Socially responsible investments among parents and adult children

European Economic Review 121 (2020) 103328

Contents lists available at ScienceDirect

European Economic Review journal homepage: www.elsevier.com/locate/euroecorev

Socially responsible investments among parents and adult children ✩ Jörgen Hellström, Nicha Lapanan∗, Rickard Olsson Umeå School of Business, Economics, and Statistics, Umeå University, Biblioteksgränd 6, 90187 Umeå, Sweden

a r t i c l e

i n f o

Article history: Received 5 June 2018 Accepted 8 October 2019 Available online 19 October 2019 Keywords: Social preferences Individual investor behavior Sustainable Investing Socially responsible mutual funds

a b s t r a c t Novel evidence is provided of a positive correlation between parents’ and their children’s socially responsible mutual fund investment behavior. Although captured parent-child correlations reflect contemporary relationships, they reveal potentially important insight into the origin of heterogeneity in individuals’ prosocial behavior. Consistent with research on socialization, the results suggest an influence from both parents, stronger for mothers, and reinforced for parents both investing in socially responsible mutual funds. Parental resources during an individual’s adolescence (financial and parental life experience) are further found to significantly explain individuals’ adult prosocial investment behavior. The results are robust to conditioning on a number of alternative explanations. © 2019 Elsevier B.V. All rights reserved.

1. Introduction The rising interest in socially responsible investment practices around the world (see, e.g., Renneboog et al., 2008) indicates a growing attention towards integrating social, environmental and ethical consideration in investment decisions. Although mounting evidence in economic research indicates that individuals, in general, are driven by concern for others in their decision-making, less is known about heterogeneity between individuals in these social preferences (see, e.g., Meier, 2007).1 To understand why some investors care about non-financial characteristics of investments and others do not, the reasons for this heterogeneity need to be examined. Many values, attitudes, and preferences are potentially formed through socialization efforts by parents, transmitting these over generations (e.g., Grusec and Hastings, 2007). While research, for example Dohmen et al. (2012), has documented parental-child correlations in measures of risk and trust attitudes, central for financial decision making, less is known about the parental transmission of prosocial behavior.2 ✩ Umeå School of Business, Economics, and Statistics, Umeå University. Financial support from the Wallander, Browald and Tom Hedelius Foundation is gratefully acknowledged. We would also like to thank the Swedish Investment Fund Association (Fondbolagens Förening) for providing the data on mutual funds. We thank Lisa Kramer, Lu Li, Bert Scholtens, Oscar Stålnacke, Vladimir Vanyushyn, and Maximilian Wimmer, as well as the participants at the Oikos Young Scholars Finance Academy (2015), the 13th International Paris Finance Meeting (2015) and the 30th Annual Congress of the European Economic Association for their useful comments and suggestions. ∗ Corresponding author. E-mail addresses: [email protected] (J. Hellström), [email protected] (N. Lapanan). 1 Actually, economic research usually treats individuals’ attitudes and value endowments as given or derived from a black box (Dohmen et al., 2012). Research on cultural transmission (e.g., Bisin and Verdier, 20 0 0) does, however, endogenize these by assuming either a parental or other role model transmission of attitudes. 2 Intergenerational transmission of attitudes, personality, and other personality traits between parents and children has been suggested as an explanation for observed correlations between behavior and outcomes across generations by, for example, Bowles and Gintis (2002), Groves (2005), and Heckman et al. (2006).

https://doi.org/10.1016/j.euroecorev.2019.103328 0014-2921/© 2019 Elsevier B.V. All rights reserved.

2

J. Hellström, N. Lapanan and R. Olsson / European Economic Review 121 (2020) 103328

In the current paper, we provide empirical evidence indicating that an individual’s propensity to invest in socially responsible mutual funds is significantly positively correlated with having parents owning these in the previous period. Our results indicate that ownership among both mothers and fathers has an influence, although this influence is stronger in the case of mothers, and that having two parents owning (rather than one) reinforces the impact on an individual’s likelihood of investing in a socially responsible manner. Parental resources (economic and life experience) during an individual’s adolescence are further found to have an impact on the strength of the individual-parental correlations. Individuals growing up with parents with greater economic resources and parental life experience during the individual’s adolescence are found, all else equal, to have a stronger correlation in their ownership of socially responsible mutual funds with that of their parents. The results are interesting, especially in light of recent research. Riedl and Smeets (2017), using administrative investor data linked to internet-based experiments, find that investors with stronger prosocial preferences are more likely to hold socially responsible funds and that expectations about returns and risks play a less important role in the decision to invest in a socially responsible manner.3 Thus, although our results capture one-year lagged individual-parental relationships in ownership of socially responsible mutual funds, they potentially reveal important insights into the origin of heterogeneity in individuals’ prosocial behavior. That both parents have an influence is consistent with research on socialization, indicating a similar importance of parents in the socialization process (e.g., Grusec and Hastings, 2007). The result, that mothers have a greater influence, further complements the findings of Dohmen et al. (2012), who find evidence indicating that mothers are more important in the transmission of trust attitudes, while both parents are of equal importance in the transmission of risk attitudes. Interestingly, our results also indicate that parental similarities in social responsible investment behavior are of importance for the strength of the parental effect on individuals’ prosocial behavior. Having two parents owning socially responsible mutual funds, rather than one, reinforces the impact on an individual’s likelihood to invest in a socially responsible manner. This is interesting and potentially associated with an assortative mating mechanism (see, e.g., Dohmen et al., 2012). Assortative mating implies, in the current context, that if one parent display prosocial behavior, then the other parent is more likely to also display a prosocial behavior, i.e., parents match on similarities in behavior to avoid distortions in the transmission of behavior to the child (assuming parents have a preference for children with similar behavior as their own). The finding, that the effect of having two parents investing in a socially responsible manner strengthens the impact on the individual’s likelihood to invest, thus provides evidence consistent with that assortative mating reinforces the transmission of prosocial behavior. In terms of heterogeneous parental impact, our results further indicate that parental resources during the individual’s adolescence are of importance for the transmission of prosocial behavior. This is consistent with research on socialization, indicating that family resources, processes and relationships are better predictors of successful socialization than family structure itself (Patterson and Hastings, 2007). Parental ownership of socially responsible mutual funds for parents observed with a relatively higher total income during the individual’s adolescence are, all else equal, found to be more strongly correlated with subsequent ownership among individuals. This tentatively indicates that parental financial resources during an individual’s youth have an influence on the transference of prosocial behavior. That our results further indicate that the correlation in parent-individual ownership of socially responsible funds strengthens with higher parental age during the individual’s adolescence emphasizes the importance of parental life experience. Comparing the effects on individuals with two employed parents (compared with one), i.e., with restrictions in parent-child time, do, however, reveal no significant differences in individuals’ prosocial investment behavior. The analysis is performed using a large sample of Swedish household investors based on detailed data derived from administrative records. The data pertain to two full cohorts of individuals (those born in 1963 and in 1973), as well as to their parents, observed annually over the period 1999–2007.4 The data, obtained from Statistics Sweden (SCB), include detailed information about individuals’ financial holdings (e.g., mutual fund holdings), taxable income and other wealth (e.g., bank holdings and real estate) derived from the Swedish tax authorities, as well as a large number of individual socioeconomic characteristics. In addition, parental variables, e.g., income, pertaining to the individual’s adolescence are also included in the data. To support the interpretation of results as reflecting intergenerational transmission of prosocial behavior, we highlight that results have been obtained while controlling for a number of alternative mechanisms that potentially could explain the observed individual-parental correlations in holdings of socially responsible mutual funds. First, individual-parental correlations may be driven by individuals inheriting a number of other characteristics from parents. Dohmen et al. (2012), for example, document intergenerational correlations in risk and trust attitudes, while Black et al. (2009), Björklund et al. (2010), Anger (2011) and Grönqvist et al. (2017) find correlations concerning cognitive ability. Individual-parental similarities in risk attitudes and cognitive ability do then potentially explain similarities in prosocial investment behavior.5 To control for this

3 Research on socially responsible investments indicates that individuals’ motives may either be profit-driven (based on expectations of a superior risk-return trade-off for socially responsible mutual funds) or value-driven (that individuals derive utility from doing good), or both (see, e.g., Fama and French, 2007; Nilsson, 2008; Riedl and Smeets, 2017). 4 The choice of the cohorts was made to increase the likelihood that individuals have parents alive during our observational period. 5 For example, since cognitive ability is a strong predictor of education, and education is a potentially important institution in the socialization process (see, e.g., Grusec and Hastings, 2007), similarities in parents’ and children’s prosocial behavior may be driven by similar educational influences.

J. Hellström, N. Lapanan and R. Olsson / European Economic Review 121 (2020) 103328

3

our models include individuals’ risky asset shares as a proxy for risk preferences and individuals’ educational attainment and income (both highly correlated with cognitive ability) as controls for cognitive ability. Second, individuals and parents may share the same financial advisor or different financial advisors from the same bank. Correlation in socially responsible mutual fund holdings may then be driven by similar mutual fund recommendations.6 To control for this we have included a “same advisor” dummy in our regressions. The dummy is based on information about which mutual funds that belong to specific mutual fund companies. Given that the mutual fund market in Sweden is dominated by mutual fund companies belonging to the four largest banks in Sweden, in which a majority of Swedish mutual fund investors are customers, we can, using information about individuals’ fund holdings, construct proxies indicating which main advisors (i.e., fund companies and banks) individuals and parents are using during our study period. Third, individual-parental correlations in socially responsible mutual fund holdings may be driven by an exposure to a common community environment. Duflo and Saez (2002), Hong et al. (2004), Brown et al. (2008) and Kaustia and Knüpfer (2012) are some among the studies documenting evidence on the importance of social interaction in individuals’ financial decision-making. To control for potential effects driven by individuals’ and parents’ common exposure to a similar community influence, we include in our analysis both a “big city” dummy, capturing whether individuals reside in one of the three largest cities in Sweden, as well as the proportion of socially responsible investors in an individual’s community. Fourth, individual-parental correlations in socially responsible mutual fund holdings may be driven by a common individual-parental exposure to a similar information environment, e.g., including advertisements about socially responsible investment opportunities, or be driven by within-family sharing of information. To lower the impact of these potential alternative explanations, we lag the parental socially responsible mutual fund ownership variable one period to avoid capturing potential contemporaneous reactions to the same information set and include the parents’ lagged socially responsible fund portfolio returns as a proxy for within-family sharing of information.7 Finally, since intergenerational transmission of prosocial behavior has been found to be, at least partly, driven by genetic inheritance (e.g., Cesarini et al., 2009), we provide an additional analysis on a smaller sample composed of adopted individuals. Interestingly, results from re-estimations using this sample indicate parental effects in line with those for our full sample and lend support for an interpretation of transmission of prosocial behavior through the process of socialization. Besides including a vast number of controls in our main analysis, we also challenge the results in a number of ways, e.g., instead using logit regression models, adding controls for unobserved heterogeneity, and instead using individuals’ and parents’ proportion of socially responsible mutual funds out of their total mutual fund portfolio as the dependent variable. Reassuringly, results and conclusions are throughout unchanged qualitatively.8 The study is, to the authors’ knowledge, the first to study intergenerational correlation in the choice of socially responsible mutual funds. Our results are interesting and contribute to the understanding of why some individuals integrate social concern into their investment decisions. The results are of interest not only for the financial industry, but also for the literature profiling socially responsible individual investors (e.g., Nilsson, 2008; Junkus and Berry, 2010; Riedl and Smeets, 2017; Hellström et al., 2018) and for the more general literature studying correlations in values, attitudes, and behavioral outcomes between parents and children (e.g., Dohmen et al., 2012 (risk and trust attitudes); Black et al., 2009; Björklund et al., 2010; Anger, 2011; Grönqvist et al., 2017 (cognitive and non-cognitive skills); Björklund and Salvanes, 2011; Black and Devereux, 2011; and Holmlund et al., 2011 (income and schooling)). The findings further contribute, at a more general level, to the literature on the formation of social preferences, see for example, Wilhelm et al. (2008), Cesarini et al. (2009), BenNer et al. (2017), and Kosse et al. (2018). The rest of the paper is organized as follows: in Section 2, we discuss the parental influence on individuals’ prosocial behavior from a socialization perspective, in Section 3 the data, control variables and summary statistics, in Section 4 our results as well as robustness tests, while Section 5 provide conclusions. 2. Parental influence on prosocial behavior An individual’s attitudes, values, and preferences are formed from childhood to adulthood through a socialization process.9 The process involves acquiring the ability to adaptively function in a larger social context, including the acceptance of societal values, standards, and customs. Key agents in this process are family members, peers, educational institutions, as well as media. The role of parents is central and many consider parents as the primary agents of socialization (see e.g., Collins et al., 20 0 0). This follows since parents constitute members of the first social group an individual comes in contact 6

