How and when do people change their minds? For example, watching a popular television series like AMC’s Mad Men seems to transport us straight to another planet. It shows the lives of advertising executives on Madison Avenue in the 1960s who spend their days drinking heavily (from 9am), chain-smoking, and fornicating. While not necessarily an accurate portrayal of corporate life in the middle of the 20th century, it reminds us how deeply cultures can be transformed in a relatively short space of time. In the Western world today, attitudes towards homosexuals, pre-marital sex, and women working outside the home are radically different from what they were a generation ago (Fernandez-Villaverde et al. 2011).
At the same time, culture seems to persist over long periods of time. Italian towns that were self-governing in the middle ages are still more prosperous, citizens give more blood, and are more trusting (Guido et al. 2008). Areas of Africa affected by the slave trade continue to show lower levels of interpersonal trust (Nunn and Watchekon 2011). What accounts for the two-faced nature of culture? Why does it change so radically some of the time, while remaining unaltered over long periods?
Anti-Semitism as an indicator
In the past, canaries were used in coal mining to detect the presence of toxic fumes. Anti-Semitism serves a similar function in our study. As an attitude, it is arguably puzzling – it is extreme, clearly-defined, and dysfunctional. Germany has not been home to Jews in any significant numbers since 1945 (despite some minor inflows after the 1990s). In recent research, we look at the persistence of Jew-hatred in Germany (see Voigtländer and Voth 2012). In earlier work, we show that towns that saw pogroms in 1350, at the time of the Black Death, were still more anti-Semitic in the 1920s and 1930s (Voigtländer and Voth 2011). In this column, we examine how much of the past still matters for the present. Specifically, we ask how much of the anti-Semitism we see based on data from 1996 and 2006 reflects attitudes in the same location as far back as the late 19th century. We also examine the conditions under which hatred of this kind can be accentuated or reduced.
Germans today are on average probably not much more anti-Semitic than other Europeans (Bergmann and Erb 1997). At the regional level, however, there are considerable differences. We use data from the German Social Survey (ALLBUS) to examine attitudes towards Jews. The survey asks a battery of questions, such as “Do you think that Jews partly brought persecution in the 20th century on themselves?” Answers range from 1 (strongly disagree) to 7 (strongly agree). Figure 1 shows regional differences (at the district level) for answers to this question, based on the proportion of the population giving a score of 5, 6, or 7.
Figure 1. Distribution of extreme anti-Semitic views
In some districts, only 8% feel that Jews are partly responsible for their own persecution; in others, some 38% of the population think so.
Other ALLBUS questions related to anti-Semitism are:
- Would you mind if you had Jewish neighbours?
- Would you mind if a Jew married into your family?
- Should Jews have equal rights?
- Do Jews have too much influence in the world?
- Jews are exploiting their victim status for their own financial gain – do you agree?
The first three of these questions are also asked about foreigners and immigrants, while the last two are specific to Jews. To construct an indicator of anti-Semitism, we aggregate the answers to these questions in two ways. We compile one variable (“broad anti-Semitism”) which simply takes the average of the answers to these questions (appropriately rescaled so that higher values correspond to more extreme views). We also define committed anti-Semites as individuals who give consistently extreme answers (5 or higher) to the three specific questions about attitudes towards Jews (influence, victim status, and responsibility for their own persecution). According to this definition, around 5% of the German population qualify as committed anti-Semites.
While both variables are designed to capture related but distinct attitudes, they identify regional patterns consistently. Figure 2 shows the correlation between broad and committed anti-Semitic views at the district level. Places such as Lower Bavaria (Niederbayern) have both a high proportion of committed anti-Semites (about three times higher than the national average), and the average answer to questions concerning Jews is strongly anti-Semitic.
