Historical Antisemitism, Ethnic Specialization, and Financial Development

Historical Antisemitism, Ethnic Specialization, and Financial Development Abstract Historically, European Jews have specialized in financial services while being the victims of antisemitism. We find that the present-day demand for finance is lower in German counties where historical antisemitism was higher, compared to otherwise similar counties. Households in counties with high historical antisemitism have similar saving rates but invest less in stocks, hold lower saving deposits, and are less likely to get a mortgage to finance homeownership after controlling for wealth and a rich set of current and historical covariates. Present-day antisemitism and supply-side forces do not fully explain the results. Households in counties where historical antisemitism was higher distrust the financial sector more—a potential cultural externality of historical antisemitism that reduces wealth accumulation in the long run. 1. Introduction Financial development varies persistently across space, and this systematic variation might contribute to spatial inequalities, because households accumulate wealth through the stock market (Guiso et al., 2004a). Households’ trust in the financial sector might help explain variation in the demand for financial services across space, including within countries whose regions have faced the same regulatory environment and the same financial institutions for decades (Gennaioli et al., 2013; Guiso et al., 2009). Measuring the size and determinants of trust in finance is challenging, because such determinants should persist for decades despite the implementation of institutional reforms and place-based policies. A potential route to measure the spatial variation in households’ trust in the financial sector is to exploit its origins in history, because inter-ethnic tensions can produce persistent anti-minority sentiment, which can survive the physical presence of minorities themselves (Voigtlaender and Voth, 2012). If different ethnic groups specialized in different economic activities in the past, ethnic tensions could lead one group to discriminate against the activity led by the other group. Parallel to the discrimination against minorities, discrimination against economic activities might persist even after ethnic specialization fades (Jha, 2013; Grosfeld et al., 2013; Jha, 2014), hence capturing the deep-rooted variation in the localized trust in economic activities across space. We build on this framework and test whether the historical specialization of Jews in financial services, paired with persistent historical antisemitism across space, helps explain the present-day regional variation in the demand for finance. Our analysis focuses on Germany—where Jewish persecution has persisted across space for centuries (Voigtlaender and Voth, 2012)—as an ideal laboratory. Crucially, the ethnic specialization of Jews in finance was an important component of historical antisemitism. It led to the emergence of negative stereotypes attached to Jews but unrelated to religious creed, which Reuveni and Wobick-Segev (2011) label “economic antisemitism”. The baseline setting for our empirical tests is the Nazi period, which represented the most dramatic peak of economic antisemitism in Germany, because Jews were blatantly and unsubstantiatedly accused by the dictatorship of manipulating the German economy and causing economic depression “by means of their predominance in the stock exchange”.1 Jewish persecution also arguably peaked during the Nazi period, when the dictatorship required the broader German population to persecute and delate Jews through widespread and pervasive propaganda. We document that present-day households in German counties with higher anti-Jewish sentiment during the Nazi period participate in the stock market less than other households. Figure 1 plots the negative correlation between stock market participation and historical antisemitism at the county level conditional on a large set of historical and present-day observables. Present-day households in counties where historical antisemitism was one standard deviation higher are about 9% less likely to hold stocks. The size of this association is similar to the effect of holding a college degree, and college education is one of the most studied determinants of stock market participation (van Rooij et al., 2011). Figure 1 View largeDownload slide Historical antisemitism and present-day stock market participation. Figure 1 View largeDownload slide Historical antisemitism and present-day stock market participation. The baseline association between local historical antisemitism and stock market participation by present-day German households is a robust feature of the data, and survives a large set of robustness tests, such as restricting the geographic variation we use in the analysis and using alternative proxies for local antisemitism during the Nazi period. Although the defamation of Jews as stock market manipulators was salient in the press and popular culture even after the Jewish presence in banking had faded after the spread of credit unions and public savings banks at the end of the eighteenth and early nineteenth centuries (Köhler, 2005), historians documented a stereotype of Jews as moneylenders and bankers that was common in the German lands for centuries. The Jewish ethnic specialization in the provision of financial services was the main driver of this stereotype, and was largely driven by the fact that Christians were banned from lending money at interest throughout the Middle Ages. Consistent with the stereotype of Jews as moneylenders, we find historical antisemitism during the Nazi period is also negatively associated with households’ use of banking services today. Households in counties with higher historical antisemitism are 10% less likely to have a mortgage, but as likely as other households to own a house. The ratio of retail deposits over total assets of the banks in counties with higher historical antisemitism is 2% lower than in other counties, even if saving propensities and the concentration of bank branches do not differ systematically across counties. We also find suggestive evidence that households in counties with higher historical antisemitism hold a larger fraction of their wealth in cash. Several channels, both on the demand side and the supply side, might explain our baseline results. We therefore exploit our setting further by focusing on our historical approach. To further disentangle the hypothesis that historical antisemitism relates to present-day financial decisions from other explanations related to localized economic shocks during the Nazi period, we build on the fact that the spatial variation in historical antisemitism within Germany has persisted for centuries since the Middle Ages (Voigtlaender and Voth, 2012). We test whether deep-rooted measures of historical antisemitism based on the violence against local Jewish communities during the Black Death of 1349 can predict the present-day stock market participation of German households similar to antisemitism during the Nazi period. Indeed, households in counties that persecuted the local Jewish communities more as far back as in the Middle Ages are less likely to invest in stocks today, and the mere presence of Jews in a county in the distant past does not explain the effect. To assess the remaining concerns that unobservable dimensions not captured by our controls and robustness tests drive both historical antisemitism and present-day financial development, we exploit a historical natural experiment (D’Acunto, 2016). We instrument the probability that a county engaged in Jewish persecution in the past with its distant from the Rhine Valley, which captures the paths of forced migrations of the first Ashkenazi communities because of the Crusades. This test confirms our baseline results. In the last part of the article, we study a set of potential supply- and demand-side channels that might have transmitted the effect of historical antisemitism to present-day financial development. On the supply side, we note that all German counties have faced the same financial-sector regulation since the 19th century,2 but locally run independent financial institutions are still an important pillar of the German banking system. Historical antisemitism might have triggered the establishment of local banks of different quality and efficiency across counties. We find the present-day supply of finance and the present-day efficiency of the local banking sector do not vary systematically with historical antisemitism. We therefore conclude that the present-day local supply of financial services cannot fully explain our results. Alternatively, economic antisemitism might have worsened the historical local supply and efficiency of financial services. Even if these differences in supply had disappeared over time, present-day households might still be less accustomed to accessing financial services in counties with higher historical antisemitism. We collect data on the supply of financial services in the past, which we can measure at the county level in the nineteenth century, and we do not find evidence that this channel is relevant to our results. On the demand side, we first test whether present-day households that are antisemitic might still associate financial services with Jews, and thus invest less in stocks and demand fewer financial services. We test this channel using data on present-day antisemitism at the county level. We find that present-day antisemitism and stock market participation are negatively associated, as predicted by the long-term persistence of local antisemitism (Voigtlaender and Voth, 2013). At the same time, however, we find no association between present-day antisemitism and stock market participation after controlling for historical antisemitism. Our measures of historical antisemitism are arguably subject to higher measurement error than the measures of present-day antisemitism, and hence this test should bias us towards detecting an effect of present-day antisemitism on top of historical antisemitism even if such an autonomous effect did not exist. Thus, the results suggest the variation in antisemitism produced in the last decades has no role in explaining the present-day demand for finance. Apart from present-day antisemitism, historical anti-Jewish sentiment might correlate with other retrograde beliefs such as xenophobia, racism, or distrust of the unfamiliar, which we label collectively “backwardness”. Using county-level data on present-day xenophobic beliefs, we propose a set of tests that suggest our results are not consistent with these alternative demand-side channels. Motivated by Gennaioli et al. (2015), we move on to test for the possibility that a persistent cultural norm of distrust in finance, transmitted across generations, has developed more in counties in which historical antisemitism was stronger. Past households in counties with higher antisemitism might have developed a negative sentiment towards the economic activity in which Jews specialized, namely, financial services. This sentiment might have persisted to the present day even if its underlying determinants—specifically, the discrimination against Jews and the association of Jews with financial services—have faded. We use novel survey data on a representative sample of 1,000 present-day Germans to elicit their distrust in finance.3 The survey also elicits measures of risk tolerance and generalized trust at the individual level, both of which are strong determinants of stock market participation (Guiso et al., 2009). Indeed, present-day distrust in finance is higher for respondents in counties with higher historical antisemitism, even after controlling for their risk tolerance and average generalized trust. Consistent with a relevant role of this channel in explaining our results, households that distrust finance more also report that they invest less in stocks and bonds. Several theoretical channels, which we discuss in more detail in Section 7, are consistent with our results and interpretation. Overall, our findings suggest historical antisemitism in the form of economic antisemitism (Reuveni and Wobick-Segev, 2011) might have started a norm of distrust in finance, which has transmitted across generations and manifests itself in lower present-day demand for finance. 1.1. Related literature This article builds on several strands of literature. First, we build on the literature that studies the non-institutional determinants of the spatial variation of economic development. Banfield (1958) and Putnam (1993), emphasize the importance of demand-side factors, such as social capital and generalized trust, in explaining persistent localized differences in development. Guiso et al. (2004b) and Algan and Cahuc (2010) investigate these determinants of present-day financial development and economic growth. Gennaioli et al. (2013) use data from 110 countries covering 97% of the world’s GDP to show human capital is crucial in accounting for regional differences in development. This literature has introduced measures of present-day social capital and trust, and has documented the robust association between these dimensions and financial development. Our contribution to this literature is to propose a determinant of local demand-side characteristics that is not prone to the concern of reverse causality because of its deep-rooted nature. Secondly, we build on recent research documenting the long-run persistence of discrimination due to historical inter-ethnic tensions. Voigtlaender and Voth (2012) and Voigtlaender and Voth (2013) show that localized historical anti-Jewish sentiment persisted for centuries and can still be detected today. Anderson et al. (2017) show that low agricultural yield explains the time-series and spatial variation of pogroms against Jews across Europe from 1100 to 1800. Grosfeld et al. (2017) argue economic specialization combined with negative shocks was crucial to the emergence of pogroms against Jews. Becker and Pascali (2017) argue the Protestant Reformation led to the entry of non-Jews in moneylending, which reduced the incentives to persecute Jews. Previous work shows persistent anti-minority sentiment due to historical inter-ethnic tensions might be rooted in the historical economic specialization of ethnic groups. Jha (2014) shows that areas in Gujarat that enjoyed inter-ethnic economic complementarity in the past were less likely to engage in ethnic violence in 2002. Grosfeld et al. (2013) show a positive effect of the Pale of Settlement—a region of present-day Ukraine in which Jews were confined—on post-Soviet electoral support