Culture and portfolios: trust, precautionary savings and home ownership

August 18, 2020 at 02:41 pm IST

Working Paper Series

Johannes Fleck, Adrian Monninger Culture and portfolios:

trust, precautionary savings and home ownership

No 2457 / August 2020

Disclaimer: This paper should not be reported as representing the views of the European Central Bank (ECB). The views expressed are those of the authors and do not necessarily reflect those of the ECB.

Household Finance and Consumption Network (HFCN)

This paper contains research conducted within the Household Finance and Consumption Network (HFCN). The HFCN consists of survey specialists, statisticians and economists from the ECB, the national central banks of the Eurosystem and a number of national statistical institutes.

The HFCN is chaired by Ioannis Ganoulis (ECB) and Oreste Tristani (ECB). Michael Haliassos (Goethe University Frankfurt), Tullio Jappelli (University of Naples Federico II) and Arthur Kennickell act as external consultants, and Juha Honkkila (ECB) and Jiri Slacalek (ECB) as Secretaries.

The HFCN collects household-level data on households' finances and consumption in the euro area through a harmonised survey. The HFCN aims at studying in depth the micro-level structural information on euro area households' assets and liabilities. The objectives of the network are:

understanding economic behaviour of individual households, developments in aggregate variables and the interactions between the two;
evaluating the impact of shocks, policies and institutional changes on household portfolios and other variables;
understanding the implications of heterogeneity for aggregate variables;
estimating choices of different households and their reaction to economic shocks;
building and calibrating realistic economic models incorporating heterogeneous agents;
gaining insights into issues such as monetary policy transmission and financial stability.

The refereeing process of this paper has been co-ordinated by a team composed of Pirmin Fessler (Oesterreichische Nationalbank), Michael Haliassos (Goethe University Frankfurt), Tullio Jappelli (University of Naples Federico II), Juha Honkkila (ECB), Jiri Slacalek (ECB), Federica Teppa (De Nederlandsche Bank) and Philip Vermeulen (ECB).

The paper is released in order to make the results of HFCN research generally available, in preliminary form, to encourage comments and suggestions prior to final publication. The views expressed in the paper are the author's own and do not necessarily reflect those of the ESCB.

ECB Working Paper Series No 2457 / August 2020

Abstract

This paper shows that individual beliefs on the effectiveness of formal and informal sources of risk sharing determine ﬁnancial precautionary behavior. We present empirical evidence demonstrating that higher trust in public insurance systems reduces net liquid wealth while higher trust in communal insurance increases it. This dichotomy is consistent with theories on access to private risk sharing networks. Moreover, we ﬁnd that both types of trust associate positively with the probability to take on ﬁnancial risk for the purpose of becoming a homeowner and the related loan-to-value ratio. Our ﬁndings are robust across a wide range of econometric controls and speciﬁcations.

Keywords: Household Saving, Portfolio Liquidity, Public and Communal Insurance

JEL Classiﬁcation: D14, D31, E71, G5

ECB Working Paper Series No 2457 / August 2020

Non-Technical Summary

The liquidity structures of household portfolios vary substantially within and across countries. We argue that cultural differences help to explain this heterogeneity as they are a key determinant of portfolio choice. Speciﬁcally, we show that trust in formal and informal insurance systems determines portfolio decisions related to precautionary behavior; more trusting households hold distinct amounts of liquid wealth dedicated to precautionary motives as they expect to receive assistance in ﬁnancial distress from public institutions and other individuals. Thus, while earlier papers studied the effect of trust in other individuals on the choice between risky and safe assets, our focus is on the effect of trust on portfolio liquidity.

Moreover, we study the relationship between this cultural trait and the most prevalent household asset - the main residence. In particular, we postulate a simple choice model regarding a household's decision to remain a renter or to become a home owner with mortgage. From these choices, we consider renting as implying less ﬁnancial risk; the household portfolio remains (more) liquid and does not feature mortgage liabilities so its leverage is low. In this set up, we investigate if households with higher trust are more willing to take ﬁnancial risks.

Our empirical strategy accounts for the interdependence between cultural norms and economic as well as ﬁnancial institutions in a given country. Due to this endogenous relationship, one cannot directly estimate the effect of culture on household portfolio choices. We study the portfolio choices of immigrants to overcome this problem, i.e. we apply the Epidemiological Approach to the Household Finance and Consumption Survey (HFCS) which provides detailed accounts of household assets and liabilities in most European countries. As we have access to conﬁdential information on country of birth, we can classify households as immigrants and assign cultural norm indicators of their countries of origin. We generate these measures using data from the European Value Study and World Value Survey (EVS and WVS).

Our ﬁndings show that households with higher trust in public institutions hold less precau- tionary savings. This is consistent with the view that they expect public institutions to function as effective safeguards in times of ﬁnancial distress. In contrast, trust in other individuals has the opposite effect. This dichotomy can be attributed to reciprocity; earlier research has shown that individuals who expect their communities to act as risk sharing networks take precautions to be able to support other network members.

We also ﬁnd that the probability to become a home owner with a mortgage and trust are positively correlated - and that trust in other individuals has a stronger correlation than trust in public institutions. The same is true when we study loan-to-value ratios of portfolios, measured as the value of the mortgage relative to the value of the main residence. This ﬁnding provides additional support to our postulated mechanism linking culture and precautionary behavior; higher trust increases the willingness to take ﬁnancial risks because trusting households have stronger beliefs that formal and informal channels reduce uninsurable risks.

ECB Working Paper Series No 2457 / August 2020

Introduction

Household portfolios vary profoundly across advanced economies such as the European Union member states. While about 22% of Belgian total household net wealth consisted of liquid assets in 2014, this share was only 6% in Poland. In the same year, French households dedicated about 10% of their wealth to private pensions. Their Italian counterparts less than 3%. Those discrepancies are even more striking for real estate assets; 70% of Italian households were home owners. In Germany, this was true for only 42%. These differences have attracted considerable attention in recent years as the composition of household portfolios has been found to be a relevant factor in determining the efﬁcacy of ﬁscal and monetary policy interventions.1

In this paper, we study the relationship between culture and the liquidity structure of household savings. Using household portfolio data of four European economies we ﬁnd that cultural norms which emphasize trust in other individuals and in public institutions affect the liquidity composition of portfolios. This is also true for the probability to become a home owner using a mortgage and the related loan-to-value (LTV) ratio. We provide evidence that culturally determined expectations on risk sharing from informal and formal channels generate these outcomes as they lead to differences in perceived uninsurable risks.

Our paper contributes to research on household portfolio heterogeneity by studying its relationship to cross-country cultural differences. Following the seminal works of Banﬁeld (1958), Putnam, Leonardi, and Nanetti (1993) and others, many papers have used cultural variation as an explanation for a wide array of outcomes such as income and growth, trade linkages, redistributive policies, entrepreneurship, ﬁnancial development and portfolio choice.2 Previous research on the relationship between culture and portfolios, most prominently Guiso, Sapienza, and Zingales (2008b), has interpreted trust in other individuals as "the subjective probability individuals attribute to the possibility of being cheated" (page 2557). Using this deﬁnition, variation in trust has been related to differences in stock market participation as the expected return of these assets depends on subjective beliefs regarding the honesty of others.

In contrast, we investigate a link between culture and portfolio liquidity based on the view that trust represents the subjective probability individuals attribute to the possibility to receive support in ﬁnancial distress. Thus, our deﬁnition emphasizes subjective views on other individuals' reliability and solidarity in times of need, i.e. their willingness to share risks. Accordingly, we focus on portfolio choice from a precautionary perspective. Hence, while the deﬁnition of Guiso, Sapienza, and Zingales (2008b) emphasizes unconditional beliefs attributable to culture, ours highlights state contingent implications; different intensities of trust in other individuals and in public institutions correspond to distinct expectations on risk sharing provided by informal

For example, using data from the United States, Kaplan and Violante (2014) demonstrated that the share of illiquid assets held by households is a critical determinant of the aggregate consumption expenditure response to ﬁscal stimulus programs. In the Euro Area, differences in household portfolios are highly important from several additional dimensions as they pose a challenge for the conduct of common monetary policy.
2See Guiso, Sapienza, and Zingales (2006) for a comprehensive survey.

ECB Working Paper Series No 2457 / August 2020

and formal sources in bad state realizations. In other words, high and low trust individuals have different subjective beliefs on uninsurable risks. Therefore, they have distinct views on the desirability of holding liquid assets, i.e. savings dedicated to precautionary motives.

We explicitly distinguish between trust individuals have in other individuals and trust individuals have in public institutions, such as a nation's justice, health care and social security system. While earlier research studied the consequences of trust differences in speciﬁc institutions such as the central bank or the ﬁnancial system, we focus on trust in public institutions as providers of insurance. Furthermore, we put particular emphasis on housing tenure choices. This is because for most households, purchasing a home - as opposed to renting - reduces asset liquidity and requires taking out a mortgage. Since assuming this speciﬁc liability has implications regarding the (ﬁnancial) risk a household exposes herself to, higher trust promotes the willingness to participate in mortgage debt markets. Accordingly, our study goes beyond pure asset choice but also considers the liability side of household portfolios.

A concern regarding our investigation is that national institutions affecting household ﬁnancial decisions are not invariant with respect to cultural norms; if the culture of a country's population favors participation in a speciﬁc asset, its institutions will be designed to ensure low access costs.3 This relationship is consequential as public choice mechanisms are responsive to expressed cultural preferences. Earlier studies facing this problem have either not been able to address it or pursued a variety of ways to account for it. The by far most common econometric approach is to instrument culture, i.e. to ﬁnd measures which are correlated to cultural variation but exogenous to variation in institutions. However, as noted by Alesina and Giu- liano (2015) all variables used for this purpose so far have been found to meet this requirement only to a limited degree. Accordingly, a growing number of recent papers follows alternative methodological avenues.

In cross-sectional studies, a well-established possibility to implement a suitable research design is to follow Fernandez (2011) in applying the Epidemiological Approach. Its idea is to study the behavior of second-generation immigrants whose parents grew up in a country where cultural norms differ sufﬁciently from those governing social behavior in their current country of res- idence. As individual cultural preferences are strongly determined by parental transmission, second generation immigrants growing up in this setting can be expected to differ from the rest of the population in that dimension. At the same time, this group faces institutions which have been determined by the culture of the native society. Moreover, their parents made the migration decision which addresses concerns related to selection bias.

Our work follows the spirit of the Epidemiological Approach as we study the portfolios of immigrant communities. We are able to do so using two speciﬁc variables of the Household Finance and Consumption Survey (HFCS); its questionnaire asks respondents for their country of birth and for how long they have already been living in their current country of residence.

These societal preferences can manifest themselves in, for example, taxation of different kinds of capital in- comes, deductibility of mortgage expenses, generosity of rent subsidies etc.

ECB Working Paper Series No 2457 / August 2020

Answers to the former question are recoded to country aggregates in the public version of the HFCS but they have been made available to us for research purposes. Hence, we can explore a novel feature of this dataset allowing us to conduct our empirical analysis in a setting which minimizes concerns regarding the endogenous relationship between culture and institutions. This is particularly true as we are able to observe immigrants from the same country of origin in several destination countries, i.e. in different institutional environments.

Our analysis provides evidence that variation in trust in other individuals and in public institutions has implications for the liquidity structure of household portfolios. First, we ﬁnd that higher public trust decreases the demand for precautionary savings (measured as the amount of net liquid wealth) while higher private trust has the opposite effect. Speciﬁcally, if public trust increases by 1%, liquid wealth decreases by up to 3.4%. For private trust, our ﬁndings indicate a 2.6% increase. This dichotomy is consistent with theories on reciprocity regarding risk sharing in private networks. Second, we estimate that a one standard deviation increase of private trust improves the probability to become a home owner with a mortgage by around 10 percentage points. This ﬁnding also applies to the intensive margin; for the same home value, households which are more trusting by a one standard deviation assume mortgages which are 4 to 6% larger.

The remainder of this paper is organized as follows; in the next section, we provide stylized facts on cross-country differences in the liquidity structures of household portfolios, outline why this heterogeneity is considered a relevant puzzle and provide a review of the most prominent explanations. In section 3, we describe how culture affects portfolio choice in detail, i.e. we postulate a mechanism and derive its testable implications. Section 4 illustrates our empirical strategy as well as the datasets we use. In section 5, we present the results of our analysis regarding the relationship between trust and portfolio choice. Section 6 is dedicated to assessing the robustness of our ﬁndings, in particular with respect to omitted variables and alternative measures of trust. Finally, section 7 concludes the paper with a discussion of our results and points out avenues for further research.

Precautionary Household Portfolio Heterogeneity

In this section, we illustrate cross-country heterogeneity of household portfolios with respect to precautionary aspects and emphasize the policy relevance of this variation as a motivation for our investigation. We focus on European countries because they are all developed economies but still vary substantially with respect to institutions and culture. As we describe below, these characteristics play a prominent role in theories explaining portfolio heterogeneity.

ECB Working Paper Series No 2457 / August 2020

2.1 Stylized Facts from Europe

The Household Finance and Consumption Survey (HFCS) allows to compare household portfolios across most European countries in a comprehensive and consistent manner as it contains detailed and ex-ante harmonized information on household assets and liabilities.4 Figure 1 displays the breakdown of total household assets and liabilities for all countries participating in its second wave. Speciﬁcally, it shows the contribution of liquid versus illiquid assets and liabilities to average household net wealth.5

Figure 1: Household portfolios in HFCS countries

While household net wealth level differences between European member states have already been documented6, ﬁgure 1 shows that net wealth varies in several additional dimensions. First, the breakdown of assets into liquid versus illiquid items is strikingly different. For ex- ample, the average household in Germany or Belgium holds a larger share in liquid form compared to countries like Greece or Poland (about 19 and 22% versus 4 and 6%). An analogous variation applies to liabilities. Second, household portfolios differ substantially with respect to the ratio of assets and liabilities. For instance, in the Netherlands, Ireland and Cyprus, liabilities constitute approximately 28 to 61% of average portfolio net wealth while households in

For a comprehensive description and documentation of the ﬁrst two waves of this dataset, see Household Finance and Consumption Network (2013) and Household Finance and Consumption Network (2016a).
5In this ﬁgure, we classify assets as liquid if they consist of deposits, funds, bonds and stocks and as illiquid if they represent real estate and voluntary and occupational pensions. For liabilities, we classify non-mortgage debt (which consists of credit card debt and non-collateralised loans) as liquid and mortgage debt as illiquid.
6For instance, the release report of the second HFCS wave contains an entire section dedicated to country speciﬁc net wealth distributions. See Household Finance and Consumption Network (2016b), section 4.

ECB Working Paper Series No 2457 / August 2020

other countries, for example Italy, Poland, Slovenia and Slovakia seem much more reluctant to take on liabilities; in these countries the share is only around 5 to 10%. As a result, the average household leverage (i.e. the ratio of assets and liabilities) also varies across countries in a non-negligible way and ranges from 2.6 in the Netherlands to 20.6 in Italy.

Figure 2: Home ownership and home acquisition in HFCS countries

The differences in the cross-country distributions of European household illiquid assets and liabilities are partly driven by drastically different home ownership rates as well as distinct ways of acquiring the household's main residence. To illustrate this point, ﬁgure 2 shows the fraction of households in each country who rent their main residence as compared to owning it. Average home ownership rates vary considerably across countries; around 42% of German and Austrian households rent their main residence while more than 80% own it in Spain, Hungary and Slovakia. Hence, it is no surprise that liabilities such as mortgages play a substantially different role in the composition of average household portfolios. A relevant dimension of heterogeneity among home owners explaining this discrepancy is the way in which households acquired their main residence. As ﬁgure 2 illustrates, more than or about half of all owners have a mortgage in Cyprus, Belgium, Ireland and the Netherlands. In Italy, Greece and Slovenia, this is true for only a small fraction of home owners. In the latter group of countries, obtaining residences as an inheritance or gift is much more common. For that reason, households are less likely to participate in mortgage markets in order to become owners. Accordingly, the distribution of mortgage debt is uneven across countries and does not necessarily correspond to average home ownership rates.

ECB Working Paper Series No 2457 / August 2020

2.2 Consequences for Fiscal and Monetary Policy

The objective of our paper is to explain why household portfolio structures - the share of liquid wealth, mortgage debt and leverage in particular - are so different across developed countries. To motivate our investigation we brieﬂy describe why this heterogeneity is relevant from the viewpoint of ﬁscal and monetary policy makers.

First, papers as Ampudia, Vlokhoven, and Zochowski (2016) demonstrated there is a link between an economy's household ﬁnances and the stability of its ﬁnancial sector. Speciﬁcally, they examined the capacity of households in Euro Area member states to absorb unexpected interest rate increases or income decreases. For identical shocks, they discovered large differences in the share of ﬁnancially distressed households across countries and found the differential outcomes to be driven by the average amount of household mortgage debt (and whether it is indexed at ﬁxed or adjustable rates). Indeed, as household mortgage defaults have triggered the failure of savings banks during the Great Recession, several publications, for instance Hart- mann (2015), argued that Euro Area macroprudential regulation needs to account for country speciﬁc household portfolio indicators and for price dynamics of national real estate markets.

Second, household portfolios determine the redistributive effects of monetary policy within and across countries of a currency union. Several recent contributions point out that for a given central bank policy change, it is the composition of their assets which distinguishes households from gaining and losing in terms of net wealth.7 Other papers showed that differences in national household portfolios determine the direction and magnitude of cross-country wealth redistribution due to policies implemented by the European Central Bank (ECB). For example, Adam and Zhu (2016) and Adam and Tzamourani (2016) found that ECB decisions which result in unexpected price level changes or alter relative asset prices lead to asymmetric wealth gains (and losses) across Euro Area member states because net nominal positions differ. Fur- thermore, Lenza and Slacalek (2018) and Slacalek, Tristani, and Violante (2020) reported that cross-country differences in portfolio interest rate exposures had implications for the across (and within) redistributive effects of the ECB's Quantitative Easing (QE) program.

Third, household portfolios regulate the effectiveness of monetary policy in providing economic stimulus. As Luetticke (2018) and Hintermaier and Koeniger (2018) demonstrated, the share of liquid versus illiquid assets and the fraction of home owners facing borrowing constraints determine the transmission of ECB monetary operations into the economies of Euro Area member states. Their ﬁndings are based on the visceral relationship between household portfolios and marginal propensities to consume (MPCs).8 The same holds true for the efﬁcacy of ﬁscal policy interventions as shown by Kaplan and Violante (2014) for the United States.9

For example Coibion et al. (2017), Auclert (2019), Gornemann, Kuester, and Nakajima (2016) and Doepke, Schneider, and Selezneva (2015).
8Jappelli and Pistaferri (2014) and Fagereng, Holm, and Natvik (2018) reported large empirical variations of MPCs across households and provide evidence that portfolio heterogeneity is among their main explanatory factors.
9As a consequence, many researchers are now working with models which include an explicit role for (nominal) portfolio heterogeneity. For illustrative examples, see Kaplan, Violante, and Weidner (2014) for ﬁscal policy and

ECB Working Paper Series No 2457 / August 2020

Finally, household portfolio heterogeneity needs to be considered when designing and reforming risk sharing mechanisms and tax codes. For the United States, the role of the ﬁnancial sector as a shock absorber and risk sharing channel has been emphasized by Asdrubali, Sorensen, and Yosha (1996) and other publications. Based on those ﬁndings, there has been a push towards greater integration of Euro Area ﬁnancial markets - culminating into the creation of the Capital Markets and Banking Union - to increase Europe's capacity to share and reduce risks through private channels. However, if households in certain member states do not participate in asset markets which provide insurance against domestic shocks, their behavior limits the scope of risk sharing integrated ﬁnancial markets can provide. Following the same reasoning, a joint ﬁscal revenue scheme based on a common ﬁnancial transaction tax as proposed by the Euro- pean Commission (2011) could imply asymmetric tax burdens for households in different Euro Area member states; the tax might fall disproportionately on households in countries where participation in stock and bond markets is more common. Hence, a better understanding of households' portfolio choices is essential to gauge the intended and unintended consequences of complementary ﬁnancial and economic Euro Area institutions.

2.3 Theories of Portfolio Heterogeneity

There is a considerable number of papers investigating why households in different countries allocate their portfolios in such heterogeneous ways. In what follows, we sort them into four groups according to the explanation they promote; institutions, shocks, product familiarity and culture.10

1. Institutions The ﬁrst explanation is based on the large variability of institutional settings across countries, where institutions are considered formal and informal rules which affect port- folio choices. Even among developed economies numerous and large differences exist with respect to those regulating precautionary ﬁnancial behavior and housing tenure choice. As an illustration, consider the generosity of minimum income beneﬁts displayed in ﬁgure 3; they range from about 60% of median disposable income in Ireland to 0% in Italy as of 2014. The same is true for institutions regulating access to home ownership which has profound impli- cations for household portfolios. For example, ﬁgure 4 illustrates that in 2015 the Netherlands spent about 2% of annual GDP on mortgage interest tax deductions while this kind of subsidy does not exist in many other European countries (which focus on subsidizing renters instead).

Kaplan, Moll, and Violante (2018) for monetary policy. While substantial progress has been made with respect to developing and solving Heterogeneous Agent New Keynesian (HANK) models, it remains a challenge to provide a framework which can capture the tremendous variation in households' asset and liability choices. See Hubmer, Krusell, and Smith (2018) for a discussion of this shortcoming and an account of the avenues researchers are currently pursuing to address it.

10Some papers even considered genetical endowments to explain differences in ﬁnancial behavior. For instance, studying identical twins, Cronqvist and Siegel (2015) found that the decision how much to save out of one's current income can be traced to shared genes. Moreover, Barth, Papageorge, and Thom (2020) demonstrated that wealth at retirement age can partly be explained by indirect effects of distinct genetic endowments.

ECB Working Paper Series No 2457 / August 2020

Figure 3: Minimum Income Beneﬁts in OECD countries

Two illustrative papers promoting explanations for portfolio heterogeneity based on institutional variation are Christelis, Georgarakos, and Haliassos (2013) and Christelis, Ehrmann, and Georgarakos (2017). Focusing on the portfolios of older households in the United States and selected European countries, they apply a counterfactual method to investigate cross-country variation in the amount and types of assets and liabilities households chose to accumulate during their lifetimes. Both papers conclude that portfolio variation is associated less with differences in cross-country household characteristics and more with different institutions.

Other contributions promoting this explanation analyze the role of speciﬁc institutions in more detail. Two recent examples are Pham-Dao (2019) and Glass, Simon, and Andersson (2016). The former emphasizes the fact that public institutions of Euro Area member states such as social security, pension and unemployment insurance systems provide distinct beneﬁts to their residents. Therefore, households face different incentives regarding their portfolio choices. The latter points out that policies governing the generosity and duration of support for families vary substantially in Europe. Applying the same reasoning as in the former paper, it follows that portfolio heterogeneity of families with children are a mere reﬂection of country speciﬁc institutions.

2. Shocks The second explanation for cross-country variation in household portfolios is based on exogenous events which only speciﬁc economies or cohorts were exposed to. In a Euro- pean context, economic examples of these are episodes of hyperinﬂation, sovereign defaults

ECB Working Paper Series No 2457 / August 2020

Figure 4: Promotion of Home Ownership

and bank failures while political examples are the Spanish Civil War, the Portuguese Carnation Revolution and German Reuniﬁcation to name only a few. The relevance of these events for portfolio choice has been documented in household ﬁnance data from different European countries. For example, Ampudia and Ehrmann (2017) discovered that individuals who lived through periods in which ﬁnancial markets yielded high and stable returns are more willing to participate in stock markets and invest larger amounts. The same has been conﬁrmed for the United States; using data on American household portfolios, Malmendier and Nagel (2011) found that the probability of a household to invest in a certain type of asset (e.g. stocks as opposed to bonds) depends on the returns which this asset realized during the life time of the household's reference person. Finally, Malmendier and Steiny (2017) have been able to trace back generational differences in home ownership rates to economic shocks, such as episodes of high inﬂation, which only certain cohorts experienced.

3. Product Familiarity The third explanation highlights national biases in ﬁnancial behav- ior related to familiarity with speciﬁc ﬁnancial instruments. For example, Guiso and Jappelli (2005) documented that stock market participation is determined by awareness and familiarity with this type of asset. Inasmuch as they are distinct across populations, familiarity patterns can lead to cross-country differences in portfolio choices. Moreover, they are highly persistent as familiarity is determined by social and geographic proximity. This ﬁnding is due to Brown et al. (2008) who documented that stock market participation is linked to the ﬁnancial behavior

ECB Working Paper Series No 2457 / August 2020

of households in the immediate neighborhood. More recently, cross-sectional and serial correlation in investment decision was also documented by Fuchs-Schundeln¨ and Haliassos (2020) who, using German reuniﬁcation as a natural experiment, showed that product familiarity can explain inertia of households' asset choices.

4. Culture The fourth explanation relates differences in household portfolios to cultural vari- ation. Since culture is a vague and multidimensional concept (comprising language, tradition, religion, etc.), scholars have focused on trust in other individuals as a particular cultural norm to explain differences in ﬁnancial choices. In fact, based on surveys such as the European Value Study (EVS), there is a striking variation of this cultural norm across Europe. For exam- ple, when asked to evaluate the statement "people can be trusted", only 9% of respondents in Cyprus agree, while this share is close to 60% in Finland. To study the implications of these cul- tural differences on portfolio choices, the most widely used deﬁnition is due to Guiso, Sapienza, and Zingales (2008b) who speciﬁed trust as the "subjective probability individuals attribute to the possibility of being cheated". Based on this deﬁnition, it has been found that more trust- ing individuals and societies are more likely to participate in formal ﬁnancial markets, and are more willing to invest in risky assets such as stocks.

