Introduction

As per the latest Gallup World Poll, if the opportunity arises, approximately 15% of the world’s adults—or more than 750 million people—would prefer to move to another country (Esipova et al. 2018). Their preferred destinations include the usual suspects of host countries, such as the USA, Canada, and Germany. As most preferred host countries are developed economies, it is safe to assume that economic motives form the basis of potential migrants’ decision to migrate. However, successful immigration depends not only on immigrants’ economic motives (Akgüç et al. 2016; Gibson and McKenzie 2011; Jaeger et al. 2010) but increasingly on the economic benefits they can bring to the host country. Consequently, many countries are designing immigration policies that maximize natives’ benefit net of immigration’s fiscal burden, as in Australia, Canada, and New Zealand (Brücker et al. 2012; de Haas et al. 2018).Footnote 1 To understand immigration’s economic impact on the host country, a better understanding of immigrants’ economic preferences and culture is mandatory. Among the economic preferences, risk and time preferences play an essential role.

To this end, this paper provides a detailed empirical investigation of the immigrant-native gap in economic preferences (EP gap hereafter) in one of the preferred host countries, Germany. Such a gap might have multiple causes, e.g., cultural or economic differences between the migrant’s home country and Germany, or a self-selection effect among migrants. Instead of attempting to distinguishing between diverse origins of the EP gap, in this paper, we restrict our focus to investigating the EP gap’s extent and evolution for Germany and list factors associated with this gap. More precisely, we first estimate the EP gap and underline its demographic and economic correlates. After that, we ask whether the EP gap has evolved in the last 70 years and list the correlates explaining its evolution. In particular, we ask whether the European Refugee Crisis in 2015, during which Germany received almost 890,000 asylum seekers (BAMF 2016; BMI 2016),Footnote 2 might have widened this gap and, if so, whether this is caused by a more considerable cultural distance of recent immigration cohorts or by reduced assimilation, caused by the arrival of a large number of immigrants within a relatively short timespan which overwhelmed usual integration programs. Following the existing research (Bonin et al. 2012; Constant et al. 2011), we highlight the pertinence of immigrants’ assimilation of host characteristics and identity in explaining the immigrant-native gap. Finally, we ask whether immigrant type (refugee vs. non-refugee, EU vs. non-EU origin) and their reason for immigration to Germany (economic vs. noneconomic) help moderate the EP gap.

Our empirical analysis employs the latest waves of the German Socio-economic Panel dataset (SOEP, v35). To estimate the EP gap, we apply the random effects estimation technique with economic preferences as dependent variables. We consider the respondents’ following two economic preferences: risk attitudes and time preference (patience). The immigrant-native gap in these economics preferences are hereafter referred to as the risk gap and the time gap, respectively.Footnote 3 The estimation sample is restricted to first-generation immigrants (FGIs hereafter) and German natives with no immigration background. Our focus on FGIs stems from the intention to shed light on immigration policy’s role in deciding the EP gap in the host country.

Before discussing our main results, in Fig. 1, we provide a plot of the raw data showing the extent of the EP gap in Germany. The figure compares the average economic preferences of German natives to those of major immigrant groups in Germany and shows that the EP gap is substantial. In other words, immigrants in Germany report lower average risk-taking levels but register as being more patient than German natives. We also notice a significant variation in the immigrant-native gap across different immigrant groups, which we exploit later in the empirical analysis.

Fig. 1
figure 1

Economic preferences of immigrants in Germany

The formal empirical investigation begins by confirming earlier observations of the existence of a non-zero EP gap in Germany. After that, we study whether the immigrant respondents’ demographic and economic characteristics help explain the narrowing of the gap in economic preferences. Regarding demographic characteristics, we find that being married further widens the risk gap compared to their non-married counterparts. Female immigrant respondents report a narrower risk gap than male immigrant respondents. Regarding economic characteristics, on the one hand, the respondents with lower education and those working in low-skilled occupations or being inactive on the labor market report a wider risk gap than their respective reference counterparts. On the other hand, compared to employed respondents, while unemployed immigrant respondents report a wider risk gap, the self-employed report a narrower risk gap. Concerning the individual-level correlates of the time gap, while the self-employed immigrant respondents report a narrower time gap, females and unemployed immigrants report a wider time gap.

Next, we ask whether the EP gap has changed during the last few decades in Germany. In Fig. 2, we plot the averages of the absolute difference in economic preferences, a proxy indicator of the EP gap against the immigrant respondents’ year of immigration to Germany. We construct this indicator by subtracting the natives’ average value of the economic preference for the survey year from the immigrant respondent’s economic preference (for details, see Subsection 3.2). The size of the bubbles indicates sample weights applied using the number of observations in each immigration year. We make the following two observations: First, the EP gap is very high but remarkably stable across immigration years, especially the gap in patience. Second, recent immigration cohorts show a substantial increase in the risk gap, but we do not find any noticeable pattern in the time gap.