for left-wing parties and on generalized trust. Our results cannot reflect the generic anti-market beliefs studied by Grosfeld et al. (2013). In untabulated results, we find no effect of historical antisemitism on the electoral support for left-wing parties of present-day Germans. Moreover, Grosfeld et al. (2013) find a positive association between generic anti-market beliefs and generalized trust. Generalized trust increases households’ likelihood of demanding financial services (Guiso et al., 2009). Hence, if the anti-market beliefs proposed by Grosfeld et al. (2013) explained our results, we should detect a positive effect of historical anti-Jewish sentiment on households’ demand for financial services, which is the opposite of what we find. We see our demand-side results as complementary to recent work on the supply-side effects of Jewish persecution (e.g. Waldinger (2010), Acemoglu et al. (2011), Akbulut-Yuksel and Yuksel (2015)). An additional contribution of our article is to bring together the two lines of research described above. The first literature has focused mainly on documenting the role of present-day determinants in explaining present-day regional differences. The second literature has focused mainly on establishing the long-run persistence of political and sociological beliefs. Our article builds on both approaches to study the deep-rooted determinants of present-day variation in economic outcomes and the channels through which these determinants affect present-day economic outcomes. This step is relevant also as a basis for informing policy makers about the demand- and supply-side dimensions on which they might intervene to modify households’ economic behaviour. The article also relates to the body of research that uses historical natural experiments to understand present-day outcomes, surveyed by Spolaore and Wacziarg (2013) and Nunn (2014). For the case of financial outcomes, D’Acunto (2016) labels this nascent approach “History & Finance”. Recent contributions include Pascali (2016), who shows Jewish-managed banking in Southern Italy triggered the establishment of competing Christian financial institutions, whose influence on financial development is detectable today. Pierce and Snyder (2017) find that firms in African countries with higher historical extraction of slaves face lower access to formal and informal credit, whereas D’Acunto (2017) shows that spatial variation in basic education has persisted for centuries, and helps explain the present-day regional differences in income and innovation across European regions. Finally, we contribute to the literature on the stock-market-participation puzzle—the fact that many households do not actively invest in stocks despite the high expected returns in the stock market. Other explanations include background risk (Paxson, 1990; Guiso et al., 1996), social interactions (Hong et al., 2004), awareness (Guiso and Jappelli, 2005), generalized trust (Guiso et al., 2009), insurance motives (Gormley et al., 2010), financial literacy (van Rooij et al., 2011), macroeconomic experiences (Malmendier and Nagel, 2011), labour-income risk (Betermier et al., 2012), and corporate scandals (Giannetti and Wang, 2016). 2. Jewish Specialization in Finance and Historical Antisemitism Our analysis is based on two features of the history of Jewish minorities across Europe since the Middle Ages. On the one hand, Jews had specialized in the provision of financial services after the fall of the Roman Empire. The sorting of Jews into the mercantile and financial sectors started around the eighth century, largely because of their human capital and their tradition in contract enforcement (Botticini and Eckstein, 2012). Bans on lending money at interest for Christians and Muslims may have contributed to crystallizing this sorting. Pope Leo IX banned Christians from lending money at interest as far back as 1049, and Gratian formalized the ban in the Corpus Iuris Canonici in 1150. The human capital Jews had accumulated since the second century facilitated their sorting into trade and finance well before 1049. At the same time, an important push for the specialization of Jews in financial services and trade came also from the contemporaneous bans on owning land Jewish communities faced, which were common all over Europe during the Middle Ages (e.g. see Roth, 1938 and Roth, 1960). Financial activities run by Christians, such as the Medici family in Italy or the Fugger family in the German lands, were active since the fifteenth century. In the German lands, the oldest non-Jewish full-service bank, called Berenberg Bank, was founded in 1590. Despite these cases, the Catholic Church maintained a formal ban on usury for centuries. For instance, Pope Benedict XIV condemned firmly the sin of usury in his encyclical letter “Vix Perveni” of 1745. At the same time, the specialization of Jews in finance persisted even after the elimination of the ban on moneylending for all Christian denominations. For instance, in 1882, 3% of German workers were Jewish, but 23% of the overall financial sector workforce, and more than 85% of the brokers in the Berlin stock exchange, were Jewish (see Glagau, 1876 and Fritsch, 1892).4$$^{,}$$5 The drop in Jewish specialization in financial services was mainly driven by the diffusion of local private credit unions such as the Raiffeisenbanken and the Volksbanken, as well as the saving-bank system known as Sparkassen, which were owned and run by local governments. Both types of banks diffused swiftly throughout the German lands in the nineteenth century and the beginning of the twentieth century. Localized inter-ethnic tensions and outright violence against Jews also accompanied European Jewish communities since the Middle Ages. On top of religious antisemitism, the Jewish specialization in trading and finance led to the emergence of “economic antisemitism”, that is, a set of negative stereotypes related to the role of Jews in the economy (Reuveni and Wobick-Segev, 2011). To date, the historiography of economic antisemitism is still debating the relationship between economic antisemitism and the discrimination and persecution of Jewish communities over the centuries. In particular, the debate is still open on whether violence against religious minorities existed beyond the minorities’ occupational specializations, or whether hatred towards specific occupations led to the persecution of ethnic minorities that specialized in those occupations. Penslar (2001) argues the distrust of trade and the mercantile economy had roots in ancient Greece and Rome, and translated into distrust of Jews once Jewish communities sorted into running financial services. In sociology, Bonacich (1972, 1973), and Horowitz (1985) propose a theory of ethnic tensions deriving from the specialization of ethnicities in different economic occupations. Ethnicities specializing in middlemen activities are especially prone to being subject to inter-ethnic violence. Other historians argue the motivations for Jewish persecutions in Europe were at first mainly cultural, political, and religious (e.g. see Flannery, 1985). This position is based on the observation that the first recorded acts of violence against Jews, such as the Alexandria pogrom in 38 CE, happened when Jews had not yet sorted into the mercantile and financial sectors (Barclay, 1996). Historians who support the non-economic roots of the early instances of persecution against Jewish minorities argue that the hatred against Jews as economic exploiters of the Christian majority appeared at a later stage (e.g. Poliakov, 1975 and Perry and Schweitzer, 2002). 3. Data Our tests require that we define proxies for local historical antisemitism across German counties, and that we assess the association between historical antisemitism and the likelihood that present-day households access financial services. 3.1. Measures of historical antisemitism We propose three proxies for historical antisemitism. The first and main proxy is the first principal component of six measures of anti-Jewish violence in Voigtlaender and Voth (2012). The measures cover several types of acts of violence perpetrated against local Jewish communities in Germany in the 1920s and 1930s, which includes the Nazi period. The variables that enter the Voigtlaender-Voth principal component (VV P.C.; Historical Antisemitism) are as follows: (i) the number of documented pogroms, that is, recorded acts of physical violence, against Jewish communities in the 1920s based on the information in Alicke (2008); (ii) the share of votes for the far-right and strongly antisemitic Deutschvölkische Freiheitspartei (DVFP) in 1924, which obtained a large share of the then-banned Nazi Party (NSDAP), based on the election data in Falter and Hänisch (1988); (iii) the share of votes for the NSDAP in 1928, which is also based on the data in Falter and Hänisch (1988); (iv) the logarithm of the number of “letters to the editor” published by the Nazi newspaper Der Stürmer from 1935 to 1938; (v) the share of Jews deported in 1933; and (vi) a dummy variable that equals 1 if a synagogue was destroyed or damaged in the 1920s and 1930s in the location. We consider the authors’ extended sample of cities, which include all cities with Jewish communities during the Weimar Republic. The only difference between the original version of the VV P.C. and the one we use in this article is the level of aggregation of the information. Voigtlaender and Voth (2012) compute their variables at the city level, which we cannot do in this article, because the finest geographic partition for which we observe financial data of present-day households is county (Kreis). We therefore compute the principal component by aggregating the city-level variables at the county level. Aggregation consists of summing up the count variables (number of pogroms in the 1920s and number of letters to Stürmer), averaging the share variables (share of DVFP votes in 1924, share of NSDAP votes in 1928, and share of Jews deported in 1933), and defining a dummy variable equal to 1 if a synagogue was destroyed or damaged in the 1920s and 1930s in any city within a county. We also propose two additional proxies for our analysis that aim to capture localized and deep-rooted historical antisemitism at the time when Jews still had the monopoly on the provision of financial services. One proxy, Pogrom 1349 (Medieval Antisemitism), is also based on observed violence against Jewish communities. It is a dummy that equals 1 if any town in the county experienced at least one anti-Jewish pogrom during the years of the Black Death around 1349. The Black Death was arguably the worst pandemic in human history, and up to one half of the European population at the time may have died. Unsubstantiated theories on the origins of the pandemic diffused all over Europe. Accusations against Jews were common and led to persecution, especially in the German lands. Voigtlaender and Voth (2012) find the incidence of pogroms during the Black Death period predict the extent of Jewish persecution during the 1920s and 1930s at the town level. Similar to the principal-component measure, the level of resolution of our financial data dictates that we depart from the city-level analysis. Our third proxy for historical antisemitism is the mere presence of a Jewish community in each county at any point in time before 1300. This measure aims to capture the possibility that historical antisemitism arose in counties even if it did not necessarily express itself through pogroms or major acts of violence against Jews. This measure also allows us to assess separately the effects of exposure to Jews before the Black Death period and the actual explosion of anti-Jewish tensions at the time of the Black Death. The two medieval proxies allow us to track the origin of historical persecution, but the variation in these dummy variables is rather coarse. At the same time, the proxy from the Nazi period allows for a more granular variation across counties and occurs at a time when the association of Jews with the stock market, our main outcome variable, was strongest. For alternative sources of variation in anti-Jewish sentiment during the Nazi period, see the discussion in Section 1 of the Online Appendix. 3.2. Other data sources To run the tests in this article, we collect data from thirteen additional sources. The characteristics of German households are from the Socio-Economic Panel (SOEP) run by the Deutches Institut für Wirtschaftsforschung Berlin (DIW). The SOEP has conducted interviews on a yearly basis since 1984. For each wave, the SOEP includes households that have been interviewed in previous waves, as well as new households. Because we are interested in the cross-sectional association between historical antisemitism and financial development, we only include non-repeated observations when running the main analysis. A drawback of the SOEP data is that they do not provide the complete financial portfolios of households; hence, we cannot document how anti-Jewish sentiment affects every component of households’ financial portfolios. Moreover, the SOEP data set does not include measures of the household head’s risk aversion, financial literacy, or generalized trust, which the literature on stock market participation identifies as important determinants for investing in stocks. To address these shortcomings of the SOEP data, we show our results are robust to using the balance sheets of the German households in the 2011 wave of the Panel of Household Finances (PHF) run by the Deutsche Bundesbank. We can match the PHF sample with the historical data for 1,256 households across ninety-nine counties, and hence this data set is too small to be our main working sample. But in the PHF data, we can control directly for households’ wealth, as well as the elicited risk aversion, financial literacy, and generalized trust of household heads. To test for the effects of historical antisemitism on present-day bank deposits, we collect information on German banks’ balance sheets from Bankscope. We obtain county-level historical characteristics from the Ifo Prussian Economic History Database, described in detail by Becker et al. (2014). We also collect a set of present-day county-level controls: socio-demographics from DeStatis; the index of land quality from Ramankutty et al. (2002); and the coordinates of the centroid of each county from Eurostat, which we use to measure the Euclidean distance of each county from the Rhine Valley in our distance-based three-stage least-squares test we describe below. We construct a placebo test on the association between the distance from the Rhine Valley and stock market participation for French households to the West of the