We thank two anonymous referees for directing our attention towards this mechanism. To further study the potential impact of similarities in individuals’ and parents’ information environment on prosocial behavior, we have in an additional analysis also estimated models capturing correlations between individuals’ holdings of socially responsible mutual funds in 2007 in relation to parents’ holdings in 1999. 8 Estimation results based on using logit regression models gives marginal effects for our variables of interest (i.e., the lagged parental socially responsible ownership variables) of similar sizes, signs, and significances, as those using linear probability models. Adding random effects to control for unobserved heterogeneity do affect the size of the marginal effects (becomes lower), but not the signs and significances. 9 Research on the formation of prosocial preferences also suggest a role of “nature”. In terms of heritability of genes, studies on twin data suggest that prosocial preferences, at least partly, is heritable (Wallace et al., 2007; Cesarini et al., 2008; Cesarini et al., 2009). Cesarini et al. (2009) do, for example, document a significant genetic effect on giving behavior. In Section 4.6, we elaborate more on the issue of “nature” versus “nurture”, when analyzing a sample of adopted individuals. 7

4

J. Hellström, N. Lapanan and R. Olsson / European Economic Review 121 (2020) 103328

with and since many individual characteristics are set in early ages.10 This is highlighted in Cunha and Heckman (2007), where parental background and investments during individuals’ childhood are identified as the primary drivers of personality formation. Research on the transmission of generosity (e.g., Wilhelm et al., 2008; Ben-Ner et al., 2017) suggest that a main mechanism through which parents install prosocial behavior is through role modelling. Ben-Ner et al. (2017) develop a theoretical framework predicting parental modelling of generosity to children, where child utility is lowered by deviations from parental norms, i.e., where children have a preference for mimicking parental behavior. Investing in socially responsible mutual funds means investing in assets managed with a focus, not only on returns, but also on environmental, social, and governance impacts. Given that it broadly focuses on “doing good” with regard to, for example, nature and society (i.e., others), we hypothesize that parental role modelling, installing prosocial values among individuals, may be a main mechanism through which individuals’ likelihood to invest in socially responsible mutual funds is affected. Research in sociology and psychology has studied the impact of family structure on the outcome of socialization (see e.g., Patterson and Hastings, 2007). Connected to differences in family structures are also differences in access to economic, social, and community resources. For example, two parents together (compared with one) generally have access to greater economic resources, more extensive social support networks, as well as provide important support and encouragement for one another. A consistent finding in this literature is that family resources, processes and relationships are better predictors of successful socialization than the family structure itself (Patterson and Hastings, 2007). While the parental role in the transmission of values, attitudes, and preferences, in general, is heterogeneous, we focus in the current paper on identification of a number of distinct aspects of the parental transmission process of prosocial values related to individuals’ socialization process.11 First, we investigate whether both mothers and fathers are of importance in the transmission of prosocial preferences. Given that mothers have been found to be of more importance than fathers in transmission of trust, but mothers and fathers of equal importance for transmission of risk attitudes (Dohmen et al., 2012), we further focus on this issue comparing mothers and fathers relative influence. Second, transmission of prosocial values may be stronger with two parents with similar, rather than dissimilar, values. This may be motivated by that the “socialization signal” from two parents may be stronger or less distorted for two agreeing parents. Evidence of this is given in Dohmen et al. (2012), who find that parents who are more similar in terms of risk attitudes are found to have a larger impact on the risk attitudes of the child. The reinforced signal from value agreeing parents is consistent with the positive assortative mating mechanism assumed by Bisin and Verdier (20 0 0) in their seminal model of cultural transmission. This mechanism posits that individuals (here parents) mate based on similarities in attitudes to minimize the distortion in the socialization process transmitting these to the child. In the analysis we provide evidence on this issue and examine whether parents with similar prosocial behavior reinforce the impact on individuals’ likelihood to invest prosocially. Third, as noted above, parental resources are of key concern for successful socialization of children (Patterson and Hastings, 2007). Research indicate that parents with fewer economic resources, compared to those more well off, show poorer socialization practices and less authoritative parenting styles (Magnuson and Duncan, 2002). They have been found to be less confident in their parenting, less warm and engaged with their children, and more verbally and physically punitive than parents with greater economic resources (Patterson and Hastings, 2007). Individual’s growing up in socioeconomically disadvantaged families are also less likely to live in homes with access to many books, less likely to have access to music and art, and less likely to visit libraries and museums (Bradley, 2002). Further, they are also less likely to participate in social activities, such as taking music or dance lessons, to participate in clubs, or to play in sports teams (Fields et al., 2001). Related to our study, DeGarmo et al. (1999) find that socioeconomic status among divorced mothers is positively correlated with better parenting practices, in turn predicting children’s adaptive and prosocial behavior at school. To what extent parents with prosocial values and relatively more economic resources during individuals’ adolescents are more successful in transmitting these values to children is examined in the empirical analysis. In terms of other parental resources, we investigate to what extent parental life experiences (proxied by the age of the parents during individuals’ adolescent) and parent-child time matters for the transmission of prosocial values. Parents with more life experience during the socialization process can, on average, be assumed to be better equipped in socializing children. For example, younger mothers tend to be less verbal, less sensitive, and less responsive to their infants than older mothers (Culp et al., 1988) and are also likely to have less previous experience with children. Further, children of relatively younger mothers have been found to perform more poorly on measures of cognitive competence, e.g., language based assessments, and to score lower on achievement tests (Moore et al., 1997; Luster et al., 20 0 0). Regarding parental-child time, we examine this from the perspective of to what extent parents were employed during individuals’ adolescents. In regard to the later, early research in sociology and psychology focused mainly on whether children would suffer with a father as the primary caregiver, compared with the traditional division of labor involving an employed father and a mother at home. Other researchers worried about that child care outside the family, in families with two employed parents, would be damaging to children. Evidence from this research did, however, dispel these concerns (Patterson and Hastings, 2007).

10 Carneiro and Heckman (2004), Heckman and Masterov (2007), and Cunha et al. (2006) do, for example, indicate that an individual’s cognitive and non-cognitive abilities are shaped in early age and that parents play an important role in the production of these. 11 Apart from the considered aspects, other important elements affecting the socialization process include the quality of the parent-child interaction, the emotional climate and stability of families, as well as availability of social support for parents (Patterson and Hastings, 2007).

J. Hellström, N. Lapanan and R. Olsson / European Economic Review 121 (2020) 103328

5

In the empirical analysis we provide evidence on to what extent parental transmission of prosocial values depends on both parental life experience and parental employment status during individuals’ adolescents. 3. Data and summary statistics 3.1. Data and sample selection The data used in this study come from three different sources. First, individual investors’ demographic and socioeconomic characteristics for the full cohorts of 1963 and 1973 and their parents are obtained from Statistics Sweden.12 Second, individual investors’ detailed wealth information including financial wealth, real estate and taxable income is obtained from the Swedish tax authorities. As Sweden abolished the wealth tax after 2007, these data are only available up to 2007. Financial wealth is categorized into investments in funds and stocks. In this study, we observe individual investors’ mutual fund holdings (number of fund units per International Securities Identification Number [ISIN]) on December 31st of each year during the period 1999–2007. Third, data on mutual funds, such as ISIN, fund name and daily net asset values (NAVs) was obtained from the Swedish Investment Fund Association (SIFA). Most of the mutual funds marketed in Sweden are members of SIFA (over 3400 funds were registered with SIFA as of 2010). We manually classify individual investors’ mutual fund holdings as socially responsible by using information from several different sources. First, mutual funds are categorized as socially responsible according to a list of ethical funds from the Swedish Pension Agency (PPM by Swedish acronym) and Swedbank (one of the largest Swedish banks). Second, we use a list of ethical funds in Sweden during the period 20 02–20 08 from Folksam (a Swedish insurance company that is one of Sweden’s largest investment managers). Apart from this matching, we screen the names of the mutual funds owned by individual investors using the following words: “sustainable”, “ethical”, “socially responsible”, “SRI”, “social” and “green”. In total, 218 funds of the available 3400 mutual funds are classified as socially responsible funds. It is important to note that this classification may suffer from look-back bias as the oldest report available is from 2002, but our data dates back to 1999. The data for each individual includes demographic characteristics such as birth year, gender, marital status, children or no children, level of education and municipality of residence. We use salary from employment as the main income source and aggregate wealth after liabilities as net wealth. Investment portfolio characteristics are calculated on the basis of reported mutual fund holdings at the end of each year. As we want to control for individual investors’ choice of funds potentially driven by a profit motive, we create a portfolio return variable using the fund portfolio weights observed on December 31st of each year and the funds’ returns during the recent year. We also create a variable for the number of funds in individual investors’ total fund portfolios as a control for the chances of selecting socially responsible funds. In our database, there are a total of 117,638 and 125,367 individuals for the 1963 and 1973 cohorts, respectively, observed during the period 1999–2007. To determine our sample, we first select individual investors with at least one mutual fund holding. A total of 69,022 and 71,654 individual investors, or around 60% of the two cohorts, invest in mutual funds. Second, since we want to observe both mothers’ and fathers’ influence on individual investors’ likelihood of holding socially responsible mutual funds, we restrict our sample to individual investors with both of their parents holding at least one mutual fund. Therefore, the final sample consists of 21,573 and 29,729 individual investors born 1963 and 1973, respectively. These numbers represent 31.26% and 41.49% of mutual fund investors from each cohort. The sample selection is summarized in Table 1, Panel A. The selection of individuals to only those owning, and observed with both parents owning, mutual funds has been made to focus on the individuals’ choice of type of mutual fund, i.e., the choice of either a conventional or a socially responsible mutual fund portfolio. Although the selection restricts the total sample, it allows us to separately analyze this choice independently of individuals’ and parents’ choices of whether or not to own mutual funds. A drawback of selecting only individual investors with both of their parents investing in mutual funds is that it may cause concerns regarding a potential selection bias. However, we assume that the participation decision (in the mutual fund market) and choice of fund type are independent (conditional on observables). This is motivated since the factors that influence parents’ and individuals’ decisions to invest in mutual funds do not necessarily influence them to invest in a socially responsible manner. 3.2. Individuals’ and parents’ holdings of socially responsible mutual funds Throughout the paper, we define socially responsible investors as individuals who have an investment in at least one socially responsible mutual fund. The proportion of identified socially responsible individual investors are presented in Table I, Panel B. Overall, this proportion seems quite stable over the years.13 For the individuals, 7.14% of all, on average, hold socially responsible mutual funds. When comparing the parents, mothers tend to invest more in socially responsible mutual 12 Since 1990, Statistics Sweden has been collecting data on all individuals aged 16 and above who are registered in Sweden on December 31st of each year. More information on the Longitudinal Integration Database for Health Insurance and Labor Market Studies (LISA by Swedish acronym) can be found on Statistics Sweden’s website www.scb.se. 13 While statistics from Eurosif show growing net asset values (NAV) in the SRI market every year, the number of individual investors in the market does not appear to increase significantly; the retail SRI market decreased from 8% in 2010 to 6% in 2012 (PI, 2012).