However, this explanation suffers from two consequential shortcomings. First, it has focused on explaining differences in the readiness to assume ﬁnancial risks but it has been mute on other portfolio aspects. Speciﬁcally, it has not been able to establish a connection between culture and the variation in precautionary portfolio structures - such as liquidity ratios - as well as in housing tenure choice. Second, this explanation has taken a precise yet narrow perspective on culture; it has focused on trust in an informal (interpersonal) context exclusively, but has neglected its formal (institutional) dimension. Yet, there is also substantial variation with respect to this aspect of trust. For example, when asking Europeans if they trust the political parliament "a great deal" or "quite a lot", less than 20% of households would agree in Poland and Latvia, while this share is more than 60% in Luxembourg. A similar pattern can be found when asking about trust in the social security system. Here 25 and 30% in Poland and Greece state they have high trust in the social security system, while in Belgium and Luxembourg more than 80% do so.

In summary, this section demonstrated that household portfolios differ profoundly in terms of precautionary aspects and housing tenure choice. While scholars have argued that institutions, shocks and familiarity play a role, culture has also been credited explanatory power. However, no link between culture and precautionary portfolio choices has been established so far.

ECB Working Paper Series No 2457 / August 2020

Culture and Portfolio Choice

3.1 Trust as the Subjective Probability to be Cheated

The human trait to which our paper relates is culture. Earlier papers studying the relationship between culture and ﬁnancial behavior focused on trust in other individuals as the central cultural norm determining portfolio choices. Common to this literature is that it considers private trust as measuring the "subjective probability individuals attribute to the possibility of being cheated" (Guiso, Sapienza, and Zingales (2008b), page 2557). Based on this operating deﬁni- tion, several papers have argued that the causal link between culture and individual as well as aggregate ﬁnancial outcomes is based on subjective beliefs determined by trust. For example, higher trust implies fewer expected incidences of check fraud, more faith in the fairness of brokers and greater conﬁdence in the accuracy of companies' proﬁt reports. In other words, low trust prevents individuals from engaging in ﬁnancial transactions which are mutually beneﬁ- cial but subject to some form of incompleteness (due to e.g. limited commitment, enforcement costs, asymmetric information and non-contingency).

The two most prominent papers in this literature are Guiso, Sapienza, and Zingales (2004) and Guiso, Sapienza, and Zingales (2008b). The former employs regional blood donations and voter turnout as a measure for social capital in Italian regions. Arguing that social capital is a critical determinant of trust among individuals, the authors approximate regional differences in private trust by these variables. Using Italian household portfolio data, they then investigate the relationship between trust and a broad set of ﬁnancial participation measures (probability to have a bank account, use of checks, applications for mortgages, etc.). Their ﬁndings indicate that trust in other individuals can indeed explain extensive and intensive variation in ﬁnancial participation. Guiso, Sapienza, and Zingales (2008b) use portfolio data of Dutch and Italian households to investigate if differences in private trust make households more likely to participate in stock markets. While controlling for a large set of characteristics such as optimism, loss and risk aversion as well as ambiguity, they ﬁnd that participation in risky assets (stocks) critically depends on trust. Again, higher trust increases the probability to participate along the extensive and intensive margin.

By characterizing the relationship between trust and ﬁnancial choices from a theoretical and empirical perspective, these papers have laid the foundations for a proliﬁc ﬁeld of research. Two examples of recent contributions are Bucciol, Cavasso, and Zarri (2019) and Jiang and Lim (2018). Answering the question if higher trust can compensate for (ﬁnancial) risk aversion is the objective of Bucciol, Cavasso, and Zarri (2019). In a sample of cross-country data on household investments, they ﬁnd that, for risk averse households, higher trust appears to compensate for the reluctance to invest in risky assets. Moreover, they argue their ﬁnding also sheds light on the cross-country dimension of the limited stock market participation puzzle. With respect to participation in debt markets Jiang and Lim (2018) report that variation in trust also explains behavior with respect to assuming ﬁnancial liabilities. Using a set of general household debt

ECB Working Paper Series No 2457 / August 2020

indicators, they document that having more trust is associated with, for example, fewer missed debt payments and fewer instances of personal bankruptcies.11

3.2 Trust as the Subjective Probability to be Supported in Financial Distress

Our paper is related but complementary to these earlier studies as our operational deﬁnition of culture is distinct; we do not presuppose that trust determines subjective beliefs on the safety of participating in formal ﬁnancial transactions and the expected return of risky assets. Instead, our view is that trust affects individual assessments of uncertainty in a more pervasive sense. Speciﬁcally, we hypothesize that trust determines subjective assessments of uninsurable risk;12 Thus, different intensities of trust make households heterogeneous in their perception of how much risk has to be borne by them.

Cultural differences generate heterogeneity in perceived uninsurable risks because trust determines subjective beliefs regarding external support available in times of need. Thus, our operational deﬁnition of trust resembles the description given by Gambetta (2000) who de- ﬁned trust as "a particular level of the subjective probability with which an agent assesses that another agent or group of agents will perform a particular action, both before he can monitor such action (...) and in a context in which it affects his own action". The particular action our hypothesized mechanism refers to is the provision of insurance by other individuals and public sources. In other words, more trusting individuals expect external sources of risk sharing to play a bigger role in mitigating idiosyncratic shocks. As a consequence, individuals who face the same objective uncertainty (e.g. due to uninsurable earning ﬂuctuations) but differ in trust endowments tend to perceive the degree of uninsurable risk they are exposed to as lower (higher). Therefore, they are less (more) concerned about the likelihood of being constrained in their choices by borrowing limits or liquidity constraints. The same applies to their capacity to service ﬁnancial obligations.

Trust in formal and informal insurance systems. In our analysis, we consider trust in other individuals as referring to a households' social network, i.e. extended family, friends, neighbors and members of the same religious and interest-basedcommunities. High informal (pri- vate) trust means that a household expects this group to provide ﬁnancial support in case of unforeseen reductions of her income, for instance due to job loss, or unexpected required ex- penditures, such as paying for the repair of a broken car.13 In other words, trust reﬂects the households' expectations regarding the availability of ﬁnancial support from risk sharing arrangements in case of unexpected changes to their ﬁnancial situation. Therefore, we interpret

They also discover a non-linear relationship of trust on net worth which they argue is due to the fact that households can be 'too' trusting, i.e. suffer ﬁnancial losses due to credulous behavior.
This kind of risk is also sometimes referred to as 'background risk' which Guiso and Paiella (2008) deﬁne as "risks that cannot be avoided or insured against".
In principle, this support can take many forms beyond ﬁnancial assistance such as sharing private goods, pro- vision of (co)residence, help with job search, free legal advice and so forth.

ECB Working Paper Series No 2457 / August 2020

trust as households' subjective beliefs in the efﬁcacy of informal insurance systems. Accordingly, higher trust is associated with lower perceived uninsurable risk.14 Thus, while the deﬁnition of Guiso, Sapienza, and Zingales (2008b) emphasizes unconditional beliefs regarding the honesty of other individuals, ours focuses on their state contingent behavior, i.e. their willingness to share risks.

These views are not mutually exclusive and there is empirical support for our interpretation. The second wave of the German Panel on Household Finances (PHF) asks respondents to what extent they trust others15 and if they believe they could obtain assistance from their social network in case of ﬁnancial need.16 As ﬁgure 5 shows, households who answer the latter question with yes (1) have more trust in others; after controlling for additional characteristics affecting access to private insurance (e.g. marital status, income, etc.) we ﬁnd that an increase in private trust by one standard deviation at the mean increases the probability to expect private risk sharing by 5.3 percentage points (see table 6 in the appendix). We interpret this ﬁnding as evidence in support of our view on private trust and perceived uninsurable risk; higher trust individuals expect more private (informal) risk sharing which reduces their subjective assessment regarding the amount of risk they are exposed to.

Figure 5: Self-reported private trust by ﬁnancial assistance

We reﬁne and extend our working deﬁnition of trust by also considering trust in formal (public)

14Note that the mechanism implied by our interpretation of trust is different from risk aversion or conﬁdence.

15The survey question reads: "Question: How do you view yourself: Are you in general a person who trusts others or do you tend to distrust people?" The respondent replies by picking an integer value from 0 ("I do not trust them at all") to 10 ("I trust others completely"). We normalize the variable to be bounded between 0 and 1.

16The survey question reads: "Question: (Could you / your household / the household) rely on ﬁnancial support from friends or family in an emergency and could they cover (your / the) cost of living for (you / your household / the household) for approximately three months? By this, we do not mean friends or family who live in (your / the) household." The response is coded as 1 if the respondent answers in the afﬁrmative and 0 otherwise.

ECB Working Paper Series No 2457 / August 2020

networks as a source of risk sharing. Analogous to our reasoning regarding trust in other in- dividuals, trust in public institutions such as a nation's social security, healthcare and judicial system corresponds to beliefs in the capacity of public institutions to provide assistance in case of economic and ﬁnancial need. Put differently, we think that public trust captures the subjective probability households assign to support arriving from formal insurance systems, i.e. publicly managed assistance programs. Low subjective probabilities can be sustained by perceptions of lengthy bureaucratic procedures, differences between legal entitlements and actual disbursements, high administrative workloads on the side of the claimant, uncovered costs in health insurance such as illicit payments demanded by service providers, or, in the case ﬁnan- cial disputes, partial court decisions.17

Earlier work has studied the role of speciﬁc social insurance policies on precautionary portfolio choices, also considering the decision between investments in risky and unrisky assets. For example, there is a literature on the respective role of unemployment and health insurance programs.18 These studies show that such programs reduce uninsurable risks from the perspective of households; more insurance makes households invest more in risky assets and reduce precautionary savings. Those ﬁndings are consistent with the theoretical relationship between trust and portfolio choices we described above. However, it should be pointed out that our notion refers to the subjective view on the efﬁcacy of these policies, not the rational expectations assessment.

Testable implications. Given that we interpret trust endowments as determining beliefs on the efﬁcacy of formal and informal insurance systems, variation in trust has implications for the degree by which a household uses her portfolio for self-insurance, i.e. how much of it is dedicated to precautionary motives. Hence, differences in trust affect household portfolios through an insurance substitution effect. This effect is easily illustrated in a standard consumption- savings model in which agents face idiosyncratic income risk, incomplete asset markets and are required to make positive consumption expenditures every period. Suppose they deviate from rational expectations due to heterogeneous trust endowments. This dimension of heterogeneity makes agents differ with respect to their perceived exposure to uninsurable risk; they vary in their beliefs regarding the transmission of income ﬂuctuations to their disposable income due to expected transfers from their informal and formal risk sharing networks. Therefore, the high trust agent assigns lower probabilities to events such as reaching borrowing or liquidity constraints. As a consequence, if agents can self-insure by accumulating non-contingent liquid and illiquid assets, it is their perceived uninsurable risk which dictates the optimal level and liquidity composition of their portfolios. In summary, if an agent expects more insurance to be provided by sources other than her own portfolio, she dedicates her savings less to precaution- ary purposes.

They can also reﬂect unwillingness to accept ﬁnancial support from public sources, e.g. due to ideological reasons.
As illustrative examples, see Engen and Gruber (2001) and Goldman and Maestas (2013).

ECB Working Paper Series No 2457 / August 2020

Based on this reasoning, we investigate the following empirical relationship:

Trust and precautionary savings: Liquid wealth mirrors precautionary savings since liq- uid wealth can be used immediately and without liquidation costs to cover unexpected required expenses and dampen the effect of adverse income shocks. Therefore, if trust in insurance systems is higher, the need to hold large amounts of liquid assets decreases.
Note, however, that the effect of higher trust in other individuals may not be identical to that of higher trust in public institutions. This is because households who expect their social network to provide ﬁnancial assistance internalize that other members expect the same. Therefore, the subjective uninsurable income risk a household takes into consid- eration is also determined by the risk of other network members. This effect, known as reciprocity in the literature, has been formalized by researchers such as, for example Cox (1987). It is based on the notion that informal risk sharing arrangements require setting aside resources. If members fail to do so, they can be excluded from access to the net- work. Hence, households with high private trust decide to hold liquid assets not for the purpose of self-insurance but to signal their readiness to share risks with others.19 Thus, our ﬁrst hypothesis has two separate components:
- H1a: Higher trust in public institutions decreases precautionary savings
- H1b: Higher trust in other individuals increases precautionary savings
Trust and mortgage-based home ownership and leverage: Distinct trust endowments can also explain decisions taken by households who have to choose between renting a residence or becoming an owner using a mortgage. The former allows an unconstrained choice of the portfolio's liquidity composition and does not require any regular debt ser- vice payments. The latter, on the other hand, implies a large swap of liquid savings for a single illiquid asset. Thus, not only does a mortgage introduce a substantial liability into a household's portfolio but the acquisition of a home also reduces its capacity to smooth consumption spending in the event of income ﬂuctuations. Moreover, both rental and mortgage obligations require regular payments but re-schedulinga mortgage contract is a much more costly and lengthy process than adjusting or terminating a rent contract; in the face of a negative income shock, a renter can informally discuss the rent schedule with the landlord or ultimately take a unilateral decision (i.e. move out) while reschedul- ing a mortgage requires lengthy negotiations with the issuing bank, possibly even ﬁling for personal bankruptcy. Hence, we test the hypothesis if a household who has higher trust in informal and formal insurance systems is more likely to become an owner using a mortgage:

19Note this reciprocity effect does not hold for public trust as the household will, arguably, not consider being ready to assist the government with liquidity.

ECB Working Paper Series No 2457 / August 2020

H2a: Higher public or private trust increase the probability to be a home owner with mortgage (as compared to being a renter)

Furthermore, a higher trusting household should also be more willing to take out a mortgage which is larger in relation to the value of the main residence, i.e. results in a larger loan-to-value (LTV) ratio. Thus, our second hypothesis related to home ownership decisions is:

H2b: Higher trust increases the portfolio's loan-to-value ratio

In the context of these hypotheses, we emphasize that we do not consider variation in trust to be associated with different preferences regarding the quality or tenure status (owned versus rented) of housing services. Instead, we associate cultural variation with differences in perceived uninsurable risk exclusively. Furthermore, a reciprocity effect does not seem applicable for home ownership and leverage decisions as purchasing a home is an investment and differs from precautionary savings which need to be available instantaneously. Therefore, we argue that trust in private and public institutions inﬂuences the home ownership decision in the same direction.

Moreover, while hypotheses H1a and H2b are related, the latter is no by-product of the former. It might appear that a household holding less liquid wealth is mechanically forced to assume a higher LTV ratio for a given house value as she cannot make the same down payment. How- ever, this proposition does not apply for two reasons. First, it is not obvious that a low trusting household would spend more of her liquid assets on the down payment as she wants to continue holding more liquid assets as a buffer even after the purchase. Yet, as we do not observe the liquidity structure of the portfolio at the time of acquisition20 we cannot directly test for how much of her liquid assets a low trust household would give up to make a down payment. Second, savings for the purpose of home ownership can be held in illiquid accounts (e.g. building savings contracts) which are liquidated for the purpose of the down payment at the time of acquisition.

Methodology and Data

4.1 Separating Cultural Norms from Institutions

Our econometric objective is to identify the effect of two cultural norms, trust in other individuals and in public institutions, on household portfolio choice. To illustrate our empirical approach, let sh denote a certain portfolio choice of household h, e.g. the share of her wealth held in liquid assets. According to the mechanism described above, we conjecture that these choices are determined by h's demographic and economic characteristics (Xh), her cultural

20Mortgage and home value are reported retrospectively in the HFCS but other portfolio items are not.

ECB Working Paper Series No 2457 / August 2020

norms (nh) and the institutions of the country c in which h resides (Wc). We consider institutions formal and informal rules determining the availability and prices of asset and liability instruments.21 Hence, our conjecture implies a true portfolio choice model speciﬁed as

s	h	= f	(X	, n , Wc)	(1)
	s	h	h

Yet, when considering to estimate equation (1), the econometrician faces the problem that a country's institutions reﬂect the culture of its population and vice versa. On the one hand, institutions are designed according to a public choice mechanism so governments need to account for societal preferences when designing them. On the other hand, institutional changes have the capacity to alter cultural norms. For example, increasing regional autonomy can lead to higher levels of trust among individuals.22 Therefore, institutions jointly determine the covariate of interest and the dependent variable.

Moreover, the full universe of these institutions is multifaceted as well as informal and unobservable so they cannot be captured by a single numerical measure. In addition, their functional relationship to our variable of interest is opaque. However, one cannot omit institutions because estimating the linear regression model

sh = gXh + dnh + #h

(2)

violates the OLS exogeneity assumption because nh is correlated with the error term #h due to its endogenous relationship with (omitted) institutions and so

E[#hjnh] 6= 0

(3)

Accordingly, if we proceed on the assumption that institutions determine the supply of assets and liabilities while cultural traits determine household demand, we cannot identify the effects of culture on portfolio choice using the research design shown above. Instead, we need to hold ﬁx one of these variables so we can study the effect of variation in the other on observed outcomes. For this reason, earlier research has focused on instrumenting culture, i.e. on ﬁnding measures which are correlated with cultural variation but uncorrelated with institutional variation. For example, Guiso, Sapienza, and Zingales (2006) used religion and ethnicity as instruments for culture while Tabellini (2010) used literacy rates. Critical to the validity of the results produced by this approach is the exclusion restriction, i.e. the assumption that the instrument is correlated with culture but not with institutions. In general, these assumptions have little theoretical underpinning and have to rely mostly on plausibility. Thus, the academic verdict on papers in this tradition is not too positive. "(...) the exclusion restriction has been

21As illustrated in section 2.3, these institutions can materialize in, for example, the tax treatment of mortgage expenses, the generosity of housing subsidies, the taxation of different kinds of capital incomes, the design of health and unemployment insurance programs as well as of minimum income support and public pension systems.

22See Guiso, Sapienza, and Zingales (2008a).

ECB Working Paper Series No 2457 / August 2020

problematic" is a characterization given by Alesina and Giuliano (2015) in a survey article on this methodological approach.

Moreover, instrumenting culture in longitudinal analyses comes with a serious limitation. When studying the relationship between trust and the evolution of GDP, Algan and Cahuc (2010) noted that pursing this approach makes it "impossible to control for speciﬁc invariant national or regional features which could codetermine both trust and economic development". Put differently, statistical estimates based on instrumenting culture with time invariant variables are likely to be contaminated by time invariant omitted variation which determines both the instrument and the variable of interest.

Since our investigation focuses on the household level, we require an approach which keeps aggregate institutions ﬁxed but features cultural variation at the individual level. An ideal environment for this kind of investigation is a culturally diverse population within a given set of institutions. To illustrate this point note that, so far, we assumed that h resides in the particular country where institutions are endogenous to her culture. Yet, if we can relax this assumption for some households, we gain an essential advantage for our investigation. Let c(h) describe the relationship between h's country of origin and her current country of residence as

c(h) =

:i if h does not reside in her country of origin

Using this notation, we can classify households h as natives (n) and immigrants (i). For the former group, the endogenous relationship between culture and institutions applies but for the latter it does not. In other words, immigrants face the same set of institutions but vary with respect to their cultural norms. Hence, by studying this group of households we can identify the parameters of our model speciﬁed in equation (2) because the covariate nh is exogenous and so we can estimate the model without concerns regarding endogeneity.

This design of our empirical investigation is inspired by the Epidemiological Approach.23 According to Fernandez (2011), it can be summarized as follows: "The Epidemiological Approach studies the variation in outcomes across different immigrant groups residing in the same coun- try. Immigrants presumably differ in their cultures but share a common institutional and economic environment. This allows one to separate the effect of culture from the original economic and institutional environment."

Accordingly, our investigation exploits two facts related to culture. First, it is considered to be close to time-invariant at the individual level. In the words of Becker (1996): "Because of the difﬁculty of changing culture and its low depreciation rate, culture is largely a 'given' to

23Other recent papers which have applied the Epidemiological Approach to study a research question related to ours are Fuchs-Schundeln,¨ Masella, and Paule-Paludkiewicz (2019), Guin (2017), Huber and Schmidt (2019), Haliassos, Jansson, and Karabulut (2017), Matha,¨ Porpiglia, and Sierminska (2011) and Bertocchi, Brunetti, and Zaiceva (2018).

ECB Working Paper Series No 2457 / August 2020

individuals throughout their lifetimes." Note that this understanding does not rule out the possibility of individual cultural adjustment but assumes this process is slow. Second, culture is assumed to be strongly serially correlated across generations. In fact, Guiso, Sapienza, and Zin- gales (2006) deﬁned culture as "(...) those customary beliefs and values that ethnic, religious, and social groups transmit fairly unchanged from generation to generation". In other words, through parental (vertical) transmission, culture can be considered as inherited so there is a close correspondence between the cultural norms of parents and their children.24 As both of these features stress the notion that culture is established during infancy, associating cultural beliefs of parents and children has become a common practice. Moreover, structural models of parental value transmissions have been proposed, for example by Bisin and Verdier (2000).

Our particular interest in this context is in parental transmission of trust as this is the speciﬁc aspect of culture we are investigating. Indeed, there exists a rich set of evidence that trust is among those cultural traits most inﬂuenced by parents. One recent example of this literature is Dohmen et al. (2012). Studying regionally dis-aggregated household information from the German Socioeconomic Panel, they provide conclusive evidence that more trusting parents raise more trusting children and that the transmission is enforced by assortative mating of parents.

First versus second generation immigrants. In the literature applying the Epidemiological Approach, second generation immigrants are the preferred units of observation. This is because members of this group have been exposed to vertical cultural transmission but have not made the decision to emigrate from their country of origin themselves. Hence, selection bias, i.e. cultural sorting with respect to the culture of the resident population, is no concern for this group. As we explain below (section 4.3), we cannot identify second generation immigrants in our dataset and so our work differs as we study ﬁrst generation immigrants.

An important aspect in which second and ﬁrst generation immigrants differ is their exposure to horizontal cultural transmission; several scholars, for example Bisin and Verdier (2001) and Benabou and Tirole (2006) emphasized that parental transmission is not the only source of cultural formation, neither at the individual nor aggregate level. They argued that sources from within the current environment, e.g. role models outside of the family and social norms, also affect cultural formation.25 To illustrate that age is a critical measure determining the effects of exposure to vertical versus horizontal cultural transmission, let the expected difference in the distribution of cultural norms n between natives n and immigrants i be deﬁned as

Nn = jE[nhn] E[nhi ]j

(4)

24We want to point out that shared genes are another channel which link cultural values between parents and children. This aspect receives little attention in economics but more in medical science and psychology. A good recent survey on this growing ﬁeld is Sanchez-Roige et al. (2018).

25Bisin and Verdier (2011) provide a review of various channels of cultural transmission.

ECB Working Paper Series No 2457 / August 2020

Presumably, for a ﬁxed distribution of natives' cultural norms, Nn increases in the age at immigration of non-natives. This is because individuals who left their country of origin at an older age had more exposure to norms prevailing in their country of origin and less to those in their current country of residence.

For this reason, we explore different thresholds for the age at immigration in our sample of ﬁrst generation immigrants. For instance, it seems likely that immigrants who arrived before or after schooling age had a distinct exposure to country of residence non-parental transmission sources. Accordingly, all of our regressions are conducted separately for the entire sample and for households who arrived prior to their 18th birthday. The latter are the key group of interest from the perspective of the Epidemiological Approach. In addition, we are controlling for years in country in all of our regressions and allow for a non-linear relationship by also considering a squared term.26

Finally, we provide external veriﬁcation that second and ﬁrst generation immigrants do not systematically differ with respect to parental transmission of the particular cultural trait we are interested in. The European Social Survey (ESS) asks respondents questions related to cultural values as well as about their own (waves 1-8) and their parents' (waves 2-3) country of birth.27 Using this data, we compute for all second generation immigrants as well as for those who arrived before their 18th birthday the deviation from the average trust reported in the current country of residence and then take the average of each country of origin. Figures 6 and 7 compare this measure for ﬁrst and second generation immigrants to average trust prevalent among the population of the parental country of origin. The ﬁgures indicate a positive relationship between the two measures. Thus, an immigrant coming from a high trust country reports higher trust values than people born in their current country of residence. Additionally, the regression line has a similar slope for ﬁrst and second generation immigrants. Hence, the difference in horizontal transmission discussed above does not make ﬁrst and second generation immigrants incomparable with respect to trust.

Individual versus countryof origin trust measures. In our data we do not observe trust in the same unit of observation for which we have portfolio data. This is a common data shortcoming in the literature related to our research question and a widely employed solution is to assign immigrants with the cultural norms of their origin country.28 This practice exploits the slow evolution of trust and is supported by robust empirical evidence. For example, Guiso,

26We acknowledge that another difference between ﬁrst and second generation relates to familiarity with their destination country. For instance, the latter might face lower language barriers, better information about ﬁnancial product markets etc. We control for this concern in the robustness section using an omitted variable approach.

27The ESS questionnaire contains the same set of questions to measure private trust as the EVS. In speciﬁc: "Most people can be trusted or you can't be too careful"; "Most people try to take advantage of you, or try to be fair?"; "Most of the time people helpful or mostly looking out for themselves?" We apply the same normalization as we describe in section 4.2.

28Some examples of papers related to ours which follow this approach are Luttmer and Singhal (2011), Giuliano (2007), Fernandez (2007) and Fernandez and Fogli (2009).

ECB Working Paper Series No 2457 / August 2020

Figure 6: First generation value transmission Figure 7: Second generation value transmission

Sapienza, and Zingales (2006) use the American General Social Survey to study if differences in trust in the US population can be traced back to the trust prevailing in the country of their ancestors. They report that a country of origin speciﬁc cultural prior "continues to affect individual beliefs even in the new environment and even several generations later." Using the same survey, Algan and Cahuc (2010) conﬁrmed and extended this ﬁnding; they found that among immigrants of different countries, "inherited trust is strongly persistent" and respective differences can be observed even after four generations. Exploiting this empirical regularity, Algan and Cahuc (2010) "proxy the inherited trust of people living in country c by the trust that the descendants of US immigrants have inherited from their ancestors coming from country c." Our approach is analogous as we associate our observations with measures of trust observed in their countries of origin.

