On Rule of Law, Civic Virtues, Trust, and Happiness

In literature, there is a long and ongoing dispute over whether free market institutions encourage or discourage civic virtues. On the one hand, the so-called doux commerce thesis states that free market institutions have a favorable influence on civic virtues, such as honesty, respect for private property, and fair dealings. This idea goes back to, amongst others, Adam Smith who famously argued that where commerce is introduced, people are more faithful to their word. Smith also believed that civic virtues foster human happiness. The policy implication would be that free market institutions increase happiness by stimulating civic virtues. On the other hand, the so-called self-destruction thesis states that free market institutions are inimical to the civic virtues, which would mean that free markets decrease human happiness by crowding out civic virtues. Although the debate on free market institutions and civic virtues is very relevant to policy makers, empirical evidence on these opposing propositions is still scarce. In this paper we test both relationships on a sample of 212,431 individuals from 80 countries by using data from the World Values Survey and the European Values Survey over the 1990–2020 period. We focus on one important dimension of free market institutions: rule of law. We find that civic virtues are positively related to rule of law and that happiness increases with civic virtues. In addition, civic virtues indirectly increase happiness by stimulating trust. In addition, trust is directly positively related to rule of law, indicating that both formal institutions (rule of law) and informal institutions (civic virtues) affect trust. An analysis of all the effects of rule of law on happiness shows, however, that the indirect effects through civic virtues and trust are of relatively minor importance in comparison to its total effect.


Introduction
In the literature on market institutions, virtues, and human happiness, two debates have attracted the attention of philosophers and economists. First, there is a long and ongoing dispute over whether free market institutions encourage or discourage civic virtues (Fourcade & Healy, 2007;Storr & Choi, 2019). Civic virtues can be defined as "those social norms, ethical commitments, and other-regarding preferences that facilitate the workings of the institutions advocated by liberals." (Bowles, 2011, p. 50) Civic virtues or "bourgeois" virtues do not include so-called "higher virtues," such as beneficence and magnanimity (Herzog, 2014), but virtues that are commonly held to be among the cultural foundations of a well-functioning liberal order, such as voluntarily paying taxes and contributing to public goods, respect for private property, honesty, and fair treatment. For most of the eighteenth century, market relations were assumed to make people more cordial. As Montesquieu states: " […] wherever there is commerce, manners are gentle." (Cited in Hirschman, 1982Hirschman, : 1464 Adam Smith largely sides with this view (Graafland & Wells, 2020). In his Lectures on Jurisprudence ( § 17) Smith (1763) states: "whenever commerce is introduced into any country, probity and punctuality always accompany it. . .. Of all the nations of Europe, the Dutch, the most commercial, are the most faithful to their word." Fourcade and Healy (2007) call this the liberal dream. Today, McCloskey (2006) is one of the best-known defenders of this so-called doux commerce thesis that commerce fosters the civic virtues. She argues that free markets nurture several bourgeois virtues, such as integrity, honesty, trustworthiness, enterprise, respect, modesty, and responsibility. The antithesis to the doux commerce thesis is the so-called parasitic liberalism (or self-destruction) thesis, which states that market institutions favor a cultural learning process that is inimical to the virtues needed for liberal institutions to function well (Bowles, 2011). Markets will crowd out traditional institutions that sustain the civic virtues, such as family life and religious communities, while endorsing the pursuit of self-interest.
The second debate that is related to, but has to be distinguished from, the debate between defenders of the liberal dream and the parasitic liberalism thesis, concerns the effect of civic virtues on human or societal happiness. In classical virtue ethics, virtues enable people to become happy. In Aristotle's view, excellence, or virtue, is an essential condition of happiness. As a virtue ethicist, Adam Smith also believed that virtues enhance human and societal happiness. For example, in his Theory of Moral Sentiments (TMS), he stated that "By acting according to the dictates of our moral faculties, we necessarily pursue the most effectual means for promoting the happiness of mankind, and may therefore be said, in some sense, to co-operate with the Deity, and to advance as far as in our power the plan of Providence." (TMS, III.5.7) Smith believed that it is not great fortunes that make people truly happy in the first place, but rather simpler pleasures such as knowing one has acted in a proper way (Rasmussen, 2006). In contrast, Mandeville argued in his Fable of the Bees that the practice of private virtues leads to societal disaster and therefore, ultimately, to human unhappiness,

