or “mostly the state”, and the value of 0 otherwise. Thus, our dependent variables indicate whether an individual has pro-state preferences. Our main explanatory variable is a dummy variable East, which is equal to 1 if the respondent lived in East Germany before reunification, independent of the current residence.

Our basic sample is constructed as follows. For the main analyses, we use data from both individual and household questionnaires for 1997, 2002, and 2017. We only include individuals born before 1989. We use the initial West and East German samples (A and C), as well as all general refreshment samples (samples E, F, H, J, and K). Appendix Tables A1 and A2 provide summary statistics of the preferences separately for the East and West samples and years, once on the full sample and once on the balanced sample that includes only individuals who answer in all three survey waves. On average, East Germans in the unbalanced sample spent roughly 30 years under Communism. For the analysis of the youngest cohorts in Section 4, we focus on individuals born between 1990 and 1999, and link them to their parents based on the biography questionnaire.

Figure 1 shows the share of the respondents from West and East who state that the government should be responsible for providing financial security for the five different vulnerable groups. The majority of the respondents agree that the state should provide financial support to the unemployed – in 2017, 74% of East Germans and 69% of West Germans had pro-state preferences with respect to this question. At the same time, only 48% of the respondents from the East and 35% of the respondents from the West think that it is the government’s duty to support the family, and the averages are in a similar range for the other outcome variables. Over time, the West German sample becomes more pro-state: the fraction of individuals who consider that the state is responsible for financial security increases between 1997 and 2017 across all five variables. The East German sample, on the other hand, becomes on average less pro-state between 1997 and 2002. However, this pattern reverses between 2002 and 2017, when the opinion moves more pro-state in both samples, with the exception of the opinion on the state taking care of unemployed individuals in the East sample.

Figure 1:

Fraction of the respondents from West and East who believe that the financial security is the responsibility of the state. Unbalanced sample.

3 Main Results

Table 1 replicates the basic regression specification from Alesina and Fuchs-Schündeln (2007), but extends the sample to include data from 2017, in addition to the years 1997 and 2002. The dependent variable is a dummy equal to one if the individual supports an active role of the state. The right-hand side variables of interest are the East dummy and the interaction terms between East and the years 2002 and 2017.

As Table 1 shows, individuals who lived in East Germany before reunification have consistently and significantly stronger pro-state preferences than individuals from West Germany in 1997, as expressed by the positive and significant coefficients on the East dummy. Having lived under different systems for up to 45 years shaped the preferences of East and West Germans. Moreover, the preferences of East and West Germans converge significantly between 1997 and 2002 – the coefficients of the interaction terms East*2002 are all significantly negative. This is in line with a common effect on preferences from living under the same system since 1990. The results so far are also quantitatively very similar to the results in Alesina and Fuchs-Schündeln (2007). Extending the time frame to include the 2017 survey has thus no effect on the results from the previous years.

What is interesting, though, is how preferences evolved between 2002 and 2017. First focusing on West Germans, there was already some movement toward stronger state preferences between 1997 and 2002 – the Year 2002 dummy is positive and significant in 3 out of 5 cases – but this movement became even stronger between 2002 and 2017: the Year 2017 dummy is positive and significant for all five variables. It is also significantly different from the Year 2002 dummy (Appendix Table A5). Thus, West German preferences moved more in favor of a strong government in the two decades between 1997 and 2017. Turning to the coefficients on the interaction terms with the East dummy, the preference differences between East and West Germans continued to narrow: the coefficients of East*2017 are not only significantly negative but also larger in absolute magnitude than the East*2002 coefficients. Yet, only in the case of taking care of unemployed individuals is the East*2017 coefficient significantly different from the East*2002 coefficient. Thus, there is significant convergence between 2002 and 2017 only in one of the five preference variables. In the other four cases, the convergence is not only not significant in this period, but according to the point estimates also substantially slower than in the period 1997 to 2002. This is true even for the only variable in which significant convergence still occurs between 2002 and 2017: taking into account that this is a 15-year period, compared to the 5-year period between 1997 and 2002, the speed of convergence slowed down significantly. The coefficients on the control variables are very close to the estimates in Alesina and Fuchs-Schündeln (2007) both in sign and magnitude. Individuals with more education and higher monthly household income, as well as the self-employed and white-collar workers, are consistently less pro-state.^[4]