Fig. 2
figure 2

Immigration cohort and the immigrant-native gap in economic preferences in Germany

A priori, the increase in the risk gap for recent immigrant cohorts might simply be an artifact caused by their recent arrival in Germany: The gap might close once they have been here for a longer time. Therefore, using the respondents’ year of immigration to Germany, we investigate whether recent immigrant cohorts report distinct levels of the immigrant-native gap compared with older cohorts, controlling for the time since their arrival. Our results find that recent immigrant cohorts to Germany indeed report a larger risk gap, but do not report a distinctly different time gap. As immigrants nowadays often originate from culturally distinct parts of the world, we additionally study the role of immigrants' assimilation of host characteristics and identity in bridging the immigrant-native gap. The results indicate that the immigrants’ host language skills, both written and oral German, are negatively associated with the risk gap, irrespective of the risk gap’s association with their year of immigration to Germany. Moreover, immigrants’ assimilation of the host identity is negatively associated with the risk gap. Interestingly, the time gap is not associated with the respondents' host language proficiency.Footnote 4

We investigate the role of the immigrants' origin, the reason for migration, and the intended duration of stay. Our results indicate that the recent cohorts of European immigrants report a smaller risk gap than others, whereas immigrants originating from other world regions report a recent widening of the risk gap. Regarding the above discussion of the 2015 European Refugee Crisis, we also ask whether recent refugee immigrants report distinct risk gap patterns. While recent refugees report a much larger risk gap than non-refugee immigrants do, we do not see any evolution patterns of refugee immigrants’ time gap.

Finally, we employ the information on immigrant respondents’ reason for migration and intended duration of stay together and conduct a policy-relevant investigation distinguishing how the EP gap differs between “stayers” and “opportunity-seeking” immigrants to Germany. For this analysis, we divide the immigrant sample into two broad categories of our own construction: We define opportunity seekers as those immigrants who immigrated to Germany for economic reasons and have a finite intended stay in Germany. We define all other immigrants with a permanent intended stay in Germany as stayers. The results indicate that opportunity seekers report a smaller risk gap than Stayers.Footnote 5

The paper is structured as follows: Sect. 2 summarizes relevant literature, particularly on immigrants’ risk and time preferences in various contexts. Section 3 describes the survey instrument, the data set, and the statistical methods we employ. Section 4 presents the main results of our paper and discusses them. Section 5 concludes.

Literature review

As immigrants constitute a sizable minority in many countries, their productive use for the natives has become an essential consideration of immigration policies. A large research body in economics and social science is devoted to understanding immigrants’ performance in the host country. This research underlines immigrants’ struggle to achieve equality in economic outcomes in the host country. For instance, Kahanec and Zaiceva (2009) show that immigrants experience lower employment and earnings than natives in many countries. Uhlendorff and Zimmermann (2014) find that unemployed immigrants need more time to find jobs than natives. Others show that the immigrant-native gap in economic outcomes can also persist among second-generation immigrants (Algan et al. 2010). A sizeable portion of this research attempts to unearth the long-debated immigration surplus (See Borjas 1995) by testing several positive theories that help understand the quantitative impact of international immigration. Researchers find a positive association between migration and new firm formation (Bettin et al. 2019; Jahn and Steinhardt 2018; Olney 2013). Others highlight that immigrants are, on average, younger and healthier than comparable native-born (Cunningham et al. 2008; Goldman et al. 2014), and studies are devoted to investigating the healthy immigrant effect (Constant et al. 2018; Kennedy et al. 2015; Maskileyson et al. 2019).

An immigration policy that facilitates immigrants’ productive use demands a better understanding of their economic preferences as they are intimately associated with immigrants’ various behaviors (Batista and Umblijs 2014, 2016; Dohmen et al. 2011; Gibson and McKenzie 2011; Vischer et al. 2013). Moreover, the immigrants’ two economic preferences, risk attitudes and patience, are also associated with their social or generalized trust (Albanese et al. 2017), which researchers deem to be a crucial determinant of their socio-economic success in the host environment (Butler et al. 2016). The next two subsections provide supporting arguments for our expectations of the existence of nonzero EP gap in Germany. To this end, we review the existing literature studying how movers (immigrants) and stayers (in this context, natives) differ in immigrants’ home country and their host country.

Willingness to take risks or risk appetite

The first economic preference that we consider is the respondent’s self-reported willingness to take risks (risk appetite). Existing research shows that individuals prone to taking more risk earn higher wages (Bonin et al. 2007). However, risk-takers can also be involved in suboptimal behaviors, such as smoking (Dohmen et al. 2011; Vischer et al. 2013). Pannenberg (2010) finds that a higher risk appetite is associated with a higher reservation wage, resulting in lower employment probability. In contrast, the research focused on immigrants’ risk appetite sheds a positive light on its importance. For example, Batista and Umblijs (2014) demonstrate a positive association between immigrants’ risk-loving behavior and their decision to be self-employed and entrepreneurs. Their risk appetite is also negatively related to their tendency to send remittances (Batista and Umblijs 2016). The two findings noted above, in addition to the finding that immigrants are likely to save more than natives (Islam et al. 2013), highlight immigrants’ potential to generate new investments and be productive members of the host society.

Immigrants may differ from natives in their risk preferences for several reasons, forming supportive arguments for a nonzero risk gap. For the residents of many countries, emigration is often arduous and is better suited for risk-takers among the populace.Footnote 6 This may make immigrants a highly selected group of risk-neutral individuals, especially compared to the natives of the host country with no migration background. Conversely, the more risk-neutral among the immigrant population in the host country may be more likely to migrate to other countries or remigrate back to their country of origin. In other words, this suggests that only the risk-averse among the immigrants decide to stay longer in the host country, predominantly forming the stayers group. Both these possibilities are inherent to immigrants’ migration decisions and contribute to the selection issue well-known in the migration literature. Although not addressed directly in this paper, it is worth mentioning that the selection issue likely biases our estimates of the EP gap. Finally, immigrants’ choices in the host environment can reinforce their risk appetite, demonstrating individual characteristics’ vital role in explaining the levels and changes in the risk gap. For example, research shows that immigrants are likely to take up riskier jobs/occupations than that of natives as these jobs allow them to skip native language requirements of the formal market and bypass the discrimination in paid employment (Clark and Drinkwater 1998; Constant and Zimmermann 2006; Orrenius and Zavodny 2009).

