Do family ties and structure matter for ethical and moral values?

Purpose – This paper seeks to determine whether family ties and structure correlate with the ethical and moral values that are important underpinnings of economic activities. Design/methodology/approach – The analysis uses data from the World Values Survey (WVS). Given the multilevelnatureofthedatainacross-countrysetting,thepaperutilizesamultilevellinearmixed-effectsmodel withmaximumlikelihoodestimation. Findings – Families with strong ties and those with traditional family structures are less tolerant of unethical conduct and have more restrictive moral values than households where ties are weak and the household is not married. There also appears to be a bi-causal relationship in the data. Originality/value – This paper considers a broad array of values in a cross-country setting and utilizes a multilevel modeling approach that has not been done in studies linking both family ties and structure.


Introduction
There are numerous pathways by which economic institutions affect the economic performance of economies (Chang, 2011;North, 1990). One pathway is through social norms, which are the shared but unwritten beliefs and attitudes that people have about how to behave within societies. Of the many types of social norms, values-that is, the beliefs and judgments of people about what is important in life-are particularly important. Social norms and values are important because they can substitute for as well as complement more formal institutions and rules of behavior, including direct economic incentives, although social norms can also work against them (Pitlik and Rode, 2017). A second pathway is through marriage and the family (Doepke and Tertilt, 2016), including the relevance of family ties and structure (Alesina and Giuliano, 2014;Guttman and Voigt, 2022). Greif (2006, p. 308) notes that "family structure" is central to "one of the most fundamental institutional changes in history [because of] its growth-related implications" and therefore calls for a greater understanding of the relationship between the emergence of corporations and the "nuclear family structure". This paper explores the connection between these two pathways. The specific research question is: do family ties and structure correlate with the ethical and moral values that are underpinnings of economic activity? Previous research has explored how family characteristics and processes moderate or impact the morality and values of children and other family members (Albanese et al., 2016;Prioste et al., 2015). Research has also examined how family characteristics and strengths contribute to economic indicators and outcomes. For example, Giuliano (2011, 2014) and Ermisch and Gambetta (2010) show that family ties negatively correlate with generalized trust in society. Other research shows how family relationships link to economic and cultural institutions and norms Giuliano, 2010, 2015;Enke, 2019;Guttman and Voigt, 2022). As significant as these studies are, few studies have examined the connection between family ties and structure and ethical and moral values, especially in a cross-country setting (see Alesina and Giuliano, 2014). In this context, family ties refer to "the strength and resilience of family loyalties, allegiances, and authority" that exist within family and household relationships (Reher, 1998, p. 203). Family structure refers to the organizational arrangement of family households, such as the marital status of parents and the presence and number of children and other family and non-family members.
To answer our research question, we use the sixth wave of the World Values Survey (WVS) in a cross-country study of 46 countries for which data are available. Dependent variables are indices of the ethical and moral values of the household head. Family ties and structure are constructed from variables provided in the WVS motivated by previous studies that have examined the characteristics of families (e.g. Alesina and Giuliano, 2014;Davis and Williamson, 2020). Controls include individual and country-level variables expected or previously shown to correlate with ethical and moral values. We also use a multilevel modeling approach that is similar to other cross-country studies examining correlates of ethical values of households (e.g. James, 2022;Martin et al., 2012).
This research is important because it supports Grief's (2006, p. 312) call to examine "the dynamic interplay between family structures and institutional development". While it is not our intent to describe how ethical and moral values are taught or transmitted within households or to explain why some families are strong or function better than others, the paper seeks to determine if family ties and structure correlate with values, which in turn can help explain why such values might be weak or strong in society. Although studies have examined the correlation between family ties or structure on some values, such as collectivist versus individualist values (Davis and Williamson, 2020;Prioste et al., 2015), this paper focuses on a broader array of values that are relevant for economic performance.