The marginal effects corresponding to Table 1 are shown in Appendix Table A7. They allow us to talk about the speed of convergence more concretely. In 1997, individuals from East Germany were between 14.8 and 17.8 percentage points more likely to have pro-government preferences than individuals from West Germany. Compared to 1997, the East-West difference in the probability of being pro-state declined by between 3.6 and 6.5 percentage points in 2002, and by between 4.4 and 11.5 percentage points in 2017. Clearly, the convergence of preferences between East and West Germans is not complete – even 27 years after reunification the preferences regarding the active role of state differ between East and West Germans. If the speed of convergence continues as it did between 2002 and 2017, then preferences for taking care of the unemployed should converge by 2025. However, for the other three variables which still show convergence between 2002 and 2017, final convergence of preferences would take significantly longer. In the case of taking care of the sick, for instance, it will take a further 98 years after 2017 for preferences to converge; both because the difference is still quite large in 2017 (East Germans are 9.8 percentage points more likely to answer that the state should take care of sick individuals), and because convergence is slow (the East-West difference declined by only 1.5 percentage points over the 15 year period 2002 to 2017). For the other two variables, the much lower speed of convergence between 2002 and 2017 implies that full convergence of preferences will take between 144 (taking care of old individuals) and 208 years (taking care of individuals requiring care). Thus, not only is the convergence between 2002 and 2017 not statistically significant, neither is it economically significant.

Individual preferences for a strong government could be affected not only by the individual economic situation, but also by general economic circumstances, either because individuals are altruistic and care for people around them, or because they could potentially be affected by these general economic circumstances themselves in the future. Thus, we include as an additional control variable the state-level unemployment rate in the survey year.^[5] As Table 2 shows, indeed individuals who reside in states with higher unemployment rates tend to exhibit stronger preferences for an active role of the state: the coefficients on the state-level unemployment rate are positive and statistically significant across all five dependent variables. The inclusion of this control variable decreases the magnitude of the East coefficients, which suggests that part of the East effect can be explained by higher unemployment rates in the Eastern states. Nevertheless, the East dummy is still a significant predictor of preferences. The convergence results are also quantitatively affected by the inclusion of this control variable: the inclusion of the state-level unemployment rate leads to stronger convergence between 1997 and 2002, and weaker convergence or even divergence between 2002 and 2017.

In Table 3, we split the Eastern sample into East German states by current residence, as in Table 8 of Alesina and Fuchs-Schündeln (2007). We drop all individuals who are originally from the East and now reside in a West state and vice versa. The table shows that the patterns we found before are quite homogeneous and are thus observed across individual states as well. First, in 1997 individuals from all Eastern states are significantly more pro-state than their West German counterparts. Secondly, all the significant coefficients of the interaction terms measuring convergence are negative (except for one coefficient for Brandenburg in 2002 and one for Mecklenburg-Vorpommern in 2017). For 2002, 15 of the 25 convergence coefficients are negative and significant, and the same is true for 2017. Convergence is stronger in Sachsen-Anhalt, Thüringen, and Sachsen than in the other three states. Generally, this analysis confirms that the baseline results are not driven by a single state.

4 Preferences of Post-Reunification Cohorts

So far, we find that convergence of preferences among individuals who were born before reunification significantly slowed down between 2002 and 2017. However, should we expect faster convergence once younger individuals born after reunification and thus never having accumulated different experiences enter the sample?

We analyze this question explicitly by focusing on the cohorts born between 1990 and 1999. These cohorts were born after the Fall of the Berlin Wall and never experienced German separation. Table 7 presents the results of the regression analysis based on the sample of these young individuals in 2017. There are two regressors of interest: Born in East States is a dummy variable equal to one if the respondent was born in one of the states that formerly belonged to East Germany; Born to Eastern parents is a dummy variable equal to one if at least one of the parents of the respondent lived in the GDR prior to 1989. Given the small number of observations and the young age of this sample, the control variables only include age, gender, and labor market variables (with Employed as the omitted category).