Finally, as immigrants nowadays originate from culturally distant parts of the world, their cultural makeup and post-migration assimilation of host characteristics and identity are also crucial in explaining the risk gap. The existing research shows that risk attitudes differ across the respondents’ ethnicity and country of origin (Bonin et al. 2012; Rivers et al. 2010). Moreover, Bonin et al. (2012) show that while immigrants’ assimilation of host identity helps bridge the risk gap, different levels of persistence of home identity are observed among various immigrant groups, crucial in the endurance of the risk gap. Their findings also indicate that immigrants originating from Turkey and East Europe are not only among the least socially assimilated population subgroups, but these immigrant groups also report the highest risk gap.

While the direction of the bias induced by sources noted above is unclear, theoretical models often assume that migrants are generally more risk-loving than natives populations of home and host home countries (Chiswick 1978; Constant and Zimmermann 2006; Heitmueller 2005; Todaro 1980). The empirical research on the topic, however, finds mixed evidence. While the assumption that risk-loving individuals among the home populace are likely to migrate finds support in the existing research (Gibson and McKenzie 2011; Goldbach and Schlüter 2018; Jaeger et al. 2010), immigrants are not consistently found to be more risk-loving than the host population. For example, Bonin et al. (2009) show that first-generation immigrants in Germany have a lower risk appetite than natives.Footnote 7Footnote 8 In this paper, we attempt to estimate the extent of the immigrant-native gap in risk appetite in Germany. In doing so, we pay special attention to the issues of immigrants’ assimilation of host characteristics and investigate their role in narrowing the risk gap.

Time preference or patience

The second economic preference we consider is the respondent’s time preference or patience. Dellavigna and Paserman (2005) find that impatience is negatively correlated with individuals’ job search efforts and the unemployment exit rate and is orthogonal to reservation wages. Dohmen et al. (2010) find that more pronounced impatience is associated with lower cognitive ability. Like risk attitudes, patience is also inherent to individuals’ migration decisions. Migrants generally face a tradeoff between the short-term costs of migration and the long-term benefits of the relocation. The time horizon of their migration, i.e., their intention to stay in the host country permanently or to re-migrate, also indicates their attitudes towards time preference (patience).

Existing empirical literature finds supporting evidence to the common belief that out-migrants have lower time discounting than that of nonmigrants of the community (Goldbach and Schlüter 2018). Gibson and McKenzie (2011) study the emigration data from Pacific countries and show that the decision to migrate is indeed associated with the respondent’s patience, especially among the high-skilled emigrants. Concerning the empirical studies focused on host countries, however, only limited research is devoted to this. For instance, Constant et al. (2011) investigate the German data and find no significant difference in patience among the unemployed second-generation immigrants and their native counterparts. However, the authors’ focus on second-generation immigrants, and the unemployed makes it difficult to draw general conclusions. Our focus on first-generation immigration and whether their individual-level characteristics and social assimilation influence the time gap is a notable contribution to the literature.

Data and methodology

Survey items and dataset

We use the latest version of the SOEP dataset (1984–2018, v35).Footnote 9 SOEP records extensive information on approximately 20,000 German households annually. To improve the SOEP’s coverage for better representativeness of the migrant population residing in Germany, in 2013, SOEP collaborated with the Institute of Employment Research (IAB) for data collection, referred to as the IAB-SOEP migration sample (Eisnecker et al. 2017). For our analysis, we consider information on German natives and foreign-born individuals currently residing in Germany.Footnote 10 Table 1 reports the summary statistics of the variables of interest.Footnote 11 For easy reference, we show these statistics as a comparison between German natives and foreign-born persons.

Table 1 Summary statistics

The survey question recording the respondent’s risk appetite asks, “How would you describe yourself: Are you generally willing to take risks, or do you try to avoid risks?” This question’s responses range from 0 (risk-averse) to 10 (fully prepared to take risks). The question was asked in survey years 2004 and 2006, and since the year 2008, it has been asked annually. Despite being a subjective measure, experimental validation studies show that it is a valid predictor of actual risky behavior (Dohmen et al. 2011). Table 1 shows that compared to German natives, immigrants in Germany are, on average, more risk-averse.

The second economic preference records the respondent's patience, asking, “How would you describe yourself: Are you generally an impatient person, or someone who always shows great patience?” The individual responses to this question range from 0 (very impatient) to 10 (very patient). The respondents’ patience is recorded once every 5 years since 2008 in the following three survey years: 2008, 2013, and 2018. Table 1 confirms the earlier observation from Fig. 1 that the immigrants in Germany are, on average, more patient than natives. Both questions have been shown in experiments to reflect actual risk and time preferences very well (Falk et al. 2016).