Figure 2. Broad and committed anti-Semitism in the cross-section
The long shadow of the past
Next, we examine what accounts for the distribution of attitudes across space. We use three historical variables, designed to capture deep-rooted historical attitudes – voting results for anti-Semitic parties in the late Imperial period (1890-1912), votes for the Nazi Party in the 1920s, and votes for the Nazis in the 1930s. After reunification in 1871, Germany experienced a rising tide of anti-Semitic sentiment. This culminated in petitions to the government to restrict Jewish migration and their access to the professions, universities, and the army, as well as in the founding of political parties whose only purpose was to reduce Jewish influence. In addition, we look at electoral support for the Nazi party (NSDAP) at the polling booth. In the 1920s, the party emphasised its radical nature, aiming at a violent overthrow of the established order, and its anti-Semitic credentials.
After 1928, it toned down this rhetoric; in a bid for power, it presented itself as an alternative for the economically disaffected, while making itself more respectable in the eyes of potential conservative allies. As a consequence, anti-Semitism was de-emphasised during election in the early 1930s (Heilbronner 2004). We combine historical voting data with the individual attitude surveys from over 260 ALLBUS municipalities.
A simple way to show how much of an areas’ past can explain the present is to divide our sample into three equal-sized groups, based on their historical voting patterns. In the lowest tercile, votes for anti-Semitic parties between 1890 and 1933 (captured by the variable ALLVOTE) were rare; in the highest tercile, they were frequent. For example, in the lowest group, the Nazi Party received only 1.8% of the vote in the 1920s; in the highest tercile, it polled 9.4%. We then ask: How do attitudes today differ across these three groups? Appendix 1 gives an overview.
We find that modern-day attitudes differ systematically. Where support for anti-Semitic parties was low in the past, Germans today are more likely to welcome Jews as neighbours, accept them as family members, support equal rights for them; they are less likely to blame them for their own persecution, to think that they have too much influence in the world, and that they exploit their victim status. Unsurprisingly, general xenophobia in these locations is also lower than in those with historically high support for anti-Semitic parties.
The extent to which the past still matters for attitudes today is surprising. After 1945, anti-Semitism went from official policy to taboo; the fact that some people confess to it to the extent that they do, even in front of an interviewer who might elicit responses that are widely approved, suggests that privately-held views are probably even more anti-Jewish. The second reason why the strong grip of the past is surprising has to do with the persistence of regional patterns. Germany experienced massive population movements during the period 1939-61, with refugees from cities fleeing the bombing first, then the Eastern expellees flooding into West Germany, and then GDR residents escaping from Communist rule (until the building of the Wall).
Next, we examine how strong the link with the past is. We find a strong and significant link between voting patterns in the same locale in the distant past and attitudes today. The strength of the link differs somewhat by explanatory variable. Support for the anti-Semitic parties in the period 1890-1912 is a good predictor of both broad anti-Semitism and the share of committed anti-Semites, and the effects are large. A one standard deviation (eight percentage point) increase in the vote share of these parties predicts a one percentage point higher share of committed anti-Semites today – relative to a nation-wide average of 4.7%, this is a large effect. Votes for the Nazi Party in the 1920s are mostly associated with more committed anti-Semites today; votes in the 1930s have more predictive power for broad anti-Semitism.
Changing minds, for better and for worse
The Nazis, once in power, pursued a single-minded policy of indoctrination. We examine if this policy was successful in instilling racial hatred that is still visible some 70 years
later. Second, we examine the impact of occupation policies after 1945. All allied forces committed to “de-Nazification”, but they implemented this policy differently. Here, we show that in one zone at least, there are consistently lower levels of modern-day anti-Semitism, independent of preferences revealed in the past.
In general, the older Germans are today, the more likely they are to hold xenophobic and anti-Semitic beliefs. One group, however, stands out from this general pattern – those born in the period 1925-34. In Appendix 2, we plot anti-Semitism and xenophobia by birth cohort. Those who were aged between 11 and 20 when the war ended are much more anti-Semitic than one would expect given their age and general hatred of foreigners (left panel); they also have a higher share of committed anti-Semites than any other group (right panel). We argue that this is a direct result of Nazi policies to indoctrinate the population in racial hatred and a belief in the superiority of the Aryan master race. Interestingly, additional results show that teaching to hate was easiest in areas where the “bourgeois” parties of the centre and right did best in the period before 1933. In principalities controlled by the left (Social Democrats and Communists), the Nazis had no success in boosting Jew-hatred among the young.