Immigration and exogenous institutions. We conclude with a ﬁnal remark regarding our econometric approach and identiﬁcation assumption. As mentioned above, we assume that immigrants arriving to a country face a supply of assets and liabilities which is exogenous to their culture as it is determined by institutions which a culturally distinct resident population has adopted. In this environment, cultural differences among immigrants materialize in different demand behavior. Variations in trust are particularly relevant for immigrants as they are generally less informed about local customs and practices and so they have to fall back on prior beliefs, in particular with respect to the behavior of other individuals.29 It is worth mentioning, however, that in the case of large inﬂows of foreigners, our identiﬁcation assumption could be questionable as newly arriving immigrants could cause a change in the institutions governing the supply of ﬁnancial instruments.

For several reasons, we think it is justiﬁable to abstract from this possibility. First, the immigrants would need to arrive from the same country of origin, or at least a set of countries

29This idea is reﬂected in the words of Porta et al. (1997) as follows: "Trust should be more essential for ensuring cooperation between strangers."

ECB Working Paper Series No 2457 / August 2020

which are culturally very similar. Second, even if respective immigration episodes actually took place, they would have needed to lead to pervasive institutional change, far beyond the scope of, say, labor market regulations or redistributive policies; as numerous formal and informal institutions affect portfolio choice, the newly arriving immigrants would need to be in a median voter position with respect to a wide array of policy topics. Hence, it appears unlikely that the institutional composition affecting portfolio choice changes drastically and abruptly, even during waves of immigration.30 Yet, to minimize concerns regarding this potential pitfall for identiﬁcation, we use country ﬁxed effects for all of our destination countries and age at immigration in all of our estimations. The former controls for heterogeneity across countries with respect to institutional change caused by immigrants and the latter accounts for portfolio choice differences across cohorts, i.e. time variation in institutions.

4.2 Measuring Trust in Other Individuals and in Public Institutions

The literature has taken two different avenues to measure trust. The ﬁrst relies on indirect evidence and uses outcome-based variables determining or correlated with trust (blood do- nations, voter turnout, literacy rates, etc.). The second is a more direct approach which elicits trust of individuals based on their actions in an experimental setting31 or via questioning in sur- veys. Given our research objective, the most suitable approach is to use survey-based measures which distinguish between trust in other individuals and in public institutions. We construct them from responses to the European Value Study (EVS) and World Value Survey (WVS). The EVS provides pertinent data from a large-scale,cross-national and longitudinal survey since 1981 at a nine-year frequency. Its fourth wave was conducted in 2008 and includes more than 22,000 households interviewed in a face-to-face setting. This EVS release32 corresponds best to the HFCS second wave from which we obtain household portfolio data for our empirical anal- ysis. To construct trust indicators for countries of origin not covered by the EVS, we use the sixth wave of the WVS which was collected between 2010 and 2013. Even though both surveys are harmonized, not every question in the EVS is asked in the WVS. Therefore, we use two slightly different trust indicators in our regressions, depending on an immigrant household's country of origin.

In the EVS, we associate trust in other individuals with the following three33 questions:

"Generally speaking, would you say that most people can be trusted or that you can't be too careful in dealing with people?"

Guiso, Sapienza, and Zingales (2006) show that the redistributive policies of US states can be traced back to the preferences for redistribution of immigrants. In Europe however, this ﬁnding cannot be replicated. For example, studying data from Germany and the UK, Dancygier and Saunders (2006) report "we observe that immigrants are no more likely to support increased social spending or redistributive measures than natives."
The 'trust game' developed by Berg, Dickhaut, and McCabe (1995) has become the standard tool for this pur- pose.
The speciﬁc EVS dataﬁle we work with is "Integrated Dataset EVS 2008 (ZA4800 Data ﬁle Version 4.0.0)"
In the WVS, only the ﬁrst two questions are available. In contrast to Guiso, Sapienza, and Zingales (2008b) who use only the ﬁrst question, we use all three to obtain a more robust measure.

ECB Working Paper Series No 2457 / August 2020

"Do you think that most people would try to take advantage of you if they got the chance, or would they try to be fair?"
"Would you say that most of the time people try to be helpful or that they are mostly looking out for themselves?"

For our investigation, trust in public institutions refers to their role as providers of insurance. Regarding these institutions, the EVS asks "How much conﬁdence you have in them: is it a great deal, quite a lot, not very much or none at all?" From a range of different options, we pick the social security, health care and justice systems as those institutions are relevant for providing insurance against adverse income and required expenditure shocks. Unfortunately, in the WVS only the last option is available.

In the EVS and WVS, possible answers depend on each question. For instance, the ﬁrst question which we associate with private trust can be answered with "Most people can be trusted" or "Can't be too careful", whereas for the other two questions more than ten different answer options are available. Therefore, we normalize the variables as shown in equation (5) where x˜ is the normalized version of x while xmin and xmax are its minimum and maximum realization:34

x˜	=	xi	xmin	(5)
xmax	xmin

As a result of this normalization, we receive values in the interval between zero (no trust) and one (full trust). To construct the trust indicators, we use the unweighted mean of all associated variables to treat each question as equally important. Figures 8 to 11 show the intensity of the indicator in each country or region. The realizations of our indicators range from around 0.25 to 0.8 with multiple increments reﬂecting the fact that trust in other individuals and in public institutions vary substantially across countries. Overall, the ranking is as expected; households in Scandinavian countries have high trust while those in Southern American countries are at the lower end of the spectrum.35

Table 1 shows the correlation between each variable which is used to construct the private and public trust indicator in the EVS. Within each trust category, the answers to the associated survey questions are closely correlated.36 The correlations between variables measuring private and public trust are smaller but uniformly positive. This relationship in our sample conforms to earlier research; using data from 20 countries, Porta et al. (1997) documented that higher pri-

Note that this normalization does not affect correlations between variables.
Throughout our investigation, we take self-reported intensities of the cultural norms of our interest as given since our research question does touch upon their generation. We note, however, that there is a rich literature which considers a wide array of phenomena explaining variations in trust. Some prominent of these are hierarchical religions (Porta et al. (1997) and Guiso, Sapienza, and Zingales (2006)), ethnic fractionalization (Knack and Keefer (1997)), episodes of regional autonomy (Guiso, Sapienza, and Zingales (2008a)), enslavement raids (Nunn and Wantchekon (2011)), traumatic experiences, discrimination, unequal opportunities (Alesina and Ferrara (2002)), class conﬂicts, famines, totalitarian or communist regimes (Algan and Cahuc (2010)).
Hence, using only the question on conﬁdence in the justice system as a measure for public trust in the WVS countries seems justiﬁable.

ECB Working Paper Series No 2457 / August 2020

Figure 8: Private Trust Indicator (constructed from EVS)

NAF: ALG, EGY, LIB, MOR, TUN. CSA: HAI, TRI, MEX, COL, ECU, PER, BRA, CHL, ARG, URU. NME: ARM, AZE, BAH, GRG, IRQ, JOR, KUW, LEB, PSE, QAT. OAS: CHN, TAW, HKG, ROK, JPN, IND, PAK, THI, MAL

Figure 9: Private Trust Indicator (constructed from WVS)

ECB Working Paper Series No 2457 / August 2020

Figure 10: Public Trust Indicator (constructed from EVS)

NAF: ALG, EGY, LIB, MOR, TUN. CSA: HAI, TRI, MEX, COL, ECU, PER, BRA, CHL, ARG, URU. NME: ARM, AZE, BAH, GRG, IRQ, JOR, KUW, LEB, PSE, QAT. OAS: CHN, TAW, HKG, ROK, JPN, IND, PAK, THI, MAL

Figure 11: Public Trust Indicator (constructed from WVS)

ECB Working Paper Series No 2457 / August 2020

vate trust is associated with higher government effectiveness (measured by indicators on the efﬁciency of the judicial system, corruption, bureaucratic quality etc.).37

			Private Trust			Public Trust
		People can	People are	People are	Social Security	Health care	Justice
		be trusted	fair	helpful	System	System	System
	People can be trusted	1.0000
Private People are fair	0.4097	1.0000
Trust
	People are helpful	0.3362	0.5365	1.0000
	Social Security System	0.0672	0.0891	0.1100	1.0000
Public	Health care System	0.0813	0.1044	0.1318	0.5295	1.0000
Trust

	Justice System	0.1292	0.1221	0.1464	0.4702	0.5537	1.0000

Table 1: Trust Indicators: Correlation Matrix

4.3 Our Sample

We use the HFCS to obtain information on household assets and liabilities as this dataset is exceptionally rich with respect to portfolio information. However, it does not allow to identify second-generation immigrants; the survey does not ask respondents for their parents' countries of birth and it does not allow to link different generations of the same family. Fortunately, we have access to a variable which allows us to identify households headed by a ﬁrst generation immigrant.38 In the public release version of the HFCS, answers to the question "In which country was (were) X (you) born?" (RA0400) are converted to aggregate country groups for non-native residents due to conﬁdentiality restrictions. From researchers and representatives of the national central banks of Austria, France, Germany39 and Ireland, we received the as- collected information of this variable and permission to use it for our research project.

Moreover, using information on years in current country40, we can compute the age at arrival of ﬁrst generation immigrants. Hence, if we restrict our attention to immigrants who arrived as young children (for example younger than 18), the assumption that their immigration decisions were exogenous is valid. Furthermore, as discussed in section 4.1, second generation immigrants and ﬁrst generation immigrants who arrived at a young age are comparable with respect to trust, i.e. our main measure of interest. In our regressions presented in section 5,

Aghion et al. (2010) provide evidence showing that low trust is associated with high governmental regulation. However, regulation in this paper refers to product and labor markets, the ease to start a business, price controls etc., and does not concern social insurance policies which we are considering.
We use the Canberra Deﬁnition to determine household reference persons.
This paper uses data from the Deutsche Bundesbank Panel on Household Finances. The results published and the related observations and analysis may not correspond to results or analysis of the data producers.
RA0500: "For how many years (have you/has X) lived in 'country in which the interview is taking place'?"

ECB Working Paper Series No 2457 / August 2020

we show results separately for all immigrants and for those who immigrated before their 18th birthday who are the focal group of our investigation.

Figure 12: HFCS countries

Figure 12 shows all wave two HFCS countries (in light blue) and those countries for which we have the non-anonymized information on country of birth (in dark blue). As we have four countries of destination, we can observe immigrants from the same country of origin in different institutional settings. This is a distinction of our paper as most other studies have been conducted in a single country context. Thus, our econometric identiﬁcation has a twofold ad- vantage. First, we observe immigrants from a large set of countries of origin and so our sample features a broad spectrum of cultural variation. Second, if the same cultural traits lead to similar ﬁnancial behavior in different institutional environments, we can consider our ﬁndings a robust and well-identiﬁed effect of culture on portfolio choices.41

Another feature of the HFCS data we need to account for is the fact that it does not provide us with a household speciﬁc measure of trust.42 Consistent with the literature (as presented in 4.1), we construct indicators measuring private trust and trust in public institutions from the EVS and WVS and assign each immigrant household its values of her country of birth. Hence,

To ensure that we can compare choices of culturally similar immigrants across our sample countries, we use ﬁxed effects to control for country speciﬁc institutional features regulating the supply of assets.
Some contributing surveys, for example the German PHF, contain a small number (one) of related questions. However, these questions are neither in the HFCS core nor non-core questionnaires. Hence, they are not harmonized across countries.

ECB Working Paper Series No 2457 / August 2020

a household with a reference person born in Italy but currently living in France receives a trust indicator from the mean of all Italian answers in the EVS (and WVS).

Table 25 in the appendix summarizes the demographics and portfolio categories of natives, all non-natives and the ﬁve largest immigrant groups in our sample countries Austria (AT), Ger- many (DE), France (FR) and Ireland (IE). The portfolios of immigrant groups differ not only from those of natives but also among each other. Hence, within a given institutional environ- ment, we observe variation both in the trust measures and also in the portfolios. Moreover, the immigrant groups are fairly similar with respect to demographic characteristics which addresses concerns that selection differs across countries of origin.

For our analysis, we focus on the working age population so we restrict the age of the household head to the range of 25 to 65 years. Moreover, we drop unbanked households because we do not have any information regarding the portfolios of these households. In other words, they could be holding large amounts of liquid savings in the form of cash or have no savings at all but these choices are unobservable for us.43 To do so, we apply the deﬁnition provided by Ampudia and Ehrmann (2017) and consider an unbanked household as "neither holding checking accounts nor savings accounts with ﬁnancial institutions". Finally, note that our non- anonymized HFCS country of birth variable (RA0400) provides limited information for some households. For example, several countries are recorded in broader country groups (such as North Africa). We assign trust indicators to these households by calculating averages over the respective country groups.

Portfolio Choice Variables

Precautionary Savings: The HFCS provides a detailed breakdown of assets which allows to classify them as liquid versus illiquid. Recall that in our research design, liquid net wealth mirrors precautionary savings as it consists of assets which are quickly converted to cash and can be used to dampen adverse economic shocks. Following standard conventions, we compute liquid assets as the sum of deposits, mutual funds, bonds and stocks and liquid debt as the sum of overdraft and credit card debt. The difference between them is liquid net wealth.

Home Ownership Status: As described above, we model the home ownership decision as the choice between staying a renter or becoming a home owner with the exigency of a mortgage.

43Guiso, Sapienza, and Zingales (2004) reported that households with low levels of trust are less likely to engage in transactions occurring in formal ﬁnancial markets. For example, they made more frequent use of cash payments and informal loan agreements. Thus, they could be overrepresented among unbanked households. To gauge the effect of our related sample selection decision, we computed the share of unbanked households in our dataset. We ﬁnd that, in our sample countries Austria, France and Germany, the share of non-banked households among natives and immigrants is extremely low (the share in Germany is largest with about 0.5%). It is also worth emphasizing that the differences between natives and immigrants are minuscule. Non-banked households also represent a very small fraction of the population in Ireland, albeit larger than in the other three countries. However, the share of unbanked households among immigrants is actually lower than among natives (3% vs. 6%).

ECB Working Paper Series No 2457 / August 2020

Hence, for this investigation, we exclude households which acquired their main residence as inheritance or gift. This approach allows us to focus on the ﬁnancial risk related to becoming an owner as opposed to renting. Using the HFCS data, we create a home ownership variable which is 1 if the household owns the main residence and took out at least one mortgage and 0 if the household is a renter. All other cases result in missing values and we exclude them from our regression as we cannot determine if they ever took out a mortgage.

Leverage Ratio: We also investigate the intensive mortgage choice, i.e. the extent to which a household is ready to assume debts relative to the value of the acquired residence. The relationship between these variables is called loan-to-value (LTV) ratio and deﬁned as

Value Mortgage	(6)
LTV = Value Main Residence

In our baseline investigation, we compute this measure using the values of the mortgage and the main residence at time of acquisition. Picking this reference period has two clear advantages over using their current values. First, it mirrors the leverage ratio at the time of the decision (when the household could choose the ratio). Second, given that the price of the residence as well as the exact mortgage amount was known to the household, it is less contaminated by measurement error. For completeness, we report the results of using current values in the robustness section.

Estimation and Results

In this section, we empirically characterize the relationship between portfolio choice and trust in other individuals and in public institutions, which reﬂect subjective expectations on the ef- ﬁcacy of informal and formal insurance systems. Hence, in what follows, we describe in detail how we test the four hypotheses H1a, H1b, H2a and H2b stated above, present the corresponding results and provide some comparisons with related ﬁndings in the literature. Our analysis distinguishes between immigrants from European and non-European (World) countries of birth because the trust indicators have to be constructed differently as mentioned earlier. Addi- tionally, we differentiate between our full sample (all) and immigrants who arrived as children (< 18). The former comprises all immigrants while the latter group is our best proxy for second generation immigrants which are the main group of interest in the context of the Epidemiolog- ical Approach as discussed in section 4.1.

5.1 Trust and Precautionary Savings

We ﬁrst test hypothesis H1. Recall it states that higher trust in public institutions reduces precautionary savings (H1a) while higher trust in other individuals increases them (H1b). In order to test this relationship, we estimate the model shown in equation (7) using OLS, i.e. we

ECB Working Paper Series No 2457 / August 2020

regress log precautionary savings44 on our (log) trust indicator and a set of controls.

˜ 0 0

log(Yi) = a + b log(Zi) + g Xi + d Wi + ei(7)

In this model, Yi measures net liquid wealth (liquid assets less liquid liabilities) of household i while Xi is a vector of household control variables45 which includes income, level of net illiquid wealth, gender, age, years in country, marital status, number of children, level of education, occupation status and risk taking behavior46. Moreover, we control for country ﬁxed effects using

i as we pool immigrants living in the four HFCS countries of our sample. ˜ i is the variable

of our main interest; it represents the value of the trust indicator (summarized in ﬁgures 8 to 11) assigned to immigrant household i based on the reference person's country of birth. Table

2 presents the coefﬁcient estimate of ˜ i and table 8 in the appendix reports the complete results

of estimating equation (7).

Columns 1 to 4 of table 2 show that we cannot reject hypothesis H1a for the group of our main interest. For ﬁrst generation immigrants who arrived as children to their current country of residence, the indicator measuring trust in public institutions is negatively correlated with pre- cautionary savings and highly signiﬁcant; if the public trust indicator increases by 1%, liquid wealth decreases between 3.4 to 1.8%. Thus, these coefﬁcient estimates provide support for our hypothesized relationship. For the entire group of immigrants, i.e. also including those who arrived as adults, the estimated sign of the relationship is identical. However, its magnitude is smaller and statistical signiﬁcance can only be established for immigrants arriving from non-European origins to our HFCS sample countries.

The estimates presented in columns 5 to 8 illustrate that this direction is reversed for precau- tionary savings and trust in other individuals. The coefﬁcient estimate of the private trust indicator is highly signiﬁcant and ranges from 2.2 to 2.6%. Both magnitude and signiﬁcance are comparable for all immigrant groups. This ﬁnding is in accordance with the mechanism described in section 3.2, i.e. it is based on reciprocity; households who trust their social networks to provide support in ﬁnancially challenging situations ready themselves to be able to do the same for other network members.47 Hence, when deciding on how much net wealth to hold in liquid form, a household considers not only the uninsurable risk faced by herself, but also risk facing other network members. As discussed earlier, the reciprocity effect does

44We choose to measure precautionary savings as current Euros held in liquid wealth for two reasons. First, this speciﬁcation allows a clear interpretation of the estimated coefﬁcients. Second, alternative measurements, such as the share of liquid versus illiquid assets, are not suitable if there are large differences of net wealth among the obser- vations; the shares of liquid assets in two portfolios with different total assets can be identical but represent distinct amounts of liquid assets in nominal terms (which is the relevant dimension from a precautionary perspective).

45We apply the Canberra Deﬁnition to identify the household reference person so that we can assign person level variables (gender, age, etc) to each household.

46As a categorical measure for risk taking behavior, we use the HFCS variable HD1800 which asks respondents about risk preferences in ﬁnancial investments.

47In fact, this behavior can be considered a requirement to be admitted to a private risk sharing network and remain a member.

ECB Working Paper Series No 2457 / August 2020

Table 2: H1: Trust and Liquid Wealth (OLS)

		(1)	(2)	(3)	(4)	(5)	(6)	(7)	(8)
		all	<18	all	<18	all	<18	all	<18
	Europe	-0.59	-3.43***
	(0.60)	(1.01)
Public Trust
			-0.81*	-1.81***
	World
			(0.41)	(0.65)


	Europe					2.44***	2.55**
					(0.63)	(1.16)
Private Trust
							2.20***	2.47***
	World
							(0.52)	(0.83)


N		1,308	454	2,111	807	1,308	454	2,111	807
Controls		X	X	X	X	X	X	X	X
Fixed Effects		X	X	X	X	X	X	X	X

Dependent variable: log liquid wealth; Variables of interest: Public/Private Trust (ln)

Countries: Austria, Germany, France, Ireland

Columns: 1-4 Public (H1a) and 5-8 Private (H1b) Trust. The ﬁrst two use only immigrants from Europe while the latter include non-European immigrants. For each speciﬁcation (all) uses the full sample while (<18) uses only immigrants who arrived before age 18.

Signiﬁcance at * p<.1, ** p<.05, *** p<.01

not apply to trust in public institutions. Hence, trust in private and public insurance systems should have a distinct impact on precautionary savings from a theoretical perspective and our ﬁndings provide empirical support for this hypothesis.

Our ﬁndings regarding hypothesis H1b also relate to results reported by other researchers. For instance, Guiso, Sapienza, and Zingales (2008b) argue that trust in other people increases the willingness to invest in risky liquid assets such as stocks. However, their mechanism is based on subjective expectations to be cheated while ours is based on perceived uninsurable risk, i.e. differences in precautionary behavior. To explore these distinct views, we follow their analysis and replace the dependent variable in equation (7) with the share of risky assets (stocks and mutual funds) in total liquid wealth. Given the scale of this alternative dependent variable, we estimate a logistic regression and ﬁnd a positive, yet insigniﬁcant relationship between private trust and stock holdings in the sample of all immigrants which we report in table 7 in the appendix.48 Thus, while Guiso, Sapienza, and Zingales (2008b) document a link between private trust and the amount of risky assets in the liquid share of the portfolio, we ﬁnd a positive association between trust and the overall amount of net liquid wealth a household owns.49

48Due to the small number of households owning stocks or mutual funds in our sample (not more than 20% across all age and country of origin groups) it is not surprising that we do not ﬁnd signiﬁcant results.

49Note that we control for income as well as net illiquid wealth which rules out a general wealth or income effect.

ECB Working Paper Series No 2457 / August 2020

Table 3: H1: Trust and Precautionary Motive to Save (Probit)

	(1)	(2)	(3)	(4)	(5)	(6)	(7)	(8)
	all	<18	all	<18	all	<18	all	<18

	-0.25	-0.63**
	Europe	(0.32)
Public Trust	(0.19)
		-0.24*	-0.23
	World
		(0.14)	(0.23)


	Europe				0.29	-0.49
				(0.22)	(0.44)
Private Trust
						0.42**	0.14
	World
						(0.18)	(0.31)


N	1,308	454	2,111	807	1,308	454	2,111	807
Controls	X	X	X	X	X	X	X	X
Fixed Effects	X	X	X	X	X	X	X	X

Dependent variable: Dummy is 1 if motive to save for precautionary reasons has highest priority; Variables of interest: Public/Private Trust

Countries: Austria, Germany, France, Ireland

Estimates are reported as margins at means; Signiﬁcance at * p<.1, ** p<.05, *** p<.01

To investigate further if the empirical evidence supports our theory regarding the effect of trust on portfolio choices, we proceed by testing it in a way which does not refer to realized portfolio choices but to reported intentions. For that purpose, we employ the HFCS question "What are your (household's) most important reasons for saving?" (HI0400x). The range of possible answers contains "Provision for unexpected events" which is closely related to the precaution- ary motive we investigate. Hence, to check if liquid wealth is accumulated for precautionary reasons, we code whether the household has chosen this particular answer (or not) as a binary variable and regress it on the trust indicators and the same controls as in equation (7) using a Probit model. As shown in table 3 (see full results in table 9 in the appendix), our earlier ﬁnd- ings remain unchanged; precautionary ﬁnancial behavior is negatively associated with public trust but positively with private trust.50 If trust in public institutions increases by one stan-

50Even though not all coefﬁcient estimates are signiﬁcant, we still consider their implications on the empirical relationship between trust and motives for precautionary saving as credible. This is because we generated the binary dependent variable used in this estimation in the most conservative way; from all twelve possible reasons

ECB Working Paper Series No 2457 / August 2020

dard deviation at the mean, the probability that a household saves for precautionary reasons decreases by 1.9 to 5.1 percentage points. However, a one standard deviation increase of trust in other individuals increases this probability by about 3 percentage points.

To sum up, we do not reject hypotheses H1a and H1b. We ﬁnd evidence that i) more trust in public institutions decreases the motive to hold liquid wealth for precautionary reasons ii) more trust in other individuals has a reverse effect which we cannot attribute to increasing participation in risky liquid assets such as stocks, but rather to a reciprocity mechanism. With respect to earlier ﬁndings, for example those presented by Guiso, Sapienza, and Zingales (2008b), we conclude that higher trust in other individuals drives up liquid savings independent of the risk associated with the asset.

5.2 Trust and Home Ownership

Hypothesis H2a states that the probability to own the main residence with a mortgage as opposed to renting it increases in trust. Accordingly, we shift the focus from liquid to illiquid assets and compare renters to home owners with a mortgage. As described above, we restrict ourselves to these two groups as their distinct willingness to take ﬁnancial risks is apparent; becoming a home owner with mortgage requires the transformation of most assets from liquid to highly illiquid for the average household. In addition, taking out a mortgage increases the ﬁnancial risk the household's portfolio is exposed to. For these reasons and because of the intrinsic differences between rent and mortgage contracts discussed in section 3.2, we consider owning with a mortgage as ﬁnancially riskier than renting. Since home ownership is a binary variable at the household level, we apply a Probit model as speciﬁed in equation (8):

Yi = a + bZ˜ i + g0Xi + d0Wi + ei

(8)

In this model, Yi assumes the value 1 for home owners with a mortgage and 0 for renters.51 As

above, the trust indicator is the variable of our main interest and is represented by ˜ i, while i

Z X

is a vector of household controls and Wi captures country ﬁxed effects.

Table 4 presents the main ﬁndings relating to hypothesis H2a (the full results are in table 10 in the appendix). With respect to trust in public institutions, we ﬁnd that the indicator is positively correlated with the probability to be an owner with mortgage (except for all immigrants from non-European countries) and has a comparable magnitude for all groups. However, only the coefﬁcient for all European immigrants is estimated with statistical signiﬁcance. For our main group of immigrants (those who arrived as children), the coefﬁcient is positive but not signiﬁcant for both sets of origin countries. Thus, we can only interpret this set of results as indicative evidence that the probability to take ﬁnancial risk for the purpose of being a home

for saving listed in HI0400x (e.g. old-age provision, travels or holidays, to leave a bequest, etc) we only classiﬁed "Provision for unexpected events" as reﬂecting a precautionary motive.