Table 1 also reports information on individual characteristics that we employ as explanatory variables in the empirical investigation. Accordingly, natives are, on average, older than immigrants (50 years vs. 43 years). Other explanatory variables are dummies representing the respondent’s gender (female), urban residence (urban), marital status (married), and current residence in East Germany (east). Most notably, only 7% of the East German population is foreign-born.

The second part of the empirical investigation considers individual-level supplementary information, as shown in Table 1. The analysis begins by studying the role played by the respondent’s demographic characteristics (age, gender, marital status, urban, and East German residence) in explaining the EP gap. For the analysis of the respondents’ age, we construct a dummy variable indicating whether the respondent is younger (aged below 35 years). As economic characteristics, the analysis considers indicator variables denoting the respondent's education (low educated vs. high educated), labor market participation (active participants vs. non-working), employment status (unemployed, self-employed, or employed), and current enrollment status as in education or training (edu training vs. others). The analysis also considers the role played by the employed respondents’ skill levels. To do this, we generate a set of dummy indicators denoting whether the respondent is currently employed in a low-skilled, medium-skilled, or high-skilled job.

After that, we consider several characteristics that are predominantly available for immigrant respondents only. Therefore, the subsequent analysis is restricted to the immigrant sample only. First, we consider social assimilation variables, i.e., the respondent’s self-reported assimilation of the German identity (GI). The survey asks respondents the following question: To what extent do you feel German? The responses to this question range from 1 (Completely) to 5 (Not at all). We reverse responses to this question for the ease of interpretation so that higher values indicate increased identification. In Table 1, we observe that immigrants report their German identity to be between 3 (in some respects) and 4 (for the most part).

Additionally, we consider information about immigrants’ oral and written language proficiency in German. In doing so, we employ dummy variables that take the value of 1 if the respondent reports having good knowledge of oral and written German and 0 otherwise. Finally, we consider several variables recording the immigrant respondents’ experience of the host country. These include three dummy variables indicating immigrants’ duration of stay in Germany as follows: short stay (less than 6 years), medium stay (between 6 and 20 years), and long stay (more than 20 years). Aside from this, we use a continuous variable indicating the immigrant respondent’s year of immigration to Germany. The following subsection justifies the inclusion of these variables in detail.

Estimation methodology

The empirical investigation begins by examining the association between the respondents’ economic preference and their immigration status. Given that the same individual is surveyed multiple times, we exploit the panel dimension of the data and employ the random effects estimation technique with the respondent's economic preferences as dependent variables.Footnote 12 The baseline empirical model is as follows:

$${EP}_{it}={\beta }_{0}+{\beta }_{1}{Immigrant}_{i}+{\beta }_{2}{X}_{it}+{\gamma }_{s}+{\gamma }_{t}+{\mu }_{ist}.$$
(1)

In this specification, the dependent variable \({EP}_{it}\) measures self-reported economic preference of the respondent i; \({Immigrant}_{i}\) is the time-invariant dummy indicator variable taking the value of 1 if the respondent is a foreign-born person and 0 otherwise; \({X}_{it}\) is the vector of the individual characteristics listed as explanatory variables in Table 1; \({\gamma }_{s}\) and \({\gamma }_{t}\) are the state and survey year fixed-effects; and \({\mu }_{ist}\) is the error term.Footnote 13Footnote 14 The main parameter of interest is \({\beta }_{1}\), which denotes the estimate of the immigrant-native gap in economic preferences.Footnote 15 As the respondents’ immigrant status is not assigned exogenously, we suggest that \({\beta }_{1}\) should not be interpreted as a causal estimate.

After that, we consider the role of the respondents' pertinent demographic and economic characteristics in moderating the immigrant-native gap denoted by the coefficient \({\beta }_{1}\). For this analysis, we consider dummy variables indicating the respondent’s demographic and economic information and modify Eq. (1) by interacting these indicators with the primary variable of interest (\({Immigrant}_{i}\)). The coefficient on the interaction term is the new association of interest.

Next, we study how the immigrant characteristics listed in Table 1 are associated with the EP gap. For this analysis, we consider several variables representing immigrants’ assimilation of the host culture and year of immigration to Germany. Given that some migration-related supplementary information, particularly German language skills and level of self-reported German identity, is predominantly available for immigrant respondents only, we perform analysis from here onward using the immigrant sample only. For this empirical investigation, we modify Eq. (1) in two significant ways. First, we consider the absolute difference in economic preferences (\({Abs\_Diff\_EP}_{it}\)) as the proxy indicator of the EP gap. We construct this variable by subtracting the average value of the economic preference of the native of similar age and gender for the survey year \({(\overline{EP} }_{native, t})\) from the immigrant respondent’s economic preference \(({EP}_{immi, it}\)), i.e., \({Abs\_Diff\_EP}_{it}={EP}_{immi, it}-{\overline{EP} }_{native, t}\). Second, in place of the dummy indicator representing the respondent’s immigration status (\({Immigrant}_{i}\)), we now consider a continuous variable indicating immigrant’s year of immigration to Germany, denoted by the term \({Immiyear}_{i}\) in Eq. (2):

$${Abs\_Diff\_EP}_{it}={\delta }_{0}+{\delta }_{1}{Immiyear}_{i}+{\delta }_{2}{X}_{immi,it}+{\delta }_{2}{X}_{it}+{\gamma }_{s}+{\gamma }_{t}+{\mu }_{ist}$$
(2)

We provide the following supporting argument for our focus on \({Immiyear}_{i}\). The respondent’s year of immigration acts as a proxy for the immigrant cohorts to Germany and allows the cohort-level evolution of immigrants’ economic preferences to be pinned down.,Footnote 16Footnote 17 We motivate this research by referring to earlier research. For instance, Bonin et al. (2009) suggest that the migration of older immigrant cohorts to Germany, especially those who arrived in the 1960s as guest workers and obtained jobs upon entering the country, might have involved a meager amount of risk compared to later cohorts. Also, these immigrants were, on average, low-skilled and thus are relatively more risk-averse than German natives. Therefore, we ask the following policy-relevant research question: Is Germany increasingly attracting immigrants with relatively distinct economic preferences compared to natives, particularly since the refugee crisis of 2015?