The relevance of values on economic performance
Values matter for economic performance because they guide, direct or constrain perspectives and behavior and form the basis for determining right and wrong. For example, North (1994, p. 58) says that some values, such as "honesty, integrity, and hard work" are "critical determinants of the costs of transacting in complex political and economic exchange". Research on the reasons and means by which values affect economic performance has progressed along several lines. One subsumes values within a broader consideration of culture (e.g. Giavazzi et al., 2019;Nikolaev and Salahodjaev, 2017). Another considers "the extent to which key values are shared in society" (Beugelsdijk and Klasing, 2016, p. 523). Accordingly, organizations and societies characterized by shared values and consistency of value systems perform better than those in which values are diverse (e.g. Ren, 2010). One way values do this is by helping societal members "overcome the free-rider problem in the pursuit of socially valuable activities" (Guiso et al., 2011, p. 419). Another way is that shared values foster trust, whereas the diversity of values correlates with lower generalized trust (Beugelsdijk and Klasing, 2016).
An important literature considers the array of values within and across societies, including how they evolve and interact with micro-and macro-level constructs (see Halman   IJSE  50,4 and Gelissen, 2019). There are two types of values that are relevant to the economic performance of societies. First, ethical values are the standards and rules relating to the consumption, exchange, property rights and cooperation, including the support of trust and trustworthiness (e.g. Uslaner, 2002). Important ethical values include keeping promises, being trustworthy, protecting property rights and respecting the rule of law. Much has been written on the relationship between ethical values and the well-being of society, especially with respect to economic performance. Adam Smith and other classical political economists made this argument (Zouboulakis, 2010). Rose (2011) explained how ethical values rather than feelings of empathy and sympathy for others are a necessary underpinning to trust and the development of effective economic institutions. There is literature documenting the relationship between indicators of ethical values and economic outcomes (e.g. James, 2015;Nunner-Walker, 2007).
Second, moral values include a broader set of considerations, such as beliefs about life and death and sexual practices (Beugelsdijk and Klasing, 2016). Scholars have noted that economic activity can impact traditional moral values and beliefs (e.g. Bowles, 1998). However, moral values can also impact economic performance. One way is that changes in moral values affect social capital. For instance, Carden (2009) suggested that changes in sexual mores can produce social unrest, which then impacts the formation and development of social capital needed for economic development. A second way is that changes in beliefs about sexuality, abortion and other morally-relevant behaviors can affect the overall culture of societies (Giavazzi et al., 2019;Greenwood and Guner, 2010), which in turn impacts economic performance.

The role and relevance of family on economic performance
Family can affect economic performance directly and indirectly. The direct effects of the family on economic performance include the provision of labor and its purchases of goods and services to meet household needs. Marriage and cohabitation relationships create new households and, as such, foster independent economic units that interact with other economic agents within the economy. While the nature and measurement of household and union formation are changing and are a source of debate among scholars (Sassler and Lichter, 2020), research considers how various facets of marital and household structures affect economic and related indicators. For example, Guttman and Voigt (2022) test Todd's (1985) claim that societies dominated by nuclear families differ significantly from extended family relationships. The authors find that families characterized by extended relationships tend to exhibit greater attitudes toward racism and preferences for economic inequality, but also support a greater level of rule of law, among other findings (some of which are contrary to Todd's assertions).
The indirect effect of family on economic activity is by orienting and transforming children and other household members into economic agents and by socializing children into roles and teaching them the norms and values that matter to society, such as cooperating, respecting the property of others and exercising self-restraint. Scholars have attempted to link the characteristics of families with cultural and social norms and values. An important development here focuses on the role of family ties (Alesina and Giuliano, 2010. For example, Davis and Williamson (2020) report that family ties are negatively correlated with various indicators of individualism, while Giuliano (2011, 2014) show that family ties and trust are negatively correlated, suggesting that family strength substitutes for rather than complements trust in society. In related work, Son and Feng (2019) show that networks with higher levels of family ties exhibit greater within-network trust than low tie networks, but the effect on generalized trust is inconclusive. However, Enke (2019) examines the relationship between family structures and ties and various cultural, social and moral Do family ties and structure matter? systems across societies. He finds that tight kinship structures correlate positively with some features of a society's moral system, such as the relative importance of communal moral values, suggesting that improved moral systems could mediate the negative relationship between family ties and trust.