The results presented in Table 7 refute the ex-ante assumption that East and West Germans born after reunification do not exhibit any difference in preferences. Individuals born to “East” parents in the period 1990 to 1999 exhibit stronger pro-state preferences compared to their Western counterparts, even though they have never lived under Communism themselves; two of the relevant coefficients are also statistically significant. By contrast, the coefficients on the dummy variable Born in East States are small, never significant, and sometimes even negative. This indicates that parents might be more important than peers or the general environment in shaping preferences. Although never directly affected by Communism themselves, respondents are more likely to support the responsibility of the state if at least one of their parents lived under Communism.

Table 8 sheds some more light on this result. In this estimation, we include the parental preferences directly as a control variable. Specifically, we create a dummy variable Parent pro-state in 2017 equal to one if at least one of the individual’s parents – mother or father – answered only the state or mostly the state to the question about the state’s responsibilities in providing financial security in 2017. The coefficient of this dummy variable is positive and statistically significant in all regressions: parents’ preferences are strongly correlated with a young person’s attitudes. This is true for both East and West Germans. At the same time, the coefficient of Born to Eastern parents is now somewhat smaller in magnitude in four of the five regressions than in Appendix Table A11, which shows the results of a regression without parental preferences on exactly the same sample (which is somewhat smaller than the sample in Table 7). This suggests that the variable Born to Eastern parents acts to a certain extent as a proxy for the previously unobserved parental preferences. Respondents with “East” parents are more likely to have parents with strong pro-state preferences, and the intergenerational transmission of preferences leads to differences in preferences also among the cohorts born after reunification. Appendix Tables A12 and A13 confirm that preferences of both parents matter. Fathers’ attitudes are equally important as mothers’ attitudes – all the coefficients of Mother pro-state in 2017 and Father pro-state in 2017 are large and statistically significant.

Summarizing, young cohorts born between 1990 and 1999 exhibit stronger pro-state preferences if at least one of their parents lived in the GDR. They are significantly more likely to hold the view that it is the state’s responsibility to provide financial security for families and older individuals. However, the pro-state attitudes of these cohorts are no longer driven by the Communist context during their upbringing, as they were born in a united Germany, but by the pro-state views of their parents, who were subject to this context in the past. The experience of the parents matters more than the experience of the peers, since growing up in an Eastern or Western state after reunification plays no significant role for preferences. A similar transmission of preferences from parents to children has been shown in other contexts, for example when it comes to the support of right-wing extremism (see Avdeenko and Siedler 2017).

5 Conclusion

In this paper, we investigate the influence of the political system on preferences about the role of the government. The German reunification experiment provides an ideal setting for such an investigation, as it constitutes an exogenous shock to the political system. East Germans spent several decades under the Communist regime, being exposed to its doctrine and values. After reunification in 1990, Communism gave way to a market-based economy. Under the assumption that there were no meaningful differences between the residents of East and West Germany before the separation, the differences in preferences that we observe after reunification can be attributed to living under the Communist regime. Even if there were differences between East and West before the separation, convergence in preferences on the individual level after reunification, as well as stronger differences in preferences for cohorts that lived under different systems for longer, point toward the effect of Communism.

Alesina and Fuchs-Schündeln (2007) quantify the East-West differences in preferences after reunification, using data from the German SOEP until 2002. Here, we replicate and extend their findings, adding the 2017 data wave. We find that even 27 years after reunification, East Germans remain more pro-state than West Germans. Our analysis suggests that in 1997 they are more likely to state it is the government’s responsibility to provide financial security for individuals who are sick, unemployed, old, or in need of care, as well as for families, compared to their West German peers. We also observe that preferences converge and adjust over time, with East Germans becoming relatively less pro-state, and West Germans appreciating the welfare state more than before. However, the pace of convergence is not uniform. The convergence of preferences was rapid in the first years after reunification, but slowed down significantly after 2002. Most of the remaining convergence comes from shifts in the cohort composition, not from convergence of preferences on the individual level.

How do preferences of individuals born after reunification compare between East and West? We find that young individuals born after reunification exhibit stronger preferences for a strong government if at least one of their parents lived in the GDR. The former place of living of the parents matters more than the current place of living of the young individual for his or her preferences. It seems that through intergenerational transmission of preferences the effects of Communism are thus truly long-lasting.