As our analysis studies immigrant-native differences in the destination country, the estimates of the EP gap crucially rely on the assumption that the EP measures are valid and capture the same thing across the populations of immigrants and natives. While we cannot assume away the systematic differences in responses to outcome measures originating from differences in linguistic and cultural backgrounds between different population subgroups, we employ numerous migration-related characteristics and investigate their independent role in the EP gap estimations. To this end, as noted earlier, we investigate the role of immigrants’ assimilation of host identity, host language proficiency, duration of residence, refugee status, origin, migration reason, and intended length of stay in explaining the EP gap. This migration-related information is denoted as \({X}_{immi,it}\) in Eq. (2).

Results and discussion

The immigrant-native gap in economic preferences (EP gap)

Baseline estimates

In Tables 2 and 3, we report the estimates of EP gap, following the estimation strategy in Eq. (1). The results in the first column of Table 2 indicate that immigrants report lower risk appetite than German natives, statistically significant at the 1% level.Footnote 18 Column (1) of Table 3 reports that immigrants are more patient than German natives, also statistically significant at the 1% level. In terms of magnitude, ceteris paribus, being an immigrant is associated with a decrease in risk preferences of 0.256 sample standard deviations and an increase in time preference of 0.151 sample standard deviations.Footnote 19

Table 2 Immigrant-native gap in risk appetite (entire sample)
Table 3 Immigrant-native gap in patience (entire sample)

Our findings that immigrants report lower risk appetite and higher time preference than natives demand auxiliary exploration along the distributional lines.Footnote 20Footnote 21 That is, we ask whether the EP gap estimates hold for the most/least risk-loving individuals in the sample. We estimate the EP gap by dividing the estimation sample below (and equal to) and above the median levels of risk (median of 5) and time preferences (median of 6).Footnote 22 The findings discussed here are available upon request. For the below (and equal to) median sample of risk preference, we find that immigrants report a lower risk appetite than natives. In comparison, immigrants register a higher risk appetite than natives for the above-median sample. We do not find any immigrant-native difference in the below (and equal to) median sample of time preference. In contrast, in the above-median sample, immigrants report higher time preferences than natives.

Supplementary analysis

Next, we perform supplementary analysis by studying the role of the respondent’s pertinent demographic and economic characteristics in narrowing the EP gap. We begin by considering the role of the respondents' following demographic characteristics: age, gender, marital status, urban, and East German residence. Columns (2)–(6) of Tables 2 and 3 present the results. We find that the risk gap is smaller among female respondents, whereas it is larger among married respondents. The results in Table 3 suggest that the time gap is wider among females, significant at the 10% level. The analysis does not find any evidence of heterogeneous association concerning the respondents’ urban and East German residence.

In Tables 4 and 5, we consider the respondents’ economic characteristics. In column (1), we begin the discussion by considering the role of the respondent’s education level. The results suggest that the risk gap is much wider for the low educated than among the highly educated. Conversely, education does not play any role in moderating the time gap.

Table 4 Economic characteristics and the immigrant-native gap in risk appetite (entire sample)
Table 5 Economic characteristics and the immigrant-native gap in patience (entire sample)

Next, we consider the respondent’s labor force status. The analysis begins by studying the immigrant-native gap among the respondents inactive on the labor market, indicated by a non-working dummy (see column 2). The reference group consists of respondents who are active in the labor force, which includes unemployed and employed respondents. In column (3), we restrict the sample to active labor market participants and consider the role of the respondent’s employment status as unemployed and those who are self-employed. In column (4), we further restrict the sample to employed respondents and study the role of their skill level. We find that the risk gap is wider among non-working respondents, unemployed respondents, and those working in low- and middle-skilled occupations, while it is narrower for the self-employed. Concerning the time gap, we find that it is wider among the unemployed respondents, while it is narrower among the self-employed.

Immigration cohorts, social assimilation, and the immigrant-native gap

The past success of immigrants’ social integration is only a good indicator for future success when factors influencing their integration are stable. These factors can be on the side of the host country—where even singular events can change integration outcomes significantly, see, e.g., Deole (2019), but also on the side of the immigrants, e.g., the role played by immigrants’ home culture (Constant et al. 2006). Therefore, it is of interest to detect potential changes in the immigrant cohorts over time, leading to changes in integration outcomes in the future. To this end, we study the association of the immigrants’ immigration year with the EP gap. We also investigate whether the immigrants’ host country experience mediates in the immigration year’s relationship with EP gap. Following the estimation strategy shown in Eq. (2), in column (1) of Tables 6 and 7, we demonstrate whether EP gap has evolved across different immigration cohorts. The results show that the risk gap shares a positive association with the immigration year, i.e., the recent immigrant cohorts report a larger risk gap, demonstrated by increased absolute difference in risk appetite. In contrast, the time gap shows a negative and statistically significant association with the immigration year.