51All other cases, such as outright owners, result in missing values and do not enter our regression.

ECB Working Paper Series No 2457 / August 2020

Table 4: H2a: Trust and Home Owner with a Mortgage (Probit)

		(1)	(2)	(3)	(4)	(5)	(6)	(7)	(8)
		all	<18	all	<18	all	<18	all	<18
	Europe	0.54***	0.22
	(0.21)	(0.40)
Public Trust
			-0.12	0.05
	World
			(0.15)	(0.29)


	Europe					1.35***	0.90
					(0.25)	(0.57)
Private Trust
							1.30***	1.22***
	World
							(0.20)	(0.42)


N		1,036	338	1,646	562	1,036	338	1,646	562
Controls		X	X	X	X	X	X	X	X
Fixed Effects		X	X	X	X	X	X	X	X

Dependent variable: Dummy is 1 if household is owner with a mortgage and 0 if renter; Variables of interest: Pub- lic/Private Trust

Control variables: log gross income, male, age, years in country, married, number of children, dummy for secondary education, dummy for currently employed, risk taking behavior

Columns: 1-4 Public and 5-8 Private Trust. The ﬁrst two use only immigrants from Europe while the latter include non- European immigrants. For each speciﬁcation (all) uses the full sample while (<18) uses only immigrants who arrived before age 18.

Estimates are reported as margins at means; Signiﬁcance at * p<.1, ** p<.05, *** p<.01

owner is increasing in trust in public institutions.

In contrast, the empirical relationship between trust in other individuals and the dependent variable is uniformly positive. Moreover, it is highly signiﬁcant with the exception of immigrants from European countries who arrived before 18. In quantitative terms, if the private trust indicator increases by 1 standard deviation, the probability to become a home owner with a mortgage increases by around 9.1 to 10.4 percentage points. This is a sizeable and precisely estimated effect in our sample.

In summary, we fail to reject hypothesis H2a for trust in other individuals, but we ﬁnd mixed empirical evidence with respect to trust in public institutions as we cannot characterize a robust relationship. We can, however, draw the conclusion that differences in risk uninsured by private sources do not only affect the amount of precautionary savings a household accumulates but they also have the capacity to affect home ownership decisions. Indeed, given the magnitude of the coefﬁcient estimates, small cultural differences with respect to trust in other individuals have large effects in this dimension.

5.3 Trust and Leverage

In a ﬁnal step, we evaluate the link between trust and willingness to assume liabilities and ﬁnancial risk. Hypothesis H2b, which we developed in section 3.2, states that a portfolio's

ECB Working Paper Series No 2457 / August 2020

leverage ratio should be higher for more trusting households. As a dependent variable for the corresponding empirical investigation, we use the ratio between the value of the mortgage and the value of the main residence which is used as a collateral, i.e. the loan-to-value (LTV) ratio. We use a Tobit speciﬁcation and retain the same controls as in the previous investigations.52 Speciﬁcally, we estimate:

	Yi = a + bZ˜ i + g0Xi + d0Wi + ei			(9)
	Table 5: H2b: Trust and LTV Ratio (Tobit)

	(1)	(2)	(3)	(4)	(5)	(6)	(7)	(8)
	all	<18	all	<18	all	<18	all	<18

	0.46***	0.25
	Europe	(0.21)
Public Trust	(0.13)
		-0.05	0.13
	World
		(0.10)	(0.16)


	Europe				0.82***	0.46
				(0.15)	(0.30)
Private Trust
						0.71***	0.56**
	World
						(0.13)	(0.23)


N	1,036	338	1,646	562	1,036	338	1,646	562
Controls	X	X	X	X	X	X	X	X
Fixed Effects	X	X	X	X	X	X	X	X

Dependent variable: Leverage ratio: Value of mortgage divided by value of household main residence (both at time of

acquisition); Variables of interest: Public/Private Trust

Control variables: log gross income, male, age, years in country, married, number of children, dummy for secondary education, dummy for currently employed, risk taking behavior

Countries: Austria, Germany, France, Ireland

Columns: 1-4 Public and 5-8Private Trust. The ﬁrst two use only immigrants from Europe while the latter include non- European immigrants. For each speciﬁcation (all) uses the full sample while (<18) uses only immigrants who arrived before turning 18.

Estimates are reported as margins at means; Signiﬁcance at * p<.1, ** p<.05, *** p<.01

Table 5 shows the coefﬁcient estimates of the trust indicators (table 11 in the appendix contains the full results). Overall, we ﬁnd similar results as for hypothesis H2a; the relationship between trust in public institutions and the LTV is positive (or weakly negative for one group) but not signiﬁcant except for immigrants from Europe. The coefﬁcient estimates of the private trust indicator, on the other hand, are uniformly positive, larger in magnitude and signiﬁcant.

52Using a linear estimation would produce biased results because renters have a leverage ratio of 0. Hence, Tobit speciﬁcations are commonly used for this kind of investigation, see for instance Georgarakos, Haliassos, and Pasini (2014).

ECB Working Paper Series No 2457 / August 2020

Speciﬁcally, if trust in other individuals is one standard deviation higher, the leverage ratio increases between 0.04 to 0.06. This means that for a given value of the main residence, a more trusting household would be ready to assume a 4% (6%) higher mortgage on average. Hence, we cannot reject hypothesis H2b and ﬁnd a sizable effect of trust in other individuals on the leverage ratio, i.e. the household's willingness to take ﬁnancial risks for the purpose of becoming a home owner.

Again, our results relate to the ﬁndings of Guiso, Sapienza, and Zingales (2008b) mentioned earlier. They reported that trust in other individuals makes households more willing to assume risks in ﬁnancial investments. Our ﬁndings generalize their conclusion as it indicates that this mechanism also applies to assuming liabilities. In this context, it is worth pointing out that Jiang and Lim (2018) hypothesize high trusting households appear to have better debt management capacities. Complementing their reasoning, our ﬁndings indicate that differences with respect to assuming liabilities are additionally driven by trust-related perceptions of uninsurable risk.

Robustness

1. Unobservable Heterogeneity in Familiarity If a household head is more familiar with her current country of residence (e.g. with respect to language, legal system, etc.) barriers to asset and debt markets might play a smaller role. Since we control for age at immigration in our baseline estimations, we already account for this concern to a large extent as the degree of familiarity is increasing in total years spent in the country and will be stronger for individuals who immigrated at an earlier age. Still, differential familiarity with respect to speciﬁc ﬁnancial products might have a major confounding effect; if certain products do not exist in some countries of origin, it might take longer to become familiar with them or to consider using them at all. We can address this concern by controlling for the status of the reference person's spouse (who can be immigrant or native). If the reference person is married to a native, the degree of familiarity is arguably stronger. In other words, our baseline regressions might suffer from an omitted variable. When extending our regressions, we ﬁnd that being married to a native increases the amount of liquid assets as well as the probability to own the main residence and also the leverage ratio. However, as shown by tables 12 to 14 in the appendix, the coefﬁcient estimates for the trust indicators remain unchanged. Hence, stronger familiarity might help to overcome barriers to speciﬁc product markets, but it does not affect the relationship between trust in other individuals and in public institutions and the ﬁnancial choice variables we are interested in.

2. Noise in Survey Trust Measures A plausible concern regarding our results is that the trust indicators are contaminated by measurement error or different interpretations of the survey questions. However, the implied ordinal ranking of countries should be unaffected by this

ECB Working Paper Series No 2457 / August 2020

concern. Hence, we split our sample and differentiate between high and low trust countries and repeat our estimation. Our results remain largely unchanged (see tables 15 to 17 in the ap- pendix), but one has to be aware of the fact that most countries are bunched close to the mean of the trust indicator variable. Hence, by splitting the immigrant sample based on an arbitrary cutoff with many countries close to this margin, interpreting the results is not as straightfor- ward. Hence, our preferred approach remains to control for speciﬁc numeric realizations of the trust indicators.

Trust and Price Expectations One could argue that purchasing real estate such as the main residence is an investment which reﬂects expectations on future house price increases. For ex- ample, at the time of acquisition, more trusting households could expect a mechanical decrease in the LTV ratio due to an increase in the value of the main residence. As a consequence, if these expectations are related to variations in trust, there would be a systematic difference be- tween measures referring to the time of acquisition and current values. However, as shown in table 18, when accounting for this concern using current values of the mortgage and the main residence, our results remain unchanged.
Robustness of the Private Trust Measure As mentioned above, scholars have used differ- ent variables to measure trust in other individuals. A prominent approach which is distinct from ours but also based on survey responses utilizes information on membership and partic- ipation in voluntary organizations to construct a "civic indicator". As explained for example by Anheier and Kendall (2002), the reasoning behind measuring trust in this way is that in- teracting with people outside the narrow personal circle teaches to cooperate with and to rely on other individuals - which ultimately improves trust. We follow Dekker and Broek (1998, 2005) and construct this type of trust measure using EVS and WVS questions on membership and voluntary work in multiple categories.53 In line with their reasoning, we do not include afﬁliations to religious and work-related organizations as the former have country speciﬁc in- terpretations and the latter (such as trade union membership) can be quasi mandatory.54

Tables 19 to 21 show the results for each hypothesis when we use these alternative trust mea- sures. As it has been argued that the intensity of participation in these organizations is important in this context, we ﬁrst generate a binary variable which is 1 if the respondent is at

53The EVS asks for a wide range of related organizations, communities and activities: social welfare services for elderly, handicapped or deprived people; religious or church; education, arts, music or culture; trade unions, political parties or groups; local community action on issues like poverty, employment, housing, racial equality; third world development or human rights; conservation; environment, ecology, animal rights; professional associations; youth work (e.g. scouts, guides, youth clubs etc.); sports or recreation; women's groups; peace movement. The WVS asks about membership and participation in organizations related to church or religion; sport or recreation; art, music or education; labor unions and political parties; environment; professional and humanitarian or charity and consumer associations, self-help and mutual aid groups.

54Note that the literature using civic indicators focuses on Western countries. As we include other parts of the world in our analysis, which might differ substantially in the concept of membership, the estimates using indicators derived from the WVS should be interpreted with caution.

ECB Working Paper Series No 2457 / August 2020

least a passive member in one organization and 0 otherwise (columns 1-4). Additionally, we follow Howard and Gilbert (2008) and distinguish between the indicator realizations 'inactive', 'passive' and 'active' members (columns 5-8). In doing so, we consider an individual an active member if she is currently doing unpaid voluntary work. The results reveal that the binary civic indicator (columns 1-4) gives similar estimates as our baseline private trust indicator in terms of signiﬁcance. Interestingly, distinguishing between different levels of engagement (columns 5-8) does not increase predictive power. In terms of magnitude, we discover a smaller association of the civic indicator with our outcome variables, presumably because being passively or actively involved in a voluntary organization measures multiple individual characteristics and not solely inter-personal trust. Hence, the variable is probably more noisy than our direct measure used in the baseline estimations.

5. Individual-SpeciﬁcUninsurable Risk Another concern regarding our results is based on individual heterogeneity with respect to income variability. It seems plausible that households who have highly variable earnings would tend to i) increase their precautionary savings ii) decrease their exposure to ﬁnancial obligations, independent of their cultural dispositions. In- deed, for some countries in our sample (France and Germany) Dossche and Hartwig (2019) have shown that workers with different income levels and ages differ with respect to the ex- posure of their income to aggregate risk. Note, however, that we already control for these variables in our baseline regression.

Moreover, in household-level data from the United States, Guvenen et al. (2017) found that earnings risk also depends on the sector in which a worker is employed. Furthermore, there is anecdotal evidence that some immigration groups tend to be systematically more self-employed than others. To address this issue, we re-estimate our baseline regressions with additional controls for individual speciﬁc income variability by adding dummies for self-employment and sector of occupation for salaried households.55 This extra set of controls also addresses concerns related to self-selection; for the United States and Germany, Schulhofer-Wohl (2011) and Fuchs-Schundeln¨ and Schundeln¨ (2005) found that more risk tolerant households select into jobs with higher earnings risk. By including these extra controls, we also reduce concerns our coefﬁcient estimates might be suffering from omitted variable bias (due to heterogeneous income risk). Tables 22 to 24 in the appendix show that, when controlling for these additional variables, the estimate of the trust coefﬁcients are unaffected. Hence, these results indicate that in our sample, trust is not (strongly) correlated with sector of occupation or entrepreneurship, i.e. objective differences in uninsurable risk.

In summary, we addressed several possible shortcomings of our empirical investigation. We did so using a variety of approaches to estimate our relationships of interest and included additional controls to account for omitted variables. Throughout all of these robustness in-

55The sectors we control for are based on the ﬁrst letters of the 2008 NACE classiﬁcation. Due to conﬁdentiality restrictions, we do not report results for each sector but treat them as ﬁxed effects in our estimation.

ECB Working Paper Series No 2457 / August 2020

vestigations, we found only small changes in the magnitude of the trust indicator coefﬁcient estimates. Hence, our baseline empirical results regarding the effect of trust on liquid wealth, home ownership with mortgage and LTV ratios seem robust.

Conclusion

In this paper, we investigate if different intensities of two cultural norms - trust in other individuals and trust in public institutions - help to explain household portfolio liquidity. Our speciﬁc interest is to explore if higher trust is associated with lower precautionary savings, measured as liquid wealth, and higher willingness to take ﬁnancial risk for the purpose of becoming a home owner. The underlying mechanism relating culture to portfolio choice is based on subjective expectations regarding the role of private and public institutions as informal and formal insurance systems; households who have more trust in them assign higher probabilities to receiving support in ﬁnancial distress, i.e. they consider uninsurable risks to be smaller.

Non-anonymized information on country of birth in the HFCS and measures on public and private trust from the EVS and WVS allow us to apply the Epidemiological Approach to isolate the effect of culture from (endogenous) institutions on ﬁnancial behavior of households. The key advantage of this approach is that we do not need to rely on exclusion restrictions of instrumental variables. Instead, we observe a culturally diverse population in an exogenous institutional environment. Accordingly, we can identify the pure effect of culture on portfolio choice. Finally, as we have a large set of origin countries as well as four destination coun- tries, our estimates result from observing individuals with the same cultural traits in several distinct environments. Hence, our identiﬁcation strategy minimizes concerns regarding the consequences of country speciﬁc ﬁnancial market features.

The results of our empirical investigation document two regularities which provide support in favor of the relationship between trust and precautionary portfolio choices we hypothesize. First, trust inﬂuences the amount of precautionary savings; an increase in trust in public institutions by 1% decreases the desire to hold liquid wealth by around 0.8-3.4%, as households believe that public insurance channels reduce uninsured risks. Contrarily, an increase in private trust by 1% increases liquid wealth by approximately 2.2-2.6%. A reciprocity effect explains this ﬁnding; to access communal risk sharing networks, households need to signal their readiness to provide ﬁnancial assistance to other network members. These results hold even when we use measures on reported purposes for saving as opposed to observable ﬁnancial choices.

The second hypothesis relates trust and home ownership decisions. In our setting, the household can either rent the main residence or purchase it by taking out a mortgage. Differences in trust in public institutions do not affect this decision signiﬁcantly. However, we ﬁnd that an increase in trust in other individuals by one standard deviation at the mean increases the probability to purchase a home using a mortgage by 9.1-10.4 percentage points. In addition, more trusting households also assume LTV ratios which are higher by 4-6 percentage points.

ECB Working Paper Series No 2457 / August 2020

Thus, private trust has a sizable effect on asset and liability choices related to a household's home ownership decision.

Our ﬁndings document that cultural variation helps to explain household portfolio heterogeneity across advanced economies. To study its quantitative effect from a general equilibrium perspective, one also needs to consider cross-country variation in other dimensions affecting portfolio choice. Put differently, controlling for economy speciﬁc features such as institutions, shocks and product familiarity allows to decompose the quantitative effect of cultural variation on cross-country portfolio differences. The model we use in section 3.2 to derive testable implications between subjective differences in uninsurable risks and portfolio choice is a starting point for this endeavor. Moreover, this exercise would also quantify the relevance of each explanation as well as residual, i.e. unexplained, variation.

As we only have access to non-anonymized information on country of birth in four HFCS national datasets (Austria, France, Germany and Ireland) our sample has a small number of observations. Due to this restriction, our sample is unbalanced with respect to the cultural norms we study; we have a large number of immigrant households from low trust countries and only few from high trust countries. Moreover, the small sample size also limits our ability to incrementally control for different ages at immigration. Finally, the HFCS does not provide geographic information of respondents so we cannot control for location sorting of different immigrant communities.

A caveat exclusive to our analysis of mortgage decisions is that observable differences among immigrants might not be due to different demand behavior but, rather, driven by varying degrees of discrimination.56 While we can control for endogeneity in aggregate supply, we do not observe the distribution of mortgage offers at the individual level. Hence, we cannot assess if it differs systematically for high and low trust immigrants. If it does, observed differences in LTV ratios could not be attributed to different demand behavior alone. One can think of two reasons why mortgage offers might vary across immigrant communities. First, for the US, there is some indicative evidence that discrimination in mortgage loan application exists despite extensive regulation promoting equal opportunities in lending.57 However, the few available studies do not provide a conclusive verdict on its quantitative aspects, in particular with respect to offered leverage. Moreover, to the best of our knowledge, there are no comparable studies on mortgage offers in Europe so we cannot address this concern.

Second, differences in mortgage offers could be related to differences in trust more directly.

56Our investigation is not the ﬁrst to face this concern. Fernandez (2011) mentioned this possibility when discussing the results of Guiso, Sapienza, and Zingales (2004): "They ﬁnd that people who were originally from provinces with higher civic capital make larger investments in stocks, rely more on checks to settle transactions, and have easier access to loans. It should be noted that while the authors interpret the latter ﬁnding as resulting from trustworthiness, it is also consistent with discrimination."

57Some papers speaking to this speciﬁc topic are Ladd (1998), Hanson et al. (2016), Ambrose, Conklin, and Lopez (2018) and Fuster et al. (2017). A common descriptive ﬁnding of these papers is that the characteristic on which lenders discriminate most often is race (Afro-American especially). Since racial characteristics are a minor difference among immigrant groups in our sample, it is not evident if and how these ﬁndings affect our analysis.

ECB Working Paper Series No 2457 / August 2020

In a seminal paper, Glaeser et al. (2000) found that high trust individuals are more trustworthy themselves. If brokers and bank loan ofﬁcers recognize and endogenize this feature of applicants, higher trust individuals would be offered more favorable terms when taking out a mortgage. Yet, there is scarce empirical evidence on this "trustworthiness discrimination" mechanism. So far, it has only been found to play a role in experimental peer-to-peer lending and in a peer-to-peer micro ﬁnance platform (see Duarte, Siegel, and Young (2012) and Chen, Foster, and Putterman (2019)). For settings represented in our sample, we have not been able to ﬁnd studies providing quantitative assessments on the extent of discrimination attributable to applicants' trust differences.

Given the tentative evidence on both of these aspects, we believe dedicated studies on discrimination in European credit markets would make for valuable contributions to this ﬁeld of research. On the one hand, their ﬁndings would provide additional insights into the causes of household portfolio heterogeneity in the Euro Area. On the other hand, they would help to sharpen the understanding of cultural determinants of household portfolio choices, those related to differences in perceived uninsurable risk in particular. In fact, the HFCS is the natural starting point for these empirical studies.

ECB Working Paper Series No 2457 / August 2020

References

Adam, Klaus and Panagiota Tzamourani (2016). "Distributional Consequences of Asset Price Inﬂation in the Euro Area." European Economic Review 89, 172 -192.

Adam, Klaus and Junyi Zhu (2016). "Price-Level Changes and the Redistribution of Nominal Wealth Across the Euro Area." Journal of the European Economic Association 14(4), 871-906.

Aghion, Philippe et al. (2010). "Regulation and Distrust." The Quarterly Journal of Economics 125(3), 1015-1049.

Alesina, Alberto and Eliana La Ferrara (2002). "Who Trusts Others?" Journal of Public Economics 85, 207-234.

Alesina, Alberto and Paola Giuliano (2015). "Culture and Institutions." Journal of Economic Literature 53(4), 898-944.

Algan, Yann and Pierre Cahuc (2010). "Inherited Trust and Growth." American Economic Review 100(5), 2060-2092.

Ambrose, Brent W., James Conklin, and Luis Lopez (2018). "Preferential Treatment in Financial Contracts: Does Broker Race Affect Mortgage Prices?" Working Paper.

Ampudia, Miguel and Michael Ehrmann (2017). "Macroeconomic Experiences and Risk Taking of Euro Area Households." European Economic Review 91(C), 146-156.

Ampudia, Miguel, Has van Vlokhoven, and Dawid Zochowski (2016). "Financial Fragility of Euro Area Households." Journal of Financial Stability 27, 250 -262.

Anheier, Helmut and Jeremy Kendall (2002). "Interpersonal Trust and Voluntary Associations: Examining Three Approaches." The British Journal of Sociology 53(3), 343-362.

Asdrubali, Pierfederico, Bent Sorensen, and Oved Yosha (1996). "Channels of Interstate Risk Sharing: United States 1963-1990." The Quarterly Journal of Economics 111(4), 1081-1110.

Auclert, Adrien (2019). "Monetary Policy and the Redistribution Channel." American Economic Review 109(6), 2333-2367.

Banﬁeld, Edward C. (1958). The Moral Basis of a Backward Society. Free Press.

Barth, Daniel, Nicholas W. Papageorge, and Kevin Thom (2020). "Genetic Endowments and Wealth Inequality." Journal of Political Economy 128(4), 1474-1522.

Becker, Gary (1996). "Preferences and Values." Accounting for Tastes. Harvard University Press.

Benabou, Roland and Jean Tirole (2006). "Belief in a Just World and Redistributive Politics." The Quarterly Journal of Economics 121(2), 699-746.

Berg, Joyce, John Dickhaut, and Kevin McCabe (1995). "Trust, Reciprocity, and Social History." Games and Economic Behavior 10(1), 122-142.

Bertocchi, Graziella, Marianna Brunetti, and Anzelika Zaiceva (2018). The Financial Decisions of Immigrant and Native Households: Evidence from Italy. CEPR Discussion Paper 13339. Centre for Economic Policy Research.

Bisin, Alberto and Thierry Verdier (2000). "Beyond the Melting Pot: Cultural Transmission, Marriage, and the Evolution of Ethnic and Religious Traits." The Quarterly Journal of Economics 115(3), 955-988.

(2001). "The Economics of Cultural Transmission and the Dynamics of Preferences." Journal of Economic Theory 97(2), 298-319.

ECB Working Paper Series No 2457 / August 2020

Bisin, Alberto and Thierry Verdier (2011). "The Economics of Cultural Transmission and Social- ization." Handbook of Social Economics, 339-416.

Brown, Jeffrey R. et al. (2008). "Neighbors Matter: Causal Community Effects and Stock Market Participation." The Journal of Finance 63(3), 1509-1531.

Bucciol, Alessandro, Barbara Cavasso, and Luca Zarri (2019). "Can Risk-Averse Households Make Risky Investments? The Role of Trust in Others." The Scandinavian Journal of Economics 121(1), 326-352.

Chen, Josie I., Andrew Foster, and Louis Putterman (2019). "Identity, trust and altruism: An experiment on preferences and microﬁnance lending." European Economic Review 120, 103304.

Christelis, Dimitris, Michael Ehrmann, and Dimitris Georgarakos (2017). Exploring Differences in Household Debt Across the United States and Euro Area Countries. CSEF Working Paper 465. University of Naples.

Christelis, Dimitris, Dimitris Georgarakos, and Michael Haliassos (2013). "Differences in Portfolios across Countries: Economic Environment versus Household Characteristics." The Review of Economics and Statistics 95(1), 220-236.

Coibion, Olivier et al. (2017). "Innocent Bystanders? Monetary Policy and Inequality." Journal of Monetary Economics 88, 70-89.

Cox, Donald (1987). "Motives for Private Income Transfers." Journal of Political Economy 95(3), 508-546.

Cronqvist, Henrik and Stephan Siegel (2015). "The Origins of Savings Behavior." Journal of Political Economy 123(1), 123-169.

Dancygier, Rafaela and Elizabeth N. Saunders (2006). "A New Electorate? Comparing Preferences and Partisanship between Immigrants and Natives." American Journal of Political Science 50(4), 962-981.

Dekker, Paul and Andries van den Broek (1998). "Civil Society in Comparative Perspective: Involvement in Voluntary Associations in North America and Western Europe." Voluntas: International Journal of Voluntary and Nonproﬁt Organizations 9(1), 11-38.

(2005). "Involvement in Voluntary Associations in North America and Western Europe: Trends and Correlates 1981-2000." Journal of Civil Society 1, 45-59.

Doepke, Matthias, Martin Schneider, and Veronika Selezneva (2015). Distributional Effects of Monetary Policy. 2015 Meeting Paper 1099. Society for Economic Dynamics.

Dohmen, Thomas et al. (2012). "The Intergenerational Transmission of Risk and Trust Atti- tudes." The Review of Economic Studies 79(2), 645-677.

Dossche, Maarten and Jacob Hartwig (2019). "Household income risk over the business cycle." ECB Economic Bulletin 6.

Duarte, Jefferson, Stephan Siegel, and Lance Young (2012). "Trust and Credit: The Role of Appearance in Peer-to-peer Lending." The Review of Financial Studies 25(8), 2455-2484.

Engen, Eric M. and Jonathan Gruber (2001). "Unemployment insurance and precautionary sav- ing." Journal of Monetary Economics 47(3), 545 -579.

European Commission (2011). Common System of Financial Transaction Tax and Amending Directive 2008/7/EC. Proposal for a Council Directive.