6

J. Hellström, N. Lapanan and R. Olsson / European Economic Review 121 (2020) 103328

Table 1 Sample selection and individuals’ holdings of socially responsible mutual funds. Panel A: Sample selection. Born 1963

Population Individuals invest in mutual funds Parents invest in mutual funds (percent out of population) Parents invest in mutual funds (percent out of mutual fund investors) Both parents invest in mutual funds (percent out of population) Both parents invest in mutual funds (percent out of mutual fund investors)

Born 1973

Total

Observation

Percent

Observation

Percent

Observation

Percent

117,638 69,022 44,038 44,038 21,573 21,573

58.67% 37.44% 63.80% 18.34% 31.26%

125,367 71,654 51,174 51,174 29,729 29,729

57.16% 40.82% 71.42% 23.71% 41.49%

243,005 140,676 95,212 95,212 51,302 51,302

57.89% 39.18% 67.68% 21.11% 36.47%

Panel B: Proportion of individuals holding socially responsible mutual funds on an annual basis.

Sustainable investors Father invest in sustainable funds. Mother invest in sustainable funds. One of the parents invest in sustainable funds. Both parents invest in sustainable funds. Number of observations

1999

20 0 0

2001

2002

2003

2004

2005

2006

2007

Total

6.74% 7.65% 8.36% 12.90% 3.11% 27,487

6.62% 7.58% 8.48% 13.07% 2.99% 35,337

7.12% 7.84% 8.69% 13.53% 3.01% 36,102

6.81% 7.32% 8.20% 12.84% 2.68% 35,738

6.84% 7.25% 8.17% 12.77% 2.65% 34,958

7.56% 8.37% 8.81% 13.96% 3.23% 33,843

7.38% 7.90% 8.40% 13.23% 3.07% 33,030

7.39% 7.66% 8.06% 12.79% 2.91% 31,644

7.87% 7.72% 8.18% 12.99% 2.91% 29,812

7.14% 7.70% 8.38% 13.13% 2.95% 297,951

Panel A reports the number of individuals in the sample after each sample selection step. First, we select individuals born 1963 and 1973 that have an investment in at least one mutual fund. Second, we keep individuals if both of their parents invest in mutual funds in the same period. Panel B reports the proportion of individuals holding socially responsible mutual fund each year. Table 2 Individuals’s and parents’ mutual fund holdings. Individual’s mutual fund investment

Conventional investor Socially responsible investor

Parent socially responsible mutual funds investment None

Mother

Father

Both Parents

Total

0.82 0.05

0.05 0.01

0.04 0.01

0.02 0.01

0.93 0.07

The table reports the cross tabulation of individuals’ and parents’ mutual fund holdings.

funds than fathers, on average, 8.38% versus 7.70%. This result is in line with previous findings that females tend to invest in socially responsible assets more than males (see, e.g., Hellström et al., 2018). Moreover, the average percentage of socially responsible individual investors with at least one parent holding socially responsible mutual funds is 13.13%. On average, only 2.95% of individuals have both of their parents investing in a socially responsible fund. From Table 2, the cross-tabulation reveals that individuals and parents without an investment in a socially responsible mutual fund is the largest group, contributing to 82% of the observations. The second-largest group is the one consisting of conventional investors whose parents have an investment in a socially responsible mutual fund, in total representing about 11% of the observations (5% for only mother, 4% for only father, and 2% for both). The group of socially responsible investors with parents having no investment in a socially responsible mutual fund contributes to 5% of the observations. In contrast, the group of socially responsible investors with socially responsible parents constitutes 3% of the observations. In Table 3, Panel A, parental characteristics are broken down by conventional and socially responsible investors (variable definitions are given in Appendix, Table A1). Focusing first on our main variables of interest, the figures indicate that the proportion of individuals with at least one parent holding a socially responsible mutual fund within the group of socially responsible investors is much higher than for the group of conventional investors. On average, 32.29% and 11.65% among socially responsible and conventional investors, respectively, have at least one parent holding socially responsible funds. In addition, it is more common among the socially responsible investors to have both parents investing in socially responsible mutual funds, 11.19% compared to 2.31% among conventional investors. In regard to parental characteristics, these differ significantly between conventional and socially responsible investors. Parents of socially responsible investors appear to be slightly older when the individual was born and to have higher incomes during the individuals’ adolescence, for both mothers and fathers. 3.3. Alternative mechanisms and investor characteristics In order to provide evidence on whether socialization play a role in explaining potential correlation between individuals’ and parents’ socially responsible fund investing, a number of competing mechanisms need to be considered. One potential mechanism concern similarities in individuals’ and parents’ risk preferences, i.e., that individuals inherit similar risk preferences as their parents. For example, if individuals and parents share perceptions that the risk-return trade-off system-

J. Hellström, N. Lapanan and R. Olsson / European Economic Review 121 (2020) 103328

7

Table 3 Summary statistics. Panel A: Parental Characteristics Conventional investor

SR investor

All

Parents SR-investment None Mother Father Both

88.34% 4.92% 4.42% 2.31%

67.70% 12.06% 9.04% 11.19%

86.87% 5.43% 4.75% 2.95%

Father Income during child adolescence Age when child was born

163,433 30.41

174,938 30.91

164,255 30.45

Mother Income during child adolescence Age when child was born

98,613 27.84

104,731 28.45

99,050 27.88

Conventional investor

SR investor

All

Socioeconomic characteristics Female Single Children Born in 1973 Income Net wealth Risky asset share

49.01% 63.63% 58.76% 58.58% 238,203 415,119 66.10%

53.10% 60.64% 57.26% 56.53% 254,223 580,755 65.40%

49.30% 63.42% 58.65% 58.43% 239,347 426,944 66.00%

Educational Attainment Compulsory school High school or lower Bachelor or Master Ph.D. Economics degree

3.74% 51.97% 43.13% 1.16% 9.08%

2.58% 37.47% 57.42% 2.53% 9.42%

3.66% 50.92% 44.17% 1.26% 9.10%

Portfolio characteristics Portfolio return Number of funds in portfolios Same fund company as parents

6.40% 2.79 43.70%

6.91% 4.88 31.40%

6.43% 2.94 42.80%

Community characteristics Big city Community share – SR investors

18.85% 6.23%

25.01% 6.90%

19.21% 6.28%

Observations

276,682

21,269

297,951

Panel B: Individual investor characteristics

In Panel A, parental characteristics are reported for conventional and socially responsible individual investors, while in Panel B for individual investors.

atically differs between conventional and socially responsible mutual funds, correlations between individuals’ and parents’ prosocial behavior may then be driven by correlation in risk preferences rather than by transmission of prosocial behavior. Dohmen et al. (2012), for example, document intergenerational correlations in risk and trust attitudes between parents and their children. To control for this competing explanation, we include individuals’ risky asset shares as a proxy for individuals’ risk preferences. In the calculation of individuals’ risky assets shares, we follow the recent literature and focus on liquid portfolios, e.g., Brunnermeier and Nagel (2008), Malmendier and Nagel (2011), and Calvet and Sodini (2014). Thus, we define the risky asset share as the proportion of risky financial assets (directly held stocks and risky mutual funds) in the complete portfolio (risky financial assets and cash, where cash means bank account balances and money market funds). Risk preferences are assumed to be a relatively stable characteristic over time and we therefore use the individuals’ average risky asset shares calculated over the full sample period as a measure of individuals’ propensity to take risk. Another mechanism potentially generating individual-parental correlation in holdings of socially responsible mutual funds is if individuals and parents share the same financial advisor or bank. Within-family correlation in socially responsible mutual fund holdings may then be driven by similar recommendations from the same advisor or from advisors from the same bank.14 To account for this potential mechanism we construct a “same advisor” dummy utilizing information about

14 That financial advice is of importance in our context is indicated, for example, in a survey by TNS Sifo Prospera (2010), (https://www.fondbolagen. se/globalassets/faktaindex/studier- o- undersokningar/fondspararundersokningen), on behalf of the Swedish Investment Fund Association, where 47% of the surveyed individual fund investors indicate taking advice from their financial advisor.

8

J. Hellström, N. Lapanan and R. Olsson / European Economic Review 121 (2020) 103328

which mutual funds that belong to specific mutual fund companies. Given that the mutual fund market in Sweden is dominated by mutual fund companies belonging to the four largest banks in Sweden (Swedbank, Nordea, Handelsbanken, and SEB) and that a majority of Swedish mutual fund investors are customers in these banks,15 we can, based on information about individuals’ mutual fund holdings, construct proxies for which main advisors (i.e., mutual fund company and bank) individuals and parents were using during our study period.16 Hence, observing an individual only holding mutual funds distributed and managed by Swedbank Robur (the largest mutual fund company in Sweden during our study period), it is quite likely that the financial advisor for this individual belonged to Swedbank. Based on this logic and assigning mutual fund companies and banks to individuals in accordance to in which mutual fund company they had their largest aggregated mutual fund holdings, we are able to assign “financial advisors” to both individuals and parents. To test to what extent our results are driven by a common financial advisor, we then create a dummy variable taking the value 1 (0 otherwise) if an individual is identified as having the same mutual fund company as either that of its father, mother, or both. A correlation between individuals and parents ownership of socially responsible mutual funds may also be driven by a common community environment. That is, both individuals and parents may be affected by peers in their community and by common social norms and attitudes. For example, a growing body of literature concerning social influence has documented peer effects on retirement savings (Duflo and Saez, 2002), on trading decisions (Kelly and O Grada, 20 0 0; Hong et al., 20 05; Shive, 2010), and stock market participation (Hong et al., 2004; Brown et al., 2008; Kaustia and Knüpfer, 2012). Furthermore, both Ivkovic´ and Weisbenner (2007) and Hvide and Östberg (2014) document similarities in portfolio compositions between individuals and their neighbors and individuals and their co-workers. Although not previously studied, one could hypothesize that peer influence may also exist and affect individuals’ decisions to invest in a prosocial manner. To condition on this potential competing explanation, we include in our regressions both a “big city” dummy (value 1, if an individual resides in one of the three largest cities in Sweden, Stockholm, Gothenburg, or Malmö, 0 otherwise) and the proportion of socially responsible investors in an individual’s community, to proxy for possible common peer effects.17 A further competing mechanism, potentially generating an individual-parental correlation in prosocial behavior, is that individuals and parents are likely to share a common information environment. This includes being exposed to the same type of general information, e.g., advertisements about socially responsible investment opportunities, as well as potential within-family sharing of information. Li (2014), for example, find evidence of the later as a driver for individuals’ stock market participation. This is supported by the findings in Hellström et al. (2014), indicating an influence of parental sharing of recent stock market experiences on individuals’ stock market entry, participation and exit. To avoid capturing potential effects from a common information environment, we lag the parental socially responsible mutual fund ownership variables one period, while to control for potential within-family sharing of information, we include the parents’ lagged socially responsible fund portfolio returns as a proxy for within-family sharing of information.18 Finally, we highlight here that potential correlations driven by inheritance of socially responsible mutual funds from parents to adult children are unlikely in our sample. The considered cohorts of individuals are born in 1963 and 1973 and their investment behaviors are observed during 20 0 0–20 07, i.e., at ages 37–44 and 27–34 for the respective cohorts. Thus, given the relative young age of the studied individuals and that the sample is restricted to individuals with both parents alive, it is unlikely that correlations are driven by parents transferring socially responsible mutual funds to adult children through heritage. It is further unlikely that parental savings decisions early in individuals’ lives, e.g., parents opening a savings account for their children at an early age, is likely to drive the estimated correlations due to the limited availability of socially responsible mutual funds on the market during the individuals’ childhood period (1960s and 1970s). In Table 3, Panel B, we report summary statistics for our control variables for the full sample as well as for the separate groups of conventional and socially responsible investors. Regarding socioeconomic characteristics, the unconditional means indicate that the proportion of females, non-singles and older investors (born in 1963) are higher in the group of socially responsible investors and furthermore that the group of conventional investors, on average, earn lower incomes from salary and have lower net wealth. Comparing the risky asset shares between groupings reveal that the group of socially responsible investors’, on average, have slightly lower risky asset shares than conventional investors (65.40% compared to 66.10%). In terms of education, individuals with education higher than high school have a greater tendency to be socially responsible investors. Among the socially responsible investors, 59.95% have a university degree or above, while the corresponding figure for conventional investors is 44.29%. The proportion of individuals with a degree in economics is 9.10% in the full sample, slightly higher in the group of socially responsible fund investors, 9.42% compared to 9.08% for conventional investors. Regarding portfolio characteristics, socially responsible investors earn slightly higher portfolio returns during the period, 6.91% compared to 6.40% for conventional investors, and hold fund portfolios with, on average, more funds than conventional 15