Table 6 Immigrants’ social assimilation and risk appetite (immigrant sample)
Table 7 Immigrants’ social assimilation and the time gap (immigrant sample)

In column (2) of Tables 6 and 7, we study whether immigrants’ German language skills show an independent association with the immigrant-native gap.Footnote 23 The results show that immigrants’ German language skills are negatively associated with the risk gap but share no association with the time gap. Column (3) of Table 6 considers the association of the immigrants’ social assimilation variable, i.e., German identity, with the risk gap. We do not perform a similar analysis for the time gap as the social assimilation question was not asked in the years where patience information is available. The results for the risk gap show that the immigrant respondent’s acquisition of German identity is unrelated with the risk gap. It is noteworthy that the magnitude of the association of Immiyear is much smaller when we control for the social assimilation variable, suggesting that the role of the immigration cohorts in explaining the risk gap becomes smaller (yet statistically significant) upon controlling for the respondent’s social assimilation.

The finding suggesting that migrants’ assimilation of German identity is, on average, unrelated to the risk gap is in direct contrast with the findings of the existing literature and needs further clarifying discussion.Footnote 24 To this end, we re-estimate the model specification in Eq. (2) by using individuals’ economic preferences as dependent variables. In other words, we now regress immigrant individuals’ risk appetite on their immigration year and other immigration-related characteristics. Tables A6–A9 in the Online Appendix present the results estimated using alternative specifications, which correspond more closely to the main analysis and have a straightforward interpretation. The results shown in columns (1)–(2) of Table A6 present evidence highlighting the role of immigration year and host language proficiency in determining immigrants’ risk preference. Notably, in column (3) of Table A6, we observe that immigrants’ German identity is positively associated with their risk appetite. This therefore resolves the seeming discrepancy to previous literature that we encountered in Table 6.

After that, we investigate whether immigrants’ more prolonged stay in the host country reduces the EP gap. In column (6), we study the association of the EP gap with indicators representing the immigrants with a shorter and medium stay in Germany. The results show that, while the risk gap exists for the recently arrived immigrants, it is inexistent for the immigrants residing for more than 6 years in Germany. The results of column (3) in Table 7 do not show any association between the immigrants’ length of stay in Germany and the time gap.Footnote 25 The results presented in Table A7 also show qualitatively similar results indicating statistically insignificant role of immigrants’ duration of residence in Germany.

Immigrants’ origin, migration motive, intended stay, and the immigrant-native gap

Tables 8 and 9 consider the role of additional pertinent migration-related information in the relationship of the variable Immiyear with EP gap. First, we use the information on the immigrant’s country of birth and designate immigrant respondents to major home regions.Footnote 26 Accordingly, the following home regions are considered: (1) Western Europe (WE); (2) Central and East Europe (CEE); (3) Turkey, the Middle East, and North Africa (TMENA); (4) and other regions (OOC). The results are estimated using Western European immigrants as the reference category. The results in column (1) of Table 8 find that TMENA immigrants report a statistically significantly larger risk gap than other immigrant groups in Germany. Interestingly, column (2) shows that the recent cohorts of TMENA, CEE, and OOC immigrants report a higher risk gap than recent cohorts of EU immigrants. Concerning the time gap, in columns (1)–(2) of Table 9, we observe a similar pattern of results for TMENA immigrants, while CEE and OOC immigrants do not show any distinguishable changes across immigration cohorts compared to that of European immigrants.

Table 8 Immigrants’ origin and the risk gap (immigrant sample)
Table 9 Immigrants’ origin, intended stay, and the immigrant-native gap in patience (immigrant sample)

Being part of the Schengen Area, Germany allows free mobility to the residents of 26 European countries. The possibility to enter the country freely reduces the costs of migrating for the citizens of these countries, which makes it possible for relatively risk-averse individuals to take up the act of migration. Alternatively, immigrants from EU countries are culturally more similar to natives than other immigrants and report a smaller EP gap. Therefore, in columns (3)–(4), we investigate whether these immigrants are distinct from other immigrants in Germany. To do this, we divide the immigrant sample into European (i.e., EU Schengen countries) and non-European immigrants. We do this by generating a dummy indicator EU origin, which takes the value 1 if the immigrant was born in a country currently a member of the EU Schengen agreement (26 European countries recorded in the SOEP) and 0 otherwise. The results of column (3) indicate that EU immigrants have a smaller risk gap than other immigrants. In column (4), we find that recent cohorts of EU immigrants have a smaller risk gap than recent cohorts of non-EU immigrants.Footnote 27 In columns (3)–(4) of Table 9, we observe similar results for the time gap for immigrants with European origin. Columns (5)–(6) of the tables consider the mediation by the immigrant respondent's refugee status. The results report that while refugee immigrants report a larger risk gap, the recent cohorts of refugees report an even larger risk gap.Footnote 28 Interestingly, as the results in columns (5)–(6) of Table 9 suggest, while refugees report a larger time gap than non-refugee immigrants, recent cohorts of refugees report a smaller time gap than earlier cohorts. The results in columns (1)–(6) of Table A8 in the Online Appendix report findings consistent with those discussed above.