ECB Working Paper Series No 2457 / August 2020

Fagereng, Andreas, Martin B. Holm, and Gisle James Natvik (2018). MPC Heterogeneity and Household Balance Sheets. Working Paper 7134. CESifo.

Fernandez, Raquel (2007). "Alfred Marshall Lecture: Women, Work, and Culture." Journal of the European Economic Association 5(2-3),305-332.

(2011). "Does Culture Matter?" Handbook of Social Economics. Ed. by Jess Benhabib, Alberto Bisin, and Matthew Jackson. Vol. 1A. Elsevier.

Fernandez, Raquel and Alessandra Fogli (2009). "Culture: An Empirical Investigation of Be- liefs, Work, and Fertility." American Economic Journal: Macroeconomics 1(1), 146-177.

Fuchs-Schundeln,¨ Nicola and Michael Haliassos (2020). "Participation Following Sudden Ac- cess." CEPR Discussion Paper.

Fuchs-Schundeln,¨ Nicola, Paolo Masella, and Hannah Paule-Paludkiewicz (2019). "Cultural Determinants of Household Saving Behavior." Journal of Money, Credit and Banking, 1-36.

Fuchs-Schundeln,¨ Nicola and Matthias Schundeln¨ (2005). "Precautionary Savings and Self- Selection: Evidence from the German Reuniﬁcation 'Experiment'." The Quarterly Journal of Economics 120(3), 1085-1120.

Fuster, Andreas et al. (2017). "Predictably Unequal? The Effects of Machine Learning on Credit Markets." Working Paper.

GESIS Data Archive, Cologne. European Values Study. Integrated Dataset EVS 2008 (ZA4800 Data ﬁle Version 4.0.0).

Gambetta, Diego (2000). "Can We Trust Trust?" Trust: Making and Breaking Cooperative Relations. University of Oxford.

Georgarakos, Dimitris, Michael Haliassos, and Giacomo Pasini (2014). "Household Debt and Social Interactions." The Review of Financial Studies 27(5), 1404-1433.

Giuliano, Paola (2007). "Living Arrangements in Western Europe: Does Cultural Origin mat- ter?" Journal of the European Economic Association 5(5), 927-952.

Glaeser, Edward L. et al. (2000). "Measuring Trust." The Quarterly Journal of Economics 115(3), 811-846.

Glass, Jennifer, Robin W. Simon, and Matthew A. Andersson (2016). "Parenthood and Happi- ness: Effects of Work-Family Reconciliation Policies in 22 OECD Countries." American Journal of Sociology 122(3).

Goldman, Dana and Nicole Maestas (2013). "Medical Expenditure Risk and Household Portfolio Choice." Journal of Applied Econometrics 28(4), 527-550.

Gornemann, Nils, Keith Kuester, and Makoto Nakajima (2016). Doves for the Rich, Hawks for the Poor? Distributional Consequences of Monetary Policy. International Finance Discussion Papers 1167. Board of Governors of the Federal Reserve System.

Guin, Benjamin (2017). Culture and Household Saving. Working Paper 2069. European Central Bank.

Guiso, Luigi and Tullio Jappelli (2005). "Awareness and Stock Market Participation." Review of Finance 9(4), 537-567.

Guiso, Luigi and Monica Paiella (2008). "Risk Aversion, Wealth, and Background Risk." Journal of the European Economic Association 6(6), 1109-1150.

ECB Working Paper Series No 2457 / August 2020

Guiso, Luigi, Paola Sapienza, and Luigi Zingales (2004). "The Role of Social Capital in Financial Development." American Economic Review 94(3), 526-556.

(2006). "Does Culture Affect Economic Outcomes?" Journal of Economic Perspectives 20(2), 23- 48.
(2008a). "Alfred Marshall Lecture: Social Capital as Good Culture." Journal of the European Economic Association 6(2-3),295-320.
(2008b). "Trusting the Stock Market." The Journal of Finance 63(6), 2557-2600.

Guvenen, Fatih et al. (2017). "Worker Betas: Five Facts about Systematic Earnings Risk." Amer- ican Economic Review 107(5), 398-403.

Haliassos, Michael, Thomas Jansson, and Yigitcan Karabulut (2017). "Incompatible European Partners? Cultural Predispositions and Household Financial Behavior." Management Science 63(11), 3780-3808.

Hanson, Andrew et al. (2016). "Discrimination in mortgage lending: Evidence from a correspondence experiment." Journal of Urban Economics 92, 48-65.

Hartmann, Philipp (2015). "Real Estate Markets and Macroprudential Policy in Europe." Journal of Money, Credit and Banking 47(S1), 69-80.

Hintermaier, Thomas and Winfried Koeniger (2018). Differences in Euro-Area Household Finances and their Relevance for Monetary-Policy Transmission. Working Paper 7088. CESifo.

Household Finance and Consumption Network (2013). The Household Finance and Consumption Survey: Methodological Report for the First Wave. Statistics Paper Series No 1. European Central Bank.

(2016a). The Household Finance and Consumption Survey: Methodological Report for the Second Wave. Statistics Paper Series No 17. European Central Bank.
(2016b). The Household Finance and Consumption Survey: Results from the Second Wave. Statistics Paper Series. European Central Bank.

Howard, Marc Morje´ and Leah Gilbert (2008). "A Cross-National Comparison of the Internal Effects of Participation in Voluntary Organizations." Political Studies 56(1), 12-32.

Huber, Stefanie J. and Tobias Schmidt (2019). Cross-Country Differences in Homeownership: A Cultural Phenomenon? Discussion Paper 40. Deutsche Bundesbank.

Hubmer, Joachim, Per Krusell, and Anthony A. Jr. Smith (2018). A Comprehensive Quantitative Theory of the U.S. Wealth Distribution. Working Paper. Yale University.

Jappelli, Tullio and Luigi Pistaferri (2014). "Fiscal Policy and MPC Heterogeneity." American Economic Journal: Macroeconomics 6(4), 107-36.

Jiang, Danling and Sonya S. Lim (2018). "Trust and Household Debt." Review of Finance 22(2), 783-812.

Kaplan, Greg, Benjamin Moll, and Giovanni L. Violante (2018). "Monetary Policy According to HANK." American Economic Review 108(3), 697-743.

Kaplan, Greg and Gianluca Violante (2014). "A Model of the Consumption Response to Fiscal Stimulus Payments." Econometrica 82(4), 1199-1239.

Kaplan, Greg, Gianluca Violante, and Justin Weidner (2014). "The Wealthy Hand-to-Mouth." Brookings Papers on Economic Activity 48(1), 77-153.

ECB Working Paper Series No 2457 / August 2020

Knack, Stephen and Philip Keefer (1997). "Does Social Capital Have an Economic Payoff? A Cross-Country Investigation." The Quarterly Journal of Economics 112(4), 1251-1288.

Ladd, Helen F. (1998). "Evidence on Discrimination in Mortgage Lending." Journal of Economic Perspectives 12(2), 41-62.

Lenza, Michele and Jiri Slacalek (2018). How does Monetary Policy affect Income and Wealth In- equality? Evidence from Quantitative Easing in the Euro Area. Working Paper 2190. European Central Bank.

Luetticke, Ralph (2018). Transmission of Monetary Policy with Heterogeneity in Household Portfolios. Discussion Paper 1819. Centre for Macroeconomics (CfM).

Luttmer, Erzo and Monica Singhal (2011). "Culture, Context, and the Taste for Redistribution." American Economic Journal: Economic Policy 3(1), 157-179.

Malmendier, Ulrike and Stefan Nagel (2011). "Depression Babies: Do Macroeconomic Experiences Affect Risk Taking?" The Quarterly Journal of Economics 126(1), 373-416.

Malmendier, Ulrike and Alexandra Steiny (2017). Rent or Buy? The Role of Lifetime Experiences of Macroeconomic Shocks within and across Countries. Manuscript. UC Berkeley.

Matha,¨ Thomas Y., Alessandro Porpiglia, and Eva Sierminska (2011). The Immigrant/Native Wealth Gap in Germany, Italy and Luxembourg. Working Paper 1302. European Central Bank.

Nunn, Nathan and Leonard Wantchekon (2011). "The Slave Trade and the Origins of Mistrust in Africa." American Economic Review 101(7), 3221-3252.

Pham-Dao, Lien (2019). "Public Insurance as a Driver of Wealth Inequality in the Euro Area." Working Paper.

Porta, Rafael La et al. (1997). "Trust in Large Organizations." The American Economic Review 87(2), 333-338.

Putnam, Robert, Robert Leonardi, and Raffaella Nanetti (1993). Making Democracy Work: Civic Traditions in Modern Italy. Princeton University Press.

Sanchez-Roige, S. et al. (2018). "The Genetics of Human Personality." Genes, Brain and Behavior 17(3).

Schulhofer-Wohl, Sam (2011). "Heterogeneity and Tests of Risk Sharing." Journal of Political Economy 119(5), 925-958.

Slacalek, Jiri, Oreste Tristani, and Gianluca Violante (2020). "Household Balance Sheet Channels of Monetary Policy: A Back of the Envelope Calculation for the Euro Area." Journal of Economic Dynamics and Control, 103879.

Tabellini, Guido (2010). "Culture and Institutions: Economic Development in the Regions of Europe." Journal of the European Economic Association 8(4), 677-716.

ECB Working Paper Series No 2457 / August 2020

Empirical Appendix

9.1 Trust and Expected Ability to obtain Financial Aid

Table 6: Trust and Expected Ability to obtain Financial Aid from Friends or Relatives (Probit)

	I
Private Trust	0.26***
	(0.04)
Gross Income (log)	0.02
	(0.01)
Net Illiquid Wealth (log)	0.02***
	(0.00)
Male	0.02
	(0.02)
Age	-0.02***
	(0.00)
Ageˆ2	0.00***
	(0.00)
Married	-0.01
	(0.03)
# Children	0.00
	(0.01)
Secondary Education	-0.11***
	(0.02)
Employed	0.02
	(0.03)
Risk Taking	-0.06***
	(0.02)
N	3,397

Dependent variable: Dummy is 1 if ability to get ﬁnancial assistance from friends or relatives is answered with 'yes'; Variable of inter- est: Private Trust (self-reported)

Control variables: log gross income, log net illiquid wealth, male, age, married, number of children, dummy for secondary education, dummy for currently employed, risk taking behavior

Estimates are reported as margins at means; Signiﬁcance at * p<.1, ** p<.05, *** p<.01

ECB Working Paper Series No 2457 / August 2020

9.2 Trust and Risky Assets

Table 7: H1b: Trust and Share of Risky Assets in Total Liquid Wealth (Logit)

		(1)	(2)	(3)	(4)
		all	<18	all	<18
	Europe	0.08	0.15
Private Trust		(0.09)	(0.18)
World			0.17**	0.28

				(0.08)	(0.18)
Germany		-0.03	0.02	-0.01	0.03
		(0.02)	(0.02)	(0.02)	(0.02)
France		0.06**	0.19***	0.06***	0.19***
		(0.03)	(0.05)	(0.02)	(0.03)
Ireland		0.00	0.03	0.02	0.04
		(0.02)	(0.02)	(0.02)	(0.02)
Gross Income (ln)		0.07***	0.06***	0.08***	0.12***
		(0.01)	(0.02)	(0.01)	(0.02)
Net Illiquid Wealth (ln)		0.01***	0.01**	0.01***	0.01**
		(0.00)	(0.00)	(0.00)	(0.00)
Male		-0.01	0.00	0.01	0.03
		(0.01)	(0.02)	(0.01)	(0.02)
Age		0.00	0.01	0.00	0.00
		(0.00)	(0.01)	(0.00)	(0.01)
Ageˆ2		0.00	0.00	0.00	0.00
		(0.00)	(0.00)	(0.00)	(0.00)
years in country		0.00	0.01	0.00	0.00
		(0.00)	(0.01)	(0.00)	(0.01)
years in country ˆ2		0.00	0.00	0.00	0.00
		(0.00)	(0.00)	(0.00)	(0.00)
Married		-0.02	-0.02	-0.02*	-0.02
		(0.01)	(0.02)	(0.01)	(0.03)
# Children		-0.01	-0.01	0.00	0.00
		(0.01)	(0.02)	(0.01)	(0.01)
Secondary Education		-0.02	-0.01	-0.02*	0.00
		(0.01)	(0.02)	(0.01)	(0.02)
Employed		-0.01	0.01	0.01	0.02
		(0.01)	(0.02)	(0.01)	(0.02)
Risk Taking		0.05***	0.07***	0.07***	0.10***
		(0.01)	(0.02)	(0.01)	(0.02)
N		1,308	454	2,111	807

Dependent variable: share of risky assets in liquid wealth; Variables of interest: Private Trust

Control variables: log gross income, log net illiquid wealth, male, age, years in coun- try, married, number of children, dummy for secondary education, dummy for currently employed, risk taking behavior.

Countries: Austria, Germany, France, Ireland

Columns: The ﬁrst two use only immigrants from Europe while the latter include non-European immigrants. For each speciﬁcation (all) uses the full sample while (<18) uses only immigrants who arrived before age 18.

Estimates are reported as margins at means; Signiﬁcance at * p<.1, ** p<.05, ***

p<.01

ECB Working Paper Series No 2457 / August 2020

9.3 Complete Baseline Regression Results

Table 8: H1: Trust and Net Liquid Wealth (OLS): Complete results

		(1)	(2)	(3)	(4)	(5)	(6)	(7)	(8)
		all	<18	all	<18	all	<18	all	<18
	Europe	-0.59	-3.43***
Public Trust		(0.60)	(1.01)
World			-0.81*	-1.81***

				(0.41)	(0.65)
	Europe					2.44***	2.55**
Private Trust						(0.63)	(1.16)
World							2.20***	2.47***

								(0.52)	(0.83)
Germany		-1.02**	-1.35**	-0.92***	-0.94*	-1.00**	-1.37**	-0.89***	-0.94*
		(0.41)	(0.64)	(0.32)	(0.56)	(0.40)	(0.65)	(0.32)	(0.55)
France		0.93**	1.32**	0.72***	1.17**	0.90**	1.27**	0.81***	1.24**
		(0.37)	(0.58)	(0.28)	(0.49)	(0.37)	(0.58)	(0.28)	(0.48)
Ireland		-0.96***	-0.50	-0.86***	-0.68	-1.37***	-1.36**	-1.09***	-1.10**
		(0.35)	(0.58)	(0.28)	(0.51)	(0.36)	(0.65)	(0.29)	(0.54)
Gross Income (ln)		0.87***	0.93***	0.76***	0.83***	0.85***	0.96***	0.75***	0.87***
		(0.12)	(0.20)	(0.08)	(0.14)	(0.12)	(0.20)	(0.08)	(0.14)
Net Illiquid Wealth (ln)		0.16***	0.15***	0.16***	0.13***	0.15***	0.15***	0.15***	0.13***
		(0.02)	(0.03)	(0.01)	(0.02)	(0.02)	(0.03)	(0.01)	(0.02)
Male		0.08	0.38	0.07	0.28	0.12	0.38	0.10	0.27
		(0.19)	(0.31)	(0.14)	(0.21)	(0.19)	(0.32)	(0.14)	(0.21)
Age		-0.23***	-0.31*	-0.25***	-0.25**	-0.25***	-0.33**	-0.26***	-0.27**
		(0.07)	(0.16)	(0.05)	(0.11)	(0.07)	(0.16)	(0.05)	(0.11)
Ageˆ2		0.00***	0.00*	0.00***	0.00**	0.00***	0.00**	0.00***	0.00**
		(0.00)	(0.00)	(0.00)	(0.00)	(0.00)	(0.00)	(0.00)	(0.00)
Years in country		-0.05**	0.06	-0.05***	0.01	-0.06**	0.01	-0.05***	-0.02
		(0.02)	(0.07)	(0.02)	(0.05)	(0.02)	(0.08)	(0.02)	(0.05)
Years in country ˆ2		0.00	0.00	0.00***	0.00	0.00**	0.00	0.00***	0.00
		(0.00)	(0.00)	(0.00)	(0.00)	(0.00)	(0.00)	(0.00)	(0.00)
Married		-0.15	-0.22	-0.04	-0.02	-0.08	-0.24	-0.05	-0.05
		(0.21)	(0.34)	(0.15)	(0.23)	(0.21)	(0.34)	(0.15)	(0.23)
# Children		-0.19	-0.27	-0.13	-0.13	-0.20	-0.38	-0.11	-0.13
		(0.16)	(0.24)	(0.10)	(0.14)	(0.16)	(0.24)	(0.10)	(0.14)
Secondary Education		-0.73***	-0.66*	-0.83***	-0.86***	-0.63***	-0.51	-0.75***	-0.72***
		(0.19)	(0.34)	(0.14)	(0.23)	(0.19)	(0.36)	(0.14)	(0.23)
Employed		0.60***	0.20	0.42***	-0.02	0.65***	0.30	0.44***	0.01
		(0.20)	(0.33)	(0.14)	(0.22)	(0.20)	(0.33)	(0.14)	(0.22)
Risk Taking		0.54***	0.37	0.59***	0.31*	0.52***	0.32	0.57***	0.30*
		(0.16)	(0.26)	(0.12)	(0.18)	(0.15)	(0.26)	(0.12)	(0.18)
Constant		1.62	-0.55	3.07**	1.89	4.97**	5.17	5.84***	5.36**
		(1.91)	(3.54)	(1.38)	(2.44)	(1.93)	(3.57)	(1.43)	(2.48)
N		1,308	454	2,111	807	1,308	454	2,111	807

Dependent variable: log liquid wealth; Variables of interest: Public/Private Trust (ln)

Countries: Austria, Germany, France, Ireland

Columns: 1-4 Public (H1a) and 5-8 Private (H1b) Trust. The ﬁrst two use only immigrants from Europe while the latter include non- European immigrants. For each speciﬁcation (all) uses the full sample while (<18) uses only immigrants who arrived before age 18. Signiﬁcance at * p<.1, ** p<.05, *** p<.01

ECB Working Paper Series No 2457 / August 2020

Table 9: H1: Trust and Precautionary Motive to Save (Probit): Complete results

		(1)	(2)	(3)	(4)	(5)	(6)	(7)	(8)
		all	<18	all	<18	all	<18	all	<18
	Europe	-0.25	-0.63**
Public Trust		(0.19)	(0.32)
World			-0.24*	-0.23

				(0.14)	(0.23)
	Europe					0.29	-0.49
Private Trust						(0.22)	(0.44)
World							0.42**	0.14

								(0.18)	(0.31)
Germany		-0.28***	-0.15	-0.29***	-0.15	-0.28***	-0.15	-0.29***	-0.14
		(0.06)	(0.10)	(0.05)	(0.09)	(0.06)	(0.10)	(0.05)	(0.09)
France		-0.48***	-0.40*-0.49-0.36-0.48*	-0.39*-0.48-0.35**
		(0.05)	(0.09)	(0.04)	(0.08)	(0.05)	(0.09)	(0.04)	(0.08)
Ireland		-0.25***	-0.14	-0.26***	-0.14	-0.28***	-0.09	-0.27***	-0.15
		(0.05)	(0.09)	(0.04)	(0.09)	(0.05)	(0.11)	(0.04)	(0.09)
Gross Income (ln)		0.01	-0.02	-0.01	-0.02	0.00	-0.02	-0.01	-0.02
		(0.02)	(0.03)	(0.01)	(0.02)	(0.02)	(0.03)	(0.01)	(0.02)
Net Illiquid Wealth (ln)		0.01***	0.02***	0.01***	0.01**	0.01***	0.02***	0.01***	0.01**
		(0.00)	(0.01)	(0.00)	(0.00)	(0.00)	(0.01)	(0.00)	(0.00)
Male		-0.01	-0.01	0.00	0.00	0.00	-0.01	0.00	-0.01
		(0.03)	(0.05)	(0.02)	(0.04)	(0.03)	(0.05)	(0.02)	(0.04)
Age		-0.01	-0.07***	-0.01	-0.06***	-0.01	-0.08***	-0.01*	-0.06***
		(0.01)	(0.03)	(0.01)	(0.02)	(0.01)	(0.03)	(0.01)	(0.02)
Ageˆ2		0.00	0.00**	0.00	0.00***	0.00	0.00***	0.00	0.00***
		(0.00)	(0.00)	(0.00)	(0.00)	(0.00)	(0.00)	(0.00)	(0.00)
Years in country		0.00	0.03**	0.01*	0.03***	0.00	0.03**	0.00	0.03***
		(0.00)	(0.01)	(0.00)	(0.01)	(0.00)	(0.01)	(0.00)	(0.01)
Years in country ˆ2		0.00	0.00	0.00	-0.00**	0.00	0.00	0.00	-0.00**
		(0.00)	(0.00)	(0.00)	(0.00)	(0.00)	(0.00)	(0.00)	(0.00)
Married		-0.06*	-0.09	-0.03	-0.05	-0.05	-0.10*	-0.04	-0.05
		(0.03)	(0.06)	(0.03)	(0.04)	(0.03)	(0.06)	(0.03)	(0.04)
# Children		0.01	0.04	0.00	0.02	0.01	0.03	0.00	0.02
		(0.03)	(0.04)	(0.02)	(0.03)	(0.03)	(0.04)	(0.02)	(0.03)
Secondary Education		-0.06**	-0.04	-0.05**	-0.01	-0.06*	-0.06	-0.04*	-0.01
		(0.03)	(0.06)	(0.02)	(0.04)	(0.03)	(0.06)	(0.02)	(0.04)
Employed		0.03	0.05	0.03	0.05	0.04	0.06	0.04	0.05
		(0.03)	(0.05)	(0.02)	(0.04)	(0.03)	(0.05)	(0.02)	(0.04)
Risk Taking		-0.01	0.02	-0.01	-0.03	-0.02	0.02	-0.02	-0.03
		(0.03)	(0.04)	(0.02)	(0.03)	(0.03)	(0.04)	(0.02)	(0.03)
N		1,308	454	2,111	807	1,308	454	2,111	807

Dependent variable: Dummy is 1 if motive to save for precautionary reasons has highest priority; Variables of interest: Public/Private Trust

Countries: Austria, Germany, France, Ireland

Columns: 1-4 Public (H1a) and 5-8 Private (H1b) Trust. The ﬁrst two use only immigrants from Europe while the latter include non- European immigrants. For each speciﬁcation (all) uses the full sample while (<18) uses only immigrants who arrived before age 18. Estimates are reported as margins at means; Signiﬁcance at * p<.1, ** p<.05, *** p<.01

ECB Working Paper Series No 2457 / August 2020

Table 10: H2a: Trust and Home owner with a Mortgage (Probit): Complete results

		(1)	(2)	(3)	(4)	(5)	(6)	(7)	(8)
		all	<18	all	<18	all	<18	all	<18
	Europe	0.54***	0.22
Public Trust		(0.21)	(0.40)
World			-0.12	0.05

				(0.15)	(0.29)
	Europe					1.35***	0.90
Private Trust						(0.25)	(0.57)
World							1.30***	1.22***

								(0.20)	(0.42)
Germany		0.12***	0.19**	0.11***	0.19***	0.15***	0.22***	0.10***	0.21***
		(0.03)	(0.07)	(0.03)	(0.07)	(0.04)	(0.08)	(0.04)	(0.08)
France		0.18***	0.29***	0.14***	0.25***	0.22***	0.33***	0.18***	0.29***
		(0.04)	(0.07)	(0.03)	(0.05)	(0.05)	(0.08)	(0.03)	(0.06)
Ireland		0.37***	0.55***	0.30***	0.51***	0.30***	0.48***	0.23***	0.41***
		(0.04)	(0.07)	(0.03)	(0.07)	(0.04)	(0.09)	(0.04)	(0.08)
Gross Income (ln)		0.18***	0.21***	0.19***	0.20***	0.17***	0.20***	0.18***	0.20***
		(0.03)	(0.05)	(0.02)	(0.03)	(0.03)	(0.05)	(0.02)	(0.03)
Male		-0.03	0.06	0.00	0.06	-0.02	0.07	0.01	0.07
		(0.03)	(0.07)	(0.03)	(0.05)	(0.04)	(0.07)	(0.03)	(0.05)
Age		0.08***	0.10***	0.08***	0.12***	0.07***	0.10***	0.07***	0.12***
		(0.01)	(0.04)	(0.01)	(0.03)	(0.01)	(0.04)	(0.01)	(0.03)
Ageˆ2		-0.00***	-0.00***	-0.00***	-0.00***	-0.00***	-0.00*-0.00-0.00**
		(0.00)	(0.00)	(0.00)	(0.00)	(0.00)	(0.00)	(0.00)	(0.00)
Years in country		0.02***	0.01	0.02***	0.00	0.02***	0.00	0.02***	0.00
		(0.00)	(0.02)	(0.00)	(0.01)	(0.00)	(0.02)	(0.00)	(0.01)
Years in country ˆ2		-0.00***	0.00	-0.00***	0.00	-0.00***	0.00	-0.00***	0.00
		(0.00)	(0.00)	(0.00)	(0.00)	(0.00)	(0.00)	(0.00)	(0.00)
Married		0.14***	0.13*	0.11***	0.14**	0.16***	0.14*	0.12***	0.14***
		(0.04)	(0.07)	(0.03)	(0.05)	(0.04)	(0.07)	(0.03)	(0.05)
# Children		0.12***	0.13**	0.03*	0.03	0.13***	0.14**	0.04**	0.04
		(0.03)	(0.06)	(0.02)	(0.03)	(0.03)	(0.06)	(0.02)	(0.03)
Secondary Education		-0.14***	-0.13*	-0.08***	-0.04	-0.11***	-0.11	-0.06**	-0.02
		(0.04)	(0.07)	(0.03)	(0.05)	(0.04)	(0.07)	(0.03)	(0.05)
Employed		-0.04	0.09	-0.04	0.06	-0.04	0.08	-0.03	0.06
		(0.04)	(0.07)	(0.03)	(0.05)	(0.04)	(0.07)	(0.03)	(0.05)
Risk Taking		-0.03	-0.06	-0.03	-0.04	-0.04	-0.06	-0.04	-0.04
		(0.03)	(0.06)	(0.02)	(0.04)	(0.03)	(0.06)	(0.02)	(0.04)
N		1,036	338	1,646	562	1,036	338	1,646	562

Dependent variable: Dummy is 1 if household is owner with a mortgage and 0 if renter; Variables of interest: Public/Private Trust

Control variables: log gross income, male, age, years in country, married, number of children, dummy for secondary education, dummy for currently employed, risk taking behavior

Countries: Austria, Germany, France, Ireland

Columns: 1-4 Public and 5-8 Private Trust. The ﬁrst two use only immigrants from Europe while the latter include non-European immigrants. For each speciﬁcation (all) uses the full sample while (<18) uses only immigrants who arrived before age 18.