The link between family characteristics and ethical and moral values
Many studies view the family unit as the most vital source of moral influence (White and Matawie, 2004). During the early years of a child's life, parents play a crucial role in fostering or inhibiting the process of moral internalization (Hardy et al., 2008). In fact, parents who tend to be more ethical provide better moral ideas and reasoning than parents who are less ethical (Parikh, 1980). Given this context, how might family ties and family structure correlate with the ethical and moral values of family or household members? Insights from the literature do not suggest a clear answer. Given the importance of ethical and moral values generally and the relevance of family ties and structure in economic analyses, and because few studies examine these considerations empirically in a cross-country setting, we examine the relationship between family ties and structure and ethical and moral values.

Conceptual framing, methods and procedures
Data are from the sixth (2010-2014) wave of the WVS (Inglehart et al., 2014) [1]. The WVS compiles responses from face-to-face interviews with citizens in many countries around the world. Interview subjects were randomly selected and stratified by region and degree of urbanization. For this study, the sample consists of 65,218 adults from 46 countries aged 18 years and older. There are no data on the values, beliefs and attitudes of all individual family or household members in the WVS. Therefore, our dependent variable is proxied by the values of the survey respondent or household head. Research suggests there is a strong connection between the values of parents and those of their children so that the values of a household head can proxy for those of family members (Albanese et al., 2016;Degner and Dalege, 2013;White and Matawie, 2004). Table 1 shows the variables from the WVS used to construct indices of ethical and moral values, family ties and other variables used in the analysis. Our measures for values are based on previous research suggesting that morally debatable behaviors can be grouped into two distinct types-those relating to attitudes about legal and illegal behaviors-that is, ethical values-and those relating to attitudes about sexuality as well as life and death-that is, moral values (see Katz et al., 1994). Previous studies have used the WVS to construct indices of these two variables (e.g. Beugelsdijk and Klasing, 2016;Marozzi, 2021;Storm, 2016). Accordingly, we construct variables for ethical values and moral values from responses to this question: "Please tell me for each of the following actions whether you think it can always be justified, never be justified, or something in between". For ethical values, five actions were presented to respondents that are presumed to reflect unethical conduct, such as stealing property and cheating on taxes. For moral values, five items presented to respondents deal with the justifiability of conduct relating to life and sexuality, such as abortion and sex before marriage. For both types of values, responses were reverse coded. The five components for ethical values were combined, and the five components for moral values were combined, respectively, by extracting the first principal component of all responses to those variables in the dataset. An increase in the variable representing ethical values indicates a greater intolerance of unethical behavior (see, for instance, James, 2022;Martin et al., 2012), while an increase in the moral values variable represents more restrictive rather than permissive beliefs about life, death and sexual behavior (see, for instance, Marozzi, 2021;Studlar and Burns, 2015)  We follow Alesina and Giuliano (2014) and Davis and Williamson (2020) in constructing a measure of family ties using variables from the WVS that attempt to capture the strength of family relationships and connections. One of the three variables utilized by these authors is in the sixth wave of the WVS [3], which asks respondents to indicate how important or not important family is. In addition, three other variables in the WVS reflect the concept of family strength. The first asks respondents to state how much they trust family. The second asks how important tradition is to the respondent and how important it is to follow the customs handed down from one's family or religion. The third asks respondents to state how much they agree with the statement that an important goal in life is to make their parents proud. We combined the individual-level responses to these four variables by extracting the first principal component to create an index variable we label family ties.
Although there are different ways of operationalizing family structure, we utilize two methods. First, numerous studies proxy family structure by the marital status of the survey respondent (see, for instance, McLanahan and Percheski, 2008;Williams and Baker, 2021). Thus, three categories are identified: the respondent is married or widowed; the respondent is living together as married; or the respondent is divorced, separated, or single. Second, we distinguish between nuclear families (where children are emancipated from their parent's household) and extended family systems (where adults with children live in their parent's home) (see, for instance, Enke, 2019;Guttman and Voigt, 2022). Two variables in the WVS allow us to determine if the household consists of an extended relationship-a question asking if the respondent lives with their parents and a question asking about the number of children the respondent has. Respondents living with a parent and indicating they have one or more children are coded as being an extended household; otherwise, they are presumed to be a nuclear household.