Rhine, using the micro data underlying the 2014 Enquete Patrimoine run by the Banque de France, which provides geo-coded information on the investment decisions of a representative set of present-day French households. To assess the association between present-day antisemitism and financial development, we use data on present-day antisemitism at the county level from the German Social Survey (ALLBUS), which gathered information on Germans’ attitudes towards Jews in 1996 and 2006. The data are described in detail in Voigtlaender and Voth (2013). We also use the ALLBUS data on present-day xenophobic attitudes to differentiate the role of antisemitism from generic xenophobia. To use these data, we arranged a special agreement between ALLBUS and DIW to merge these two proprietary data sources. Moreover, we use the micro data underlying the World Value Survey’s 2006 wave, in which households were asked about the importance of religion for their life, and other questions related to religiosity. This survey allows us to create regional-level measures of the importance of religion to present-day households irrespective of their creed or denomination. In the analysis of the channels that transmitted the long-run association between localized historical antisemitism and financial decisions, we also introduce three sources of data that are in large part new to research in economics: (i) data on the market structure, competition, and efficiency in German banking at the county level from the German Council of Economic Experts6, —an advisory institution to the German administration similar to the U.S. Council of Economic Advisers (see Koetter, 2013); (ii) data on the foundation dates of German’s Volksbanken and Raiffeisenbanken from the Hoppenstedt database, which allow us to construct the spatial diffusion pattern of credit unions across the German lands in the second half of the nineteenth century; and (iii) our own survey aimed at eliciting present-day German households’ distrust in financial services. We ran the survey through the company Clickworker because we are not aware of any data on a representative set of German households regarding their trust in financial services. We describe the survey design and characteristics in more detail in Section 7. These data include elicitation of several types of financial beliefs and attitudes, and are publicly available to any authors interested in their use. 3.3. Summary statistics The full sample of non-repeated households in the SOEP county-level data set includes 29,680 observations. The county of residence is not available for 2,655 households. Moreover, we are missing the county-level historical information for 9,207 households. The remaining missing observations are due to refusal to answer demographic questions, such as the income or age of the household head. We report the basic summary statistics for the variables in the main analysis in Table 1. The top panel of Table 1 describes the measures of historical antisemitism at the county level. We observe the emergence of pogroms during the Black Death and a county’s exposure to Jewish communities before 1300 in 307 counties, whereas we can compute the VV P.C. of Jewish persecution for 298 counties. During the Black Death period, 54% of counties faced a pogrom against the local Jewish community, whereas 92% of the counties were exposed to local Jewish presence at least once before 1300. In the regression analysis, we assign the county-level value of each variables to all SOEP households residing in the county. Table 1. Summary statistics       Obs.  Mean  Std  Min  Max        (1)  (2)  (3)  (4)  (5)  Antisemitism                    Historical antisemitism (VV P.C.)     298  –0.61  1.02  –3.84  2.25  Medieval antisemitism (Pogrom1349)     307  0.54  0.50  0.00  1.00  Exposure to Jews pre-1300     307  0.92  0.27  0.00  1.00  County characteristics                    Log Jews 1933     307  5.43  1.76  0.00  11.99  Percentage unemployed 1933     307  16.24  8.26  2.62  40.52  Percentage blue collars 1933     307  42.86  11.32  16.49  72.40  Percentage self-employed 1933     307  21.01  4.55  9.10  32.74  Percentage Catholics 1925     307  36.75  34.62  0.50  98.77  Latitude     443  50.64  1.72  47.95  54.03  Land-quality index     442  0.56  0.14  0.31  0.87  Income per capita 2005     413  17,294  2,280  12,846  27,253  Population density 2005     425  2,308  2,549  546  29,036  Percentage college graduates 2005     430  24.49  5.03  17.60  34.60  Household characteristics                    Holds stocks     26,761  0.16  0.37  0.00  1.00  Homeowner     27,064  0.39  0.49  0.00  1.00  Has life insurance     26,761  0.47  0.50  0.00  1.00  Income     26,761  31,522  26,614  $$-$$36  986,400  Age     21,981  48.62  17.64  17.00  97.00  Female     21,982  0.49  0.50  0.00  1.00  Single     27,064  0.18  0.38  0.00  1.00  High school or higher     27,079  0.77  0.42  0.00  1.00  Eastern Germany     27,079  0.13  0.34  0.00  1.00        Obs.  Mean  Std  Min  Max        (1)  (2)  (3)  (4)  (5)  Antisemitism                    Historical antisemitism (VV P.C.)     298  –0.61  1.02  –3.84  2.25  Medieval antisemitism (Pogrom1349)     307  0.54  0.50  0.00  1.00  Exposure to Jews pre-1300     307  0.92  0.27  0.00  1.00  County characteristics                    Log Jews 1933     307  5.43  1.76  0.00  11.99  Percentage unemployed 1933     307  16.24  8.26  2.62  40.52  Percentage blue collars 1933     307  42.86  11.32  16.49  72.40  Percentage self-employed 1933     307  21.01  4.55  9.10  32.74  Percentage Catholics 1925     307  36.75  34.62  0.50  98.77  Latitude     443  50.64  1.72  47.95  54.03  Land-quality index     442  0.56  0.14  0.31  0.87  Income per capita 2005     413  17,294  2,280  12,846  27,253  Population density 2005     425  2,308  2,549  546  29,036  Percentage college graduates 2005     430  24.49  5.03  17.60  34.60  Household characteristics                    Holds stocks     26,761  0.16  0.37  0.00  1.00  Homeowner     27,064  0.39  0.49  0.00  1.00  Has life insurance     26,761  0.47  0.50  0.00  1.00  Income     26,761  31,522  26,614  $$-$$36  986,400  Age     21,981  48.62  17.64  17.00  97.00  Female     21,982  0.49  0.50  0.00  1.00  Single     27,064  0.18  0.38  0.00  1.00  High school or higher     27,079  0.77  0.42  0.00  1.00  Eastern Germany     27,079  0.13  0.34  0.00  1.00  Notes: This table reports summary statistics for measures of historical antisemitism in Germany, and for the characteristics of households and counties where households live. Each observation is a German household interviewed by SOEP any time between 1984 and 2011. For each variable, the table reports the number of observations for which the variable is observed, its mean, standard deviation, minimal, and maximal values. The table reports statistics for households for which we observe the county of residence. We exclude repeated observations. Historical Antisemitism (VV P.C.) is the principal component of Voigtlaender and Voth (2012) of their different measures of Jewish persecution in the 1920s–1930s, Medieval Antisemitism (Pogrom 1349) is a dummy variable that equals 1 if a pogrom against Jews occurred during the Black Death, and Exposure to Jews pre-1300 is a dummy variable that equals 1 if a county hosted a Jewish community before 1300. Table 1. Summary statistics       Obs.  Mean  Std  Min  Max        (1)  (2)  (3)  (4)  (5)  Antisemitism                    Historical antisemitism (VV P.C.)     298  –0.61  1.02  –3.84  2.25  Medieval antisemitism (Pogrom1349)     307  0.54  0.50  0.00  1.00  Exposure to Jews pre-1300     307  0.92  0.27  0.00  1.00  County characteristics                    Log Jews 1933     307  5.43  1.76  0.00  11.99  Percentage unemployed 1933     307  16.24  8.26  2.62  40.52  Percentage blue collars 1933     307  42.86  11.32  16.49  72.40  Percentage self-employed 1933     307  21.01  4.55  9.10  32.74  Percentage Catholics 1925     307  36.75  34.62  0.50  98.77  Latitude     443  50.64  1.72  47.95  54.03  Land-quality index     442  0.56  0.14  0.31  0.87  Income per capita 2005     413  17,294  2,280  12,846  27,253  Population density 2005     425  2,308  2,549  546  29,036  Percentage college graduates 2005     430  24.49  5.03  17.60  34.60  Household characteristics                    Holds stocks     26,761  0.16  0.37  0.00  1.00  Homeowner     27,064  0.39  0.49  0.00  1.00  Has life insurance     26,761  0.47  0.50  0.00  1.00  Income     26,761  31,522  26,614  $$-$$36  986,400  Age     21,981  48.62  17.64  17.00  97.00  Female     21,982  0.49  0.50  0.00  1.00  Single     27,064  0.18  0.38  0.00  1.00  High school or higher     27,079  0.77  0.42  0.00  1.00  Eastern Germany     27,079  0.13  0.34  0.00  1.00        Obs.  Mean  Std  Min  Max        (1)  (2)  (3)  (4)  (5)  Antisemitism                    Historical antisemitism (VV P.C.)     298  –0.61  1.02  –3.84  2.25  Medieval antisemitism (Pogrom1349)     307  0.54  0.50  0.00  1.00  Exposure to Jews pre-1300     307  0.92  0.27  0.00  1.00  County characteristics                    Log Jews 1933     307  5.43  1.76  0.00  11.99  Percentage unemployed 1933     307  16.24  8.26  2.62  40.52  Percentage blue collars 1933     307  42.86  11.32  16.49  72.40  Percentage self-employed 1933     307  21.01  4.55  9.10  32.74  Percentage Catholics 1925     307  36.75  34.62  0.50  98.77  Latitude     443  50.64  1.72  47.95  54.03  Land-quality index     442  0.56  0.14  0.31  0.87  Income per capita 2005     413  17,294  2,280  12,846  27,253  Population density 2005     425  2,308  2,549  546  29,036  Percentage college graduates 2005     430  24.49  5.03  17.60  34.60  Household characteristics                    Holds stocks     26,761  0.16  0.37  0.00  1.00  Homeowner     27,064  0.39  0.49  0.00  1.00  Has life insurance     26,761  0.47  0.50  0.00  1.00  Income     26,761  31,522  26,614  $$-$$36  986,400  Age     21,981  48.62  17.64  17.00  97.00  Female     21,982  0.49  0.50  0.00  1.00  Single     27,064  0.18  0.38  0.00  1.00  High school or higher     27,079  0.77  0.42  0.00  1.00  Eastern Germany     27,079  0.13  0.34  0.00  1.00  Notes: This table reports summary statistics for measures of historical antisemitism in Germany, and for the characteristics of households and counties where households live. Each observation is a German household interviewed by SOEP any time between 1984 and 2011. For each variable, the table reports the number of observations for which the variable is observed, its mean, standard deviation, minimal, and maximal values. The table reports statistics for households for which we observe the county of residence. We exclude repeated observations. Historical Antisemitism (VV P.C.) is the principal component of Voigtlaender and Voth (2012) of their different measures of Jewish persecution in the 1920s–1930s, Medieval Antisemitism (Pogrom 1349) is a dummy variable that equals 1 if a pogrom against Jews occurred during the Black Death, and Exposure to Jews pre-1300 is a dummy variable that equals 1 if a county hosted a Jewish community before 1300. The middle panel of Table 1 reports the other observables measured at the county level, whereas the bottom panel describes household-level variables. The average fraction of households owning stocks between 1984 and 2011 is 16%. The average age of the person who makes financial decisions is forty-nine years. Thirty-nine percent of responding households are homeowners, and the average self-reported income is 31,355 euros. The SOEP survey does not ask households for an estimate of their overall wealth. We use income and homeownership to proxy for wealth. About 77% of responding households have a high school degree or higher levels of education. Moreover, about 13% of the households we observe reside in Eastern Germany.7 Figure 2 depicts the properties of historical antisemitism and present-day stock market participation at the county level. To allow easier interpretation of the magnitudes and variation of historical antisemitism in the cross-section of counties, we consider the ratio of the local Jewish population as of 1933 that was deported during the Nazi period as a proxy for historical antisemitism. Panels (a) and (b) of Figure 2 show the spatial distribution of the share of the deported Jewish population during the Nazi period and of the average ratio of households that own stocks from 1984 to 2011. In both maps, the darker the county, the higher the value of the variable. The data are not available for blank counties. Relative deportations of Jews during the Nazi period were higher in western counties. Stock market participation is higher in the south and in the north. As expected, participation is systematically lower in Eastern Germany. Panels (c) and (d) of Figure 2 plot the densities of the ratio of deported Jews, which obtains over the full range of [0,100], and present-day stock market participation, both measured at the county level. Panel (e) plots the correlation between the ratio of deported Jews and the average ratio of households that own stocks from 1984 to 2011, which is negative ($$-$$0.13, $$p$$-value=0.03). Note Panel (e) corresponds to the figure we reported in the “Introduction”, but we replace the VV P.C. of Jewish persecution with the ratio of Jews deported during the Nazi period. Panel (f) of Figure 2 shows the average participation across counties with and without pogroms during the Black Death. Participation is higher in counties with no pogroms, but a t-test for the difference between the two means does not reject the null that the means are equal. We find the negative, although statistically insignificant, association between experiencing a pogrom around 1349 and present-day stock participation encouraging. Of course, the non-significant difference in stock market participation in the raw data might reflect substantial variation in important determinants of participation across counties, such as income, age, or education, which is why we can only assess the precision of this negative association by running a multivariate analysis that keeps constant other determinants of participation across counties. Figure 2 View largeDownload slide Data properties: historical antisemitism and stock market participation. Notes: In Panels (a) and (b), the darker a county is, the higher the value of the depicted variable. Blank counties are those for which the data are not available. Panel (a) plots the ratio of Jews deported during the Nazi period over the Jewish population in the county as of 1933. Panel (b) plots the average yearly ratio of households who have invested in stocks from 1984 to 2011. Panels (c) and (d) plot the sample distributions of the same measures as above. Panel (e) depicts the unconditional correlation between stock market participation and the ratio of Jews deported during the Nazi period over the Jewish population in the county as of 1933. Panel (f) shows the mean stock market participation in counties that experienced and did not experience a pogrom in 1349. Figure 2 View largeDownload slide Data properties: historical antisemitism and stock market participation. Notes: In Panels (a) and (b), the darker a county is, the higher the value of the depicted variable. Blank counties are those for which the data are not available. Panel (a) plots the ratio of Jews deported during the Nazi period over the Jewish population in the county as of 1933. Panel (b) plots the average yearly ratio of households who have invested in stocks from 1984 to 2011. Panels (c) and (d) plot the sample distributions of the same measures as above. Panel (e) depicts the unconditional correlation between stock market participation and the ratio of Jews deported during the Nazi period over the Jewish population in the county as of 1933. Panel (f) shows the mean stock market participation in counties that experienced and did not experience a pogrom in 1349. The Online Appendix describes additional characteristics of the raw data. Figure A.1 shows the spatial distribution of pogroms against Jews during the Black Death. In Figure A.2, we plot the correlations between stockholdings and additional proxies for historical antisemitism in the raw data, all of which are negative. We provide code and data we use in the analysis in the Supplementary Appendix. 