For example, in 1999 the mutual fund companies owned by the four largest banks in Sweden held 85% of the outstanding fund wealth in Sweden. Notable features of importance prevailing during our study period were that mutual fund companies to a large extent only marketed and distributed their own mutual funds (Swedish Competition Authority, Report 2013:4, p. 135) and that individuals’ customer mobility were low. (http: //www.konkurrensverket.se/globalassets/aktuellt/nyheter/konkurrensen- pa- den- finasiella- markanden.pdf). 17 Furthermore, models including community fixed effects or using robust standard errors clustered at the community level have also been considered. Throughout, effects from accounting for this in the models (results unreported, but available on request) have no effects on the estimated correlations (i.e., same magnitudes, signs and significances) between individuals’ and parents’ ownership of socially responsible mutual funds. 18 Hellström et al. (2014) utilize one period lagged stock portfolio outcomes for parents and partners to capture recent parental and partner stock market experiences. Results indicate that parents’ and partners’ one-period lagged stock portfolio outcomes are of significant importance for individuals’ entry, participation, as well as for exit from the stock market. 16

J. Hellström, N. Lapanan and R. Olsson / European Economic Review 121 (2020) 103328

9

investors, 4.88 funds compared with 2.79.19 In total 42.8% of the individuals in our sample are identified to have the same main fund company (i.e., same “financial advisor”) as their parents. Notable, this figure is significantly larger for the group of conventional investors, 43.7% compared to 31.4% for socially responsible investors. Concerning controls for community characteristics, the figures indicate that among the group of socially responsible investors it is more common to reside in a big city and in a community with a higher proportion of peers holding socially responsible mutual funds than it is for the group of conventional investors. 4. Results To study the parental influence on an individual’s likelihood to hold socially responsible mutual funds, we split the sample of individuals, conditional on owning mutual funds at time t, between conventional (coded as “0”) and socially responsible investors (coded as “1”). Socially responsible investors are defined as individuals owning at least one socially responsible mutual fund at time t.20 The binary indicator variable is related to regressors by use of linear probability models. In our analysis, the main regressors of interest are binary indicator variables capturing the lagged parental ownership of socially responsible mutual funds (e.g., whether mothers, fathers, or both, own socially responsible mutual funds). Although this captures a contemporary correlation between individuals’ and parents’ lagged ownership of socially responsible mutual funds, we regard this correlation to mainly reflect similarities in individuals’ and parents’ prosocial behavior. To strengthen this claim, we condition in the analysis on proxies capturing the previously discussed alternative mechanisms (see Section 3.3) and a host of control variables (e.g., individuals’ net wealth, income, education, gender, fund portfolio return, and number of funds in individuals’ fund portfolios) as well as on annual fixed effects. To test the claim that the individual-parental correlations in holdings of socially responsible mutual funds reflect transmission of behavior originating during the individual’s youth, the parental ownership variables are further interacted with parental characteristics pertaining to the individual’s adolescence (age 18–19). Systematic patterns revealed by these interactions are unlikely to be explained by contemporary factors, but seem more likely to be associated with conditions pertaining to the individual’s socialization process. In the following, we report our empirical results. Throughout, coefficients and corresponding standard errors (cluster robust at the individual level) are reported for a number of model specifications. 4.1. Intergenerational correlation in holdings of socially responsible mutual funds Table 4, Models 1 and 2, present results for a baseline model (excluding parental variables), as well as for a model including a parental ownership indicator of socially responsible mutual funds, respectively. The indicator variable indicates whether either the mother or the father, or both, own socially responsible mutual funds in the previous period (dummy with the value 1 for ownership, 0 otherwise). As indicated in Model 2, our results suggest that an individual’s likelihood of holding a socially responsible mutual fund in her fund portfolio increases by 10.1% (statistically significant at the 1% level) with parental ownership. Given that the captured parental effect could potentially reflect both a transmission of prosocial behavior as well as an effect driven by contemporary information-sharing, we include in Model 3 the past one-year return on the parental socially responsible mutual fund portfolio. This is included as a proxy to broadly capture information-sharing effects; here it is assumed that a positive (negative) past year socially responsible mutual fund performance creates a positive (negative) parental sentiment towards investing in these funds, potentially shared with individuals through social interaction. Interestingly, the estimates for the parental socially responsible mutual fund ownership dummy do not change when including this broad control for social interaction effects (cf. Models 2 and 3). Notably, the estimated coefficient for the past year parental mutual fund portfolio return is positive and statistically significant (at the 1% level). This indicates that relatively more positive parental socially responsible mutual fund experiences (during the past year) seem to encourage individuals’ subsequent likelihood to hold these. The results are conditioned on a vast number of characteristics as well as annual fixed effects. Given that these controls are included to capture competing mechanisms (e.g., intergenerational transmission of risk preferences, effects from common financial advisors, community effects, and sharing of a common information environment), potentially generating parental and individual correlation in socially responsible mutual fund ownership, we interpret the results as mainly pertaining to effects from a parental transmission of prosocial behavior through socialization. 4.2. Importance of mothers and fathers To study the separate influence of the mother’s and father’s prosocial investment behavior on an individual’s likelihood to hold socially responsible mutual funds, the (lagged) parental ownership variable (in Table 4) is separated into indicators 19 The number of funds in the individuals’ fund portfolio is included as a control since individuals with a preference for holding a larger number of funds in their portfolio may, for purely probabilistic reasons, have a larger chance of selecting a socially responsible mutual fund. 20 In our sample, we have on average (over the full sample period) about 93% zeros (i.e., individuals who do not own any socially responsible mutual funds), about 1.8% who only own socially responsible mutual funds (i.e., no conventional funds) and about 5.2% who hold mixed mutual fund portfolios with both conventional and socially responsible mutual funds. The division into conventional and socially responsible mutual fund investors is somewhat broad, but in line with Riedl and Smeets (2017).

10

J. Hellström, N. Lapanan and R. Olsson / European Economic Review 121 (2020) 103328 Table 4 Overall parental influence on individuals’ likelihood to hold socially responsible mutual funds. Model 1

Model 2

Model 3

Model 4

Coef.

s.e.

Coef.

s.e.

Coef.

s.e.

Coef.

s.e.

Socio-economic characteristics Female Single Children Born 1973 Income Net wealth Risky asset share

0.012∗∗∗ −0.007∗∗∗ 0.002 −0.014∗∗∗ −0.001∗∗∗ −0.001 −0.015∗∗∗

(0.002) (0.002) (0.002) (0.003) (0.000) (0.001) (0.003)

0.012∗∗∗ −0.006∗∗∗ 0.003 −0.015∗∗∗ −0.001∗∗ −0.002 −0.014∗∗∗

(0.002) (0.002) (0.002) (0.003) (0.000) (0.001) (0.003)

0.012∗∗∗ −0.006∗∗∗ 0.003 −0.015∗∗∗ −0.001∗∗ −0.002 −0.014∗∗∗

(0.002) (0.002) (0.002) (0.003) (0.000) (0.001) (0.003)

0.012∗∗∗ −0.006∗∗∗ 0.003 −0.015∗∗∗ −0.001∗∗∗ −0.002 −0.014∗∗∗

(0.002) (0.002) (0.002) (0.002) (0.000) (0.001) (0.002)

Educational attainment High school Bachelor or Master Ph.D. Economics degree

−0.007 0.012∗∗ 0.042∗∗∗ −0.013∗∗∗

(0.005) (0.006) (0.014) (0.004)

−0.008 0.009 0.038∗∗∗ −0.012∗∗∗

(0.005) (0.005) (0.014) (0.004)

−0.008 0.009 0.038∗∗∗ −0.012∗∗∗

(0.005) (0.005) (0.014) (0.004)

−0.008 0.009 0.038∗∗∗ −0.012∗∗∗

(0.005) (0.005) (0.014) (0.004)

Portfolio characteristics SR-fund portfolio return indicatort-1 Number of funds in portfolio Same mutual fund company as parents

0.076∗∗∗ 0.022∗∗∗ −0.021∗∗∗

(0.003) (0.001) (0.002)

0.075∗∗∗ 0.022∗∗∗ −0.013∗∗∗

(0.002) (0.001) (0.002)

0.075∗∗∗ 0.022∗∗∗ −0.013∗∗∗

(0.002) (0.001) (0.002)

0.075∗∗∗ 0.022∗∗∗ −0.013∗∗∗

(0.002) (0.001) (0.002)

Community characteristics Big city Community share of SR investors

−0.003 0.009∗∗∗

(0.003) (0.001)

−0.003 0.008∗∗∗

(0.003) (0.001)

−0.003 0.008∗∗∗

(0.003) (0.001)

−0.003 0.008∗∗∗

(0.003) (0.001)

0.101∗∗∗

(0.005)

0.101∗∗∗

(0.005)

−0.089∗∗∗

(0.007)

0.018∗∗∗ −0.091∗∗∗

(0.005) (0.007)

0.101∗∗∗ −0.001 −0.002 0.018∗∗∗ −0.091∗∗∗

(0.005) (0.013) (0.042) (0.005) (0.007)

Parental socially responsible investment Parental ownershipt-1 Adopted Parental ownershipt-1 × Adopted Parental SR-fund portfolio returnt-1 Constant

−0.081∗∗∗

Annual fixed effects R2 Number of Observations

Yes 0.070 258,374

(0.007)

Yes 0.087 258,374

Yes 0.087 258,374

Yes 0.087 258,374

The table report estimated coefficients and corresponding standard errors in parenthesis for linear probability models using cluster robust standard errors at the individual level. The dependent variable is a binary variable, where 1 indicates individuals’ ownership of socially responsible mutual funds, 0 otherwise (ownership of conventional mutual funds). The main variable of interest, parental ownership, is an indicator with the value 1 if either the mother or the father, or both, own socially responsible mutual funds in the previous period, 0 otherwise. Significance levels:. ∗∗∗ p<0.01. ∗∗ p<0.05∗ p<0.10.

of whether only mothers, only fathers, or both, own socially responsible mutual funds. In Table 5, we report results for different model specifications using these indicators. As seen in Model 1, an individual’s likelihood to hold socially responsible mutual funds is significantly higher (statistically significant at the 1% level) for individuals when either only mothers or only fathers, or both, own socially responsible mutual funds in the previous period. The likelihood is 6.5% higher if only fathers own, 8.5% higher if only mothers own, and 19.0% higher if both parents own socially responsible mutual funds in the previous period. Controlling for the past one-year returns on mothers’ and fathers’ socially responsible mutual fund portfolios, reported in Model 2, does not change these estimates. Neither does adding control variables regarding parental characteristics during the individual’s adolescence, reported in Model 3. Overall, the results in regard to parental ownership of socially responsible mutual funds are interesting and tentatively suggest that both mothers and fathers, separately, are of importance in the transmission of prosocial behavior. Interestingly, the effect of “only mother” owning socially responsible mutual funds is somewhat stronger than that of “only father” (8.5% for “only mother” is significantly different at the 5% level from 6.5% for “only father” based on a Wald test). This result is consistent with Dohmen et al. (2012), who find that mothers are of higher importance than fathers in transmission of trust and of equal importance in the transmission of risk attitudes. Notable, for the controls capturing contemporary social interaction effects, i.e., the socially responsible mutual fund portfolio returns for mothers and fathers, only those for mothers have a weak influence (statistically significant at the 5% level). A higher return on the mother’s socially responsible mutual fund portfolio during the past year, i.e., a more positive sentiment towards social responsible investments, slightly increases the individual’s likelihood of investing in socially responsible mutual funds.