Consistent with other studies on ethical judgments, individual control variables include the religiosity of the respondent, the number of children in the household, the degree to which respondents are happy, the respondent's age, gender, income level and level of education (e.g. Craft, 2013;Pan and Sparks, 2012). Observations with missing values were replaced with variable means, and country-level means were differenced from individual-level covariates. Country-level controls, also utilized by Martin et al. (2012) and James (2022) reverse causation, we use data from 2008 rather than later years. The WGI ranges from À2.5 (weak institutions) to 2.5 (strong institutions), while the CPI ranges from 0 (highly corrupt) to 10 (highly clean). Both indexes were rescaled to range from one to 10 and then averaged to create a country-level index of institutional quality.
Because of the multilevel structure of the data, in which both individual-and country-level effects can correlate with the dependent variables, we use a multilevel linear mixed-effects model with maximum likelihood estimation. The estimated equation takes the following form: where Y ij is the dependent variable for ethical values or moral values for respondent i in country j; family tiesij and family structureij are the variables for family ties or family structure (i.e. marital status and nuclear or extended family); X ij represent individual-level controls; W j is a country-level predictor; and e ij þ μ j are individual-level and country-level error terms, respectively (see Snijders and Bosker (2012), for additional details). Table 2  Models 1a and 1b in Table 2 introduce the mixed-effects analysis and are presented with a random intercept and without covariates in order to calculate the intraclass correlation coefficient [7]. The percent of variability of values accounted for by country-level effects is 18.6% for ethical values and 37.7% for moral values, suggesting there is clustering in the data at the country level, especially in the case of moral values, thus confirming the need for a multilevel regression analysis. Models 2a and 2b show the results of the individual level analysis, which indicate that family ties are positively correlated with the ethical values and moral values of the household head, consistent with the OLS findings. In other words, households with stronger family ties are less accepting of unethical conduct and tend to hold more restrictive attitudes relating to conduct affecting life, death and sexuality than households with lower family ties. Table 3 introduces the country-level control for institutional quality as well as interaction effects between family ties and family structure. The introduction of the country-level control does not affect the relationship between family ties and ethical and moral values. While the country-level control has a positive and substantive effect on ethical values (see Model 3a), it has a significantly negative effect on moral values (see Model 3b), suggesting that ethical values tend to be higher on average in countries with strong economic institutions while moral values tend to be lower as institutional quality increases. Similarly, including the institutional-level control does not change the relationship between the family structure variables (marital status and extended family relationships), which remain negative and significant except for the correlation between extended families and moral values.    institutional quality works against family ties in the ethical values model but works with family ties in the moral values model. Models 5a and 5b add interactions with the family structure variables. Institutional quality has a negative interaction with family households where the respondent is not married for both the ethical values and moral values models. The interaction between institutional quality and the presence of an extended household relationship is oppositive in both models; in the case of ethical values, the relationship is negative (Model 5a) but positive in the case of moral values (Model 5b).

Main analysis
Overall, the effect of the control variables is not surprising. Religiosity, happiness and age have a consistently positive effect on ethical and moral values, while income has a consistently negative effect on the dependent variables. Being female and having postsecondary education is positively correlated with ethical values but negatively correlated with moral values.

Robustness check on dependent variables
Previous cross-country studies examining correlates of attitudes toward morally debatable behavior typically construct the dependent variable as the average of individual components from the WVS (e.g. James, 2022;Martin et al., 2012). For comparison purposes, Table 4 provides the results of our multilevel analyses using this (averaged) method of constructing the dependent variables of ethical values and moral values (reported as Models 6a and 6b). However, Marozzi (2021) states that such studies may not be sufficiently rigorous if constructed indices are computed as unweighted averages of individual components. Thus, we provide an additional check of our results by calculating new measures for ethical values and moral values as weighted averages, using the procedure outlined in Marozzi (2021). Models 7a and 7b in Table 4 provide the results of this analysis. All models presented in Table 4  Model fit AIC 239,554 À61,565 256,652 À70,691 Note(s): All models estimated with random intercept and fixed slopes, including control variables used previously. In Models 6a and 6b, the dependent variables are the average of each of the five components comprising ethical values and moral values respectively. In Models 7a and 7b, the dependent variables are weighted using the technique described in the text. For marital status, married is the excluded category. Standard errors in parentheses. N 5 65,218. *** 5 p < 0.01, ** 5 p < 0.05, * 5 p < 0.10  Tables 2 and 3.