Now we consider the role of immigrants’ main reasons for immigration to Germany and the intended duration of stay. The results for the risk gap are reported in Table 10. The analysis begins by employing the SOEP question recording immigrants’ main reasons for immigration to Germany. The survey records a total of seven individual responses to the question, which we group into four broad categories of migration reasons: (1) family (partnership and other family reasons), (2) economic reasons (own economic perspective, economic perspective for children, other economic reasons), (3) political reasons, (4) other reasons.Footnote 29 The results in column (1) show that immigrants who immigrated for political and other reasons report a higher risk gap than that of those who immigrated for family (and economic) reasons. Column (2) reports that recent immigrant cohorts who immigrated for political and other reasons show an even larger risk gap than others. The results in columns (3)–(4) consider the role of immigrants’ intended length of stay in Germany (return to home country within this year, in a few years, or never). A broad reading of the estimates suggests that the immigrant respondents’ intended length of stay does not mediate in the relationship presented in column (1) of Table 6.

Table 10 Migration motive, intended stay and the risk gap (immigrant sample)

Finally, in columns (5)–(6) of Table 10, we reconsider the variables indicating the immigrants’ migration reasons and intended length of stay in a novel way. We want to distinguish between immigrants who have a short-term horizon and had migrated for study or job-related reasons (economic reasons), which we refer to as opportunity seekers, and those who have a permanent intended stay in the host country and had migrated for economic as well as noneconomic reasons (stayers). A motivation for this distinction is that the latter group is the one that is typically in the political discussion, while there is little opposition to the former group. We call those economic immigrants opportunity seekers who plan to return to their home country in the future, i.e., within this year or in a few years. We call all other immigrants stayers, including all noneconomic immigrants and those economic immigrants who plan to stay in Germany permanently. We study whether stayers and opportunity seekers report a distinct EP gap. Column (5) finds that opportunity seekers register a smaller risk gap than stayers, whereas, in column (6), the results indicate that recent cohorts of opportunity seekers are not distinct in their risk gap to older cohorts.

Dealing with unobserved heterogeneity

Earlier, we argued that our results should not be interpreted as causal estimates. For instance, the selection of immigrants to Germany, or more narrowly in our survey, is not exogenous and depends on many foreseeable factors, not accounted for directly in our analysis. Other effects how omitted variables impact our results also can not be ruled out. These factors include innate/pre-migration levels of economic preferences, the original reason for migration and decision to migrate to Germany specifically, and their post-migration decision to (or not to) remigrate to other countries or back to their home countries, which again largely depend on their economic preferences and acquisition of host characteristics and identity. These unobservable factors may bias the results to the extent that it overturns our main finding. Therefore, we conduct the bounding exercise described in Oster (2019) and estimate the level of bias in the unobserved variables necessary to overturn our EP gap estimates.

The method attempts to measure the level of unobserved heterogeneity that would invalidate our point estimates of the EP gap. It assumes a maximum R-squared value of 1 in a hypothetical regression, including all observed and unobserved covariates. The method then estimates a ratio (referred to as delta) that indicates how large the relationship between unobservables and the immigrant dummy has to be relative to the relationship between the observables and the immigrant dummy to overturn the observed EP gap estimates. The method estimates the delta after assuming proportionality between these relationships. To be precise, the estimated delta indicates the relative size of the relationship needed between unobserved omitted variables and immigrant indicator relative to the relationship between observed covariates and immigrant indicator to result in the coefficient of 0.

We employ this method on the baseline estimates presented in column (1) of Table 2 using the Stata command: psacalc. As the command is incompatible with the random effects panel estimator, we perform this analysis using the OLS estimates. The OLS estimates for the risk gap are − 0.120 (0.010), and the delta ratio estimated using psacalc command is approx. 5.52. In other words, the delta value indicates that to eliminate any observed effect of the immigrant indicator, the relationship between risk appetite and unobserved heterogeneity must be 5.52 times the size of the relationship between the observables and risk appetite. As our baseline model included detailed information on individual’s pertinent characteristics, we find it challenging to imagine sources of unobserved heterogeneity with such great predictive power so that the risk gap estimates can be overturned. In other words, we argue that the risk gap estimates noted above are likely to hold even in the presence of substantial unobserved heterogeneity. The OLS estimates for the time gap are 0.140 (0.016) with a delta ratio of − 0.041, indicating that, unlike earlier results, unobserved heterogeneity may overturn our estimates of the time gap. This potentially limits the viability of the findings for the time gap, which therefore requires further studies.

European refugee crisis (ERC) and the immigrant-native gap in economic preferences

In this subsection, we consider the episode of the 2015 European Refugee Crisis (ERC hereafter). In particular, we verify the robustness of our earlier findings when the effect associated with the ERC is taken into account. To do this, we generate a dummy indicator ERC taking the value of 1 if the immigrant respondent immigrated to Germany from 2015 onward and 0 otherwise. In column (5) of Table 6, we present the results and confirm that all earlier findings hold. We also demonstrate that those who immigrated during the ERC report a much larger risk gap than others. These results support the claim made earlier that the 2015 ERC changed the pattern of immigration to Germany regarding the risk gap: Our results suggest that lower degrees of assimilation in this immigrant cohort led to larger differences in the risk gap. What causes these lower degrees of assimilation cannot be answered with our data, but natural explanations would be inherent differences in the self-selection of migrants and larger integration problems due to the sheer number of immigrants arriving during a relatively short time. In column (6), we study whether the results in column (5) hold if we control for the immigrants’ assimilation of German identity. While the findings reiterate that the main message holds (column (5)), coefficients for the variables indicating German identity and FGIs shorter duration of stay in Germany are not statistically significant. However, we refrain from concluding that this implies a mediator effect of assimilation on the risk gap due to the smaller sample size.