J. Hellström, N. Lapanan and R. Olsson / European Economic Review 121 (2020) 103328

11

Table 5 Mother versus father influence on individuals’ likelihood to hold socially responsible funds. Model 1

Parental ownership of socially responsible funds Only Fathert-1 Only Mothert-1 Both Mother and Father

Model 2

Coef.

s.e.

Coef.

s.e.

Coef.

s.e.

0.065∗∗∗ 0.085∗∗∗ 0.190∗∗∗

(0.007) (0.007) (0.011)

0.065∗∗∗ 0.085∗∗∗ 0.190∗∗∗

(0.007) (0.007) (0.011)

0.065∗∗∗ 0.085∗∗∗ 0.189∗∗∗

(0.007) (0.007) (0.011)

0.006 0.016∗∗

(0.007) (0.007)

0.005 0.016∗∗

(0.007) (0.007)

0.001∗∗∗ 0.000 0.001∗∗ −0.000 −0.006∗

(0.000) (0.000) (0.000) (0.000) (0.003)

Social responsible fund portfolio return Father Mother Parental characteristics during individuals youth Mother Income Father Income Mother age Father age Parental equality – both parents work Explanatory variables (same as in Table 4) Annual fixed effects R2 Number of Observations

Model 3

Yes Yes 0.091 258,374

Yes Yes 0.091 258,374

Yes Yes 0.092 258,374

The table report estimated coefficients and corresponding standard errors in parenthesis for linear probability models using cluster robust standard errors at the individual level. The dependent variable is a binary variable, where 1 indicates individuals’ ownership of socially responsible mutual funds, 0 otherwise (ownership of conventional mutual funds). The main parental variables of interest include dummy variables indicating if only the mother, only the father, or both, own socially responsible mutual funds, the socially responsible mutual fund returns for mothers and fathers during the past year (separately), as well as parental characteristics pertaining to the individuals adolescent. While the reporting of results is restricted to the parental variables of main interest, the full model (as those reported in Table 4) have been used in all estimations. The estimates for these other explanatory variables are similar in sign and significance, as those reported in Table 4. Significance levels:. ∗∗∗ p<0.01. ∗∗ p<0.05. ∗ p<0.10.

For parental characteristics during the individual’s adolescence, it is interesting to note that higher incomes among mothers (but not among fathers) during individuals’ adolescences have a statistical significant (at the 1% level) positive direct effect on individuals’ likelihood of investing in a prosocial manner. Furthermore, a higher age of the mother (but not the father) during an individual’s adolescence also increases (statistically significant at the 5% level) the individual’s likelihood to invest in socially responsible mutual funds. In terms of parent-child time, the indicator of whether both or only one parent worked during the individual’s youth (a binary variable equals to 1 when both parents worked, 0 otherwise) indicates a negative (statistically significant at the 10% level) impact on the individual’s adult prosocial behavior, i.e., individuals having both parents working during adolescence are less likely to invest in socially responsible mutual funds. Thus, parental resources (financial and life experience) as well as availability in time (whether both parents worked) during the individual’s adolescence seem to matter for individuals’ adult prosocial behavior. 4.3. Impact of assortative parental mating In Table 5 above, the marginal effect for the indicator of dual-parental ownership of socially responsible mutual funds is positive (statistically significant at the 1% level) and in size substantially larger than each of the individual influences from the mother and the father. This indicates that the correlation in parental and individual holdings of socially responsible mutual funds seems to be reinforced when both parents invest in a prosocial manner. To study this further, we report in Table 6 estimation results, separating the effects for individuals with one parent investing in socially responsible mutual funds (regardless whether it is the mother or the father), from that of those with two. As seen in Model 1, having two parents holding socially responsible mutual funds in the previous period substantially increases the individual’s subsequent likelihood of investing in a prosocial manner, relative to when only one parent (mother or father) invest prosocially. In terms of size, the effect of having both parents investing in a prosocial manner, compared to one, more than doubles the positive impact on individuals’ prosocial investing (2.5 times larger effect when both parents invest prosocially). The results also hold when controlling for the past-year parental household social responsible mutual fund returns, as well as when including parental characteristics from the individual’s adolescence (Models 2 and 3). A possible mechanism behind this substantially stronger effect of having both parents investing prosocially is assortative mating. Assortative mating indicates that (parental) relationships are not exogenously formed, but are instead based on similarities in values, attitudes and preferences. Assuming parents have preferences to pass their values on to children, assortative

12

J. Hellström, N. Lapanan and R. Olsson / European Economic Review 121 (2020) 103328 Table 6 Effect of parental similarities in prosocial values on individuals’ likelihood to hold socially responsible funds. Model 1 Coef. Parental ownership of socially responsible funds Only one parentt-1 0.076∗∗∗ 0.190∗∗∗ Both parentst-1 Social responsible fund portfolio return Parentst-1

Model 2 s.e.

Coef.

s.e.

Coef.

s.e.

(0.005) (0.011)

0.076∗∗∗ 0.190∗∗∗

(0.005) (0.011)

0.075∗∗∗ 0.190∗∗∗

(0.005) (0.011)

0.015∗∗∗

(0.005)

0.015∗∗∗

(0.005)

0.004∗∗ . 0.001∗∗ −0.004

(0.002) (0.000) (0.003)

Parental characteristics during individuals youth Parental household income Mother’s age Parental equality – both parents work Explanatory variables (same as in Table 4) Annual fixed effects R2 Number of Observations

Yes Yes 0.091 258,374

Model 3

Yes Yes 0.091 258,374

Yes Yes 0.091 258,374

The table report estimated coefficients and corresponding standard errors in parenthesis for linear probability models using cluster robust standard errors at the individual level. The dependent variable is a binary variable, where 1 indicates individuals’ ownership of socially responsible mutual funds, 0 otherwise (ownership of conventional mutual funds). The main parental variables of interest include dummy variables indicating if only one (regardless of mother or father), or both of the parents own socially responsible mutual funds, the total socially responsible mutual fund returns for mothers and fathers during the past year, as well as variables pertaining to the individuals adolescent. While the reporting of results is restricted to the parental variables of main interest, the full model (as those reported in Table 4) have been used in all estimations. The estimates for these other explanatory variables are similar in sign and significance, as those reported in Table 4. Significance levels:. ∗∗∗ p<0.01. ∗∗ p<0.05∗ p<0.10.

parental mating (including consideration to prosocial values) then minimizes the potential distortion in parental transference of prosocial values. Having two parents agreeing in prosocial behavior then potentially explains the reinforced effect from parents on individuals’ prosocial behavior. This result is consistent with Dohmen et al. (2012), who find that parents with more similar risk attitudes have a larger impact on the risk attitudes of the child. 4.4. Parental resources during individuals’ adolescence Research in sociology and psychology indicates that parental resources are of key concern in the socialization process (e.g., Grusec and Hastings, 2007). To study to what extent the strength of the correlation between parents’ and individuals’ ownership of socially responsible mutual funds depends on parental access to resources during the individuals’ youth, we interact the general parental ownership variable (value 1, if at least one parent owns socially responsible mutual funds, 0 otherwise) with proxies for parental resources during the individuals’ adolescence. To capture the potential effect of parental financial resources, the total income of the parents during the individual’s adolescence is utilized. To measure parental life experience during socialization, the age of the mother when the child was born, and to capture the potential impact of parent-child time, an indicator of whether both or only one parent worked during the individual’s youth, are utilized. The results from including these interaction terms capturing parental differences in resources are shown in Table 7 (benchmark model reported in Model 1). The results in Model 1 indicate that an individual’s likelihood to hold socially responsible mutual funds is 10.1% higher if the parents hold (only mother, only father, or both) socially responsible funds in the previous period. Interacting the parental effect with total parental income during an individual’s adolescence (Model 2), indicates that an individual’s likelihood of holding socially responsible mutual funds increases significantly (at the 5% level) by 1.6% for increasing total parental income during the individual’s youth. This tentatively indicates that parental financial resources during the socialization process (in the individual’s adolescence) are of importance for the parental transmission of prosocial behavior. This is interesting and consistent with research in sociology and psychology indicating the importance of parental financial resources for successful socialization (e.g., Fields et al., 2001; Bradley, 2002; Magnuson and Duncan, 2002; Patterson and Hastings, 2007) and transference of prosocial behavior (e.g., DeGarmo et al., 1999). In Model 3, we report the extension including parental experiences during the individual’s youth (proxied by age of the mother when the child was born). The results indicate that the correlation between parental and subsequent individual ownership in socially responsible mutual funds increases slightly by 0.4% (statistically significant at the 1% level) for individuals that have been raised by relatively more experienced (older) mothers. This is interesting and consistent with research suggesting that parents with more life experience during the socialization process, on average, are better equipped to socialize children (e.g., Culp et al., 1988; Moore et al., 1997; Luster et al., 20 0 0). In Model 4, we further extend the analysis by including a parental equality interaction variable. This variable captures potential effects of having both parents working during the individual’s adolescence. As seen, the effect from this variable is

J. Hellström, N. Lapanan and R. Olsson / European Economic Review 121 (2020) 103328

13

Table 7 Parental resources during socialization and individuals’ likelihood to hold socially responsible funds.