4. Historical Antisemitism and Stock Market Participation In the baseline analysis, we estimate the association between historical antisemitism measured at the county level and stock market participation by German households from 1984 to 2011. The following is our most general specification:   \begin{align}\label{reducedform} Pr(Holds Stocks_{ikt} | X_{ikt}, K_{k})& = \Phi(\alpha + \beta Historical~Antisemitism_{k} + X_{ikt}'\gamma + K_{k}'\delta\notag\\ &\quad + Income\_deciles + \eta_{t} + \epsilon_{ikt}), \end{align} (1) where $$Holds Stocks_{ikt}$$ is a dummy that equals 1 if household $$i$$ in county $$k$$ and surveyed in wave $$t$$ holds any stocks, and $$Historical~Antisemitism_{k}$$ is one of the proxies for historical antisemitism we describe in Section 3. $$X_{ikt}$$ includes the following individual-level controls: gender, single status, age (second-degree polynomial), and dummies for college education, homeownership, and investment in life insurance. $$K_{k}$$ includes the following county-level current and historical controls: latitude, income per capita, share of college-educated population, index of quality of cultivable land, log of population in 1933, log of Jewish population in 1933, share of population employed in the retail sector in 1933, share of population employed in manufacturing in 1933, and share of Catholic population in 1925. $$Income\_deciles$$ are dummies indicating the decile of the income distribution to which the household belongs, and $$\Phi$$ is the standard normal cdf.8$$\eta_{t}$$ are a set of survey-wave group fixed effects, each capturing a group of four adjacent years.9 We allow for correlation of unknown form across residuals at the county level, because attributing county-level measures to each household induces a mechanical correlation of residuals across households in the same county. Table 2 reports the average marginal effects for our baseline specification. All the variables are standardized, with the exception of dummy variables. Columns (1) and (2) report the results for the baseline specification on the full sample. In column (1), we only include the logarithm of the number of Jews residing in each county in 1933 to scale the persecution measure by the size of the local Jewish community, and hence the scope for persecution, on the right-hand side. A one-standard-deviation increase in the VV P.C. (1.02) is associated with 0.9-percentage-point-lower stock market participation. In column (2), we add the full set of historical and present-day controls, a dummy that equals 1 for households in Eastern Germany, and survey-wave group fixed effects. Adding this set of controls increases the size of the negative association between historical antisemitism and stock market participation to 1.4 percentage points. Because the average stock market participation rate in our sample is 16%, this association corresponds to about 9% of the average participation. Table 2. Historical antisemitism and present-day stock market participation       Full sample  Full sample  Only Western Germany  Adding longitude  Excl. least populated  Excl. high inequality  Excl. poorest counties  Excl. least educated        (1)  (2)  (3)  (4)  (5)  (6)  (7)  (8)  Historical antisemitism     –0.009**  –0.014***  –0.015***  –0.010**  –0.013***  –0.016***  –0.020***  –0.013**  (VV P.C.)     (0.005)  (0.004)  (0.005)  (0.004)  (0.005)  (0.005)  (0.005)  (0.005)  Log Jews 1933     0.007***  0.003  0.000  –0.001  0.000  0.000  –0.003  0.004        (0.003)  (0.005)  (0.005)  (0.005)  (0.007)  (0.006)  (0.007)  (0.006)  Percentage        0.000  0.000  0.000  0.000  0.000  0.000  0.000      Catholics 1925        (0.000)  (0.000)  (0.000)  (0.000)  (0.000)  (0.000)  (0.000)  Age        –0.002*  –0.002  –0.002**  –0.002  –0.003***  –0.003**  –0.003**           (0.001)  (0.001)  (0.001)  (0.001)  (0.001)  (0.001)  (0.001)  Age$${^2}$$/100        0.043***  0.044***  0.043***  0.041***  0.045***  0.048***  0.047**           (0.011)  (0.012)  (0.011)  (0.012)  (0.012)  (0.012)  (0.012)  Female        –0.001  0.000  –0.002  0.000  0.000  –0.001  0.004           (0.006)  (0.006)  (0.006)  (0.007)  (0.007)  (0.007)  (0.007)  Single        0.060***  0.066***  0.053***  0.056***  0.047***  0.065***  0.050***           (0.010)  (0.010)  (0.009)  (0.010)  (0.010)  (0.010)  (0.011)  College        0.011***  0.011***  0.012***  0.013***  0.013***  0.013***  0.010**           (0.004)  (0.004)  (0.004)  (0.004)  (0.004)  (0.004)  (0.004)  Eastern Germany        –0.024     –0.033*  –0.006  –0.030  –0.045*  –0.020           (0.021)     (0.018)  (0.028)  (0.019)  (0.024)  (0.020)  Income p.c. 2005        0.005***  0.005***  0.005***  0.008***  0.003  0.007***  0.007***           (0.002)  (0.002)  (0.002)  (0.002)  (0.003)  (0.002)  (0.002)  Percentage College        0.001  0.001  –0.001  0.000  0.001  –0.001  0.002      graduates 2005        (0.001)  (0.001)  (0.001)  (0.002)  (0.002)  (0.002)  (0.002)  Income deciles        X  X  X  X  X  X  X  Other individual controls        X  X  X  X  X  X  X  Other historical controls        X  X  X  X  X  X  X  Other regional controls        X  X  X  X  X  X  X  Wave groups f.e.        X  X  X  X  X  X  X  Observations     13,599  13,599  12,701  13,599  11,168  11,012  11,144  10,128  N. of clusters     261  261  226  261  188  206  200  191  (Pseudo-) R$$^{2}$$     0.001  0.105  0.106  0.093  0.095  0.094  0.094  0.090        Full sample  Full sample  Only Western Germany  Adding longitude  Excl. least populated  Excl. high inequality  Excl. poorest counties  Excl. least educated        (1)  (2)  (3)  (4)  (5)  (6)  (7)  (8)  Historical antisemitism     –0.009**  –0.014***  –0.015***  –0.010**  –0.013***  –0.016***  –0.020***  –0.013**  (VV P.C.)     (0.005)  (0.004)  (0.005)  (0.004)  (0.005)  (0.005)  (0.005)  (0.005)  Log Jews 1933     0.007***  0.003  0.000  –0.001  0.000  0.000  –0.003  0.004        (0.003)  (0.005)  (0.005)  (0.005)  (0.007)  (0.006)  (0.007)  (0.006)  Percentage        0.000  0.000  0.000  0.000  0.000  0.000  0.000      Catholics 1925        (0.000)  (0.000)  (0.000)  (0.000)  (0.000)  (0.000)  (0.000)  Age        –0.002*  –0.002  –0.002**  –0.002  –0.003***  –0.003**  –0.003**           (0.001)  (0.001)  (0.001)  (0.001)  (0.001)  (0.001)  (0.001)  Age$${^2}$$/100        0.043***  0.044***  0.043***  0.041***  0.045***  0.048***  0.047**           (0.011)  (0.012)  (0.011)  (0.012)  (0.012)  (0.012)  (0.012)  Female        –0.001  0.000  –0.002  0.000  0.000  –0.001  0.004           (0.006)  (0.006)  (0.006)  (0.007)  (0.007)  (0.007)  (0.007)  Single        0.060***  0.066***  0.053***  0.056***  0.047***  0.065***  0.050***           (0.010)  (0.010)  (0.009)  (0.010)  (0.010)  (0.010)  (0.011)  College        0.011***  0.011***  0.012***  0.013***  0.013***  0.013***  0.010**           (0.004)  (0.004)  (0.004)  (0.004)  (0.004)  (0.004)  (0.004)  Eastern Germany        –0.024     –0.033*  –0.006  –0.030  –0.045*  –0.020           (0.021)     (0.018)  (0.028)  (0.019)  (0.024)  (0.020)  Income p.c. 2005        0.005***  0.005***  0.005***  0.008***  0.003  0.007***  0.007***           (0.002)  (0.002)  (0.002)  (0.002)  (0.003)  (0.002)  (0.002)  Percentage College        0.001  0.001  –0.001  0.000  0.001  –0.001  0.002      graduates 2005        (0.001)  (0.001)  (0.001)  (0.002)  (0.002)  (0.002)  (0.002)  Income deciles        X  X  X  X  X  X  X  Other individual controls        X  X  X  X  X  X  X  Other historical controls        X  X  X  X  X  X  X  Other regional controls        X  X  X  X  X  X  X  Wave groups f.e.        X  X  X  X  X  X  X  Observations     13,599  13,599  12,701  13,599  11,168  11,012  11,144  10,128  N. of clusters     261  261  226  261  188  206  200  191  (Pseudo-) R$$^{2}$$     0.001  0.105  0.106  0.093  0.095  0.094  0.094  0.090  Notes: This table reports average marginal effects computed after estimating the following probit specification:   \begin{equation*} Pr(Holds Stocks_{ik} | X_{ik}, K_{k}) = \Phi(\alpha + \beta \times Historical~Antisemitism_{k} + X_{ik}' \times \gamma + K_{k}' \times \delta + Income\_deciles + \eta_{t} + \epsilon_{ik}). \end{equation*} Each observation is a German household interviewed by SOEP between 1984 and 2011. In all columns, the dependent variable is a dummy that equals 1 if the household holds stocks. The main covariate of interest, $$Historical~Antisemitism$$, is the VV P.C. of Jewish persecution in the 1920s–1930s,. $$X_{ik}$$ includes the following individual-level controls: gender, single-status dummy, age (second-degree polynomial), college-education dummy. Other individual controls include: homeownership dummy and life and social insurance dummy. Other historical controls include: log of population in 1933, ratio of population employed in the retail sector in 1933, and ratio of population employed in manufacturing in 1933. Other regional controls include: population density, latitude, index of quality of cultivable land. $$Income\_deciles$$ are dummies indicating the decile of the income distribution to which the household belongs, and $$\Phi$$ is the standard normal cdf. We cluster standard errors at the county level. $$^{*} p<0.10, {}^{**} p<0.05, {}^{***} p<0.01$$. Table 2. Historical antisemitism and present-day stock market participation       Full sample  Full sample  Only Western Germany  Adding longitude  Excl. least populated  Excl. high inequality  Excl. poorest counties  Excl. least educated        (1)  (2)  (3)  (4)  (5)  (6)  (7)  (8)  Historical antisemitism     –0.009**  –0.014***  –0.015***  –0.010**  –0.013***  –0.016***  –0.020***  –0.013**  (VV P.C.)     (0.005)  (0.004)  (0.005)  (0.004)  (0.005)  (0.005)  (0.005)  (0.005)  Log Jews 1933     0.007***  0.003  0.000  –0.001  0.000  0.000  –0.003  0.004        (0.003)  (0.005)  (0.005)  (0.005)  (0.007)  (0.006)  (0.007)  (0.006)  Percentage        0.000  0.000  0.000  0.000  0.000  0.000  0.000      Catholics 1925        (0.000)  (0.000)  (0.000)  (0.000)  (0.000)  (0.000)  (0.000)  Age        –0.002*  –0.002  –0.002**  –0.002  –0.003***  –0.003**  –0.003**           (0.001)  (0.001)  (0.001)  (0.001)  (0.001)  (0.001)  (0.001)  Age$${^2}$$/100        0.043***  0.044***  0.043***  0.041***  0.045***  0.048***  0.047**           (0.011)  (0.012)  (0.011)  (0.012)  (0.012)  (0.012)  (0.012)  Female        –0.001  0.000  –0.002  0.000  0.000  –0.001  0.004           (0.006)  (0.006)  (0.006)  (0.007)  (0.007)  (0.007)  (0.007)  Single        0.060***  0.066***  0.053***  0.056***  0.047***  0.065***  0.050***           (0.010)  (0.010)  (0.009)  (0.010)  (0.010)  (0.010)  (0.011)  College        0.011***  0.011***  0.012***  0.013***  0.013***  0.013***  0.010**           (0.004)  (0.004)  (0.004)  (0.004)  (0.004)  (0.004)  (0.004)  Eastern Germany        –0.024     –0.033*  –0.006  –0.030  –0.045*  –0.020           (0.021)     (0.018)  (0.028)  (0.019)  (0.024)  (0.020)  Income p.c. 2005        0.005***  0.005***  0.005***  0.008***  0.003  0.007***  0.007***           (0.002)  (0.002)  (0.002)  (0.002)  (0.003)  (0.002)  (0.002)  Percentage College        0.001  0.001  –0.001  0.000  0.001  –0.001  0.002      graduates 2005        (0.001)  (0.001)  (0.001)  (0.002)  (0.002)  (0.002)  (0.002)  Income deciles        X  X  X  X  X  X  X  Other individual controls        X  X  X  X  X  X  X  Other historical controls        X  X  X  X  X  X  X  Other regional controls        X  X  X  X  X  X  X  Wave groups f.e.        X  X  X  X  X  X  X  Observations     13,599  13,599  12,701  13,599  11,168  11,012  11,144  10,128  N. of clusters     261  261  226  261  188  206  200  191  (Pseudo-) R$$^{2}$$     0.001  0.105  0.106  0.093  0.095  0.094  0.094  0.090        Full sample  Full sample  Only Western Germany  Adding longitude  Excl. least populated  Excl. high inequality  Excl. poorest counties  Excl. least educated        (1)  (2)  (3)  (4)  (5)  (6)  (7)  (8)  Historical antisemitism     –0.009**  –0.014***  –0.015***  –0.010**  –0.013***  –0.016***  –0.020***  –0.013**  (VV P.C.)     (0.005)  (0.004)  (0.005)  (0.004)  (0.005)  (0.005)  (0.005)  (0.005)  Log Jews 1933     0.007***  0.003  0.000  –0.001  0.000  0.000  –0.003  0.004        (0.003)  (0.005)  (0.005)  (0.005)  (0.007)  (0.006)  (0.007)  (0.006)  Percentage        0.000  0.000  0.000  0.000  0.000  0.000  0.000      Catholics 1925        (0.000)  (0.000)  (0.000)  (0.000)  (0.000)  (0.000)  (0.000)  Age        –0.002*  –0.002  –0.002**  –0.002  –0.003***  –0.003**  –0.003**           (0.001)  (0.001)  (0.001)  (0.001)  (0.001)  (0.001)  (0.001)  Age$${^2}$$/100        0.043***  0.044***  0.043***  0.041***  0.045***  0.048***  0.047**           (0.011)  (0.012)  (0.011)  (0.012)  (0.012)  (0.012)  (0.012)  Female        –0.001  0.000  –0.002  0.000  0.000  –0.001  0.004           (0.006)  (0.006)  (0.006)  (0.007)  (0.007)  (0.007)  (0.007)  Single        0.060***  0.066***  0.053***  0.056***  0.047***  0.065***  0.050***           (0.010)  (0.010)  (0.009)  (0.010)  (0.010)  (0.010)  (0.011)  College        0.011***  0.011***  0.012***  0.013***  0.013***  0.013***  0.010**           (0.004)  (0.004)  (0.004)  (0.004)  (0.004)  (0.004)  (0.004)  Eastern Germany        –0.024     –0.033*  –0.006  –0.030  –0.045*  –0.020           (0.021)     (0.018)  (0.028)  (0.019)  (0.024)  (0.020)  Income p.c. 2005        0.005***  0.005***  0.005***  0.008***  0.003  0.007***  0.007***           (0.002)  (0.002)  (0.002)  (0.002)  (0.003)  (0.002)  (0.002)  Percentage College        0.001  0.001  –0.001  0.000  0.001  –0.001  0.002      graduates 2005        (0.001)  (0.001)  (0.001)  (0.002)  (0.002)  (0.002)  (0.002)  Income deciles        X  X  X  X  X  X  X  Other individual controls        X  X  X  X  X  X  X  Other historical controls        X  X  X  X  X  X  X  Other regional controls        X  X  X  X  X  X  X  Wave groups f.e.        