3
On Rule of Law, Civic Virtues, Trust, and Happiness because without vices a nation will not prosper. In his view, private vices, like pride and vanity, have built more hospitals than all the virtues put together.
Empirical evidence on the relationships between free market institutions, civic virtues and happiness is still very scant. Previous research has mostly focused on studying how formal institutions affect life satisfaction (Veenhoven, 2000;Ovaska & Takashima, 2006;Gropper et al., 2011;Graafland & Compen, 2015), without reference to informal institutions, such as culture and morality. The first attempt to come close is Graafland and Compen (2015). They showed that generalized trust (defined as trusting people you do not know) positively mediates the influence of the quality of the legal system (in short, rule of law) on happiness. The ingredients of such a legal system are rule of law, security of property rights, an independent and unbiased judiciary, and impartial and effective enforcement of the law. It is a key dimension of the so-called economic freedom. Individuals have economic freedom when their property rights are well protected and when "they are free to use, exchange, or give their property as long as their actions do not violate the identical rights of others" (https:// www. frase rinst itute. org/ econo mic-freed om/ econo mic-freed om-basics). Economic freedom comprises four other subdimensions (small size of government, sound money, free trade, and freedom from government regulation), but it is particularly rule of law that has been found to affect trust (Knack & Keefer, 1997;Zak & Knack, 2001;Berggren & Jordahl, 2006;Graafland & Compen, 2015). Indeed, it is well known that generalized trust is affected by the quality of the law, as it makes it harder for people to deceive others (Beugelsdijk, 2006). In our paper, we therefore focus on this dimension of free market institutions. However, it is likely that trust is also influenced by civic virtues, such as honesty and justice. If many members of a society are reliable, people living in that society will be more inclined to trust others. Besides rule of law, it is therefore likely that civic virtues affect happiness by increasing trust.
The research questions that this paper focuses on are therefore threefold: 1 How does rule of law influence civic virtues? 2. To what extent is trust affected by rule of law, as an indicator of formal institutions, and by civic virtues, as an indicator of informal institutions? 3. How do civic virtues influence happiness? To answer these research questions, we will test the relationships between rule of law, civic virtues, trust, and happiness using data from the World and European Value Studies between 1990 and 2020. In what follows, we first present the hypotheses. Section 3 describes the methodology of our empirical research. Section 4 reports the results of the empirical analysis. In Section 5 we summarize and discuss the main findings.

Rule of Law, Trust and Civic Virtues
Previous research has shown that free market institutions may increase trust (Helliwell & Barrington-Leigh, 2010;Oishi et al., 2011;Graafland & Compen, 2015). Trust can be defined as a psychological state comprising the intention to accept vulnerability, based on positive expectations of the intentions or behavior of another person (Rousseau et al., 1998). It is an expectation of the reliability of others to live up to their obligations, the predictability of their behavior and the fairness of their actions (Beugelsdijk, 2006). The literature distinguishes between particularized trust and generalized trust (Berggren & Jordahl, 2006). The former entails trusting people you know, the latter trusting people you do not know. Most economic literature uses the concept of generalized trust. It constitutes an evaluation of individuals' attitudes towards their surroundings and the community of which they are a part and affects how people interact with each other. It fosters so-called bridging social capital, which refers to the connections between dissimilar people (Jen et al., 2010). Generalized trust is a prerequisite for a well-functioning economy, as it enables people to engage in trade and other constructive activities, lowers transaction costs, and enhances stability and peaceful interaction among members of a society.
Research by Knack and Keefer (1997), Zak and Knack (2001), Berggren and Jordahl (2006), Graafland and Compen (2015), and Sapsford et al. (2019) has shown that the rule of law, which is key in the concept of free market institutions, enhances trust. The rule of law refers to a legal system that is durable, complies with the rules and is accountable. Rules that have been established and announced in advance make it possible for individuals to predict with a fair degree of precision how the government will use the coercive forces at its disposal in certain circumstances. In this way, everyone is free, within the known rules of the game, to pursue one's personal purposes in the certainty that the powers of the government will not be used to destroy the activities of the individual (Hayek, 1976). A well-developed legal system that protects property rights and contracts not only reinforces a climate of trust in the government, it also increases trust in other people. The institutions of law and order detect and punish people who break contracts or steal from others. If these institutions work well, then people will have reason to believe that the chance of people getting away with such noncooperative behavior is small. Hence, people will have good reason to refrain from untruthful behavior and, as a result, most people will believe that most other people can be trusted (Rothstein, 2000). This leads to our first hypothesis: Hypothesis 1 Trust is positively related to rule of law.
The trust generated by rule of law is not necessarily evidence of the presence of civic virtues. People may act in a trustful way out of fear of punishment by the law. The reason that people act faithfully is then a matter of self interest, because cheating one's trading partners may generate more loss than profit in the long term. Civic virtues, such as honesty and reliability, are intrinsically motivated. This follows from the definition of virtue as an acquired disposition that is socially valued as part of the character of a morally good human being, exhibited in the person's habitual behavior (Velasquez, 1998). Virtues go deeper than mere behavior and habits (although this may be important for their development) as they are constitutive of how a person perceives situations and reasons for actions. The reason why the person acted is crucial: persons who act in a virtuous way but intensely dislike considering the interests of others should be judged deficient in virtue.