Parental ownership of socially responsible funds Parental ownershipt-1

Model 1

Model 2

Model 3

Model 4

Model 5

0.101∗∗∗ (0.005)

0.119∗∗∗ (0.011) 0.016∗∗ (0.008)

−0.065 (0.045)

0.101∗∗∗ (0.010)

0.000 (0.011)

0.043 (0.048) 0.017∗∗∗ (0.008) 0.004∗∗∗ (0.001) −0.002 (0.011)

Parental ownershipt-1 × parental household income

0.004∗∗∗ (0.001)

Parental ownershipt-1 × parental experience (mother’s age) Parental ownershipt-1 × parental equality (both parents work) Social responsible fund portfolio return Parentst-1 Parental characteristics during individuals youth Parental household income Parental experience (mother’s age) Parental equality (both parents work) Explanatory variables (same as in Table 4) Annual fixed effects R2 Number of Observations

0.018∗∗∗ (0.005)

0.018∗∗∗ (0.005)

0.018∗∗∗ (0.005)

0.018∗∗∗ (0.005)

0.018∗∗∗ (0.005)

0.004∗∗ (0.002) 0.001∗∗ (0.000) −0.004 (0.003)

0.002 (0.002) 0.001∗∗ (0.000) −0.004 (0.003)

0.004∗∗ (0.002) 0.000 (0.000) −0.004 (0.003)

0.004∗∗ (0.002) 0.001∗∗ (0.000) −0.004 (0.003)

0.001 (0.002) 0.000 (0.000) −0.003 (0.002)

Yes Yes 0.087 258,374

Yes Yes 0.087 258,374

Yes Yes 0.087 258,374

Yes Yes 0.087 258,374

Yes Yes 0.087 258,374

The table report estimated coefficients and corresponding standard errors in parenthesis for linear probability models using cluster robust standard errors at the individual level. The dependent variable is a binary variable, where 1 indicates individuals’ ownership of socially responsible mutual funds, 0 otherwise (ownership of conventional mutual funds). The main parental variables of interest include dummy variables indicating if only one (regardless of mother or father), or both of the parents own socially responsible mutual funds, the total socially responsible mutual fund returns for mothers and fathers during the past year, as well as variables pertaining to the individuals adolescent. While the reporting of results is restricted to the parental variables of main interest, the full model (as those reported in Table 4) have been used in all estimations. The estimates for these other explanatory variables are similar in sign and significance, as those reported in Table 4. Significance levels:. ∗∗∗ p<0.01. ∗∗ p<0.05∗ p<0.10.

insignificant. Thus, having both parents at work during the individual’s youth does not seem to affect the individual’s adult investing behavior. Overall, in line with the previous results, including all interaction terms (reported in Model 5) indicate a significant (at the 1% level) positive effect of parental financial resources and parental experience on individuals’ likelihood to hold socially responsible mutual funds. In terms of other variables, we note that our broad control for contemporary parental social interaction effects, i.e., the parental return on their socially responsible mutual fund portfolio during the previous year, has a positive significant (at the 1% level) effect on an individual’s likelihood of investing in a socially responsible manner. This indicates that recent positive parental experiences from investing in socially responsible mutual funds seem to be shared and to affect individuals’ subsequent decisions. 4.5. Socialization or inherited genetically driven effects? So far we have not touched on the questions of whether prosocial behavior may be genetically inherited. Recent research on the formation of generosity suggest a role of both “nature” and “nurture”. In terms of heritability, studies on twin data suggest that prosocial preferences, at least partly, is heritable (Wallace et al., 2007; Cesarini et al., 20 08, 20 09). Cesarini et al. (2009) do, for example, document a significant genetic effect on giving behavior, based on comparisons of twins sharing the same set of genes (monozygotic) with those with imperfectly correlated genes (dizygotic). Thus, part of the prosocial correlations found in our study may potentially derive from individuals inheriting genes from their parents. Although we have no information to perform a classical twin study, i.e., to study identical and fraternal twins, our data allow us to identify a sample of adopted individuals. Based on information about the country of origin for individuals’ and their birth-parents as well as about the parents they grew up with, we are able to identify a small sample of 348 individuals fulfilling our sample requirements, whereof 34 invests in socially responsible mutual funds.21 In total the sample is composed of 1791 observations. Given that individuals and parents in our sample of adopted individuals do not directly 21 Note here that we are only able to identify individuals who are adopted from outside Sweden. Thus, we have no information to identify adoptions of children born in Sweden.

14

J. Hellström, N. Lapanan and R. Olsson / European Economic Review 121 (2020) 103328

share genetic compositions, but only the social environment during upbringing, we perform additional analysis using this restricted sample. In Model 4, Table 4, we report estimation results for a model specification including an interaction term capturing the parent-child correlation in prosocial investing separately for the sample of adopted individuals. The insignificant result for the interaction term indicates a similar parental-child correlation also for the sample of adopted individuals as for the full sample. Re-estimation of the models reported in Table 4 on only the sample of adopted individuals further indicate that the lagged parental socially responsible ownership dummy is positively significant (at the 5% level) and in size comparable to the marginal effects based on the full sample. Although these results do not rule out potential effects from inherited genes, it still provide some assurance that effects, at least, do not seem solely driven by “nature”, but to a large extent seem consistent with our interpretation of socialization driven effects. To provide further evidence on whether our identified parental-child correlations actually seem to correspond to effects driven by socialization, we have also performed a comparison of whether siblings’ socially responsible mutual fund investments look more alike than those of randomly paired individuals.22 In our data we are able to identify individuals born by the same parents in our cohorts, i.e., either real twins or closely born siblings (during the same year) or siblings identified between cohorts. Focusing on pairs of twins and siblings, we identify a “siblings” sample of 1182 pairs satisfying our final sample requirements. To compare behavior, we obtain the predicted probability for socially responsible mutual fund investments for each individual (based on estimation of Model 3, Table 4). We then compare the average absolute difference in predicted probability for socially responsible investing between pairs in the “siblings” sample with that of randomly paired individuals in a randomly drawn comparison sample of similar size.23 The average absolute difference in predicted probability for prosocial investments in the “siblings” sample is 2.73 percent (s.d. 0.036). The corresponding average absolute difference in the sample of randomly paired individuals is 8.39% (s.d. 0.071). Based on a t-test, the difference in the “siblings” sample is significantly smaller (at the 1% level) than that for the randomly paired individuals. This indicates that our estimated model predicts prosocial mutual fund investment behavior that are more similar among “siblings” than among randomly paired individuals. This loosely speaks in favor of socialization driven effects.

4.6. On alternative mechanisms and other control variables Of particular interest are the proxies that are included to condition on the main competing explanations for the finding of a correlation between individuals’ and parents’ prosocial mutual fund investment behavior, i.e., regarding intergenerational transmission of risk preferences, effects driven by a common financial advisor, potential community driven effects, and potential effects from sharing a common information environment. The individuals’ risky asset share, included to capture intergenerational transmission of risk preferences, is significantly negative (at the 1% level) in all models. This indicates that, on average, more risk averse individuals are more likely to invest in a prosocial manner.24 Regarding whether effects are driven by individual-parental sharing of the same financial advisor or advisors from the same bank, the included dummy for having the same fund company as parents is significantly negative (at the 1% level) in all regressions. Thus, individuals sharing financial advisors with their parents are less likely to invest in socially responsible mutual funds. This effect may be driven by that advisors, on average, may be more prone to give similar advices concerning investments in non-socially responsible funds. Alternatively, the negative effect may reflect inactivity among individuals. Assuming that most individuals start out with the same bank as their parents and that picking socially responsible mutual funds require a more active search, then less active individuals would be more likely to have the same advisor (bank) as their parents and also to hold non-socially responsible mutual funds. To provide further evidence on the effect of having the same financial advisor or advisors from the same bank, we have also interacted the “same fund company” dummy with the lagged parental socially responsible mutual fund ownership dummy. The results from this regression (unreported) indicate that the direct effect on individuals’ likelihood to hold socially responsible mutual funds from parental prosocial investing reduces from 10.1% (Model 3, Table 4) to 8.6% (significant at the 1% level), while the interaction effect, i.e., having the same financial advisor as parents and parents that invest prosocially, is 0.053 (significant at the 1% level). Thus, part, but far from all, of the correlation between individuals’ and parents’ prosocial investing behavior is driven by a common financial advisor. Controlling for effects of a common financial advisor by direct inclusion of fund advisor fixed effects in our regression models have also been considered. Results from this analysis (unreported, but available on request) indicate only small changes in the estimated parental-child correlation in prosocial investment behavior. In terms of community driven effects, i.e., that the correlation between individuals and parents ownership of socially responsible mutual funds may be driven by a common community environment (e.g., through community social norms and attitudes), the big city dummy (Stockholm, Gothenburgh, and Malmö) is throughout insignificant, while the community

22

We thank an anonymous referee for pointing us in this direction. Several comparison samples were randomly drawn and used for comparison. Results are similar as that reported in the paper when altering the comparison samples. 24 The risky asset shares of mothers and fathers have also been considered in the empirical analysis. Including these did not change any of the reported results. 23

J. Hellström, N. Lapanan and R. Olsson / European Economic Review 121 (2020) 103328

15

proportion of socially responsible mutual fund investors is significantly positive (at the 1% level), in most models. The estimated effect of the community proportion indicate that individuals’ likelihood to hold socially responsible mutual funds increases slightly (0.8 percent) for a marginal increase in the proportion of socially responsible investors in an individuals’ community. More importantly, including the community proportion in the regression do only marginally affect the estimated correlation between individuals’ and parents’ ownership of socially responsible mutual funds (the marginal effect decreases from 10.3% to 10.1%). Concerning potential effects driven by a common information environment, this has mainly been dealt with by including the lagged parental socially responsible mutual fund ownership variables, lagged parental socially responsible mutual fund returns, and by including annual fixed effects. To further strengthen our evidence supporting an interpretation that correlations are driven by an intergenerational transmission of prosocial behavior, we have also estimated models on the cross-section of individuals pertaining to the year 2007, but with parental holdings pertaining to the year 1999, i.e., with a longer lag. In this analysis it is less likely that individual-parental correlations in holdings of socially responsible mutual funds are driven by common reactions to the same information or that decisions to hold are influenced by within-family sharing of information. Results (c.f. Model 3, Table 4) from this analysis (unreported) indicate that the likelihood that an individual hold socially responsible mutual funds in 2007 increase with 5.94% (significant at the 1% level) if parents (either mother, father, or both) invest prosocially in 1999. In regard to the other control variables included in all regression models (see Table 4), the results are stable over all model specifications. In terms of socioeconomic characteristics, females, non-singles, and older individuals (belonging to the older cohort born in 1963) are (respectively, and all else equal) all more likely to invest in socially responsible mutual funds (results are statistically significant at the 1% level). In terms of education, the estimation results indicate that individuals with higher educational attainment (Ph.D., Master or Bachelor degree as highest) are more likely to invest in a prosocial manner (statistically significant at the 1% level). This is an interesting result since the educational system is usually considered to be one of the main institutions in the socialization process. Worth noting is that individuals who have a degree in economics, in all models, are found to have a lower likelihood (significant at the 1% level) to hold socially responsible mutual funds. In regard to portfolio characteristics, all models include an indicator variable of whether the previous period return on individuals’ socially responsible fund part of their total fund portfolio is larger than that on an index capturing returns for conventional funds (value 1, 0 otherwise) as well as the total number of mutual funds included in the fund portfolio. The results imply that if the past period return performance of socially responsible funds is higher than that of conventional funds, then the likelihood that an individual invest in a prosocial manner increases with 7.5% (significant at the 1% level). An increasing number of mutual funds in an individual’s mutual fund portfolio also increases the likelihood to hold socially responsible mutual funds with 2.2% (significant at the 1% level). 4.7. Additional robustness tests Our identification strategy of intergenerational transmission of prosocial behavior relies on controlling for competing explanations to individual-parental correlations in holdings of socially responsible mutual funds. Thus, exclusion of alternative explanations is of key importance.25 Even though, in the main analysis, we have included controls for a number of important mechanisms, e.g., control to capture effects from intergenerational transmission of risk preferences, to capture effects from having the same financial advisor, and for potential community driven effects, we have paid some additional attention towards further testing the robustness of our results. First, in our main analysis presented within the paper linear probability models are utilized. To test whether the results are robust also when controlling for unobserved heterogeneity, we re-estimate linear probability models including individual-specific random effects.26 In Appendix, Table A2, Model 1, we report, as an example, estimation results for the model specification in Table 4, Model 3. The results from this model indicate a smaller (3.9% instead of 10.1%), but still positive and significant (at the 1% level), effect of parental ownership of socially responsible mutual funds on individuals’ subsequent prosocial investment behavior. Thus, the results seem to hold also when accounting for unobserved heterogeneity, although with smaller effects. In terms of economic significance, we highlight here that the conditional mean likelihood (evaluated as the average of fitted values) to invest in socially responsible mutual funds is 7.23% (Model 3 in Table 4). An increase of 3.9% for individuals with parents investing in a prosocial manner is thus in relation to the conditional mean of sizable magnitude. In comparison with effects for other characteristics, e.g., being female or having a higher educational attainment, results are also of considerable size. Second, all analysis have also been performed using logit regression models, which better adhere to the binary nature of our dependent variable.27 Reassuring, estimation results, presented in Appendix, Table A2, both for logit (Model 2) and logit 25 As noted by an anonymous referee, the "adding control variables approach" to identify causal effects is likely to be insufficient to rule out all competing explanations. We do, however, in lack of pure exogenous variation for identification, try to provide evidence “beyond reasonable doubt”, i.e., evidence consistent with that no other logical explanation can be derived, although not being proof with absolute certainty. 26 Limited time variation in a number of explanatory variables generally restrict the use of fixed effects models. 27 Actually, all analysis, apart from the models including interaction variables, were presented in the main analysis in an earlier version of the paper. Results based on the logit regression models are available on request.