Bi-causality analysis
Although we have established a correlation between family ties and structure and ethical and moral values, it is useful to determine if there is causality among the explanatory and dependent variables. For example, do households with stronger family ties or who live as nuclear families cause ethical and moral values to be higher, or do individuals who are less tolerant of unethical conduct and who have more restrictive than permissive moral values have families with strong ties or avoid extended family relationships? Or is the causal connection more complicated? Some research suggests a more complicated relationship. For example, while families impact economic conditions, there is also literature showing how economic factors affect family structure and functioning (White and Rogers, 2000) and relevance (Guttman and Yacouel, 2007), including in cross-country settings (e.g. Majeed and Kanwal, 2019).
In order to tease out causal pathways, we present two analyses in Table 5. The first utilizes two-stage least squares techniques, and the second employs a three-square regression approach. Because our proxies for family structure are dummy variables, we consider only the potential bi-causal relationship between family ties and ethical and moral values. The results in Table 5 show that not only do family ties continue to have a positive and significant effect on ethical and moral values, but these values also correlate with higher levels of family ties. Thus, there appears to be a bi-causal relationship between family ties and ethical and moral values.  for ethical values and moral values, whether respondent sees themself as part of the nation; for family ties, whether respondent trusts people they know personally. Analyses include country dummies. For marital status, married is the excluded category. Standard errors in parentheses. N 5 65,218. *** 5 p < 0.01, ** 5 p < 0.05, * 5 p < 0.10 Table 5. Bicausality analysis using two-stage and three-stage least squares regressions Do family ties and structure matter?

Discussion
This study has demonstrated that family ties correlate positively with the ethical and moral values of individuals in a cross-country setting, even after controlling for individual and country level effects. However, previous researchers have documented that family ties correlate negatively with generalized trust. These findings might seem counterintuitive given that Rose (2011) claims ethical and moral values are essential for the development of trust in society. How can family ties correlate positively with ethical and moral values while at the same time correlate negatively with trust, if ethical and moral values are in turn necessary for trust? There are two possible reasons. First, strong family ties foster an enclosed sense of security within households, so that outside the bounds of these relationships trust is jeopardized (Alesina and Giuliano, 2014;Ermisch and Gambetta, 2010). In contrast, weak family ties provide a stronger incentive for family members to get along with strangers in order to fulfill the needs that are not being met by family, thereby contributing to generalized trust. However, Enke (2019) concludes that improved moral systems could mediate the negative relationship between family ties and trust. If the family fosters within network security as well as socializes members on cooperative behavior outside the family through the transmission of ethical and moral values, such as those evaluated here, then the aggregate nature of cross-country studies may not fully capture the idiosyncratic dynamics of how values, family characteristics, and other factors interact at the household level. Second, James (2015) shows that the generalized morality of citizens has a more pronounced effect on trust when economic institutions are weak than when they are strong. Thus, accounting for institutional quality may partly resolve the apparent counterintuitive relationships among generalized trust, family ties and ethical and moral values. There appears to be some empirical support for this conjecture. Table 6 reports results of a simple logistic regression with individual trust as the dependent variable for countries with low (e.g. Azerbaijan, Haiti), moderate (e.g. China, Thailand) and high (e.g. Australia, Sweden) quality of economic institutions [8]. Table 7 reports results where models 3a and 3b are re-estimated for countries with low, moderate and high institutional quality. The tables exhibit several interesting findings. First, the effect of family ties on trust is not consistently negative. Rather, it is neutral for countries with low institutional quality, negative for countries of moderate institutional quality and positive for countries with high institutional quality (see the model for family ties in Table 6). Second, ethical values have a negative effect on trust for low quality countries, no effect in the case of moderate countries and a positive effect in countries with good institutions. However, the effect of moral values on trust is negative for all levels of institutional quality. Third, consistent with