X  X  X  X  X  X  X  Observations     13,599  13,599  12,701  13,599  11,168  11,012  11,144  10,128  N. of clusters     261  261  226  261  188  206  200  191  (Pseudo-) R$$^{2}$$     0.001  0.105  0.106  0.093  0.095  0.094  0.094  0.090  Notes: This table reports average marginal effects computed after estimating the following probit specification:   \begin{equation*} Pr(Holds Stocks_{ik} | X_{ik}, K_{k}) = \Phi(\alpha + \beta \times Historical~Antisemitism_{k} + X_{ik}' \times \gamma + K_{k}' \times \delta + Income\_deciles + \eta_{t} + \epsilon_{ik}). \end{equation*} Each observation is a German household interviewed by SOEP between 1984 and 2011. In all columns, the dependent variable is a dummy that equals 1 if the household holds stocks. The main covariate of interest, $$Historical~Antisemitism$$, is the VV P.C. of Jewish persecution in the 1920s–1930s,. $$X_{ik}$$ includes the following individual-level controls: gender, single-status dummy, age (second-degree polynomial), college-education dummy. Other individual controls include: homeownership dummy and life and social insurance dummy. Other historical controls include: log of population in 1933, ratio of population employed in the retail sector in 1933, and ratio of population employed in manufacturing in 1933. Other regional controls include: population density, latitude, index of quality of cultivable land. $$Income\_deciles$$ are dummies indicating the decile of the income distribution to which the household belongs, and $$\Phi$$ is the standard normal cdf. We cluster standard errors at the county level. $$^{*} p<0.10, {}^{**} p<0.05, {}^{***} p<0.01$$. 4.1. Robustness German counties are likely to differ along several dimensions, such as geography, history, and the quality of current and historical institutions. For these reasons, in the rest of the Table 2, we assess the robustness of the baseline negative association between historical antisemitism at the county level and the likelihood that present-day households living in those counties participate in the stock market today. We first verify that the baseline results hold when only considering counties in West Germany. Note that the baseline specifications already include a dummy variable for whether a county was part of Eastern Germany after the Second World War, but one might still be concerned about systematic nonlinear differences in the effect across the two areas. In column (3) of Table 2, we find our results do not change if we look only at Western counties. In our second test, we add the longitude of the counties’ centroids to the baseline specification as a direct control. This control is motivated by the fact that important shocks related to counties’ longitude had differential long-run effects on the growth of German regions. For instance, Acemoglu et al. (2011) show that institutions imposed by the French on German areas closer to the French border after the French Revolution had a long-run effect on growth through their effect on institutions. In column (4) of Table 2, we find our results are similar if we include longitude explicitly as a control in the baseline specification. Note that the shocks that had long-run effects on growth and were correlated with longitude happened after the Black Death of 1349. Such shocks then would only be able to explain the results if their geographic dispersion were highly correlated with the geographic distribution of medieval pogroms. Finally, in columns (5)–(8) of Table 2, we exclude groups of counties, which perform worst based on economic indicators in the present day. In column (5), we exclude the bottom quarter of counties by population density and hence the most rural counties; in column (6), we exclude the top quarter of counties by income inequality; in column (7), we exclude the bottom quarter of counties by average income and hence the poorest counties; and in column (8), we exclude the bottom quarter of counties by share of college-educated inhabitants and hence the least educated counties. Across all subsamples, we do not detect substantial differences compared to our baseline results. We propose additional robustness tests in Table A.1 and Table A.2 of the Online Appendix, and confirm the negative association between historical antisemitism and present-day stock market participation is a robust feature of the data. 4.2. Alternative samples and sources of variation The SOEP sample does not allow us to keep constant dimensions that previous research has shown to be important determinants of financial decision-making. Prior research shows financial literacy (van Rooij et al., 2011), risk aversion (Samuelson, 1969), and household wealth are first-order determinants of stock market participation. Moreover, an important determinant of historical antisemitism could be households’ religiosity, which might have also persisted over time irrespective of households’ religious denomination, and hence might confound our interpretation of the baseline results. To assess the extent to which any of these dimensions might explain our results, we replicate the cross-sectional analysis on the German households in the 2011 wave of the PHF. The size of the PHF sample is more than one order of magnitude lower than the SOEP sample, and we can only exploit the variation in Jewish persecution across ninety-nine German counties for which we have both historical data on persecution and PHF observations. For these reasons, we cannot use the PHF sample as the main sample in our analysis, but we believe it provides a useful alternative data set to assess the robustness of our baseline results. The PHF questionnaire asks households to provide an estimate of their overall wealth. It also elicits households’ financial literacy and risk aversion using qualitative scales, as well as the frequency with which respondents attend religious functions, irrespective of their religious denomination. All the results based on the PHF sample are reported in Table A.3 in the Online Appendix. In column (1) of Table A.3, we replicate our baseline results by estimating the specification in equation (1) and augmenting the right-hand side with direct measures of financial literacy, risk tolerance, and the religiosity of the respondent, as well as a full set of dummies for wealth deciles. As expected, the measures of risk tolerance and financial literacy are positively associated with the likelihood of holding stocks, on top of the effect of being male and holding a college degree. We estimate a larger negative association between historical antisemitism and the likelihood that respondents hold stocks in the PHF sample than in the SOEP sample, even after controlling for additional important determinants of stock market participation. One-standard-deviation-higher historical antisemitism in the county decreases the likelihood that the household owns stocks by 7 percentage points, which is about 24% of the average likelihood of holding stocks in this sample. The likelihood of holding stocks in the PHF sample is 29%, which is similar to the likelihood of holding stocks for the SOEP households in the 2010 wave (28%). In section 1 of the Online Appendix, we also propose an alternative test to address the concern that historical persecution against Jews might have been perpetrated due to incentives unrelated to antisemitism. For instance, individuals and political leaders may have hoped to seize Jewish property if they took part in or promoted the attacks against Jews, which would have affected historical persecution against Jews as well as the long-run wealth of local households. The test exploits political support for the Nazi party as an alternative proxy for historical antisemitism, because antisemitism was a major pillar of the Nazi party’s ideology in the late 1920s and early 1930s. But motivations other than antisemitism contributed to the political support for the Nazi party. In particular, the prolonged economic recession that hit Germany after 1929 was famously a major determinant of Nazi support. We therefore conjecture that voting support for the Nazi party should be a more direct proxy for antisemitism in counties in which unemployment was lower than the national average, whereas it should be a noisier proxy for antisemitism in counties in which unemployment was high, and hence motivations other than antisemitism might have increased Nazi votes. Armed with this interpretation, we estimate the effect of county-level Nazi votes in the general elections of September 1930 and of 1933 on present-day stock market participation. Consistent with our conjecture, we find Nazi votes are strongly negatively associated with present-day stockholdings in counties at the bottom of the distribution by unemployment, whereas this association stays negative but smaller in size and statistically insignificant for counties at the top of the distribution by unemployment. The negative association declines monotonically as the share of a county’s unemployment decreases, as depicted in Figure A.3 of the Online Appendix. Contrary to Nazi votes, all the other dimensions we measure at the county level in the early 1930s do not produce the pattern described above, including the vote shares for non-antisemitic parties (see Figure A.4 and discussion in Section 1 of the Online Appendix). These results corroborate our baseline analysis by using a different source of variation and proxy for historical antisemitism than the ones we used above. 5. Historical Antisemitism and Banking: Mortgages and Deposits So far, we have focused on the likelihood that German households hold stocks. Focusing on stock market participation is meaningful, because the defamation of Jews as stock market manipulators survived in the press and popular culture even after the Jewish presence in other financial institutions, such as banking, had faded. The role of Jews in banking services started to decrease substantially with the foundation of the first Raiffeisenbank in 1843 and the subsequent diffusion of Volksbanken across German counties. Several generations of Germans have accessed banking services run by the non-Jewish population. But if the historical association between Jews and financial services affects current financial decisions through channels other than current antisemitism, we would expect also to find an effect of historical antisemitism on present-day Germans’ access to banking services. We first look at the decision to get a mortgage to finance homeownership. This decision allows us to observe whether households increase their debt through bank financing, or if they use their own savings, keeping constant the likelihood that they are homeowners. For the case of Germany, looking at this margin is quite relevant: in 2001, 43% of German households owned their home, but only 20% of households have ever held mortgages; that is, only 47% of homeowners had financed their homeownership via a mortgage (Georgarakos et al., 2010). In Table 3, we find that historical antisemitism is unrelated to households’ decision to buy their home, but it significantly decreases the likelihood that households hold a mortgage. In columns (1) and (2), we report the coefficients for estimating two probit specifications whose outcome variable is a dummy equal to 1 if the household owns any real estate properties. The effect of antisemitism on the likelihood of homeownership is economically and statistically insignificant. In columns (3) and (4), we report the coefficients for the same specifications, but now the outcome variable is a dummy equal to 1 if the household has ever held a mortgage. A one-standard-deviation increase in historical antisemitism reduces the likelihood of holding a mortgage by 0.7 percentage points, which is 10% of the average likelihood of holding mortgages in our sample (6.9%).10 The size and statistical significance of this association are in line with the effect of antisemitism on present-day stockholdings, which we documented above. Table 3. Historical antisemitism, mortgages, and deposits       Home Owner  Has a mortgage  Regularly saves part of income  Deposits/assets local banks     Probit  Probit  Probit  Probit  Probit  Probit  OLS  OLS     (1)  (2)  (3)  (4)  (5)  (6)  (7)  (8)  Historical antisemitism  –0.009  –0.008  –0.007**  –0.007**  0.006  0.005  –0.015**  –0.015**  (VV P.C.)  (0.010)  (0.011)  (0.003)  (0.003)  (0.007)  (0.020)  (0.007)  (0.006)  Log Jews 1933  –0.007  –0.011  0.004  0.004  –0.001  –0.001  –0.025**  –0.021**     (0.011)  (0.012)  (0.003)  (0.003)  (0.009)  (0.007)  (0.011)  (0.010)  Percentage  0.002  0.000  0.000  –0.001  0.000  0.000  0.000  0.000      Catholics 1925  (0.002)  (0.001)  (0.001)  (0.001)  (0.001)  (0.001)  (0.001)  (0.001)  Age  0.000  –0.001  0.011***  0.01***  –0.001  –0.005***  –0.002        (0.002)  (0.002)  (0.001)  (0.001)  (0.002)  (0.002)  (0.002)     Age$$^{2}$$ /100  0.000  –0.001  –0.001***  –0.001***  0.001***  0.001***           (0.001)  (0.001)  (0.000)  (0.000)  (0.000)  (0.000)        Female  0.019*  0.019*  –0.005  –0.004  –0.034***  –0.032***     0.005     (0.010)  (0.010)  (0.004)  (0.004)  (0.009)  (0.008)     (0.044)  Single  0.153***  0.147***  –0.001  0.006  0.077***  0.092***     –0.035     (0.013)  (0.013)  (0.006)  (0.007)  (0.017)  (0.018)     (0.072)  College  0.036***  0.033***  –0.001  0.003  –0.016***  –0.012***     0.230     (0.006)  (0.005)  (0.002)  (0.002)  (0.005)  (0.005)     (0.177)  Eastern Germany  0.068  0.065  –0.019*  –0.019*  0.005  0.005     0.079***     (0.049)  (0.048)  (0.011)  (0.010)  (0.030)  (0.028)     (0.020)  Income p.c. 2005  –0.004  –0.003  0.001  0.002  0.003  0.001     –0.001     (0.004)  (0.004)  (0.001)  (0.001)  (0.003)  (0.003)     (0.007)  Percentage College  0.001  –0.003  0.001  0.001  0.001  0.001            graduates 2005  (0.030)  (0.003)  (0.001)  (0.001)  (0.002)  (0.002)        Income deciles  X  X  X  X  X  X        Other individual controls     X     X     X     X  Other historical controls     X     X     X     X  Other regional controls     X     X     X     X  Wave groups f.e.     