16

J. Hellström, N. Lapanan and R. Olsson / European Economic Review 121 (2020) 103328

regression with individual-specific random effects (Model 3), specified as Model 3 in Table 4, are similar to those for our linear probability models concerning our main variables of interest. The marginal effect for the parental socially responsible mutual fund ownership variable is 8.9% in the logit regression, while 4.4% in the logit regression accounting for unobserved heterogeneity. Third, in our main analysis we do not distinguish between individuals investing a small or a large proportion of their total fund portfolio in socially responsible mutual funds. To test whether results change when accounting for how much individuals invest in a prosocial manner, we have re-estimated models using instead the proportion invested in socially responsible mutual funds as the dependent variable. In Appendix, Table A2, Model 4, we present estimates for the model specification corresponding to Model 3, Table 4, within the paper. The results indicate that individuals’ proportions of socially responsible mutual funds (out of their total mutual fund portfolio) increase significantly (at the 1% level) with the parental proportions invested in socially responsible mutual funds. Overall, using the individuals’ and parents’ proportions give results similar to those reported within the paper.

5. Conclusions In this paper we have analyzed the correlation between parental and individual subsequent ownership of socially responsible mutual funds. Given that the results are robust towards a vast number of alternative explanations and given the recent evidence indicating that individuals’ investing in socially responsible mutual funds mainly seem driven by prosocial preferences (Riedl and Smeets, 2017), we favor an interpretation of the effects as reflecting a parental transmission of prosocial behavior. Interestingly, studying parental-individual correlations in detail reveals distinct patterns. First, ownership of socially responsible mutual funds among mothers is found to have a greater influence on individuals’ subsequent likelihood of owning, although fathers also have an influence. Second, parents agreeing in prosocial behavior have a larger impact on individuals’ prosocial behaviors. Third, greater resources (both economical and life experiences) held by parents during the individual’s adolescence are found to strengthen the parental-individual correlations in holdings of socially responsible mutual funds. Overall, these results are consistent with research in sociology and psychology on the parental role in the individual’s socialization process, tentatively indicating that parental effects seem to pertain to a transmission of prosocial behavior originating from socialization. This conclusion is reinforced by the finding of similar parental effects using a restricted sample of adopted individuals. A general conclusion from our study is thus that individuals’ prosocial behavior, affecting investing behavior, seem to be established early in life and that parents are an important influence in this process. Given that behaviors well rooted in an individual’s youth may be harder to change, this has implications for policies trying to promote socially responsible investing practices. Recent efforts among governments (see, e.g., Steurer et al., 2008) also include initiatives directed towards private investors, mainly in terms of informational guidelines and resources. Given our findings, these efforts may, however, prove to be of limited importance. An indication of this is given by the fact that over our observational period the proportion, out of individuals owning mutual funds, holding socially responsible mutual funds is fairly stable around 7.15%. Given that awareness about the existence of, and information related to, socially responsible mutual funds has likely increased during this period, this seems to have had a limited impact on individuals’ prosocial investment behavior. If governmental efforts to also promote prosocial investment practices among private investors are serious, then our results suggest that economic incentives (tax deductions or subsidies) or legal instruments are of better use. Understanding the reasons for investors’ socially responsible investing is important. This distinction is essential because previous research (e.g., Heinkel et al., 2001; Fama and French, 2007; Statman et al., 2008; Hong and Kacperczyk, 2009) indicates that the effect of prosocial investment behavior on asset prices differs depending on whether it is profit-driven or prosocially value-driven. If deviations from the market portfolio (due to restricting investments to socially responsible assets) are driven by prosocial preferences, they could influence stock prices in the long run, while if they are driven by differences in performance expectations, they could pertain to short-run effects on asset prices (Derwall et al., 2011). Given our empirical evidence, we favor the conclusion that individual investors’ socially responsible investments, at least partly, are prosocially value-driven. While our results are of direct interest for the financial literature, they are also of interest for the more general literature considering individuals’ social or other-regarding preferences. Recent research in this broader literature has begun to focus on the development of social preferences for children and adolescents (e.g., Cesarini et al., 2009; Sutter et al., 2018). Given that these studies provide experimental evidence related to the origin of prosocial preferences, the evidence within our study complements this research with results based on individuals’ actual observed behavior.

Supplementary materials Supplementary material associated with this article can be found, in the online version, at doi:10.1016/j.euroecorev.2019. 103328.

J. Hellström, N. Lapanan and R. Olsson / European Economic Review 121 (2020) 103328

17

Appendix

Table A1 Variable definitions. Variable

Definition

Socially responsible (SR) investor Female Single Children Born 1973 Income Net wealth

Equal to 1 if an individual hold at least one SR mutual fund in year t, 0 otherwise. Equal to 1 if individuals are female, 0 otherwise. Equal to 1 if individuals are not married or cohabitate, 0 otherwise. Equal to 1 if individuals have children, 0 otherwise. Equal to 1 if individuals are born in 1973, 0 otherwise. Annual income before tax from labor (in million SEK). Net wealth calculated from cash in bank, investment in funds, bonds, stocks, real estate wealth, minus the total value of liabilities (in million SEK). Proportion of risky financial assets (directly held stocks and risky mutual funds) in the complete portfolio (risky financial assets and cash). Dummy variables for highest level of education for each individual: "Compulsory school", "High school or lower", "Bachelor or Master degree", and "Ph.D. degree”. Equal to 1 if individuals have a degree in business or economics, 0 otherwise. Value-weighted portfolio returns on total mutual fund holdings of each individual using current mutual fund choice, current mutual fund weights and mutual fund returns during the year. Equal to 1 if individuals return on their SR mutual fund part of their total fund portfolio is higher than the return on a conventional mutual fund index, 0 otherwise. The number of funds held by each individual each year. Equal to 1 if individuals are identified to have the same main mutual fund company as their parents, 0 otherwise. Equal to 1 if individuals reside in either “Stockholm", “Gothenburg" or

Risky asset share Educational attainment Economics degree Portfolio return SR-fund portfolio return indicator Number of funds in portfolio Same mutual fund company as parents Big city “Malmö", 0 otherwise. Community share of SR investors One Parent

The proportion of SR mutual fund investors in an individual’s community. Equal to 1 if at least one of the parent invests in socially responsible mutual funds in the same period, 0 otherwise. Equal to 1 if both the mother and father invest in socially responsible mutual funds in the same period, 0 otherwise. Categorical variables for characteristics of parents’ socially responsible mutual fund holdings: equal to 0 if neither holds, 1 if only mother holds, 2 if only father holds and 3 if both mother and father hold. Parents’ value-weighted SR mutual fund portfolio return (based on the portfolio of both mother and father). Mother’s value-weighted SR mutual fund portfolio return. Father’s value-weighted SR mutual fund portfolio return. Mother’s income during the period when the child is 17 years old (in million SEK). Father’s income during the period when the child is 17 years old (in million SEK). Mother’s age when the child was born. Father’s age when the child was born. Equal to 1 if both parents work, 0 otherwise.

Both Parents Parent SR Investment

Parent SR portfolio return Mother SR portfolio return Father SR portfolio return Mother Income Father Income Mother age Father age Parental equality

Table A2 Overall parental influence on individuals’ likelihood to hold socially responsible mutual funds. Model 1 Coef.

Model 2

Model 3

Model 4

s.e.

Marg. eff.

s.e.

Marg. eff.

s.e.

Coef.

s.e.

Socio-economic characteristics Female 0.013∗∗∗ Single 0.000 Children 0.002 Born 1973 −0.014∗∗∗ Income −0.000 Net wealth 0.000 Risky asset share −0.002

(0.002) (0.002) (0.001) (0.002) (0.000) (0.001) (0.001)

0.013∗∗∗ −0.006∗∗∗ 0.002 −0.014∗∗∗ −0.001∗∗∗ −0.001 −0.013∗∗∗

(0.002) (0.002) (0.002) (0.002) (0.000) (0.001) (0.003)

0.009∗∗∗ −0.001 0.002 −0.011∗∗∗ 0.000 0.000 −0.003

(0.002) (0.002) (0.002) (0.002) (0.000) (0.001) (0.002)

0.005∗∗∗ −0.003∗∗ 0.000 −0.009∗∗∗ −0.000∗∗∗ −0.001∗∗ −0.008∗∗∗

(0.001) (0.001) (0.001) (0.001) (0.000) (0.001) (0.002)

Educational attainment High school Bachelor or Master Ph.D. Economics degree

(0.005) (0.005) (0.009) (0.004)

−0.005 0.014∗∗∗ 0.033∗∗∗ −0.011∗∗∗

(0.006) (0.006) (0.011) (0.004)

−0.004 0.008 0.008 −0.007∗∗∗

(0.008) (0.008) (0.010) (0.003)

−0.000 0.009∗∗∗ 0.026∗∗∗ −0.003

(0.002) (0.003) (0.008) (0.002)

0.001 0.012∗∗∗ 0.007 −0.007∗

(continued on next page)

18

J. Hellström, N. Lapanan and R. Olsson / European Economic Review 121 (2020) 103328 Table A2 (continued) Model 1 Coef. Portfolio characteristics SR-fund portfolio return indicatort-1 Number of funds in portfolio Same fund company as parents Community characteristics Big city Community share of SR investors

0.020∗∗∗ 0.020∗∗∗ −0.003∗∗∗ −0.004∗ 0.006∗∗∗

Parental socially responsible investment Parental ownershipt-1 0.039∗∗∗ Parental SR-fund portfolio returnt-1 0.009∗∗∗ Constant −0.046∗∗∗ Annual fixed effects Random effects R2 Number of Observations

Yes Yes 0.075 258,374

Model 2

Model 3

Model 4

s.e.

Marg. eff.

s.e.

Marg. eff.

s.e.

Coef.

s.e.