the interactions shown in Models 4a, 4b, 5a and 5b in Table 3, improved institutional quality reduces the effect of family ties on ethical values and increases modestly the effect of family ties on moral values (see Table 7). Together these findings reveal that in countries with strong economic institutions, family ties can reinforce ethical values and trust, suggesting that family ties, generalized trust and ethical values complement economic institutions in some contexts. In contrast, ethical and moral values, family ties and generalized trust can interfere with each other when economic institutions are relatively weak, as implied by explanations for the negative relationship between family ties and trust, which might explain in part why economic institutions are weak. Given findings by Alesina and Giuliano (2014) that family ties negatively correlate with indicators of institutional quality, more work is clearly needed to understand better the extent to which family ties either complement or interfere with trust and institutional quality.  Note(s): All models estimated with random intercept and fixed slopes. Controls are the same as those listed in Table 3 models 3a and 3b. Standard errors in parentheses.
Countries with low (high) institutional quality had an index value that was at least one standard deviation below (above) the mean. N for low quality institutions is 13,418; N for moderate quality institutions is 37,713; N for high quality institutions is 14,087. *** 5 p < 0.01, ** 5 p < 0.05, * 5 p < 0.10 6. Conclusions Do family ties and family structure correlate with the ethical and moral values that are important underpinnings of economic activity? We find evidence in the affirmative in our cross-country study. In particular, we find that households with strong family ties, married or widowed households and households characterized by a nuclear family structure are less tolerant of unethical conduct and have less permissive moral values when compared to single and living together households and households with weaker levels of family functioning. The results are generally robust after controlling for individual and country-level effects and a potential for bi-causality. Our findings that family ties and structure correlate with ethical and moral values underscore the continuing importance of the family as an economic institution in society. The creation of strong families and the transmission of ethical and moral values within households are crucial and support the indirect way in which families impact the economic performance of societies-through the orientation and transformation of children into productive and ethical economic agents. This suggests that a strategy for strengthen the ethical and moral values of society generally and for encouraging the ethical judgments and behaviors of individuals should focus in part on strengthening traditional family units and the relationships among family members. Furthermore, if future scholarship confirms evidence that family characteristics correlate with values that promote economic activity, then a strategy for strengthening promoting economic performance should focus in part on strengthening traditional family units.
If family structure and ties matter for economic performance, then there might be a reverse effect whereby the breakdown of the nuclear family has real economic consequences. There appears to be evidence for this conjecture (e.g. Iceland, 2003). For example, Aguirre (2001) showed how rising divorce, illegitimacy and single and no parent households are linked to greater welfare costs, higher opportunity costs from money wasted in family courts, and deteriorating human capital development of children, each of which correlates with lower economic growth and development. Our paper points to an additional avenue by which a deterioration of the structure and a weakening of family ties can have adverse economic impacts on societies.
Notes 1. Data are used from the sixth rather than the more recent seventh wave of the WVS because too many countries in the seventh wave had missing variables or data.
2. In stating how one might interpret what an increase or decrease in the moral values variable represents, we do not make or imply any judgment on the morality of any of the specific components used in constructing that variable.
3. The missing variables reflect the duties of parents and children and the love and respect for one's parents. 7. The intraclass correlation coefficient is calculated by dividing the intercept error variance by the total error variance (Snijders and Bosker, 2012). The Akaike information criterion (AIC) is used to Do family ties and structure matter?
compare regression models. Because the absolute value of the AIC is meaningless, when comparing comparable models (e.g. Model 2a relative to Model 1a or Model 2b relative to Model 1b), the model with the lower AIC is the better model (see Baguley, 2012, p. 402).
8. Countries were identified as having low (high) quality economic institutions if their institutional quality index was at least one standard deviation below (above) the mean.