X     X     X        Observations  11,484  11,484  11,484  11,484  10,900  10,900  236  236  N. of clusters  236  236  236  236  236  236        (Pseudo-) R$$^{2}$$  0.05  0.06  0.16  0.18  0.09  0.11  0.07  0.19        Home Owner  Has a mortgage  Regularly saves part of income  Deposits/assets local banks     Probit  Probit  Probit  Probit  Probit  Probit  OLS  OLS     (1)  (2)  (3)  (4)  (5)  (6)  (7)  (8)  Historical antisemitism  –0.009  –0.008  –0.007**  –0.007**  0.006  0.005  –0.015**  –0.015**  (VV P.C.)  (0.010)  (0.011)  (0.003)  (0.003)  (0.007)  (0.020)  (0.007)  (0.006)  Log Jews 1933  –0.007  –0.011  0.004  0.004  –0.001  –0.001  –0.025**  –0.021**     (0.011)  (0.012)  (0.003)  (0.003)  (0.009)  (0.007)  (0.011)  (0.010)  Percentage  0.002  0.000  0.000  –0.001  0.000  0.000  0.000  0.000      Catholics 1925  (0.002)  (0.001)  (0.001)  (0.001)  (0.001)  (0.001)  (0.001)  (0.001)  Age  0.000  –0.001  0.011***  0.01***  –0.001  –0.005***  –0.002        (0.002)  (0.002)  (0.001)  (0.001)  (0.002)  (0.002)  (0.002)     Age$$^{2}$$ /100  0.000  –0.001  –0.001***  –0.001***  0.001***  0.001***           (0.001)  (0.001)  (0.000)  (0.000)  (0.000)  (0.000)        Female  0.019*  0.019*  –0.005  –0.004  –0.034***  –0.032***     0.005     (0.010)  (0.010)  (0.004)  (0.004)  (0.009)  (0.008)     (0.044)  Single  0.153***  0.147***  –0.001  0.006  0.077***  0.092***     –0.035     (0.013)  (0.013)  (0.006)  (0.007)  (0.017)  (0.018)     (0.072)  College  0.036***  0.033***  –0.001  0.003  –0.016***  –0.012***     0.230     (0.006)  (0.005)  (0.002)  (0.002)  (0.005)  (0.005)     (0.177)  Eastern Germany  0.068  0.065  –0.019*  –0.019*  0.005  0.005     0.079***     (0.049)  (0.048)  (0.011)  (0.010)  (0.030)  (0.028)     (0.020)  Income p.c. 2005  –0.004  –0.003  0.001  0.002  0.003  0.001     –0.001     (0.004)  (0.004)  (0.001)  (0.001)  (0.003)  (0.003)     (0.007)  Percentage College  0.001  –0.003  0.001  0.001  0.001  0.001            graduates 2005  (0.030)  (0.003)  (0.001)  (0.001)  (0.002)  (0.002)        Income deciles  X  X  X  X  X  X        Other individual controls     X     X     X     X  Other historical controls     X     X     X     X  Other regional controls     X     X     X     X  Wave groups f.e.     X     X     X        Observations  11,484  11,484  11,484  11,484  10,900  10,900  236  236  N. of clusters  236  236  236  236  236  236        (Pseudo-) R$$^{2}$$  0.05  0.06  0.16  0.18  0.09  0.11  0.07  0.19  Notes: Columns (1)–(6) report average marginal effects computed after estimating the following probit specification:   \begin{equation*} Pr(Depvar_{ik} | X_{ik}, K_{k}) = \Phi(\alpha + \beta \times Historical~Antisemitism_{k} + X_{ik}' \times \gamma + K_{k}' \times \delta + Income\_deciles + \eta_{t} + \epsilon_{ik}). \end{equation*} Each observation is a German household interviewed by SOEP between 1984 and 2011. The main covariate of interest, $$Historical~Antisemitism$$, is the VV P.C. of measures of Jewish persecution in the 1920s–1930s. $$X_{ik}$$ includes the following individual-level controls: gender, single status dummy, age (second degree polynomial), college education dummy. Other individual controls include: homeownership dummy and life and social insurance dummy. Other historical controls include: log of population in 1933, ratio of population employed in the retail sector in 1933, and ratio of population employed in manufacturing in 1933. Other regional controls include: population density, latitude, index of quality of cultivable land. $$Income\_deciles$$ are dummies indicating the decile of the income distribution to which the household belongs, and $$\Phi$$ is the standard normal cdf. The outcome variable, $$Depvar_{ik}$$, is a dummy that equals 1 if the household owns any real estate property in columns (1) and (2), a dummy that equals 1 if the household has ever had a mortgage outstanding in columns (3) and (4), and a dummy that equals 1 if the household declares they save part of their monthly income regularly in columns (5) and (6). Columns (7) and (8) report the results for a county-level OLS regression of the ratio of total bank deposits in each county to total assets of banks in the county. We aggregate the branch-level deposits and assets from Bankscope at the county level. We cluster standard errors at the county level. $$^{*} p<0.10, {}^{**} p<0.05, {}^{***} p<0.01$$. Table 3. Historical antisemitism, mortgages, and deposits       Home Owner  Has a mortgage  Regularly saves part of income  Deposits/assets local banks     Probit  Probit  Probit  Probit  Probit  Probit  OLS  OLS     (1)  (2)  (3)  (4)  (5)  (6)  (7)  (8)  Historical antisemitism  –0.009  –0.008  –0.007**  –0.007**  0.006  0.005  –0.015**  –0.015**  (VV P.C.)  (0.010)  (0.011)  (0.003)  (0.003)  (0.007)  (0.020)  (0.007)  (0.006)  Log Jews 1933  –0.007  –0.011  0.004  0.004  –0.001  –0.001  –0.025**  –0.021**     (0.011)  (0.012)  (0.003)  (0.003)  (0.009)  (0.007)  (0.011)  (0.010)  Percentage  0.002  0.000  0.000  –0.001  0.000  0.000  0.000  0.000      Catholics 1925  (0.002)  (0.001)  (0.001)  (0.001)  (0.001)  (0.001)  (0.001)  (0.001)  Age  0.000  –0.001  0.011***  0.01***  –0.001  –0.005***  –0.002        (0.002)  (0.002)  (0.001)  (0.001)  (0.002)  (0.002)  (0.002)     Age$$^{2}$$ /100  0.000  –0.001  –0.001***  –0.001***  0.001***  0.001***           (0.001)  (0.001)  (0.000)  (0.000)  (0.000)  (0.000)        Female  0.019*  0.019*  –0.005  –0.004  –0.034***  –0.032***     0.005     (0.010)  (0.010)  (0.004)  (0.004)  (0.009)  (0.008)     (0.044)  Single  0.153***  0.147***  –0.001  0.006  0.077***  0.092***     –0.035     (0.013)  (0.013)  (0.006)  (0.007)  (0.017)  (0.018)     (0.072)  College  0.036***  0.033***  –0.001  0.003  –0.016***  –0.012***     0.230     (0.006)  (0.005)  (0.002)  (0.002)  (0.005)  (0.005)     (0.177)  Eastern Germany  0.068  0.065  –0.019*  –0.019*  0.005  0.005     0.079***     (0.049)  (0.048)  (0.011)  (0.010)  (0.030)  (0.028)     (0.020)  Income p.c. 2005  –0.004  –0.003  0.001  0.002  0.003  0.001     –0.001     (0.004)  (0.004)  (0.001)  (0.001)  (0.003)  (0.003)     (0.007)  Percentage College  0.001  –0.003  0.001  0.001  0.001  0.001            graduates 2005  (0.030)  (0.003)  (0.001)  (0.001)  (0.002)  (0.002)        Income deciles  X  X  X  X  X  X        Other individual controls     X     X     X     X  Other historical controls     X     X     X     X  Other regional controls     X     X     X     X  Wave groups f.e.     X     X     X        Observations  11,484  11,484  11,484  11,484  10,900  10,900  236  236  N. of clusters  236  236  236  236  236  236        (Pseudo-) R$$^{2}$$  0.05  0.06  0.16  0.18  0.09  0.11  0.07  0.19        Home Owner  Has a mortgage  Regularly saves part of income  Deposits/assets local banks     Probit  Probit  Probit  Probit  Probit  Probit  OLS  OLS     (1)  (2)  (3)  (4)  (5)  (6)  (7)  (8)  Historical antisemitism  –0.009  –0.008  –0.007**  –0.007**  0.006  0.005  –0.015**  –0.015**  (VV P.C.)  (0.010)  (0.011)  (0.003)  (0.003)  (0.007)  (0.020)  (0.007)  (0.006)  Log Jews 1933  –0.007  –0.011  0.004  0.004  –0.001  –0.001  –0.025**  –0.021**     (0.011)  (0.012)  (0.003)  (0.003)  (0.009)  (0.007)  (0.011)  (0.010)  Percentage  0.002  0.000  0.000  –0.001  0.000  0.000  0.000  0.000      Catholics 1925  (0.002)  (0.001)  (0.001)  (0.001)  (0.001)  (0.001)  (0.001)  (0.001)  Age  0.000  –0.001  0.011***  0.01***  –0.001  –0.005***  –0.002        (0.002)  (0.002)  (0.001)  (0.001)  (0.002)  (0.002)  (0.002)     Age$$^{2}$$ /100  0.000  –0.001  –0.001***  –0.001***  0.001***  0.001***           (0.001)  (0.001)  (0.000)  (0.000)  (0.000)  (0.000)        Female  0.019*  0.019*  –0.005  –0.004  –0.034***  –0.032***     0.005     (0.010)  (0.010)  (0.004)  (0.004)  (0.009)  (0.008)     (0.044)  Single  0.153***  0.147***  –0.001  0.006  0.077***  0.092***     –0.035     (0.013)  (0.013)  (0.006)  (0.007)  (0.017)  (0.018)     (0.072)  College  0.036***  0.033***  –0.001  0.003  –0.016***  –0.012***     0.230     (0.006)  (0.005)  (0.002)  (0.002)  (0.005)  (0.005)     (0.177)  Eastern Germany  0.068  0.065  –0.019*  –0.019*  0.005  0.005     0.079***     (0.049)  (0.048)  (0.011)  (0.010)  (0.030)  (0.028)     (0.020)  Income p.c. 2005  –0.004  –0.003  0.001  0.002  0.003  0.001     –0.001     (0.004)  (0.004)  (0.001)  (0.001)  (0.003)  (0.003)     (0.007)  Percentage College  0.001  –0.003  0.001  0.001  0.001  0.001            graduates 2005  (0.030)  (0.003)  (0.001)  (0.001)  (0.002)  (0.002)        Income deciles  X  X  X  X  X  X        Other individual controls     X     X     X     X  Other historical controls     X     X     X     X  Other regional controls     X     X     X     X  Wave groups f.e.     X     X     X        Observations  11,484  11,484  11,484  11,484  10,900  10,900  236  236  N. of clusters  236  236  236  236  236  236        (Pseudo-) R$$^{2}$$  0.05  0.06  0.16  0.18  0.09  0.11  0.07  0.19  Notes: Columns (1)–(6) report average marginal effects computed after estimating the following probit specification:   \begin{equation*} Pr(Depvar_{ik} | X_{ik}, K_{k}) = \Phi(\alpha + \beta \times Historical~Antisemitism_{k} + X_{ik}' \times \gamma + K_{k}' \times \delta + Income\_deciles + \eta_{t} + \epsilon_{ik}). \end{equation*} Each observation is a German household interviewed by SOEP between 1984 and 2011. The main covariate of interest, $$Historical~Antisemitism$$, is the VV P.C. of measures of Jewish persecution in the 1920s–1930s. $$X_{ik}$$ includes the following individual-level controls: gender, single status dummy, age (second degree polynomial), college education dummy. Other individual controls include: homeownership dummy and life and social insurance dummy. Other historical controls include: log of population in 1933, ratio of population employed in the retail sector in 1933, and ratio of population employed in manufacturing in 1933. Other regional controls include: population density, latitude, index of quality of cultivable land. $$Income\_deciles$$ are dummies indicating the decile of the income distribution to which the household belongs, and $$\Phi$$ is the standard normal cdf. The outcome variable, $$Depvar_{ik}$$, is a dummy that equals 1 if the household owns any real estate property in columns (1) and (2), a dummy that equals 1 if the household has ever had a mortgage outstanding in columns (3) and (4), and a dummy that equals 1 if the household declares they save part of their monthly income regularly in columns (5) and (6). Columns (7) and (8) report the results for a county-level OLS regression of the ratio of total bank deposits in each county to total assets of banks in the county. We aggregate the branch-level deposits and assets from Bankscope at the county level. We cluster standard errors at the county level. $$^{*} p<0.10, {}^{**} p<0.05, {}^{***} p<0.01$$. The second decision that relates households to banking services is their likelihood of saving through bank deposits. In the SOEP data set, we observe whether households declare that they regularly save part of their income. Reassuringly, in columns (5) and (6) of Table 3, we find historical antisemitism is unrelated to the likelihood that the households in our sample declare they regularly save part of their monthly income. This non-result suggests households in counties with higher or lower historical antisemitism do not differ in their wealth or overall saving behaviour. Ideally, we would like to observe the share of households’ savings in bank deposits. Unfortunately, we do not observe this information in the SOEP sample. Our second source of household-level data, the PHF, does include information on whether households declare they have a checking/ savings account. We find that 99.35% of respondents declare they have a checking/ savings account, which does not provide us with enough variation in this outcome to compare the behaviour of households across counties with different levels of historical antisemitism. Because aggregate deposits of bank customers appear as liabilities in the balance sheets of banks, we can use aggregate data on the ratio of deposits to total assets for the banks that operate in each county. We obtain this information from Bankscope, and we regress this ratio on historical antisemitism and the other observables. This test aims to check the amount of money households deposit in local banks, keeping constant the size of the local banks’ activities. In columns (7) and (8) of Table 3, we find that a one-standard-deviation increase in antisemitism reduces the county-level ratio of deposits over the sum of local bank assets by 1.5 percentage points, which is 2% of the average ratio of deposits over assets across counties (76%). This result is consistent with the notion that households in counties with higher historical antisemitism tend to use bank services less than other households. For robustness purposes, in Table A.3 of the Online Appendix, we replicate the results described above in the PHF sample of German households surveyed in 2010. Historical antisemitism is unrelated to the likelihood that households save a part of their monthly income regularly, and it is unrelated to the likelihood that the household is a homeowner. Instead, higher antisemitism is associated with a lower likelihood of holding a mortgage, even after controlling for wealth and for the elicited risk tolerance, financial literacy, and religiosity of the household head. In addition, we find historical antisemitism is unrelated to outcomes that do not require accessing financial services (see columns (6)–(11)). In column (12) of Table A.3 of the Online Appendix, we also find suggestive evidence that households in counties with higher historical antisemitism keep a higher fraction of their wealth in cash form, although this effect is barely statistically significant. Overall, the PHF data also provide results consistent with the notion that present-day households in counties with higher historical antisemitism access financial services less than other households. 