If Germans could be influenced strongly in their beliefs during the Nazi period, is there any evidence of the opposite once racial hatred became an official taboo after 1945? We compare the level of anti-Semitism in the different zones of occupation. The former British zone today has by far the least anti-Semitic beliefs, even after controlling for pre-1945 differences. The American zone, on the other hand, has strong levels of support for anti-Jewish views.
Figure 3. Broad anti-Semitism in Germany today, by former zone of occupation
Based on a detailed examination of occupation policies, we argue that these differences probably reflect different approaches to de-Nazification. The American authorities ran a highly ambitious and punitive programme which resulted in many incarcerations and convictions, with numerous, low-ranking officials banned and punished. Citizens were confronted with German crimes, forced to visit concentration camps, and attend education films about the Holocaust. There was a considerable backlash, and perceived fairness was low. The Jewish Advisor to the American Military Government concluded in 1948 that “... if the United States Army were to withdraw tomorrow, there would be pogroms on the following day.” In contrast, the British authorities pursued a limited and pragmatic approach that focused on major perpetrators. Public support was substantial, perceived fairness was higher, and intelligence reports concluded that the population even wanted more done to pursue and punish Nazi officials.
A challenge for social science
How can the Middle East be made safe for peace? How can Southern Europeans be convinced to pay their taxes? If there is one conclusion that can be drawn from nearly a decade of intense research in cultural economics, it is that incentives are not all that there is behind how people behave. Seemingly irrational preferences are pursued at high costs; cultural distinctiveness is preserved in the face of substantial drawbacks. How to change people’s minds is arguably the ultimate frontier for social science.
We examine what it takes to change beliefs by looking at the case of Jew-hatred in modern-day Germany. Across one of the greatest discontinuities in social norms in recorded history – the change of racism from encouraged attitude to banned belief – we find that policy can make a difference. The young can be manipulated by massive indoctrination, but only to the extent that the new, radical beliefs are not completely at variance with pre-existing norms. Policies designed to change the views of the population largely failed after 1945; we conclude that low-pressure interventions with substantial public support may be best to generate “buy-in“ from the population.
Bergmann, W and R Erb (1997), Anti-Semitism in Germany: the post-Nazi epoch since 1945, trans. B. Cooper and A. Brown. London: Transaction Publishers.
Fernández-Villaverde, J, J Greenwood and N Guner (2011), "From Shame to Game in One Hundred Years: The Rise in Premarital Sex and its Destigmitization", CEPR Discussion Papers No. 8667.
Goschler, C (1991), “The Attitude towards Jews in Bavaria after the Second World War”, Leo Baeck Institute Yearbook, 36(1):443-458.
Guiso, L, P Sapienza, and L Zingales (2008), “Long Term Persistence”, NBER Working Papers No.14278.
Heilbronner, O (2004), “German or Nazi Anti-Semitism?”, in Dan Stone (ed.), The Historiography of the Holocaust, Palgrave Macmillan.
Nunn, N and L Wantchekon (2011), “The Slave Trade and the Origins of Mistrust in Africa”, American Economic Review, 101(7): 3221-3252.
Voigtländer, N and H-J Voth (2011), “Persecution Perpetuated: The Medieval Origins of Anti-Semitic Violence in Nazi Germany”, Working Paper Series.
Voigtländer, N and H-J Voth (2012), “Shaping Hatred: Anti-Semitic Attitudes in Germany, 1890-2006”, CEPR DP 8935.
Overview for main Variables, by Tercile of ALLVOTE
Notes: ASVOTE19C is the average vote for anti-Semitic parties between 1890 and 1912. NSVOTE20s is the average vote for the Nazi party in the 1920s (DVFP in 1924 and NSDAP in 1928). NSVOTE30s is the average vote for the Nazi party in the 1930s (elections in September 1930 and March 1933). ALLVOTE is the average of (standardized) ASVOTE19C , NSVOTE20s, NSVOTE30s. See Section III for details.
Anti-Semitism and xenophobia, by age cohort