Therefore, in column (7) of Table 6, we re-estimate the baseline model with GI as the dependent variable to test the mediator effect. The model also controls for dummies indicating the respondent’s immigration duration and sheds light on the role of immigrants’ duration of stay in Germany. The results suggest that Immiyear is negatively associated with GI, suggesting that the increase over time in the risk gap can be attributed to recent drops in newcomers’ social assimilation (after controlling for the time of their stay in Germany). The findings also provide supporting evidence of the argument made earlier that immigrants’ recent cohorts report lower assimilation outcomes, including those who arrived in Germany around the 2015 European Refugee Crisis.

Limitations

SOEP data is widely considered of high quality, and the sample is representative of the German population. Nevertheless, of course, there are some limitations worth mentioning. First, the dataset was boosted multiple times since its inception to improve its coverage of the immigrant sample, most notably, in 2013, near the onset of the European Refugee Crisis. The frequent survey boosting can make it challenging to generalize some of the findings to all immigrants. However, according to the survey information document, SOEP’s particular attention to gathering in-depth information on the labor-market integration processes of migrants in Germany makes it better suitable for the exploration of long-time societal changes, notably, the speed of convergence between migrants and natives (Goebel et al. 2019), like the one performed in this paper. On the other hand, Germany’s unique situation in the aftermath of the refugee crisis, when the country observed a substantial increase in immigration inflows, poses exciting questions and challenges, some of which we addressed with our research. Another limitation when studying recent migration waves is obviously that long-term outcomes are not known yet. While we acknowledge this limitation, our study can help fill in some gaps in the knowledge about immigrants’ relatively recent cohorts.

Any investigation of survey responses to individuals’ economic preferences mandates discussion of item nonresponse to sometimes complicated survey questions. Existing research shows that older, female, low-income, and less educated respondents are disproportionately likely to skip answers to survey questions recording individuals’ economic preferences (Wärneryd 1996). Moreover, the issue becomes severe given immigrant-native differences in host language proficiency needed to answer complex survey questions. However, as Coppola (2014) states, compared to lottery questions recording the respondents’ risk preferences, nonresponse rates are significantly lower for general risk-taking questions, such as those used in this analysis, as they are easier to comprehend. In the SOEP version used for this study, the nonresponse rate for the risk appetite question was 0.52% (i.e., only 1720 of total 331,293 did not provide an answer to the question), whereas that for the patience question was a mere 0.16% (i.e., 102 of total 64,795 abstained from answering this question). Finally, there are certain limitations regarding the measurements of risk and time preferences. The SOEP survey is not a laboratory experiment, so the measurement of economic preferences is necessarily less precise than we would wish for. This distinction leaves room for further studies, but these simple measures are undoubtedly useful as a general indication.

Conclusion

These days, a sizeable portion of the Western population shares an immigrant background. According to Eurostat (2019), around 22.3 million non-EU citizens are currently living in EU countries. However, increasingly, Europeans consider immigration as the EU’s primary concern (Eurobarometer 91, 2019). The recent successes of the anti-immigration political rhetoric (Trump, Brexit) have given traction to economists investigating immigration’s economic impact on the host countries. As migrants’ economic preferences are intimately associated with their economic success in the host country, the study of migrants’ economic preferences, such as their risk and time preferences, is topical.

Our analysis observed a substantial immigrant-native gap in economic preferences (what we call the EP gap) between natives and immigrants. However, we also found that the EP gap differs across immigrant groups and their pertinent characteristics and performance in the host country, especially for risk preferences. While male immigrants, those with lower education and low-skilled or no work are more risk-averse than natives, the differences tend to disappear for other groups. We also find that integration reduces this gap. Interestingly, when comparing different immigrant cohorts, our results indicated that the gap in risk preferences has increased over time (even when controlling for the time immigrants have stayed already in Germany). Notably, our findings suggest that those immigrants who arrived in Germany around the 2015 European Refugee Crisis tend to have a larger gap in risk preferences than older cohorts of immigrants. We further find that the difference of results between cohorts is mediated by lower degrees of their social assimilation of German identity. The gap in time preferences is, on the other hand, remarkably stable and does not respond to the immigrant respondents’ demographic and economic characteristics, their time spent in the host country, or their assimilation to the host language and culture.

Our results suggest that the increase in risk aversion among recent migration cohorts should be taken seriously. For instance, future research should be devoted to investigating how immigrants’ economic preferences mediate in their various behaviors that are important for their economic integration in the host society, i.e., entrepreneurial activity, job perspectives, and ultimately integration outcomes. Future research can also study whether these recent increases in the immigrant-native gap in economic preferences have macroeconomic and fiscal costs for the host country. Another relevant question is whether other economic preferences, such as intra- and interpersonal trust or honesty, also show similarly large immigrant-native gaps and whether similar factors contribute to this. Finally, it would be pertinent to understand the drivers of differences in economic preferences among different immigrant groups: To what extent are these based on cultural differences of the home countries, self-selection effects, or migration experiences?