Estimates are reported as margins at means; Signiﬁcance at * p<.1, ** p<.05, *** p<.01

ECB Working Paper Series No 2457 / August 2020

Table 11: H2b: Trust and LTV Ratio (Tobit): Complete results

		(1)	(2)	(3)	(4)	(5)	(6)	(7)	(8)
		all	<18	all	<18	all	<18	all	<18
	Europe	0.46***	0.25
Public Trust		(0.13)	(0.21)
World			-0.05	0.13

				(0.10)	(0.16)
	Europe					0.82***	0.46
Private Trust						(0.15)	(0.30)
World							0.71***	0.56**

								(0.13)	(0.23)
Germany		0.08*	0.16**	0.08**	0.16***	0.08**	0.16**	0.08**	0.15**
		(0.04)	(0.07)	(0.04)	(0.06)	(0.04)	(0.07)	(0.04)	(0.06)
France		0.09**	0.15**	0.09***	0.17***	0.10**	0.15**	0.11***	0.18***
		(0.04)	(0.06)	(0.03)	(0.05)	(0.04)	(0.06)	(0.03)	(0.05)
Ireland		0.26***	0.39***	0.23***	0.38***	0.21***	0.34***	0.19***	0.32***
		(0.04)	(0.06)	(0.03)	(0.06)	(0.04)	(0.07)	(0.03)	(0.06)
Gross Income (ln)		0.08***	0.09***	0.08***	0.08***	0.07***	0.08***	0.07***	0.07***
		(0.01)	(0.02)	(0.01)	(0.02)	(0.01)	(0.02)	(0.01)	(0.02)
Male		0.00	0.05	0.01	0.05*	0.00	0.06*	0.01	0.05*
		(0.02)	(0.04)	(0.02)	(0.03)	(0.02)	(0.04)	(0.02)	(0.03)
Age		0.05***	0.05**	0.05***	0.06***	0.04***	0.05***	0.05***	0.06***
		(0.01)	(0.02)	(0.01)	(0.01)	(0.01)	(0.02)	(0.01)	(0.01)
Ageˆ2		-0.00***	-0.00***	-0.00***	-0.00***	-0.00***	-0.00*-0.00-0.00**
		(0.00)	(0.00)	(0.00)	(0.00)	(0.00)	(0.00)	(0.00)	(0.00)
Years in country		0.01***	0.01	0.02***	0.01*	0.01***	0.01	0.01***	0.01
		(0.00)	(0.01)	(0.00)	(0.01)	(0.00)	(0.01)	(0.00)	(0.01)
Years in country ˆ2		-0.00***	0.00	-0.00***	0.00	-0.00***	0.00	-0.00***	0.00
		(0.00)	(0.00)	(0.00)	(0.00)	(0.00)	(0.00)	(0.00)	(0.00)
Married		0.06**	0.00	0.05**	0.02	0.07***	0.01	0.05***	0.02
		(0.02)	(0.04)	(0.02)	(0.03)	(0.02)	(0.04)	(0.02)	(0.03)
# Children		0.06***	0.06**	0.02	0.02	0.06***	0.06**	0.02*	0.02
		(0.02)	(0.03)	(0.01)	(0.02)	(0.02)	(0.03)	(0.01)	(0.02)
Secondary Education		-0.09***	-0.10**	-0.06***	-0.02	-0.07***	-0.08*	-0.05***	-0.02
		(0.02)	(0.04)	(0.02)	(0.03)	(0.02)	(0.04)	(0.02)	(0.03)
Employed		-0.01	0.06	-0.02	0.04	-0.01	0.06	-0.01	0.04
		(0.02)	(0.04)	(0.02)	(0.03)	(0.02)	(0.04)	(0.02)	(0.03)
Risk Taking		0.00	0.01	0.00	0.03	0.01	0.02	0.00	0.03
		(0.02)	(0.03)	(0.01)	(0.02)	(0.02)	(0.03)	(0.01)	(0.02)
N		1,036	338	1,646	562	1,036	338	1,646	562

Dependent variable: Leverage ratio: Value of mortgage divided by value of household main residence (both at time of acquisition);

Variables of interest: Public/Private Trust

Control variables: log gross income, male, age, years in country, married, number of children, dummy for secondary education, dummy for currently employed, risk taking behavior

Countries: Austria, Germany, France, Ireland

Columns: 1-4 Public and 5-8 Private Trust. The ﬁrst two use only immigrants from Europe while the latter include non-European immigrants. For each speciﬁcation (all) uses the full sample while (<18) uses only immigrants who arrived before age 18.

Estimates are reported as margins at means; Signiﬁcance at * p<.1, ** p<.05, *** p<.01

ECB Working Paper Series No 2457 / August 2020

9.4 Robustness Regression Results

Table 12: H1: Trust and Liquid Wealth (OLS): Robustness: Native Spouse

		(1)	(2)	(3)	(4)	(5)	(6)	(7)	(8)
		all	<18	all	<18	all	<18	all	<18
	Europe	-0.85	-3.47***
Public Trust		(0.61)	(1.00)
World			-0.81**	-1.76***

				(0.41)	(0.65)
	Europe					2.07***	2.07*
Private Trust						(0.65)	(1.19)
World							1.92***	2.19***

								(0.53)	(0.84)
Spouse		0.74***	0.85**	0.63***	0.67***	0.51**	0.68*	0.52***	0.59**
		(0.22)	(0.37)	(0.16)	(0.25)	(0.23)	(0.38)	(0.16)	(0.25)
Germany		-0.89**	-1.20*	-0.82**	-0.82	-0.92**	-1.25*	-0.81**	-0.82
		(0.41)	(0.64)	(0.32)	(0.56)	(0.41)	(0.65)	(0.32)	(0.55)
France		1.03***	1.38**	0.78***	1.21**	0.95***	1.32**	0.85***	1.27***
		(0.37)	(0.58)	(0.28)	(0.48)	(0.37)	(0.58)	(0.28)	(0.48)
Ireland		-0.95***	-0.58	-0.85***	-0.72	-1.33***	-1.30**	-1.05***	-1.08**
		(0.35)	(0.58)	(0.28)	(0.51)	(0.37)	(0.65)	(0.29)	(0.54)
Gross Income (ln)		0.81***	0.83***	0.72***	0.76***	0.81***	0.87***	0.72***	0.80***
		(0.12)	(0.21)	(0.09)	(0.14)	(0.12)	(0.21)	(0.08)	(0.14)
Net Illiquid Wealth (ln)		0.15***	0.15***	0.15***	0.13***	0.14***	0.14***	0.14***	0.12***
		(0.02)	(0.03)	(0.01)	(0.02)	(0.02)	(0.03)	(0.01)	(0.02)
Male		0.07	0.31	0.06	0.22	0.11	0.32	0.08	0.22
		(0.19)	(0.31)	(0.14)	(0.21)	(0.19)	(0.32)	(0.14)	(0.21)
Age		-0.23***	-0.28*	-0.25***	-0.24**	-0.25***	-0.31*	-0.27***	-0.26**
		(0.07)	(0.16)	(0.05)	(0.11)	(0.07)	(0.16)	(0.05)	(0.11)
Ageˆ2		0.00***	0.00*	0.00***	0.00**	0.00***	0.00**	0.00***	0.00**
		(0.00)	(0.00)	(0.00)	(0.00)	(0.00)	(0.00)	(0.00)	(0.00)
Years in country		-0.05**	0.05	-0.05***	-0.01	-0.06***	0.00	-0.05***	-0.02
		(0.02)	(0.07)	(0.02)	(0.05)	(0.02)	(0.08)	(0.02)	(0.05)
Years in country ˆ2		0.00	0.00	0.00***	0.00	0.00*	0.00	0.00***	0.00
		(0.00)	(0.00)	(0.00)	(0.00)	(0.00)	(0.00)	(0.00)	(0.00)
Married		-0.33	-0.54	-0.18	-0.26	-0.21	-0.50	-0.17	-0.26
		(0.21)	(0.37)	(0.16)	(0.25)	(0.21)	(0.37)	(0.16)	(0.25)
# Children		-0.19	-0.28	-0.14	-0.13	-0.20	-0.38	-0.12	-0.13
		(0.16)	(0.24)	(0.10)	(0.14)	(0.16)	(0.24)	(0.10)	(0.14)
Secondary Education		-0.71***	-0.62*	-0.82***	-0.85***	-0.63***	-0.51	-0.75***	-0.72***
		(0.19)	(0.34)	(0.14)	(0.22)	(0.19)	(0.36)	(0.14)	(0.23)
Employed		0.62***	0.25	0.47***	0.03	0.67***	0.33	0.48***	0.05
		(0.20)	(0.32)	(0.15)	(0.22)	(0.20)	(0.33)	(0.14)	(0.22)
Risk Taking		0.53***	0.37	0.58***	0.31*	0.51***	0.32	0.57***	0.30
		(0.16)	(0.26)	(0.12)	(0.18)	(0.15)	(0.26)	(0.12)	(0.18)
Constant		1.99	0.20	3.55**	2.51	5.01***	5.37	5.98***	5.64**
		(1.90)	(3.55)	(1.38)	(2.44)	(1.93)	(3.57)	(1.43)	(2.47)
N		1,308	454	2,111	807	1,308	454	2,111	807

Dependent variable: log liquid wealth; Variables of interest: Public/Private Trust (log)

Control variables: native spouse, log gross income, log net illiquid wealth, male, age, years in country, married, number of children, dummy for secondary education, dummy for currently employed, risk taking behavior.

Countries: Austria, Germany, France, Ireland

Columns: 1-4 Public (H1a) and 5-8 Private (H1b) Trust. The ﬁrst two use only immigrants from Europe while the latter include non- European immigrants. For each speciﬁcation (all) uses the full sample while (<18) uses only immigrants who arrived before age 18. Signiﬁcance at * p<.1, ** p<.05, *** p<.01

ECB Working Paper Series No 2457 / August 2020

Table 13: H2a: Trust and Home Owner with Mortgage (Probit): Robustness: Native Spouse

		(1)	(2)	(3)	(4)	(5)	(6)	(7)	(8)
		all	<18	all	<18	all	<18	all	<18
	Europe	0.42**	0.21
Public Trust		(0.21)	(0.40)
World			-0.11	0.07

				(0.15)	(0.29)
	Europe					1.15***	0.87
Private Trust						(0.26)	(0.58)
World							1.11***	1.14***

								(0.20)	(0.42)
Spouse		0.15***	0.05	0.17***	0.09	0.11***	0.02	0.14***	0.06
		(0.04)	(0.08)	(0.03)	(0.06)	(0.04)	(0.08)	(0.03)	(0.06)
Germany		0.13***	0.20**	0.12***	0.21***	0.15***	0.22***	0.11***	0.21***
		(0.04)	(0.08)	(0.03)	(0.07)	(0.04)	(0.08)	(0.04)	(0.08)
France		0.18***	0.30***	0.14***	0.25***	0.22***	0.33***	0.17***	0.29***
		(0.04)	(0.07)	(0.03)	(0.05)	(0.05)	(0.08)	(0.03)	(0.06)
Ireland		0.37***	0.55***	0.30***	0.51***	0.31***	0.48***	0.24***	0.42***
		(0.04)	(0.07)	(0.03)	(0.07)	(0.04)	(0.09)	(0.04)	(0.08)
Gross Income (ln)		0.16***	0.20***	0.17***	0.19***	0.15***	0.20***	0.16***	0.20***
		(0.03)	(0.05)	(0.02)	(0.04)	(0.03)	(0.05)	(0.02)	(0.04)
Male		-0.03	0.06	0.00	0.05	-0.02	0.07	0.01	0.06
		(0.04)	(0.07)	(0.03)	(0.05)	(0.04)	(0.07)	(0.03)	(0.05)
Age		0.08***	0.10***	0.08***	0.12***	0.07***	0.10***	0.07***	0.12***
		(0.01)	(0.04)	(0.01)	(0.03)	(0.01)	(0.04)	(0.01)	(0.03)
Ageˆ2		-0.00***	-0.00***	-0.00***	-0.00*-0.00-0.00-0.00*	-0.00***
		(0.00)	(0.00)	(0.00)	(0.00)	(0.00)	(0.00)	(0.00)	(0.00)
Years in country		0.02***	0.01	0.02***	0.00	0.02***	0.00	0.02***	0.00
		(0.00)	(0.02)	(0.00)	(0.01)	(0.00)	(0.02)	(0.00)	(0.01)
Years in country ˆ2		-0.00***	0.00	-0.00***	0.00	-0.00***	0.00	-0.00***	0.00
		(0.00)	(0.00)	(0.00)	(0.00)	(0.00)	(0.00)	(0.00)	(0.00)
Married		0.11***	0.11	0.08***	0.11*	0.14***	0.13*	0.10***	0.12**
		(0.04)	(0.08)	(0.03)	(0.06)	(0.04)	(0.08)	(0.03)	(0.06)
# Children		0.12***	0.14**	0.03*	0.03	0.13***	0.14**	0.03*	0.04
		(0.03)	(0.06)	(0.02)	(0.03)	(0.03)	(0.06)	(0.02)	(0.03)
Secondary Education		-0.13***	-0.13*	-0.08***	-0.04	-0.11***	-0.11	-0.06**	-0.02
		(0.04)	(0.07)	(0.03)	(0.05)	(0.04)	(0.07)	(0.03)	(0.05)
Employed		-0.03	0.09	-0.02	0.07	-0.03	0.08	-0.02	0.07
		(0.04)	(0.07)	(0.03)	(0.05)	(0.04)	(0.07)	(0.03)	(0.05)
Risk Taking		-0.04	-0.06	-0.04	-0.04	-0.04	-0.06	-0.04*	-0.04
		(0.03)	(0.06)	(0.02)	(0.04)	(0.03)	(0.06)	(0.02)	(0.04)
N		1,036	338	1,646	562	1,036	338	1,646	562

Dependent variable: Dummy is 1 if household is owner with a mortgage and 0 if renter; Variables of interest: Public/Private Trust

Control variables: native spouse, log gross income, male, age, years in country, married, number of children, dummy for secondary education, dummy for currently employed, risk taking behavior

Countries: Austria, Germany, France, Ireland

Columns: 1-4 Public and 5-8 Private Trust. The ﬁrst two use only immigrants from Europe while the latter include non-European immigrants. For each speciﬁcation (all) uses the full sample while (<18) uses only immigrants who arrived before age 18.

Estimates are reported as margins at means; Signiﬁcance at * p<.1, ** p<.05, *** p<.01

ECB Working Paper Series No 2457 / August 2020

Table 14: H2b: Trust and LTV Ratio (Tobit): Robustness: Native Spouse

		(1)	(2)	(3)	(4)	(5)	(6)	(7)	(8)
		all	<18	all	<18	all	<18	all	<18
	Europe	0.39***	0.25
Public Trust		(0.13)	(0.21)
World			-0.04	0.14

				(0.09)	(0.16)
	Europe					0.72***	0.46
Private Trust						(0.15)	(0.31)
World							0.60***	0.53**

								(0.13)	(0.23)
Spouse		0.09***	0.01	0.11***	0.04	0.07**	0.00	0.09***	0.03
		(0.03)	(0.04)	(0.02)	(0.03)	(0.03)	(0.05)	(0.02)	(0.03)
Germany		0.09**	0.17**	0.09**	0.17***	0.09**	0.16**	0.08**	0.15**
		(0.04)	(0.07)	(0.04)	(0.06)	(0.04)	(0.07)	(0.04)	(0.06)
France		0.09**	0.15**	0.09***	0.17***	0.10**	0.15**	0.11***	0.18***
		(0.04)	(0.06)	(0.03)	(0.05)	(0.04)	(0.06)	(0.03)	(0.05)
Ireland		0.26***	0.39***	0.23***	0.38***	0.22***	0.34***	0.20***	0.32***
		(0.04)	(0.06)	(0.03)	(0.06)	(0.04)	(0.07)	(0.03)	(0.06)
Gross Income (ln)		0.07***	0.08***	0.07***	0.07***	0.07***	0.08***	0.07***	0.07***
		(0.01)	(0.02)	(0.01)	(0.02)	(0.01)	(0.02)	(0.01)	(0.02)
Male		-0.01	0.05	0.01	0.05	0.00	0.06*	0.01	0.05*
		(0.02)	(0.04)	(0.02)	(0.03)	(0.02)	(0.04)	(0.02)	(0.03)
Age		0.04***	0.05**	0.05***	0.06***	0.04***	0.05***	0.05***	0.06***
		(0.01)	(0.02)	(0.01)	(0.01)	(0.01)	(0.02)	(0.01)	(0.01)
Ageˆ2		-0.00***	-0.00***	-0.00***	-0.00*-0.00-0.00-0.00*	-0.00***
		(0.00)	(0.00)	(0.00)	(0.00)	(0.00)	(0.00)	(0.00)	(0.00)
Years in country		0.01***	0.01	0.01***	0.01	0.01***	0.01	0.01***	0.01
		(0.00)	(0.01)	(0.00)	(0.01)	(0.00)	(0.01)	(0.00)	(0.01)
Years in country ˆ2		-0.00***	0.00	-0.00***	0.00	-0.00***	0.00	-0.00***	0.00
		(0.00)	(0.00)	(0.00)	(0.00)	(0.00)	(0.00)	(0.00)	(0.00)
Married		0.04*	0.00	0.03	0.01	0.05**	0.01	0.04*	0.02
		(0.02)	(0.04)	(0.02)	(0.03)	(0.02)	(0.04)	(0.02)	(0.03)
# Children		0.06***	0.06**	0.02	0.02	0.06***	0.06**	0.02	0.02
		(0.02)	(0.03)	(0.01)	(0.02)	(0.02)	(0.03)	(0.01)	(0.02)
Secondary Education		-0.09***	-0.09**	-0.06***	-0.02	-0.07***	-0.08*	-0.05***	-0.02
		(0.02)	(0.04)	(0.02)	(0.03)	(0.02)	(0.04)	(0.02)	(0.03)
Employed		-0.01	0.06	-0.01	0.04	-0.01	0.06	-0.01	0.04
		(0.02)	(0.04)	(0.02)	(0.03)	(0.02)	(0.04)	(0.02)	(0.03)
Risk Taking		0.00	0.01	0.00	0.03	0.01	0.02	0.00	0.03
		(0.02)	(0.03)	(0.01)	(0.02)	(0.02)	(0.03)	(0.01)	(0.02)
N		1,036	338	1,646	562	1,036	338	1,646	562

Dependent variable: Leverage ratio: Value of mortgage divided by value of household main residence (both at time of acquisition);

Variables of interest: Public/Private Trust

Control variables: native spouse, log gross income, male, age, years in country, married, number of children, dummy for secondary education, dummy for currently employed, risk taking behavior

Countries: Austria, Germany, France, Ireland

Columns: 1-4 Public and 5-8 Private Trust. The ﬁrst two use only immigrants from Europe while the latter include non-European immigrants. For each speciﬁcation (all) uses the full sample while (<18) uses only immigrants who arrived before age 18.

Estimates are reported as margins at means; Signiﬁcance at * p<.1, ** p<.05, *** p<.01

ECB Working Paper Series No 2457 / August 2020

Table 15: H1: Trust and Liquid Wealth (OLS): Robustness: Ordinal Trust Scale

		(1)	(2)	(3)	(4)	(5)	(6)	(7)	(8)
		all	<18	all	<18	all	<18	all	<18
	Europe	-0.23	-1.26***
Public Trust		(0.21)	(0.39)
World			-0.20	-0.54**

				(0.14)	(0.23)
	Europe					0.56*	0.11
Private Trust						(0.28)	(0.44)
World							0.33**	0.35

								(0.16)	(0.26)
Germany		-0.92**	-0.87	-0.86***	-0.77	-0.91**	-1.33**	-0.82**	-0.77
		(0.42)	(0.66)	(0.32)	(0.55)	(0.41)	(0.66)	(0.33)	(0.56)
France		0.89**	1.09*	0.77***	1.35***	0.85**	1.27**	0.71**	1.21**
		(0.37)	(0.58)	(0.28)	(0.49)	(0.37)	(0.59)	(0.28)	(0.49)
Ireland		-0.87**	0.03	-0.76***	-0.29	-0.83**	-0.70	-0.95***	-0.77
		(0.37)	(0.63)	(0.29)	(0.52)	(0.37)	(0.61)	(0.28)	(0.52)
Gross Income (ln)		0.87***	0.94***	0.76***	0.85***	0.87***	0.96***	0.76***	0.86***
		(0.12)	(0.20)	(0.08)	(0.14)	(0.12)	(0.20)	(0.08)	(0.14)
Net Illiquid Wealth (ln)		0.16***	0.15***	0.16***	0.13***	0.16***	0.15***	0.16***	0.13***
		(0.02)	(0.03)	(0.01)	(0.02)	(0.02)	(0.03)	(0.01)	(0.02)
Male		0.07	0.36	0.07	0.27	0.08	0.34	0.08	0.26
		(0.19)	(0.31)	(0.14)	(0.21)	(0.19)	(0.32)	(0.14)	(0.21)
Age		-0.23***	-0.32**	-0.24***	-0.26**	-0.24***	-0.34**	-0.26***	-0.28**
		(0.07)	(0.16)	(0.05)	(0.11)	(0.07)	(0.16)	(0.05)	(0.11)
Ageˆ2		0.00***	0.00**	0.00***	0.00**	0.00***	0.00**	0.00***	0.00**
		(0.00)	(0.00)	(0.00)	(0.00)	(0.00)	(0.00)	(0.00)	(0.00)
Years in country		-0.05**	0.07	-0.04***	0.02	-0.05**	0.04	-0.06***	-0.01
		(0.02)	(0.07)	(0.02)	(0.05)	(0.02)	(0.07)	(0.02)	(0.05)
Years in country ˆ2		0.00	0.00	0.00***	0.00	0.00	0.00	0.00***	0.00
		(0.00)	(0.00)	(0.00)	(0.00)	(0.00)	(0.00)	(0.00)	(0.00)
Married		-0.16	-0.26	-0.05	-0.05	-0.12	-0.29	-0.04	-0.04
		(0.21)	(0.34)	(0.15)	(0.23)	(0.21)	(0.34)	(0.15)	(0.23)
# Children		-0.19	-0.29	-0.12	-0.12	-0.20	-0.37	-0.13	-0.14
		(0.16)	(0.24)	(0.10)	(0.14)	(0.16)	(0.24)	(0.10)	(0.14)
Secondary Education		-0.74***	-0.76**	-0.83***	-0.86***	-0.68***	-0.66*	-0.80***	-0.78***
		(0.19)	(0.35)	(0.14)	(0.23)	(0.19)	(0.36)	(0.14)	(0.23)
Employed		0.61***	0.24	0.42***	-0.01	0.63***	0.29	0.44***	0.03
		(0.20)	(0.33)	(0.15)	(0.22)	(0.20)	(0.33)	(0.15)	(0.22)
Risk Taking		0.55***	0.40	0.59***	0.30*	0.52***	0.31	0.57***	0.28
		(0.16)	(0.26)	(0.12)	(0.18)	(0.16)	(0.26)	(0.12)	(0.18)
Constant		2.02	2.27	3.63***	3.00	2.08	2.83	3.97***	3.49
		(1.82)	(3.40)	(1.35)	(2.39)	(1.81)	(3.44)	(1.35)	(2.39)
N		1,308	454	2,111	807	1,308	454	2,111	807

Dependent variable: log liquid wealth; Variables of interest: Public/Private Trust (dummy is 1 for high trust country)

Countries: Austria, Germany, France, Ireland

Columns: 1-4 Public (H1a) and 5-8 Private (H1b) Trust. The ﬁrst two use only immigrants from Europe while the latter include non- European immigrants. For each speciﬁcation (all) uses the full sample while (<18) uses only immigrants who arrived before age 18. Signiﬁcance at * p<.1, ** p<.05, *** p<.01

ECB Working Paper Series No 2457 / August 2020

Table 16: H2a: Trust and Home Owner with Mortgage (Probit): Robustness: Ordinal Trust Scale

		(1)	(2)	(3)	(4)	(5)	(6)	(7)	(8)
		all	<18	all	<18	all	<18	all	<18
	Europe	0.11***	0.11
Public Trust		(0.04)	(0.08)
World			0.01	-0.05

				(0.03)	(0.05)
	Europe					0.13**	0.05
Private Trust						(0.06)	(0.11)
World							0.24***	0.21***

								(0.03)	(0.06)
Germany		0.10***	0.17**	0.11***	0.19***	0.13***	0.19***	0.16***	0.25***
		(0.04)	(0.08)	(0.03)	(0.07)	(0.03)	(0.07)	(0.04)	(0.08)
France		0.21***	0.33***	0.14***	0.26***	0.16***	0.29***	0.15***	0.28***
		(0.05)	(0.08)	(0.03)	(0.05)	(0.04)	(0.07)	(0.03)	(0.05)
Ireland		0.34***	0.51***	0.30***	0.53***	0.40***	0.56***	0.23***	0.42***
		(0.04)	(0.08)	(0.04)	(0.07)	(0.04)	(0.07)	(0.04)	(0.07)
Gross Income (ln)		0.18***	0.21***	0.19***	0.20***	0.18***	0.20***	0.18***	0.20***
		(0.03)	(0.05)	(0.02)	(0.03)	(0.03)	(0.05)	(0.02)	(0.03)
Male		-0.02	0.06	0.00	0.06	-0.02	0.06	0.01	0.06
		(0.03)	(0.07)	(0.03)	(0.05)	(0.03)	(0.07)	(0.03)	(0.05)
Age		0.07***	0.10***	0.08***	0.12***	0.08***	0.10***	0.07***	0.12***
		(0.01)	(0.04)	(0.01)	(0.03)	(0.01)	(0.04)	(0.01)	(0.03)
Ageˆ2		-0.00***	-0.00***	-0.00***	-0.00***	-0.00***	-0.00*-0.00-0.00**
		(0.00)	(0.00)	(0.00)	(0.00)	(0.00)	(0.00)	(0.00)	(0.00)
Years in country		0.02***	0.01	0.02***	0.01	0.03***	0.01	0.02***	-0.01
		(0.00)	(0.02)	(0.00)	(0.01)	(0.00)	(0.02)	(0.00)	(0.01)
Years in country ˆ2		-0.00***	0.00	-0.00***	0.00	-0.00***	0.00	-0.00***	0.00
		(0.00)	(0.00)	(0.00)	(0.00)	(0.00)	(0.00)	(0.00)	(0.00)
Married		0.15***	0.12*	0.11***	0.14**	0.15***	0.13*	0.13***	0.15***
		(0.04)	(0.07)	(0.03)	(0.05)	(0.04)	(0.07)	(0.03)	(0.06)
# Children		0.12***	0.13**	0.03*	0.04	0.13***	0.14**	0.04**	0.04
		(0.03)	(0.05)	(0.02)	(0.03)	(0.03)	(0.06)	(0.02)	(0.03)
Secondary Education		-0.14***	-0.13*	-0.08***	-0.04	-0.12***	-0.12	-0.07***	-0.02
		(0.04)	(0.07)	(0.03)	(0.05)	(0.04)	(0.08)	(0.03)	(0.05)
Employed		-0.05	0.09	-0.03	0.06	-0.05	0.08	-0.02	0.07
		(0.04)	(0.07)	(0.03)	(0.05)	(0.04)	(0.07)	(0.03)	(0.05)
Risk Taking		-0.04	-0.07	-0.03	-0.04	-0.03	-0.06	-0.04*	-0.04
		(0.03)	(0.06)	(0.02)	(0.04)	(0.03)	(0.06)	(0.02)	(0.04)
N		1,036	338	1,646	562	1,036	338	1,646	562

Dependent variable: Dummy is 1 if household is owner with a mortgage and 0 if renter; Variables of interest: Public/Private Trust (dummy is 1 for high trust country)

Control variables: log gross income, male, age, years in country, married, number of children, dummy for secondary education, dummy for currently employed, risk taking behavior

Countries: Austria, Germany, France, Ireland

Columns: 1-4 Public and 5-8 Private Trust. The ﬁrst two use only immigrants from Europe while the latter include non-European immigrants. For each speciﬁcation (all) uses the full sample while (<18) uses only immigrants who arrived before age 18.