6. The Deep Roots of Historical Antisemitism Our measure of historical antisemitism during the Nazi period might raise concerns it captures spatial variation in economic conditions in the Inter-war period, which might have persisted for decades. To further disentangle the hypothesis that historical antisemitism relates to present-day financial decisions from other explanations related to localized economic shocks during the Nazi period, we build on the fact that the spatial variation in historical antisemitism within Germany has persisted for centuries since the Middle Ages, and especially since the Black Death of 1349 (Voigtlaender and Voth, 2012). The Black Death was arguably the worst pandemic in human history, and up to one half of the European population at the time may have died. Unsubstantiated theories on the origins of the pandemic diffused all over Europe. Accusations against Jews were common and led to persecution, especially in the German lands. Voigtlaender and Voth (2012) document that areas of Germany in which more pogroms against the local Jewish communities occurred during the Black Death of 1349 also displayed higher levels of antisemitism during the Nazi period. In this section, we first test whether deep-rooted measures of historical antisemitism based on the violence against local Jewish communities during the Black Death of 1349 can predict the present-day stock market participation of German households in a similar manner as antisemitism during the Nazi period. Then, we exploit the deep roots of historical antisemitism and build on an historical natural experiment—the forced migrations of the first Ashkenazi Jewish communities through the German Lands after the First Crusade—to design a strategy that exploits quasi-exogenous variation in the likelihood that German counties engaged in anti-Jewish violence in the past. 6.1. Medieval pogroms, historical antisemitism, and stock market participation We first replicate our baseline analysis by regressing the likelihood that households hold stocks today on the medieval persecution proxy: the dummy equals 1 if a pogrom was documented in the county during the Black Death around 1349 (column (1) of Table 4). Pogroms in 1349 are associated with a 2-percentage-point-lower stock market participation by present-day households, which is 12.5% of the average participation rate. Thus, the association between our proxy for antisemitism in the Middle Ages and present-day stock market participation is statistically significant and economically large. Table 4. The deep roots of antisemitism       Medieval Antisemitism     Exposure to Jews        Full sample  Adding longitude  Excluding Bishop seats  Excluding Hanseatic League     Exposed pre- 1300  Exposed pre- Black Death  Exposed pre- 1300  Exposed pre- Black Death        (1)  (2)  (3)  (4)     (5)  (6)  (7)  (8)  Medieval antisemitism     –0.020**  –0.016*  –0.018*  –0.019*     –0.019**  –0.019**        (Pogrom1349)     (0.009)  (0.009)  (0.010)  (0.010)     (0.009)  (0.009)        Exposure to Jews                    –0.020**  –0.024**  –0.025**  –0.030**                       (0.010)  (0.010)  (0.012)  (0.012)  Historical Antisemitism                          –0.014***  –0.014***  (VV P.C.)                          (0.004)  (0.004)  Log Jews 1933     –0.001  –0.003  0.001  –0.002     –0.001  –0.001  0.001  0.001        (0.004)  (0.004)  (0.004)  (0.004)     (0.004)  (0.004)  (0.005)  (0.005)  Percentage     0.000  0.000  –0.000  –0.000     –0.000  –0.000  0.000  0.000      Catholics 1925     (0.001)  (0.000)  (0.000)  (0.000)     (0.000)  (0.000)  (0.000)  (0.000)  Age     –0.002**  –0.003**  –0.002*  –0.003**     –0.002**  –0.002**  –0.002*  –0.002*        (0.001)  (0.001)  (0.001)  (0.001)     (0.001)  (0.001)  (0.001)  (0.001)  Age$${^2}$$ /100     0.046***  0.045***  0.044***  0.048***     0.046***  0.046***  0.043***  0.043***        (0.011)  (0.011)  (0.012)  (0.012)     (0.011)  (0.011)  (0.011)  (0.011)  Female     –0.002  –0.003  –0.004  –0.003     –0.002  –0.002  –0.001  –0.005        (0.006)  (0.006)  (0.006)  (0.007)     (0.006)  (0.006)  (0.006)  (0.006)  Single     0.060***  0.053***  0.060***  0.059***     0.060***  0.060***  0.060***  0.060***        (0.009)  (0.009)  (0.010)  (0.010)     (0.009)  (0.009)  (0.010)  (0.010)  College     0.010**  0.011***  0.012***  0.010**     0.010***  0.010***  0.011***  0.011***        (0.004)  (0.004)  (0.004)  (0.004)     (0.004)  (0.004)  (0.004)  (0.004)  Eastern Germany     –0.021  –0.031*  0.006  –0.001     –0.021  –0.021  –0.025  –0.026        (0.020)  (0.017)  (0.018)  (0.021)     (0.020)  (0.020)  (0.021)  (0.021)  Income p.c. 2005     0.003*  0.004**  0.004**  0.005**     0.003*  0.003*  0.005***  0.004***        (0.002)  (0.002)  (0.002)  (0.002)     (0.002)  (0.002)  (0.002)  (0.002)  Percentage College     0.001  –0.001  0.001  0.001     0.001  0.001  0.001  0.001      graduates 2005     (0.001)  (0.001)  (0.001)  (0.001)     (0.001)  (0.001)  (0.001)  (0.001)  Income deciles     X  X  X  X     X  X  X  X  Other historical controls     X  X  X  X     X  X  X  X  Wave groups f.e.     X  X  X  X     X  X  X  X  Regional controls     X  X  X  X     X  X  X  X  Observations     13,870  13,870  12,423  11,560     13,870  13,870  13,599  13,599  N. of clusters     270  270  249  244     270  270  261  261  (Pseudo-) R$$^2$$     0.106  0.093  0.093  0.094     0.106  0.106  0.106  0.106        Medieval Antisemitism     Exposure to Jews        Full sample  Adding longitude  Excluding Bishop seats  Excluding Hanseatic League     Exposed pre- 1300  Exposed pre- Black Death  Exposed pre- 1300  Exposed pre- Black Death        (1)  (2)  (3)  (4)     (5)  (6)  (7)  (8)  Medieval antisemitism     –0.020**  –0.016*  –0.018*  –0.019*     –0.019**  –0.019**        (Pogrom1349)     (0.009)  (0.009)  (0.010)  (0.010)     (0.009)  (0.009)        Exposure to Jews                    –0.020**  –0.024**  –0.025**  –0.030**                       (0.010)  (0.010)  (0.012)  (0.012)  Historical Antisemitism                          –0.014***  –0.014***  (VV P.C.)                          (0.004)  (0.004)  Log Jews 1933     –0.001  –0.003  0.001  –0.002     –0.001  –0.001  0.001  0.001        (0.004)  (0.004)  (0.004)  (0.004)     (0.004)  (0.004)  (0.005)  (0.005)  Percentage     0.000  0.000  –0.000  –0.000     –0.000  –0.000  0.000  0.000      Catholics 1925     (0.001)  (0.000)  (0.000)  (0.000)     (0.000)  (0.000)  (0.000)  (0.000)  Age     –0.002**  –0.003**  –0.002*  –0.003**     –0.002**  –0.002**  –0.002*  –0.002*        (0.001)  (0.001)  (0.001)  (0.001)     (0.001)  (0.001)  (0.001)  (0.001)  Age$${^2}$$ /100     0.046***  0.045***  0.044***  0.048***     0.046***  0.046***  0.043***  0.043***        (0.011)  (0.011)  (0.012)  (0.012)     (0.011)  (0.011)  (0.011)  (0.011)  Female     –0.002  –0.003  –0.004  –0.003     –0.002  –0.002  –0.001  –0.005        (0.006)  (0.006)  (0.006)  (0.007)     (0.006)  (0.006)  (0.006)  (0.006)  Single     0.060***  0.053***  0.060***  0.059***     0.060***  0.060***  0.060***  0.060***        (0.009)  (0.009)  (0.010)  (0.010)     (0.009)  (0.009)  (0.010)  (0.010)  College     0.010**  0.011***  0.012***  0.010**     0.010***  0.010***  0.011***  0.011***        (0.004)  (0.004)  (0.004)  (0.004)     (0.004)  (0.004)  (0.004)  (0.004)  Eastern Germany     –0.021  –0.031*  0.006  –0.001     –0.021  –0.021  –0.025  –0.026        (0.020)  (0.017)  (0.018)  (0.021)     (0.020)  (0.020)  (0.021)  (0.021)  Income p.c. 2005     0.003*  0.004**  0.004**  0.005**     0.003*  0.003*  0.005***  0.004***        (0.002)  (0.002)  (0.002)  (0.002)     (0.002)  (0.002)  (0.002)  (0.002)  Percentage College     0.001  –0.001  0.001  0.001     0.001  0.001  0.001  0.001      graduates 2005     (0.001)  (0.001)  (0.001)  (0.001)     (0.001)  (0.001)  (0.001)  (0.001)  Income deciles     X  X  X  X     X  X  X  X  Other historical controls     X  X  X  X     X  X  X  X  Wave groups f.e.     X  X  X  X     X  X  X  X  Regional controls     X  X  X  X     X  X  X  X  Observations     13,870  13,870  12,423  11,560     13,870  13,870  13,599  13,599  N. of clusters     270  270  249  244     270  270  261  261  (Pseudo-) R$$^2$$     0.106  0.093  0.093  0.094     0.106  0.106  0.106  0.106  Notes: This Table reports average marginal effects computed after estimating the following probit specification:   \begin{equation*} Pr(Holds Stocks_{ik} | X_{ik}, K_{ik}) = \Phi(\alpha + \beta \times Medieval~Antisemitism_{k} + X_{ik}' \times \gamma + K_{ik}' \times \delta + Income\_deciles + \eta_{t} + \epsilon_{ik}), \end{equation*} Each observation is a German household interviewed by the SOEP any time between 1984 and 2011. In all columns, the dependent variable is a dummy that equals 1 if the household holds stocks. The main covariate of interest, $$Medieval~Antisemitism$$, is a dummy variable that equals 1 if a pogrom happened in county $$k$$ during the Black Death period (around 1349), and zero otherwise. $$Historical~Antisemitism$$ is the VV P.C. of measures of Jewish persecution in the 1920s-1930s. $$X_{ik}$$ includes the following individual-level controls: gender, single/marital status, income (second-degree polynomial), age (second-degree polynomial); college education, homeownership, and life and social insurance. $$K_{k}$$ includes the following county-level current and historical controls: latitude, income per capita, share of college-educated population, index of quality of cultivable land, log of population in 1933, log of Jewish population in 1933, ratio of population employed in the retail sector in 1933, ratio of population employed in manufacturing in 1933, and ratio of Catholic population in 1925. $$Income\_deciles$$ are dummies indicating the decile of the income distribution to which the household belongs, and $$\Phi$$ is the standard normal cdf. We cluster standard erorrs at the county level. $$^{*} p<0.10, {}^{**} p<0.05, {}^{***} p<0.01$$. Table 4. The deep roots of antisemitism       Medieval Antisemitism     Exposure to Jews        Full sample  Adding longitude  Excluding Bishop seats  Excluding Hanseatic League     Exposed pre- 1300  Exposed pre- Black Death  Exposed pre- 1300  Exposed pre- Black Death        (1)  (2)  (3)  (4)     (5)  (6)  (7)  (8)  Medieval antisemitism     –0.020**  –0.016*  –0.018*  –0.019*     –0.019**  –0.019**        (Pogrom1349)     (0.009)  (0.009)  (0.010)  (0.010)     (0.009)  (0.009)        Exposure to Jews                    –0.020**  –0.024**  –0.025**  –0.030**                       (0.010)  (0.010)  (0.012)  (0.012)  Historical Antisemitism                          –0.014***  –0.014***  (VV P.C.)                          (0.004)  (0.004)  Log Jews 1933     –0.001  –0.003  0.001  –0.002     –0.001  –0.001  0.001  0.001        (0.004)  (0.004)  (0.004)  (0.004)     (0.004)  (0.004)  (0.005)  (0.005)  Percentage     0.000  0.000  –0.000  –0.000     –0.000  –0.000  0.000  0.000      Catholics 1925     (0.001)  (0.000)  (0.000)  (0.000)     (0.000)  (0.000)  (0.000)  (0.000)  Age     –0.002**  –0.003**  –0.002*  –0.003**     –0.002**  –0.002**  –0.002*  –0.002*        (0.001)  (0.001)  (0.001)  (0.001)     (0.001)  (0.001)  (0.001)  (0.001)  Age$${^2}$$ /100     0.046***  0.045***  0.044***  0.048***     0.046***  0.046***  0.043***  0.043***        (0.011)  (0.011)  (0.012)  (0.012)     (0.011)  (0.011)  (0.011)  (0.011)  Female     –0.002  –0.003  –0.004  –0.003     –0.002  –0.002  –0.001  –0.005        (0.006)  (0.006)  (0.006)  (0.007)     (0.006)  (0.006)  (0.006)  (0.006)  Single     0.060***  0.053***  0.060***  0.059***     0.060***  0.060***  0.060***  0.060***        (0.009)  (0.009)  (0.010)  (0.010)     (0.009)  (0.009)  (0.010)  (0.010)  College     0.010**  0.011***  0.012***  0.010**     0.010***  0.010***  0.011***  0.011***        (0.004)  (0.004)  (0.004)  (0.004)     (0.004)  (0.004)  (0.004)  (0.004)  Eastern Germany     –0.021  –0.031*  0.006  –0.001     –0.021  –0.021  –0.025  –0.026        (0.020)  (0.017)  (0.018)  (0.021)     (0.020)  (0.020)  (0.021)  (0.021)  Income p.c. 2005     0.003*  0.004**  0.004**  0.005**     0.003*  0.003*  0.005***  0.004***        (0.002)  (0.002)  (0.002)  (0.002)     (0.002)  (0.002)  (0.002)  (0.002)  Percentage College     0.001  –0.001  0.001  0.001     0.001  0.001  0.001  0.001      graduates 2005     (0.001)  (0.001)  (0.001)  (0.001)     (0.001)  (0.001)  (0.001)  (0.001)  Income deciles     X  X  X  X     X  X  X  X  Other historical controls     X  X  X  X     X  X  X  X  Wave groups f.e.     X  X  X  X     X  X  X  X  Regional controls     X  X  X  X     X  X  X  X  Observations     13,870  13,870  12,423  11,560     13,870  13,870  13,599  13,599  N. of clusters     270  270  249  244     270  270  261  261  (Pseudo-) R$$^2$$     0.106  0.093  0.093  0.094     0.106  0.106  0.106  0.106        Medieval Antisemitism     Exposure to Jews        Full sample  Adding longitude  Excluding Bishop seats  Excluding Hanseatic League     Exposed pre- 1300  Exposed pre- Black Death  Exposed pre- 1300  Exposed pre- Black Death        (1)  (2)  (3)  (4)     (5)  (6)  (7)  (8)  Medieval antisemitism     –0.020**  –0.016*  –0.018*  –0.019*     –0.019**  –0.019**        (Pogrom1349)     (0.009)  (0.009)  (0.010)  (0.010)     (0.009)  (0.009)        Exposure to Jews                    –0.020**  –0.024**  –0.025**  –0.030**                       (0.010)  (0.010)  (0.012)  (0.012)  Historical Antisemitism                          –0.014***  –0.014***  (VV P.C.)                          (0.004)  (0.004)  Log Jews 1933     –0.001  –0.003  0.001  –0.002     –0.001  –0.001  0.001  0.001        (0.004)  (0.004)  (0.004)  (0.004)     (0.004)  (0.004)  (0.005)  (0.005)  Percentage     0.000  0.000  –0.000  –0.000     –0.000  –0.000  0.000  0.000      Catholics 1925     (0.001)  (0.000)  (0.000)  (0.000)     (0.000)  (0.000)  (0.000)  (0.000)  Age     –0.002**  –0.003**  –0.002*  –0.003**     –0.002**  –0.002**  –0.002*  –0.002*        (0.001)  (0.001)  (0.001)  (0.001)     (0.001)  (0.001)  (0.001)  (0.001)  Age$${^2}$$ /100     0.046***  0.045***  0.044***  0.048***     0.046***  0.046***  0.043***  0.043***        (0.011)  (0.011)  (0.012)  (0.012)     (0.011)  (0.011)  (0.011)  (0.011)  Female     –0.002  –0.003  –0.004  –0.003     –0.002  –0.002  –0.001  –0.005        (0.006)  (0.006)  (0.006)  (0.007)     (0.006)  (0.006)  (0.006)  (0.006)  Single     0.060***  0.053***  0.060***  0.059***     0.060***  0.060***  0.060***  0.060***        (0.009)  (0.009)  (0.010)  (0.010)     (0.009)  (0.009)  (0.010)  (0.010)  College     0.010**  0.011***  0.012***  0.010**     0.010***  0.010***  0.011***  0.011***        (0.004)  (0.004)  (0.004)  (0.004)     (0.004)  (0.004)  (0.004)  (0.004)  Eastern Germany     –0.021  –0.031*  0.006  –0.001     –0.021  –0.021  –0.025  –0.026        (0.020)  (0.017)  (0.018)  (0.021)     (0.020)  (0.020)  (0.021)  (0.021)  Income p.c. 2005     0.003*  0.004**  0.004**  0.005**     0.003*  0.003*  0.005***  0.004***        (0.002)  (0.002)  (0.002)  (0.002)     (0.002)  (0.002)  (0.002)  (0.002)  Percentage College     0.001  –0.001  0.001  0.001     0.001  0.001  0.001  0.001      graduates 2005     (0.001)  (0.001)  (0.001)  (0.001)     (0.001)  (0.001)  (0.001)  (0.001)  Income deciles     X  X  X  X     X  X  X  X  Other historical controls     X  X  X  X     X  X  X  X  Wave groups f.e.     