(0.001) (0.001) (0.001)

0.060∗∗∗ 0.013∗∗∗ −0.014∗∗∗

(0.002) (0.000) (0.002)

0.013∗∗∗ 0.012∗∗∗ −0.005∗∗∗

(0.001) (0.000) (0.002)

0.032∗∗∗ 0.000 −0.010∗∗∗

(0.002) (0.000) (0.001)

(0.002) (0.000)

−0.004 0.008∗∗∗

(0.003) (0.001)

−0.004∗∗∗ 0.006∗∗∗

(0.001) (0.001)

0.001 0.004∗∗∗

(0.002) (0.000)

(0.003) (0.003) (0.006)

0.089∗∗∗ 0.009∗∗∗

(0.004) (0.003)

0.044∗∗∗ 0.007∗∗∗

(0.003) (0.003)

0.129∗∗∗ 0.006∗ −0.015∗∗∗

(0.009) (0.003) (0.004)

Yes – 0.134 258,374

Yes Yes 0.112 258,374

Yes – 0.031 258,374

The table report estimated coefficients (marginal effects for logit models) and corresponding standard errors in parenthesis for a linear probability model (Model 1), logit regression models (Model 2 and 3), and a linear regression model (Model 4), using cluster robust standard errors at the individual level. The dependent variable in Model 1–3 is a binary indicator variable, where 1 indicates individuals’ ownership of socially responsible mutual funds, 0 otherwise (ownership of conventional mutual funds). The dependent variable in Model 4 is the individuals’ proportion of socially responsible mutual funds out of their total mutual fund portfolio. The main variable of interest, parental ownership, is an indicator with the value 1 if either the mother or the father, or both, own socially responsible mutual funds in the previous period, 0 otherwise. Significance levels:. ∗∗∗ p<0.01. ∗∗ p<0.05. ∗ p<0.10.

References Anger, S., 2011, The intergenerational transmission of cognitive and non-cognitive skills during adolescence and young adulthood, SOEP paper No. 473-2012. Ben-Ner, A., List, J.A., Putterman, L., Samek, A., 2017. Learned generosity? An artefactual field experiment with parents and their children. J. Econ. Behav. Organ. 143, 28–44. Bisin, A., Verdier, T., 20 0 0. Beyond the melting pot: cultural transmission, marriage, and the evolution of ethnic and religious traits. Q. J. Econ. 115, 955–988. Björklund, A., Hederos Eriksson, K., Jäntti, M., 2010. IQ and family background: are associations strong or weak? BE J. Econ. Anal. Policy 10. Björklund, A., Salvanes, K.G., 2011. Education and family background: mechanisms and policies. In: Handbook of the Economics of Education, 3. Elsevier, pp. 201–247. Black, S.E., Devereux, P.J., 2011. Recent developments in intergenerational mobility. In: Handbook of Labor Economics„ 4. Elsevier, pp. 1487–1541. Black, S.E., Devereux, P.J., Salvanes, K.G., 2009. Like father, like son? A note on the intergenerational transmission of IQ scores. Econ. Lett. 105, 138–140. Bowles, S., Gintis, H., 2002. The inheritance of inequality. J. Econ. Perspect. 16, 3–30. Bradley, R.H., 2002. Environment and parenting. In: Handbook of Parenting, 2. Lawrence Erlbaum Associates, pp. 281–314. ´ Z., Smith, P.A., Weisbenner, S., 2008. Neighbors matter: causal community effects and stock market participation. J. Financ. 63, Brown, J.R., Ivkovic, 1509–1531. Brunnermeier, M.K., Nagel, S., 2008. Do wealth fluctuations generate time-varying risk aversion? Micro-evidence on individuals. Am. Econ. Rev. 98, 713–736. Calvet, L.E., Sodini, P., 2014. Twin picks: disentangling the determinants of risk-taking in household portfolios. J. Financ. 69, 867–906. Carneiro, P., Heckman, J., 2004. Human capital policy, in inequality in America: what role for human capital policies. Inequality in America: What Role For Human Capital Policies. (MIT). Cesarini, D., Dawes, C.T., Fowler, J.H., Johannesson, M., Lichtenstein, P., Wallace, B., 2008. Heritability of cooperative behavior in the trust game. Proc. Natl. Acad. Sci. 104 (15) 631–15, 634. Cesarini, D., Dawes, C., Johannesson, M., Lichtenstein, P., Wallace, B., 2009. Genetic variation in preferences for giving and risk taking. Q. J. Econ. 124, 809–842. Collins, W.A., Maccoby, E.E., Steinberg, L., Hetherington, E.M., Bornstein, M.H., 20 0 0. Contemporary research on parenting: the case for nature and nurture. American Psychologist 55, 218–232. Culp, R.E., Appelbaum, M.I., Osofsky, J.D., Levy, J.A., 1988. Adolescent and older mothers: comparison between prenatal maternal variables and newborn interaction measures. Infant Behav. Dev. 11, 353–362. Cunha, F., Heckman, J.J., Navarro, S., 2006. Counterfactual analysis of inequality and social mobility. In: Mobility and inequality: Frontiers of Research in Sociology and Economics. Stanford University Press, Stanford, CA, pp. 290–348. Cunha, F., Heckman, J., 2007. The technology of skill formation. Am. Econ. Rev. 97, 31–47. DeGarmo, D.S., Forgatch, M.S., Martinez Jr, C.R., 1999. Parenting of divorced mothers as a link between social status and boys’ academic outcomes: unpacking the effects of socioeconomic status. Child Dev. 70, 1231–1245. Derwall, J., Koedijk, K., Ter Horst, J., 2011. A tale of values-driven and profit-seeking social investors. J. Bank. Financ. 35, 2137–2147. Dohmen, T., Falk, A., Huffman, D., Sunde, U., 2012. The intergenerational transmission of risk and trust attitudes. Rev. Econ. Stud. 79, 645–677. Duflo, E., Saez, E., 2002. Participation and investment decisions in a retirement plan: the influence of colleagues choices. J. Public Econ. 85, 121–148. Fama, E.F., French, K.R., 2007. Disagreement, tastes, and asset prices. J. Financ. Econ. 83, 667–689. Fields, J., Smith, K., Bass, L.E., Lugaila, T., 2001. A child’s day: home, school, and play (selected indicators of child well-being). Curr. Popul. Rep. 70, 68. Grönqvist, E., Ockert, B., Vlachos, J., 2017. The intergenerational transmission of cognitive and non-cognitive abilities. J. Hum. Resour. 52, 887–918. Groves, M.O., 2005. Personality and the intergenerational transmission of economic status. Unequal chances: Family background and Economic success, 208–231. Princeton University Press, Princeton. Grusec, J.E., Hastings, P.D., 2007. Handbook of socialization: Theory and Research. Guilford Press, New York.

J. Hellström, N. Lapanan and R. Olsson / European Economic Review 121 (2020) 103328

19

Heckman, J.J., Masterov, D.V., 2007. The productivity argument for investing in young children. Appl. Econ. Perspect. Policy 29, 446–493. Heckman, J.J., Stixrud, J., Urzua, S., 2006. The effects of cognitive and noncognitive abilities on labor market outcomes and social behavior. J. Labor Econ. 24, 411–482. Heinkel, R., Kraus, A., Zechner, J., 2001. The effect of green investment on corporate behavior. J. Financ. Quant. Anal. 36, 431–449. Hellström, J., Lapanan, N., and Olsson, R., 2018, Socially responsible investments among individuals. Working Paper. Hellström, J., Zetterdahl, E., and Hanes, N., 2014, Loved ones matter: family effects and stock market participation. Working Paper. Holmlund, H., Lindahl, M., Plug, E., 2011. The causal effect of parents’ schooling on children’s schooling: a comparison of estimation methods. J. Econ. Lit. 49, 615–651. Hong, H., Kacperczyk, M., 2009. The price of sin: the effects of social norms on markets. J. Financ. Econ. 93, 15–36. Hong, H., Kubik, J.D., Stein, J.C., 2004. Social interaction and stock-market participation. J. Financ. 59, 137–163. Hong, H., Kubik, J.D., Stein, J.C., 2005. Thy neighbor’s portfolio: word-of-mouth effects in the holdings and trades of money managers. J. Financ. 60, 2801–2824. Hvide, H.K., Östberg, P., 2014. Stock investments at work. J. Financ. Econ. 117, 628–652. ´ Z., Weisbenner, S., 2007. Information diffusion effects in individual investors’ common stock purchases: covet thy neighbors’ investment choices. Ivkovic, Rev. Financ. Stud. 20, 1327–1357. Junkus, J.C., Berry, T.C., 2010. The demographic profile of socially responsible investors. Manag. Financ. 36, 474–481. Kaustia, M., Knüpfer, S., 2012. Peer performance and stock market entry. J. Financ. Econ. 104, 321–338. Kelly, M., Grada, C.O., 20 0 0. Market contagion: evidence from the panics of 1854 and 1857. Am. Econ. Rev. 90, 1110–1124. Kosse, F., Deckers, T., Pinger, P.R., Schildberg-Hörisch, H., and Falk, A., 2018, The formation of prosociality: causal evidence on the role of social environment, CESifo Working Paper Series No. 7068. Li, G., 2014. Information sharing and stock market participation: evidence from extended families. Rev. Econ. Stat. 96, 151–160. Luster, T., Bates, L., Fitzgerald, H., Vandenbelt, M., Key, J.P., 20 0 0. Factors related to successful outcomes among preschool children born to low-income adolescent mothers. J. Marriage Fam. 62, 133–146. Magnuson, K.A., and Duncan, G.J., 2002, Parents in poverty. In: Handbook of Parenting, Lawrence Erlbaum Associates, 4, 95–121. Malmendier, U., Nagel, S., 2011. Depression babies: do macroeconomic experiences affect risk taking? Q. J. Econ. 126, 373–416. Meier, S., 2007. A survey of economic theories and field evidence on pro-social behavior. Econ. Psychol.: Promis. New Cross-Discipl. Field 51–88. Moore, K.A., Morrison, D.R., Greene, A.D., 1997. Effects on the children born to adolescent mothers. Kids Having Kids: Economic Costs and Social Consequences of Teen Pregnancy. Urban Institute Press, Washington DC. Nilsson, J., 2008. Investment with a conscience: examining the impact of pro-social attitudes and perceived financial performance on socially responsible investment behavior. J. Bus. Ethics 83, 307–325. Patterson, C.J., Hastings, P.D., 2007. Socialization in the context of family diversity. Handbook of socialization: Theory and research, 328–351. Guilford Press, New York. PI, 2012, European SRI study, https://www.eurosif.org/sri- study- 2012/, Accessed: 2017-0830. Renneboog, L., Ter Horst, J., Zhang, C., 2008. Socially responsible investments: institutional aspects, performance, and investor behavior. J. Bank. Financ. 32, 1723–1742. Riedl, A., Smeets, P., 2017. Why do investors hold socially responsible mutual funds? J. Financ. 72, 2505–2550. Shive, S., 2010. An epidemic model of investor behavior. J. Financ. Quant. Anal. 45, 169–198. Statman, M., Fisher, K.L., Anginer, D., 2008. Affect in a behavioral asset-pricing model. Financ. Anal. J. 64, 20–29. Steurer, R., Margula, S., and Martinuzzi, A., 2008, Socially responsible investment in EU member states: overview of government initiatives and SRI experts expectations towards governments. Final report to the EU high-level group on CSR. Tender No. VT/2005/063. Sutter, M., Feri, F., Glutzle-Rutzler, D., Kocher, M.G., Martinsson, P., Nordblom, K., 2018. Social preferences in childhood and adolescence. A large-scale experiment to estimate primary and secondary motivations. J. Econ. Behav. Organ. 146, 16–30. Wallace, B., Cesarini, D., Lichtenstein, P., Johannesson, M., 2007. Heritability of ultimatum game responder behavior. Proc. Natl. Acad. Sci. 104 (15), 631–634. Wilhelm, M.O., Brown, E., Rooney, P.M., Steinberg, R., 2008. The intergenerational transmission of generosity. J. Public Econ. 92, 2146–2156.