Estimates are reported as margins at means; Signiﬁcance at * p<.1, ** p<.05, *** p<.01

ECB Working Paper Series No 2457 / August 2020

Table 17: H2b: Trust and LTV Ratio (Tobit): Robustness: Ordinal Trust Scale

		(1)	(2)	(3)	(4)	(5)	(6)	(7)	(8)
		all	<18	all	<18	all	<18	all	<18
	Europe	0.10***	0.09**
Public Trust		(0.02)	(0.04)
World			0.03**	0.03

				(0.02)	(0.03)
	Europe					0.07*	0.02
Private Trust						(0.04)	(0.06)
World							0.17***	0.10***

								(0.02)	(0.04)
Germany		0.04	0.13*	0.08**	0.15**	0.10**	0.17**	0.11***	0.18***
		(0.04)	(0.07)	(0.04)	(0.06)	(0.04)	(0.07)	(0.04)	(0.06)
France		0.10**	0.15**	0.08**	0.17***	0.10**	0.15**	0.09***	0.18***
		(0.04)	(0.06)	(0.03)	(0.05)	(0.04)	(0.06)	(0.03)	(0.05)
Ireland		0.22***	0.34***	0.21***	0.36***	0.29***	0.40***	0.19***	0.33***
		(0.04)	(0.07)	(0.03)	(0.06)	(0.04)	(0.06)	(0.03)	(0.06)
Gross Income (ln)		0.08***	0.09***	0.08***	0.08***	0.08***	0.08***	0.07***	0.07***
		(0.01)	(0.02)	(0.01)	(0.02)	(0.01)	(0.02)	(0.01)	(0.02)
Male		0.00	0.05	0.01	0.05*	0.00	0.06	0.01	0.05*
		(0.02)	(0.04)	(0.02)	(0.03)	(0.02)	(0.04)	(0.02)	(0.03)
Age		0.04***	0.05**	0.05***	0.06***	0.05***	0.05***	0.04***	0.06***
		(0.01)	(0.02)	(0.01)	(0.01)	(0.01)	(0.02)	(0.01)	(0.01)
Ageˆ2		-0.00***	-0.00***	-0.00***	-0.00***	-0.00***	-0.00*-0.00-0.00**
		(0.00)	(0.00)	(0.00)	(0.00)	(0.00)	(0.00)	(0.00)	(0.00)
Years in country		0.01***	0.01	0.01***	0.01	0.02***	0.01	0.01***	0.01
		(0.00)	(0.01)	(0.00)	(0.01)	(0.00)	(0.01)	(0.00)	(0.01)
Years in country ˆ2		-0.00***	0.00	-0.00***	0.00	-0.00***	0.00	-0.00***	0.00
		(0.00)	(0.00)	(0.00)	(0.00)	(0.00)	(0.00)	(0.00)	(0.00)
Married		0.06**	0.00	0.05**	0.02	0.06**	0.01	0.06***	0.03
		(0.02)	(0.04)	(0.02)	(0.03)	(0.02)	(0.04)	(0.02)	(0.03)
# Children		0.06***	0.06**	0.02	0.02	0.06***	0.06**	0.02*	0.02
		(0.02)	(0.03)	(0.01)	(0.02)	(0.02)	(0.03)	(0.01)	(0.02)
Secondary Education		-0.09***	-0.09**	-0.06***	-0.02	-0.09***	-0.09**	-0.05***	-0.02
		(0.02)	(0.04)	(0.02)	(0.03)	(0.02)	(0.04)	(0.02)	(0.03)
Employed		-0.01	0.06	-0.01	0.04	-0.02	0.06	-0.01	0.04
		(0.02)	(0.04)	(0.02)	(0.03)	(0.02)	(0.04)	(0.02)	(0.03)
Risk Taking		0.00	0.01	0.00	0.03	0.01	0.02	0.00	0.03
		(0.02)	(0.03)	(0.01)	(0.02)	(0.02)	(0.03)	(0.01)	(0.02)
N		1,036	338	1,646	562	1,036	338	1,646	562

Dependent variable: Leverage ratio: Value of mortgage divided by value of household main residence (both at time of acquisition);

Variables of interest: Public/Private Trust (dummy is 1 for high trust country)

Control variables: log gross income, male, age, years in country, married, number of children, dummy for secondary education, dummy for currently employed, risk taking behavior

Countries: Austria, Germany, France, Ireland

Columns: 1-4 Public and 5-8 Private Trust. The ﬁrst two use only immigrants from Europe while the latter include non-European immigrants. For each speciﬁcation (all) uses the full sample while (<18) uses only immigrants who arrived before age 18.

Estimates are reported as margins at means; Signiﬁcance at * p<.1, ** p<.05, *** p<.01

ECB Working Paper Series No 2457 / August 2020

Table 18: H2b: Trust and LTV Ratio (Tobit): Robustness: Current Values

		(1)	(2)	(3)	(4)	(5)	(6)	(7)	(8)
		all	<18	all	<18	all	<18	all	<18
	Europe	0.37***	0.15
Public Trust		(0.10)	(0.17)
World			-0.02	0.09

				(0.08)	(0.13)
	Europe					0.70***	0.42*
Private Trust						(0.12)	(0.24)
World							0.61***	0.52***

								(0.10)	(0.18)
Germany		0.04	0.11**	0.05*	0.11**	0.04	0.11**	0.04	0.10**
		(0.03)	(0.06)	(0.03)	(0.05)	(0.03)	(0.06)	(0.03)	(0.05)
France		0.05	0.10*	0.06**	0.12***	0.06*	0.10*	0.07***	0.12***
		(0.03)	(0.05)	(0.02)	(0.04)	(0.03)	(0.05)	(0.02)	(0.04)
Ireland		0.18***	0.30***	0.17***	0.29***	0.14***	0.26***	0.13***	0.24***
		(0.03)	(0.05)	(0.03)	(0.05)	(0.03)	(0.06)	(0.03)	(0.05)
Gross Income (ln)		0.07***	0.07***	0.06***	0.05***	0.06***	0.06***	0.06***	0.05***
		(0.01)	(0.02)	(0.01)	(0.01)	(0.01)	(0.02)	(0.01)	(0.01)
Male		-0.01	0.04	0.00	0.04	0.00	0.05*	0.01	0.04*
		(0.02)	(0.03)	(0.01)	(0.02)	(0.02)	(0.03)	(0.01)	(0.02)
Age		0.04***	0.03**	0.04***	0.04***	0.03***	0.03**	0.04***	0.04***
		(0.01)	(0.02)	(0.00)	(0.01)	(0.01)	(0.02)	(0.00)	(0.01)
Ageˆ2		-0.00***	-0.00**	-0.00***	-0.00*-0.00-0.00-0.00**	-0.00***
		(0.00)	(0.00)	(0.00)	(0.00)	(0.00)	(0.00)	(0.00)	(0.00)
Years in country		0.01***	0.01	0.01***	0.01**	0.01***	0.01	0.01***	0.01
		(0.00)	(0.01)	(0.00)	(0.01)	(0.00)	(0.01)	(0.00)	(0.01)
Years in country ˆ2		-0.00***	0.00	-0.00***	-0.00*	-0.00***	0.00	-0.00***	0.00
		(0.00)	(0.00)	(0.00)	(0.00)	(0.00)	(0.00)	(0.00)	(0.00)
Married		0.04**	0.00	0.03**	0.01	0.05***	0.01	0.03**	0.01
		(0.02)	(0.03)	(0.01)	(0.02)	(0.02)	(0.03)	(0.01)	(0.02)
# Children		0.03**	0.04*	0.01	0.01	0.04**	0.04*	0.01	0.01
		(0.02)	(0.02)	(0.01)	(0.02)	(0.02)	(0.02)	(0.01)	(0.02)
Secondary Education		-0.07***	-0.08**	-0.05***	-0.02	-0.06***	-0.07*	-0.03**	-0.01
		(0.02)	(0.03)	(0.01)	(0.02)	(0.02)	(0.03)	(0.01)	(0.02)
Employed		-0.01	0.05	-0.02	0.02	0.00	0.04	-0.01	0.02
		(0.02)	(0.03)	(0.01)	(0.02)	(0.02)	(0.03)	(0.01)	(0.02)
Risk Taking		0.00	0.02	0.00	0.03	0.01	0.03	0.00	0.03*
		(0.01)	(0.02)	(0.01)	(0.02)	(0.01)	(0.02)	(0.01)	(0.02)
N		1,036	338	1,646	562	1,036	338	1,646	562

Dependent variable: Leverage ratio: Value of mortgage divided by value of household main residence (both present values); Vari-

ables of interest: Public/Private Trust

Control variables: log gross income, male, age, years in country, married, number of children, dummy for secondary education, dummy for currently employed, risk taking behavior

Countries: Austria, Germany, France, Ireland

Columns: 1-4 Public and 5-8 Private Trust. The ﬁrst two use only immigrants from Europe while the latter include non-European immigrants. For each speciﬁcation (all) uses the full sample while (<18) uses only immigrants who arrived before age 18. Estimates are reported as margins at means; Signiﬁcance at * p<.1, ** p<.05, *** p<.01

ECB Working Paper Series No 2457 / August 2020

Table 19: H1b: Trust and Liquid Wealth (OLS): Robustness: Civic Indicator

	(1)	(2)	(3)	(4)	(5)	(6)	(7)	(8)
	all	<18	all	<18	all	<18	all	<18
	Europe 1.68***	0.94
Civic Indicator	(0.62)	(1.27)
World		1.97***	2.04**

			(0.46)	(0.82)
Passive	Europe				2.33	5.14
					(2.62)	(4.92)
	World						0.54	1.25
							(1.80)	(3.28)
Active	Europe				1.05	-2.93
					(2.57)	(4.51)
	World						3.11**	2.63
							(1.45)	(2.45)
Germany	-0.91**	-1.33**	-0.77**	-0.86	-0.92**	-1.37**	-0.77**	-0.86
	(0.41)	(0.65)	(0.32)	(0.55)	(0.41)	(0.65)	(0.32)	(0.55)
France	0.90**	1.27**	0.76***	1.20**	0.91**	1.35**	0.72***	1.18**
	(0.37)	(0.59)	(0.28)	(0.49)	(0.37)	(0.59)	(0.28)	(0.49)
Ireland	-1.10***	-0.87	-0.98***	-0.92*	-1.12***	-1.05	-0.95***	-0.89
	(0.36)	(0.62)	(0.28)	(0.53)	(0.37)	(0.67)	(0.29)	(0.55)
Gross Income (ln)	0.86***	0.95***	0.75***	0.85***	0.87***	0.96***	0.75***	0.85***
	(0.12)	(0.21)	(0.08)	(0.14)	(0.12)	(0.21)	(0.08)	(0.14)
Net Illiquid Wealth (ln)	0.15***	0.15***	0.15***	0.13***	0.15***	0.15***	0.15***	0.13***
	(0.02)	(0.03)	(0.01)	(0.02)	(0.02)	(0.03)	(0.01)	(0.02)
Male	0.08	0.34	0.08	0.25	0.08	0.35	0.08	0.25
	(0.19)	(0.32)	(0.14)	(0.21)	(0.19)	(0.32)	(0.14)	(0.21)
Age	-0.25***	-0.34**	-0.27***	-0.29**	-0.26***	-0.34**	-0.26***	-0.29**
	(0.07)	(0.16)	(0.05)	(0.11)	(0.07)	(0.16)	(0.05)	(0.11)
Ageˆ2	0.00***	0.00**	0.00***	0.00**	0.00***	0.00**	0.00***	0.00**
	(0.00)	(0.00)	(0.00)	(0.00)	(0.00)	(0.00)	(0.00)	(0.00)
Years in country	-0.05**	0.03	-0.05***	-0.01	-0.06**	0.01	-0.05***	-0.01
	(0.02)	(0.08)	(0.02)	(0.05)	(0.02)	(0.08)	(0.02)	(0.05)
Years in country ˆ2	0.00*	0.00	0.00***	0.00	0.00*	0.00	0.00***	0.00
	(0.00)	(0.00)	(0.00)	(0.00)	(0.00)	(0.00)	(0.00)	(0.00)
Married	-0.09	-0.28	-0.04	-0.06	-0.09	-0.26	-0.05	-0.07
	(0.21)	(0.34)	(0.15)	(0.23)	(0.21)	(0.35)	(0.15)	(0.23)
# Children	-0.19	-0.37	-0.11	-0.14	-0.20	-0.37	-0.11	-0.14
	(0.16)	(0.24)	(0.10)	(0.14)	(0.16)	(0.24)	(0.10)	(0.14)
Secondary Education	-0.66***	-0.62*	-0.71***	-0.69***	-0.66***	-0.62*	-0.71***	-0.69***
	(0.19)	(0.36)	(0.14)	(0.23)	(0.19)	(0.36)	(0.14)	(0.23)
Employed	0.64***	0.30	0.47***	0.03	0.64***	0.30	0.46***	0.03
	(0.20)	(0.33)	(0.14)	(0.22)	(0.20)	(0.33)	(0.15)	(0.22)
Risk Taking	0.51***	0.31	0.54***	0.27	0.51***	0.30	0.54***	0.27
	(0.16)	(0.26)	(0.12)	(0.18)	(0.16)	(0.26)	(0.12)	(0.18)
Constant	2.39	2.92	3.75***	3.41	2.42	3.14	3.69***	3.39
	(1.80)	(3.43)	(1.34)	(2.39)	(1.81)	(3.44)	(1.34)	(2.39)
N	1,308	454	2,111	807	1,308	454	2,111	807

Dependent variable: log liquid wealth; Variables of interest: Civic Indicator (member in voluntary organizations), Passive (passive

member), Active (active member: does voluntary work)

Countries: Austria, Germany, France, Ireland

Columns: 1-4 binary civic indicator and 5-8 inactive, passive and active. The ﬁrst two use only immigrants from Europe while the latter include non-European immigrants. For each speciﬁcation (all) uses the full sample while (<18) uses only immigrants who arrived before age 18.

Signiﬁcance at * p<.1, ** p<.05, *** p<.01

ECB Working Paper Series No 2457 / August 2020

Table 20: H2a: Trust and Home Owner with Mortgage (Probit): Robustness: Civic Indicator

	(1)	(2)	(3)	(4)	(5)	(6)	(7)	(8)
	all	<18	all	<18	all	<18	all	<18
	Europe 0.60***	0.37
Civic Indicator	(0.12)	(0.27)
World		0.46***	0.56***

			(0.09)	(0.20)
Passive	Europe				2.05***	2.57**
					(0.47)	(1.04)
	World						2.17***	2.25***
							(0.34)	(0.78)
Active	Europe				-0.87*	-1.73*
					(0.47)	(1.00)
	World						-0.94***	-0.78
							(0.28)	(0.62)
Germany	0.15***	0.21***	0.13***	0.22***	0.15***	0.21**	0.13***	0.22***
	(0.04)	(0.08)	(0.04)	(0.08)	(0.04)	(0.08)	(0.04)	(0.08)
France	0.19***	0.31***	0.15***	0.27***	0.24***	0.38***	0.20***	0.31***
	(0.04)	(0.07)	(0.03)	(0.06)	(0.05)	(0.08)	(0.03)	(0.06)
Ireland	0.34***	0.51***	0.26***	0.44***	0.31***	0.45***	0.22***	0.37***
	(0.04)	(0.08)	(0.03)	(0.07)	(0.04)	(0.09)	(0.03)	(0.08)
Gross Income (ln)	0.16***	0.20***	0.18***	0.20***	0.17***	0.21***	0.18***	0.20***
	(0.03)	(0.05)	(0.02)	(0.03)	(0.03)	(0.05)	(0.02)	(0.03)
Male	-0.02	0.07	0.01	0.06	-0.02	0.07	0.01	0.06
	(0.04)	(0.07)	(0.03)	(0.05)	(0.04)	(0.07)	(0.03)	(0.05)
Age	0.07***	0.10***	0.07***	0.12***	0.07***	0.10***	0.07***	0.11***
	(0.01)	(0.04)	(0.01)	(0.03)	(0.01)	(0.04)	(0.01)	(0.03)
Ageˆ2	-0.00***	-0.00*-0.00-0.00-0.00*	-0.00**	-0.00***	-0.00***
	(0.00)	(0.00)	(0.00)	(0.00)	(0.00)	(0.00)	(0.00)	(0.00)
Years in country	0.03***	0.01	0.02***	0.00	0.02***	0.00	0.02***	0.00
	(0.00)	(0.02)	(0.00)	(0.01)	(0.00)	(0.02)	(0.00)	(0.01)
Years in country ˆ2	-0.00***	0.00	-0.00***	0.00	-0.00***	0.00	-0.00***	0.00
	(0.00)	(0.00)	(0.00)	(0.00)	(0.00)	(0.00)	(0.00)	(0.00)
Married	0.17***	0.13*	0.12***	0.13**	0.17***	0.14*	0.13***	0.14***
	(0.04)	(0.07)	(0.03)	(0.05)	(0.04)	(0.07)	(0.03)	(0.06)
# Children	0.13***	0.13**	0.03**	0.04	0.12***	0.14**	0.03*	0.04
	(0.03)	(0.06)	(0.02)	(0.03)	(0.03)	(0.05)	(0.02)	(0.03)
Secondary Education	-0.11***	-0.12	-0.06**	-0.02	-0.12***	-0.12*	-0.06**	-0.02
	(0.04)	(0.07)	(0.03)	(0.05)	(0.04)	(0.07)	(0.03)	(0.05)
Employed	-0.04	0.08	-0.03	0.07	-0.04	0.09	-0.03	0.07
	(0.04)	(0.07)	(0.03)	(0.05)	(0.04)	(0.07)	(0.03)	(0.05)
Risk Taking	-0.04	-0.06	-0.04*	-0.04	-0.04	-0.07	-0.04*	-0.04
	(0.03)	(0.06)	(0.02)	(0.04)	(0.03)	(0.06)	(0.02)	(0.04)
N	1,036	338	1,646	562	1,036	338	1,646	562

Dependent variable: Dummy is 1 if household is owner with a mortgage and 0 if renter; Variables of interest: Civic Indicator (member

in voluntary organizations), Passive (passive member), Active (active member: does voluntary work)

Control variables: log gross income, male, age, years in country, married, number of children, dummy for secondary education, dummy for currently employed, risk taking behavior

Countries: Austria, Germany, France, Ireland

Estimates are reported as margins at means; Signiﬁcance at * p<.1, ** p<.05, *** p<.01

ECB Working Paper Series No 2457 / August 2020

Table 21: H2b: Trust and LTV Ratio (Tobit): Robustness: Civic Indicator

	(1)	(2)	(3)	(4)	(5)	(6)	(7)	(8)
	all	<18	all	<18	all	<18	all	<18
	Europe 0.37***	0.25*
Civic Indicator	(0.07)	(0.14)
World		0.27***	0.29***

			(0.06)	(0.11)
Passive	Europe				1.25***	0.77
					(0.29)	(0.55)
	World						1.38***	0.97**
							(0.22)	(0.41)
Active	Europe				-0.51*	-0.24
					(0.29)	(0.52)
	World						-0.64***	-0.24
							(0.18)	(0.33)
Germany	0.10**	0.17**	0.10***	0.16***	0.10**	0.16**	0.09***	0.15**
	(0.04)	(0.07)	(0.04)	(0.06)	(0.04)	(0.07)	(0.04)	(0.06)
France	0.10**	0.14**	0.10***	0.17***	0.12***	0.15**	0.13***	0.18***
	(0.04)	(0.06)	(0.03)	(0.05)	(0.04)	(0.06)	(0.03)	(0.05)
Ireland	0.25***	0.36***	0.21***	0.33***	0.23***	0.33***	0.19***	0.30***
	(0.04)	(0.07)	(0.03)	(0.06)	(0.04)	(0.07)	(0.03)	(0.06)
Gross Income (ln)	0.07***	0.08***	0.08***	0.07***	0.07***	0.08***	0.08***	0.07***
	(0.01)	(0.02)	(0.01)	(0.02)	(0.01)	(0.02)	(0.01)	(0.02)
Male	0.00	0.06	0.01	0.05*	-0.01	0.06	0.01	0.05*
	(0.02)	(0.04)	(0.02)	(0.03)	(0.02)	(0.04)	(0.02)	(0.03)
Age	0.04***	0.05***	0.05***	0.06***	0.04***	0.05***	0.05***	0.06***
	(0.01)	(0.02)	(0.01)	(0.01)	(0.01)	(0.02)	(0.01)	(0.01)
Ageˆ2	-0.00***	-0.00*-0.00-0.00-0.00*	-0.00*-0.00-0.00**
	(0.00)	(0.00)	(0.00)	(0.00)	(0.00)	(0.00)	(0.00)	(0.00)
Years in country	0.01***	0.01	0.01***	0.01	0.01***	0.01	0.01***	0.01
	(0.00)	(0.01)	(0.00)	(0.01)	(0.00)	(0.01)	(0.00)	(0.01)
Years in country ˆ2	-0.00***	0.00	-0.00***	0.00	-0.00***	0.00	-0.00***	0.00
	(0.00)	(0.00)	(0.00)	(0.00)	(0.00)	(0.00)	(0.00)	(0.00)
Married	0.07***	0.01	0.05***	0.02	0.07***	0.01	0.05***	0.02
	(0.02)	(0.04)	(0.02)	(0.03)	(0.02)	(0.04)	(0.02)	(0.03)
# Children	0.06***	0.06**	0.02*	0.02	0.06***	0.06**	0.02	0.02
	(0.02)	(0.03)	(0.01)	(0.02)	(0.02)	(0.03)	(0.01)	(0.02)
Secondary Education	-0.07***	-0.08**	-0.05***	-0.01	-0.07***	-0.08**	-0.05***	-0.01
	(0.02)	(0.04)	(0.02)	(0.03)	(0.02)	(0.04)	(0.02)	(0.03)
Employed	-0.01	0.06	-0.01	0.04	0.00	0.06	-0.01	0.04
	(0.02)	(0.04)	(0.02)	(0.03)	(0.02)	(0.04)	(0.02)	(0.03)
Risk Taking	0.00	0.02	0.00	0.03	0.00	0.02	0.00	0.03
	(0.02)	(0.03)	(0.01)	(0.02)	(0.02)	(0.03)	(0.01)	(0.02)
N	1,036	338	1,646	562	1,036	338	1,646	562

Dependent variable: Leverage ratio: Value of mortgage divided by value of household main residence (both at time of acquisition);

Variables of interest: Civic Indicator (member in voluntary organizations), Passive (passive member), Active (active member: does voluntary work)

Control variables: log gross income, male, age, years in country, married, number of children, dummy for secondary education, dummy for currently employed, risk taking behavior

Countries: Austria, Germany, France, Ireland

Estimates are reported as margins at means; Signiﬁcance at * p<.1, ** p<.05, *** p<.01