X  X  X  X     X  X  X  X  Regional controls     X  X  X  X     X  X  X  X  Observations     13,870  13,870  12,423  11,560     13,870  13,870  13,599  13,599  N. of clusters     270  270  249  244     270  270  261  261  (Pseudo-) R$$^2$$     0.106  0.093  0.093  0.094     0.106  0.106  0.106  0.106  Notes: This Table reports average marginal effects computed after estimating the following probit specification:   \begin{equation*} Pr(Holds Stocks_{ik} | X_{ik}, K_{ik}) = \Phi(\alpha + \beta \times Medieval~Antisemitism_{k} + X_{ik}' \times \gamma + K_{ik}' \times \delta + Income\_deciles + \eta_{t} + \epsilon_{ik}), \end{equation*} Each observation is a German household interviewed by the SOEP any time between 1984 and 2011. In all columns, the dependent variable is a dummy that equals 1 if the household holds stocks. The main covariate of interest, $$Medieval~Antisemitism$$, is a dummy variable that equals 1 if a pogrom happened in county $$k$$ during the Black Death period (around 1349), and zero otherwise. $$Historical~Antisemitism$$ is the VV P.C. of measures of Jewish persecution in the 1920s-1930s. $$X_{ik}$$ includes the following individual-level controls: gender, single/marital status, income (second-degree polynomial), age (second-degree polynomial); college education, homeownership, and life and social insurance. $$K_{k}$$ includes the following county-level current and historical controls: latitude, income per capita, share of college-educated population, index of quality of cultivable land, log of population in 1933, log of Jewish population in 1933, ratio of population employed in the retail sector in 1933, ratio of population employed in manufacturing in 1933, and ratio of Catholic population in 1925. $$Income\_deciles$$ are dummies indicating the decile of the income distribution to which the household belongs, and $$\Phi$$ is the standard normal cdf. We cluster standard erorrs at the county level. $$^{*} p<0.10, {}^{**} p<0.05, {}^{***} p<0.01$$. In columns (2)–(4) of Table 4, we assess the robustness of this result. First, we find that controlling for counties’ longitude does not modify the result substantially. Then, we consider two subsamples of households when excluding areas in which one might believe historical antisemitism was either particularly high or particularly low in the past. In column (3), we exclude counties whose cities hosted at least one bishop seat. The rationale for this exclusion is that in counties with bishop seats, the ban on locals from engaging in moneylending might have been enforced more strictly than in other counties, and at the same time, a culture of distrust of moneylending might have been stronger. We find our baseline result is robust to this exclusion. In column (4), instead, we exclude cities that were part of the Hanseatic League. Hanseatic towns were more open to commerce and hence potentially more cultural tolerant than other counties, and at the same time, their trade activities required a large amount of financing compared to local economic activities in other counties. Even in this case, we find our baseline result is replicated. In Table A.4 of the Online Appendix, we propose a large set of additional robustness tests and find that the negative associations between medieval antisemitism and current-day stock market participation are a robust feature of the data. An important point to assess is whether our proxies for historical antisemitism are merely capturing Jewish settlements in the Middle Ages, because unobservables that favoured the settlement of Jewish communities in the past could drive both historical antisemitism and present-day stockholdings. To address this point, we estimate specifications that include a dummy that equals 1 if the county was exposed to a Jewish community before the Black Death. Note the exposure to a Jewish community in the Middle Ages might be interpreted as a proxy for historical anti-Jewish sentiment by itself, as we discuss in Section 3. This dummy might capture the potential for localized anti-Jewish sentiment that did not necessarily express itself in pogroms or major acts of violence against Jews. We compute the dummy at two horizons—exposure before 1300 and exposure just before the Black Death of 1349. In columns (5) and (6) of Table 4, we find the baseline associations between historical antisemitism and present-day stock market participation do not change in terms of magnitude or statistical significance once we add the dummy for exposure to Jewish communities before the Black Death. Exposure is negatively associated with present-day stock market participation, which suggests that either historical antisemitism that did not erupt in violence against Jews also helps explain stock market participation, or that counties with medieval exposure to Jews became less financially developed in the long run irrespective of antisemitism. In both cases, controlling directly for exposure to Jews does not change the baseline result that historical antisemitism is negatively related to present-day financial development. Because medieval persecution and medieval exposure to Jewish communities aim to capture the deep roots of persecution, as opposed to unobservables related to the historical presence of Jews in a county, we would also expect that the baseline results do not change when using our measure of historical antisemitism during the Nazi period controlling for medieval exposure to Jewish communities. Indeed, in columns (7) and (8) of Table 4, we replicate our baseline results on the negative association between historical antisemitism and present-day stock market participation when controlling directly for the medieval exposure of counties to Jewish communities. 6.2. Forced migrations of Ashkenazi Jews and three-stage least squares Unobservable characteristics of German counties may have jointly determined Jewish persecution and long-run financial development. Ideally, we would have assigned anti-Jewish sentiment across similar German counties randomly before the Black Death of 1349, because the variation of historical antisemitism across counties has persisted since the Middle Ages. To get close to such an experiment, we look at the forced migrations of Ashkenazi Jews out of the Rhine Valley after the eleventh century. We provide intuition for this strategy in Figure 3. In the top map of Figure 3, the darker is a county, the earlier is the first Jewish community documented in the county. Blank counties are those with missing data. The earliest Jewish presence in the German lands was found in the cities of Trier, along the Mosel, and Cologne, along the Rhine. Archaeologists date this presence to the ninth century. Research has found evidence of Jewish communities in the tenth century along the entire Rhine Valley.11 The Jewish population in other areas of current Germany was sparse before the eleventh century (Engelman, 1944). At the onset of the Crusades, Christian knights travelling from England and France to the Holy Land persecuted Jewish communities. Several towns on the Rhine expelled Jews, causing a massive Jewish migration towards Eastern, Northern, and Southern Germany. Evidence of sizeable Jewish communities dates back to the late thirteenth and fourteenth centuries in Munich (south) and Berlin (east) (Toch, 2012).12 The bottom maps of Figure 3 show the location of the cities of Trier, on the Mosel, and Emmerich, on the northern end of the German Rhine. The age of the first-documented Jewish community in a county increases as one moves towards each of these cities. Figure 3 View largeDownload slide Settlement of Jewish communities and distance from the Rhine Valley. Notes: The maps document the foundation dates of Jewish communities at the county level. The darker a county is, the earlier a Jewish community was documented in that county. Blank counties are those for which the data are not available. The bottom maps show the location of the cities of Trier, on the Mosel, and Emmerich, on the northern end of the river Rhine in Germany. The isodistance curves centred around those two cities emphasize which counties are at the same distance from Trier or Emmerich. (a) Year when the first Jewish community was documented (b) Counties at same distance from Trier (c) Counties at same distance from Emmerich. Figure 3 View largeDownload slide Settlement of Jewish communities and distance from the Rhine Valley. Notes: The maps document the foundation dates of Jewish communities at the county level. The darker a county is, the earlier a Jewish community was documented in that county. Blank counties are those for which the data are not available. The bottom maps show the location of the cities of Trier, on the Mosel, and Emmerich, on the northern end of the river Rhine in Germany. The isodistance curves centred around those two cities emphasize which counties are at the same distance from Trier or Emmerich. (a) Year when the first Jewish community was documented (b) Counties at same distance from Trier (c) Counties at same distance from Emmerich. We argue the distance of counties from the Rhine Valley determined the existence of Jewish communities at the time of medieval persecutions. In a first step, we use the distance of a county from the Rhine Valley to predict the probability that a Jewish community existed in the county before the Black Death. In a second step, we use the existence to predict the extent of Jewish persecution. The rationale is as follows: in counties with no Jewish communities before the Black Death, violence against Jews cannot have emerged, because no targets for such violence existed. In counties where early Jewish communities existed, the probability of an historical pogrom against the local Jews is strictly positive ex ante, because of the mere presence of Jews.13 In a third step, we use the persecution to predict present-day stock market participation. Note this source of variation does not capture the different attitudes towards Jewish persecution across the counties that hosted Jewish communities, but only the variation in the likelihood of persecution between the counties that hosted a community and those that did not. Both margins of variation in persecution are relevant to the effects we document. For this test, we consider five measures of the distance of a county from the Rhine Valley. They are the Euclidean distances of a county’s centroid from five large cities that lie at different longitudes on the Rhine and Mosel rivers, namely, Mainz, Worms, Speyer, Trier, and Emmerich, on the northern end of the German Rhine. The shortest distance is about 2 km, whereas the greatest distance is 1100 km. The alternative measures aim to capture alternative gradients of the distance from the Rhine Valley, ranging from the southwest/northeast gradient to the northwest/southeast gradient. Across all gradients, the likelihood that a Jewish community existed in the Middle Ages increases toward the Rhine Valley. If we wanted to interpret this strategy as a causal test for the effect of Jewish persecution on present-day financial development, we should assume a demanding exclusion restriction. The distance of a county from the Rhine Valley should not affect current stock market participation through channels different from the county-level historical persecution against Jewish communities. Moreover, Jewish communities escaping from the Rhine Valley should have been equally likely to settle in any counties at the same distance from the Rhine Valley. Note if the latter condition did not hold, we would expect, if anything, that Jewish communities were more likely to settle in counties with higher demand for financial services; hence, the selection would bias our reduced-form coefficients downwards. We propose two tests to assess the extent to which this exclusion restriction may be economically plausible. First, in Panel A of Table 5, we estimate the reduced-form effect of the distances from the Rhine Valley on the ratio of households that own stocks when the distance, instead of historical antisemitism, enters as a regressor, and when both the distance and the VV P.C. enter jointly. All the coefficients refer to ordinary least squares (OLS) regressions. In odd columns, all five distances are positively associated with the likelihood that households hold stocks. Once the VV P.C. enters the reduced-form specifications, the estimated autonomous associations between the distances and stockholdings drop in magnitude, whereas the estimated standard errors attached to coefficients barely change. This result suggests the distance from the Rhine Valley is unlikely to capture unobserved determinants of present-day stockholdings, which are not already captured by historical antisemitism. Table 5. Exclusion restriction: distance from Rhine, and historical antisemitism       Distance Mainz     Distance Worms     Distance Speyer     Distance Trier     Distance Emmerich        (1)  (2)     (3)  (4)     (5)  (6)     (7)  (8)     (9)  (10)        Panel A: Reduced form  Log distance     0.0117**  0.0076     0.0127**  0.0089     0.0149**  0.0111*     0.0129**  0.0104     0.0165**  0.0109        (0.006)  (0.006)     (0.006)  (0.006)     (0.006)  (0.006)     (0.007)  (0.007)     (0.007)  (0.008)  Historical antisemitism        –0.0133***        –0.0132***        –0.0129***        –0.0132***        –0.0127***  (VV P.C.)        (0.004)        (0.004)        (0.004)        (0.004)        (0.005)  Income deciles     X  X     X  X     X  X     X  X     X  X  Individual controls     X  X     X  X     X  X     X  X     X  X  Historical controls     X  X     X  X     X  X     X  X     X  X  Wave group f.e.     X  X     X  X     X  X     X  X     X  X  Regional controls     X  X     X  X     X  X     X  X     X  X  Observations     13,870  13,870     13,870  13,870     13,870  13,870     13,870  13,870     13,870  13,870  Adjusted R$$^{2}$$     0.068  0.068     0.068  0.068     0.068  0.068     0.068  0.068     0.068  0.068        Distance Mainz     Distance Worms     Distance Speyer     Distance Trier     Distance Emmerich        (1)  (2)     (3)  (4)     (5)  (6)     (7)  (8)     (9)  (10)        Panel A: Reduced form  Log distance     0.0117**  0.0076     0.0127**  0.0089     0.0149**  0.0111*     0.0129**  0.0104     0.0165**  0.0109        (0.006)  (0.006)     (0.006)  (0.006)     (0.006)  (0.006)     (0.007)  (0.007)     (0.007)  (0.008)  Historical antisemitism        –0.0133***        –0.0132***        –0.0129***        –0.0132***        –0.0127***  (VV P.C.)        (0.004)        (0.004)        (0.004)        (0.004)        (0.005)  Income deciles     X  X     X  X     X  X     X  X     X  X  Individual controls     X  X     X  X     X  X     X  X     X  X  Historical controls     X  X     X  X     X  X     X  X     X  X  Wave group f.e.     X  X     X  X