ECB Working Paper Series No 2457 / August 2020

Table 22: H1: Trust and Liquid Wealth (OLS): Robustness: Uninsurable Risk

		(1)	(2)	(3)	(4)	(5)	(6)	(7)	(8)
		all	<18	all	<18	all	<18	all	<18
	Europe	-0.76	-3.38***
Public Trust		(0.60)	(1.02)
World			-0.78*	-1.77***

				(0.41)	(0.66)
	Europe					2.26***	2.23*
Private Trust						(0.64)	(1.19)
World							2.17***	2.39***

								(0.53)	(0.84)
Germany		-1.00**	-1.26*	-0.90***	-0.92	-1.00**	-1.23*	-0.87***	-0.89
		(0.42)	(0.67)	(0.33)	(0.58)	(0.41)	(0.67)	(0.33)	(0.57)
France		0.89**	1.22**	0.70**	1.08**	0.85**	1.19**	0.78***	1.16**
		(0.37)	(0.58)	(0.28)	(0.49)	(0.37)	(0.59)	(0.28)	(0.49)
Ireland		-1.01***	-0.60	-0.91***	-0.77	-1.42***	-1.35**	-1.13***	-1.15**
		(0.36)	(0.59)	(0.28)	(0.52)	(0.37)	(0.66)	(0.29)	(0.54)
Gross Income (ln)		0.85***	0.81***	0.72***	0.77***	0.84***	0.83***	0.71***	0.81***
		(0.12)	(0.21)	(0.09)	(0.14)	(0.12)	(0.21)	(0.09)	(0.14)
Illiquid Assets (ln)		0.16***	0.14***	0.15***	0.13***	0.14***	0.13***	0.14***	0.12***
		(0.02)	(0.03)	(0.01)	(0.02)	(0.02)	(0.03)	(0.01)	(0.02)
Male		0.07	0.52	0.03	0.30	0.10	0.49	0.06	0.28
		(0.20)	(0.33)	(0.14)	(0.22)	(0.20)	(0.33)	(0.14)	(0.22)
Age		-0.25***	-0.36**	-0.26***	-0.29**	-0.26***	-0.38**	-0.27***	-0.30***
		(0.07)	(0.16)	(0.05)	(0.11)	(0.07)	(0.16)	(0.05)	(0.11)
Ageˆ2		0.00***	0.00**	0.00***	0.00**	0.00***	0.00**	0.00***	0.00***
		(0.00)	(0.00)	(0.00)	(0.00)	(0.00)	(0.00)	(0.00)	(0.00)
years in country		-0.04*	0.08	-0.04**	0.02	-0.05**	0.03	-0.05***	0.00
		(0.02)	(0.08)	(0.02)	(0.05)	(0.02)	(0.08)	(0.02)	(0.05)
years in country ˆ2		0.00	0.00	0.00**	0.00	0.00*	0.00	0.00***	0.00
		(0.00)	(0.00)	(0.00)	(0.00)	(0.00)	(0.00)	(0.00)	(0.00)
Married		-0.10	-0.06	0.04	0.08	-0.04	-0.06	0.02	0.05
		(0.21)	(0.35)	(0.15)	(0.24)	(0.21)	(0.35)	(0.15)	(0.24)
# Children		-0.18	-0.29	-0.12	-0.13	-0.20	-0.38	-0.10	-0.13
		(0.16)	(0.24)	(0.10)	(0.14)	(0.16)	(0.24)	(0.10)	(0.14)
Secondary Education		-0.60***	-0.37	-0.71***	-0.66***	-0.55***	-0.26	-0.65***	-0.53**
		(0.20)	(0.36)	(0.15)	(0.24)	(0.20)	(0.37)	(0.14)	(0.24)
Employed		0.72	1.77	0.46	0.75	0.77	2.06	0.53	0.74
		(0.87)	(1.27)	(0.64)	(0.90)	(0.87)	(1.28)	(0.64)	(0.91)
Risk Taking		0.51***	0.37	0.56***	0.29	0.50***	0.31	0.55***	0.28
		(0.16)	(0.26)	(0.12)	(0.18)	(0.16)	(0.26)	(0.12)	(0.18)
Self-Employed		0.87**	1.07*	0.72***	0.80**	0.79**	1.10**	0.68***	0.79**
		(0.34)	(0.55)	(0.24)	(0.36)	(0.34)	(0.55)	(0.24)	(0.36)
Constant		1.67	1.01	3.49**	2.81	4.82**	6.33*	6.11***	6.13**
		(1.92)	(3.62)	(1.40)	(2.49)	(1.95)	(3.66)	(1.44)	(2.53)
Sector FE		X	X	X	X	X	X	X	X
N		1,308	454	2,111	807	1,308	454	2,111	807

Dependent variable: log liquid wealth; Variables of interest: Public/Private Trust (ln)

Control variables: log gross income, log net illiquid wealth, male, age, years in country, married, number of children, dummy for secondary education, dummy for currently employed, risk taking behavior, dummy for self-employed, dummies for sector of occupation (suppressed due to possible conﬁdential issues)

Countries: Austria, Germany, France, Ireland

Signiﬁcance at * p<.1, ** p<.05, *** p<.01

ECB Working Paper Series No 2457 / August 2020

Table 23: H2a: Trust and Home Owner with Mortgage (Probit): Robustness: Uninsurable Risk

		(1)	(2)	(3)	(4)	(5)	(6)	(7)	(8)
		all	<18	all	<18	all	<18	all	<18
	Europe	0.55***	0.31
Public Trust		(0.21)	(0.36)
World			-0.06	0.03

				(0.15)	(0.27)
	Europe					1.32***	0.85
Private Trust						(0.25)	(0.54)
World							1.29***	1.24***

								(0.20)	(0.42)
Germany		0.15***	0.21***	0.12***	0.23***	0.17***	0.23***	0.11***	0.24***
		(0.04)	(0.08)	(0.04)	(0.08)	(0.04)	(0.09)	(0.04)	(0.08)
France		0.16***	0.26***	0.12***	0.22***	0.19***	0.29***	0.16***	0.26***
		(0.04)	(0.07)	(0.03)	(0.05)	(0.05)	(0.08)	(0.03)	(0.06)
Ireland		0.37***	0.54***	0.30***	0.51***	0.30***	0.47***	0.23***	0.41***
		(0.04)	(0.09)	(0.03)	(0.07)	(0.04)	(0.10)	(0.04)	(0.08)
Gross Income (ln)		0.17***	0.17***	0.17***	0.17***	0.16***	0.17***	0.17***	0.17***
		(0.03)	(0.05)	(0.02)	(0.04)	(0.03)	(0.05)	(0.02)	(0.04)
Male		-0.02	0.02	0.00	0.03	-0.01	0.03	0.01	0.04
		(0.04)	(0.06)	(0.03)	(0.05)	(0.04)	(0.06)	(0.03)	(0.05)
Age		0.07***	0.09**	0.08***	0.11***	0.07***	0.10***	0.07***	0.11***
		(0.01)	(0.04)	(0.01)	(0.03)	(0.01)	(0.04)	(0.01)	(0.03)
Ageˆ2		-0.00***	-0.00**	-0.00***	-0.00***	-0.00***	-0.00*-0.00-0.00**
		(0.00)	(0.00)	(0.00)	(0.00)	(0.00)	(0.00)	(0.00)	(0.00)
Years in country		0.02***	0.00	0.02***	0.00	0.02***	0.00	0.02***	0.00
		(0.00)	(0.02)	(0.00)	(0.01)	(0.00)	(0.02)	(0.00)	(0.01)
Years in country ˆ2		-0.00***	0.00	-0.00***	0.00	-0.00***	0.00	-0.00***	0.00
		(0.00)	(0.00)	(0.00)	(0.00)	(0.00)	(0.00)	(0.00)	(0.00)
Married		0.13***	0.07	0.11***	0.10*	0.15***	0.08	0.12***	0.10*
		(0.04)	(0.07)	(0.03)	(0.05)	(0.04)	(0.07)	(0.03)	(0.05)
# Children		0.12***	0.13**	0.04**	0.04	0.13***	0.13**	0.04**	0.05
		(0.03)	(0.05)	(0.02)	(0.03)	(0.03)	(0.05)	(0.02)	(0.03)
Secondary Education		-0.13***	-0.15**	-0.07**	-0.07	-0.11***	-0.13*	-0.05**	-0.06
		(0.04)	(0.07)	(0.03)	(0.05)	(0.04)	(0.07)	(0.03)	(0.05)
Employed		0.23*	0.52***	0.15	0.40**	0.20	0.52***	0.15	0.37**
		(0.13)	(0.16)	(0.11)	(0.18)	(0.13)	(0.16)	(0.11)	(0.19)
Risk Taking		-0.04	-0.05	-0.03	-0.04	-0.04	-0.04	-0.04*	-0.04
		(0.03)	(0.05)	(0.02)	(0.04)	(0.03)	(0.05)	(0.02)	(0.04)
Self-Employed		0.23***	0.21*	0.21***	0.23***	0.21***	0.20*	0.20***	0.23**
		(0.06)	(0.11)	(0.05)	(0.09)	(0.06)	(0.11)	(0.05)	(0.09)
Sector FE		X	X	X	X	X	X	X	X
N		1,030	330	1,635	559	1,030	330	1,635	559

Dependent variable: Dummy is 1 if household is owner with a mortgage and 0 if renter; Variables of interest: Public/Private Trust

Control variables: log gross income, male, age, years in country, married, number of children, dummy for secondary education, dummy for currently employed, risk taking behavior, dummy for self-employed, dummies for sector of occupation (suppressed due to possible conﬁdential issues)

Countries: Austria, Germany, France, Ireland

Columns: 1-4 Public and 5-8 Private Trust. The ﬁrst two use only immigrants from Europe while the latter include non-European immigrants. For each speciﬁcation (all) uses the full sample while (<18) uses only immigrants who arrived before age 18. Estimates are reported as margins at means; Signiﬁcance at * p<.1, ** p<.05, *** p<.01

ECB Working Paper Series No 2457 / August 2020

Table 24: H2b: Trust and LTV Ratio (Tobit): Robustness: Uninsurable Risk

		(1)	(2)	(3)	(4)	(5)	(6)	(7)	(8)
		all	<18	all	<18	all	<18	all	<18
	Europe	0.45***	0.27
Public Trust		(0.13)	(0.21)
World			-0.01	0.12

				(0.10)	(0.15)
	Europe					0.75***	0.48
Private Trust						(0.15)	(0.30)
World							0.66***	0.59***

								(0.13)	(0.23)
Germany		0.09**	0.16**	0.09**	0.18***	0.09**	0.16**	0.08**	0.16***
		(0.04)	(0.07)	(0.04)	(0.06)	(0.04)	(0.07)	(0.04)	(0.06)
France		0.07*	0.15**	0.08***	0.17***	0.08**	0.15**	0.10***	0.17***
		(0.04)	(0.06)	(0.03)	(0.05)	(0.04)	(0.06)	(0.03)	(0.05)
Ireland		0.26***	0.39***	0.23***	0.38***	0.21***	0.34***	0.20***	0.32***
		(0.04)	(0.06)	(0.03)	(0.06)	(0.04)	(0.07)	(0.03)	(0.06)
Gross Income (ln)		0.08***	0.09***	0.07***	0.06***	0.07***	0.08***	0.07***	0.06***
		(0.01)	(0.03)	(0.01)	(0.02)	(0.01)	(0.02)	(0.01)	(0.02)
Male		0.00	0.02	0.00	0.03	0.00	0.03	0.01	0.03
		(0.02)	(0.04)	(0.02)	(0.03)	(0.02)	(0.04)	(0.02)	(0.03)
Age		0.04***	0.05***	0.05***	0.06***	0.04***	0.06***	0.05***	0.06***
		(0.01)	(0.02)	(0.01)	(0.01)	(0.01)	(0.02)	(0.01)	(0.01)
Ageˆ2		-0.00***	-0.00***	-0.00***	-0.00***	-0.00***	-0.00*-0.00-0.00**
		(0.00)	(0.00)	(0.00)	(0.00)	(0.00)	(0.00)	(0.00)	(0.00)
Years in country		0.01***	0.00	0.01***	0.01	0.01***	0.00	0.01***	0.01
		(0.00)	(0.01)	(0.00)	(0.01)	(0.00)	(0.01)	(0.00)	(0.01)
Years in country ˆ2		-0.00***	0.00	-0.00***	0.00	-0.00***	0.00	-0.00***	0.00
		(0.00)	(0.00)	(0.00)	(0.00)	(0.00)	(0.00)	(0.00)	(0.00)
Married		0.06**	-0.02	0.05**	0.00	0.06***	-0.01	0.05***	0.00
		(0.02)	(0.04)	(0.02)	(0.03)	(0.02)	(0.04)	(0.02)	(0.03)
# Children		0.06***	0.06**	0.02*	0.02	0.07***	0.07**	0.02*	0.03
		(0.02)	(0.03)	(0.01)	(0.02)	(0.02)	(0.03)	(0.01)	(0.02)
Secondary Education		-0.08***	-0.11***	-0.05**	-0.05	-0.07***	-0.10**	-0.04**	-0.04
		(0.02)	(0.04)	(0.02)	(0.03)	(0.02)	(0.04)	(0.02)	(0.03)
Employed		0.19**	0.22	0.11	0.18*	0.16*	0.17	0.11	0.15
		(0.09)	(0.14)	(0.08)	(0.11)	(0.09)	(0.14)	(0.08)	(0.11)
Risk Taking		0.00	0.02	0.00	0.03	0.01	0.02	0.00	0.03
		(0.02)	(0.03)	(0.01)	(0.02)	(0.02)	(0.03)	(0.01)	(0.02)
Self-Employed		0.12***	0.04	0.13***	0.12**	0.10**	0.04	0.12***	0.11**
		(0.04)	(0.07)	(0.03)	(0.05)	(0.04)	(0.07)	(0.03)	(0.05)
Sector FE		X	X	X	X	X	X	X	X
N		1,036	338	1,646	562	1,036	338	1,646	562

Dependent variable: Leverage ratio: Value of mortgage divided by value of household main residence (both at time of acquisition);

Variables of interest: Public/Private Trust

Countries: Austria, Germany, France, Ireland

Columns: 1-4 Public and 5-8 Private Trust. The ﬁrst two use only immigrants from Europe while the latter include non-European immigrants. For each speciﬁcation (all) uses the full sample while (<18) uses only immigrants who arrived before age 18.

Estimates are reported as margins at means; Signiﬁcance at * p<.1, ** p<.05, *** p<.01

ECB Working Paper Series No 2457 / August 2020

2020 August / 2457 No Series Paper Working ECB

9.5 Sample Summary Statistics (unweighted)

						AT								DE								FR								IE
1	Reference	Assets / Income				2014Q4 / 2013							2014Q3 / 2013							2014Q4 / 2014						2013Q2 / 2012Q1-2013Q3

				Natives	Non-Natives							Natives	Non-Natives							Natives	Non-Natives							Natives	Non-Natives
						I	II	III	IV	V				I	II	III	IV	V				I	II	III	IV	V				I	II	III	IV	V
						DE	TR	PL	HR	OAS				PL	RU	TR	KZK	RO				NAF	PT	OAS	IT	DE				GB	PL	OAS	LT	USA
2	N		2,686	240	60	25	20	15	14	4,016	411	68	42	33	29	25	10,511	1,234	593	113	106	75	57	4,319	978	406	149	102	47	39

3	Married		47	54	40	72	50	73	50	65	67	63	64	85	x	60	55	62	61	72	68	64	63	53	54	54	53	75	62	62
4	hh size		2.0	2.2	1.9	3.0	1.8	2.5	2.3	2.3	2.5	2.2	2.2	3.2	3.2	2.0	2.4	2.5	2.6	2.5	2.6	2.0	2.2	2.7	2.7	2.7	2.7	2.9	2.9	2.9
			(1.2)	(1.3)	(1.1)	(1.7)	(1.0)	(1.2)	(1.4)	(1.1)	(1.3)	(1.1)	(1.0)	(1.3)	(1.2)	(0.7)	(1.3)	(1.4)	(1.6)	(1.2)	(1.3)	(1.0)	(1.1)	(1.5)	(1.4)	(1.5)	(1.1)	(1.3)	(1.2)	(1.5)
5	Age		53.0	51.6	51.9	43.7	58.2	50.3	43.0	56.2	51.2	51.8	52.7	42.7	42.9	56.5	54.3	57.2	57.9	57.3	52.7	65.7	59.2	49.4	40.1	47.9	32.8	35.3	32.7	42.1
		At immigration			22.6	21.8	16.0	29.5	21.4	29.6			21.1	21.2	32.2	14.7	24.6	26.5			18.3	17.3	16.0	18.5	16.9	14.0			22.9	19.1	25.8	26.2	25.0	21.5
		Years in country			28.9	30.1	27.7	28.8	28.9	13.4			30.1	30.6	20.5	28.1	18.3	30.0			38.8	40.6	41.1	33.9	48.8	45.3			17.3	28.8	7.1	9.1	7.7	20.5
6	Education	Secondary (or less)	83	74	70	100	80	93	36	67	70	84	71	91	48	52	57	55	57	83	46	67	40	65	51	59	62	19	55	23
		Tertiary	17	26	30	0	20	7	64	33	30	16	29	9	52	48	43	45	43	17	54	33	60	35	49	41	38	81	45	77

7	Employment	Employed	45	40	38	52	20	53	57	48	54	50	63	52	86	44	50	43	40	45	57	31	49	43	59	46	68	75	68	41
		Unemployed	3	10	8	20	15	20	7	3	5	x	x	18	x	0	4	7	9	3	4	3	2	11	16	15	23	7	21	15
		Self-employed	6	7	7	4	5	7	14	9	8	12	x	x	x	x	10	11	9	14	13	5	9	12	9	12	3	8	2	21
		Retired	40	35	38	16	55	20	7	36	27	32	26	9	x	44	32	34	36	35	25	59	39	20	8	17	1	1	0	10
		Other	6	8	8	8	5	0	14	4	6	x	x	x	x	x	4	5	6	4	2	3	2	14	9	10	6	10	9	13
8	Income		42.4	39.4	42.7	42.1	35.6	37.2	37.6	72.8	57.0	45.9	44.3	48.5	49.6	46.7	71.0	71.5	58.8	53.4	63.6	117.0	92.3	60.7	51.3	59.9	36.0	60.1	36.3	71.1
			(30.5)	(29.5)	(32.9)	(37.2)	(28.2)	(23.8)	(19.8)	(93.0)	(73.3)	(43.2)	(41.9)	(29.8)	(39.2)	(25.6)	(118.1)	(185.0)	(83.3)	(63.5)	(92.9)	(520.0)	(90.9)	(86.0)	(57.3)	(74.4)	(17.5)	(47.2)	(18.5)	(75.0)

9	Real Assets	HMR	297.6	460.0	279.1	180.0	373.3	130.0	280.0	332.5	265.7	294.6	188.7	213.6	176.4	225.0	298.7	381.8	375.1	268.1	435.8	283.6	578.7	207.2	210.4	208.4	199.4	212.4	113.0	281.0
		Other	45.9	29.2	43.2	60.0	25.4	5.1	0.0	70.9	12.7	4.3	0.0	x	x	x	119.6	70.2	63.2	37.1	39.5	231.5	57.1	84.0	24.2	40.9	0.7	2.6	0.1	21.0
10	Financial Assets	Liquid
		Risky	5.9	2.8	5.9	0.4	2.0	1.4	0.6	23.3	6.4	x	0.0	x	x	x	59.5	64.6	21.8	5.0	36.1	70.7	124.0	7.6	6.8	9.0	0.0	17.0	0.0	18.9
		Non-risky	26.8	18.6	17.0	13.0	13.5	9.8	13.4	60.4	27.9	18.4	9.9	4.9	9.6	19.5	39.8	40.7	31.5	17.2	29.6	33.5	54.7	26.1	16.5	23.2	5.0	8.5	5.5	38.4
		Illiquid	5.8	4.7	5.9	10.4	0.6	4.0	3.7	154.3	85.6	80.6	16.5	31.8	63.5	85.0	115.0	95.6	76.8	24.2	117.7	51.4	271.2	17.2	14.5	25.3	2.0	1.0	0.1	37.6
11	Liabilities	Mortgage	88.6	116.0	130.5	105.0	144.5	?	138.3	147.0	133.4	95.0	134.5	103.0	135.3	117.6	167.1	174.6	165.7	116.1	179.1	243.1	135.7	209.7	205.5	179.0	163.4	247.1	122.7	270.7
		Other	2.8	0.7	0.0	0.0	0.5	?	0.0	4.6	3.5	x	8.8	8.4	x	x	10.0	6.4	8.7	8.6	3.1	3.7	8.1	7.9	6.5	5.1	1.9	13.0	0.9	28.1
12	Housing	Owner (outright)	14	9	12	4	10	0	14	30	23	16	21	24	31	12	23	18	14	27	23	7	18	37	25	44	5	13	13	31
		Owner (mortgage)	38	21	34	4	20	7	0	34	21	18	17	9	10	36	52	44	37	45	46	69	53	42	17	32	1	8	2	23
		Renter	48	70	53	92	70	93	86	35	56	66	62	67	59	52	25	38	49	27	31	24	30	21	58	25	95	79	85	46

Assets: AT, DE, IE ﬁeldwork midpoint, FR 31.12.2014 / Income: IE last twelve months; 2: number of hhs; 3: %; 4: persons per hh; 5: years; 6, 7: %; 8: Income from all sources, thousand Euros; 9: "Other" excludes business wealth, thousand Euros; 10: thousand Euros; 11: %; 12: thousand Euros Standard deviations in parentheses; GB excludes Northern Ireland; NAF (North Africa) comprises: Algeria, Egypt, Libya, Morocco, Tunisia; OAS (Other Asia) comprises: China, Taiwan, Hong Kong, South Korea, Japan, India, Pakistan, Thailand, Malaysia, Singapore, Philippines
DE: The Bundesbank's data and conﬁdentiality protection provision requires the following adjustments: x suppressed; AT: ? missing data; summary statistics of the HFCS immigrant sample before applying sample selection criteria (see section 4.3)

Table 25: Households in the unrestricted HFCS immigrant sample

Acknowledgements

We are deeply indebted to the members of the European Central Bank's Household Finance and Consumption Network (HFCN), in particular Jirka Slacacek, Oreste Tristani and Philip Vermeulen. We also thank an anonymous referee, Pascal Achard, Miguel Ampudia, Axelle Ferriere, Nicola Fuchs-Schündeln, Luigi Guiso, Ana Lamo, Ramon Marimon, Stefan Nagel, Fabiano Schivardi and Kjetil Storesletten for helpful comments and Juha Honkkila for answering numerous questions on country specific details of the HFCS data. Peter Lindner, Reamonn Lydon, Frédérique Savignac and the Bundesbank's RDSC-Team provided valuable support in obtaining access to the country of origin variable of the Austrian, Irish, French and German HFCS datasets. We also benefited from comments of audiences at the ECB, the EUI, the HFCN meeting in March 2020 as well as the SEM 2019 and CEF 2020 conferences. All errors are our own. Fleck did some of the work on this paper while he was visiting the economics department at the University of Oslo. Financial support of Oslo Fiscal Studies (OFS) is gratefully acknowledged.

Discussion Papers represent the authors' personal opinion and do not necessarily reflect the view of the Deutsche Bundesbank or the Eurosystem.

Johannes Fleck

European University Institute, Fiesole, Italy; email: johannes.fleck@eui.eu

Adrian Monninger

Deutsche Bundesbank, Frankfurt am Main, Germany; email: adrian.monninger@bundesbank.de

Postal address 60640 Frankfurt am Main, Germany

Telephone +49 69 1344 0

Website www.ecb.europa.eu

All rights reserved. Any reproduction, publication and reprint in the form of a different publication, whether printed or produced electronically, in whole or in part, is permitted only with the explicit written authorisation of the ECB or the authors.

This paper can be downloaded without charge from www.ecb.europa.eu, from the Social Science Research Network electronic library or from RePEc: Research Papers in Economics. Information on all of the papers published in the ECB Working Paper Series can be found on the ECB's website.

PDF

ISBN 978-92-899-4374-1

ISSN 1725-2806

doi:10.2866/988878

QB-AR-20-109-EN-N

Attachments

Original document
Permalink

Disclaimer

ECB - European Central Bank published this content on 18 August 2020 and is solely responsible for the information contained therein. Distributed by Public, unedited and unaltered, on 18 August 2020 09:10:10 UTC

VODAFONE : Berenberg remains Neutral	08:31pm	ZD
Boeing commercial jet deliveries drop by two in April	08:31pm	RE
BOEING REPORTS APRIL DELIVERIES OF FOUR 787 DREAMLINER JETS…	08:30pm	RE
Vodafone gets Spanish green light for its $5.41 billion deal with Zegona	08:28pm	RE
Brenntag shares fall as chemicals price pressures hit forecast	08:23pm	RE
Britain rejects call to ban gagging orders in finance	08:23pm	RE
Eurozone's Weak Productivity Risks Higher Inflation, ECB's Schnabel Says	08:22pm	DJ
Eight killed, eight critically injured in Florida bus crash - local media	08:22pm	RE
Bouygues: old modems transformed into new boxes	08:20pm	CF
India's Dalmia Bharat Sugar fourth-quarter profit falls on weak sales	08:19pm	RE
BHP CEO Says Still Confident in Merits of Rejected Anglo American Takeover Proposal	08:19pm	DJ
VEOLIA : Q1 24: Portfolio refocusing continues while water and waste lead earnings	08:18pm
SONOVA HOLDING : In-line operational performance; markets focus on planned product launches	08:18pm
Citadel's Griffin wants to know Trump's running mate before deciding to donate	08:18pm	RE
Exclusive-German panel of economic advisors see growth of 0.2% for 2024, says source	08:17pm	RE

Stock Market News in real time

Culture and portfolios: trust, precautionary savings and home ownership

All News: More news

Headlines

Most Read News