3
On Rule of Law, Civic Virtues, Trust, and Happiness Although generalized trust is likely to be connected to civic virtues at the macro level (see the next Section), at the micro level it is conceptually different. As noted above, people may trust others, not because those others are intrinsically motivated to be honest, but because they are extrinsically motivated because of fear of punishment or reputation harm. The difference can also be illustrated by the example of a vicious person who operates in an environment where more people have internalized civic virtues, such as reliability and honesty. Because of the reliability of other people, this person is likely to trust others, without being virtuous him or herself. 1 Notwithstanding the conceptual difference between trust and civic virtues, we expect that the influence of rule of law on civic virtues is similar to its influence on trust. We base this argument on habit formation. According to Aristotle (Ethica Nicomachea) (n.d.), virtues are learned through a tedious process of trial and error. They become internalized through the regular repetition of certain (good) actions and thinking, often originally motivated by extrinsic motives. When we repeatedly are honest towards others out of self-interest, we develop the habit of being honest, and gradually this habit may crowd in and develop into the civic virtue of honesty. Once the habit of being honest towards others is internalized and becomes a civic virtue, honesty stems from a genuine intrinsic motivation (Paganelli, 2010). Extrinsic motives will therefore slowly crowd in and become internalized. Relations with others that were entered into as means to economic ends become transformed into ends in themselves. Besides habit formation, Aristotle also acknowledges the value of the external pressure from a good and reasonable government to induce virtues. The rule of law may provide cues for appropriate moral behavior, like being reliable, speaking the truth, and keeping your promises, and trigger moral engagement (Bowles & Polania-Reyes, 2012). Governments are, in fact, important institutional players with the ability to influence social norms, values, and societal expectations on behavior (Weaver et al., 1999), thereby enhancing intrinsic motivations and civic virtues. This leads to our second hypothesis: Hypothesis 2 Civic virtues are positively related to rule of law.

Civic Virtues and Trust
Whereas trust may depend on the quality of formal institutions such as rule of law, as discussed above, it will also be informed by the civic virtues of people in the environment in which an individual agent operates. Because they are intrinsically motivated, virtuous people keep their promises even when it would be profitable to break them. The more people in an agent's social environment exhibit civic virtues, the more often this agent confronts honest behavior, and the more likely it is that (s) he trusts others. Moreover, as virtues lead to cross-situational consistency in people's behavior (Alzola, 2008;Bright et al., 2014), it might reduce uncertainty, thus further increasing trust. Also, empirical research has shown that virtues are important antecedents of trust (Hendriks et al., 2020;Fleckenstein, 2008;Morgan & Hunt, 1994). For example, Van Dalen and Henkens (2018) found that trust in Dutch pension providers depends on their perceived integrity, competence, and benevolence. Van Esterik-Plasmeijer and Van Raaij (2017) found that integrity plays a key role in explaining trust in Dutch banks. Hence, we hypothesize that: Hypotheses 3 An agent's trust is positively related to civic virtues of people in the agent's social environment.

Civic Virtues, Trust and Happiness
According to Adam Smith, knowing that one has acted in a praiseworthy manner gives peace of mind (Rasmussen, 2006): "To be amiable and to be meritorious; that is, to deserve love and to deserve reward, are the great characters of virtue… The consciousness that it is the object of such favorable regards, is the source of that inward tranquility and self-satisfaction with which it is naturally attended, as the suspicion of the contrary gives occasion to the torments of vice." (Smith, 1759: iii.i.7). This also holds for civic virtues, such as honesty in economic affairs: "In all the middling and inferior professions, real and solid professional abilities, joint to prudent, just, firm, and temperate conduct, can very seldom fail of success…. honesty is the best policy, holds, in such situations, almost always perfectly true." (TMS, I.ii.3.5) Hence, Smith held that civic virtues, such as justice and honesty, contribute to the individual's well-being. Smith's perception that civic virtues foster life satisfaction is also evident in his Wealth of Nations. In their semantic-network approach, Graafland and Wells (2020) found that the majority (over 90%) of 240 citations in which Smith relates human flourishing to virtues, the association is positive. This shows that Smith clearly disagreed with Mandeville.
Previous empirical research provides support for the positive relationship between civic virtues and happiness. Doing well is inherently rewarding. An abundant literature shows that prosocial behavior makes a person happier (Dunn et al., 2008). For example, on a sample of US and Swiss citizens, Peterson et al. (2007) and Abasimi and Xiaosong (2016) found that fairness and honesty are significantly related to life satisfaction. Furthermore, Wang and Hackett (2016) provided empirical evidence that virtuous leaders have higher hedonic well-being. Therefore, our fourth hypothesis is: Hypothesis 4 Happiness is positively related to civic virtues.
Literature has argued and shown that trust also affects life satisfaction (Helliwell, 2003(Helliwell, , 2006Bjørnskov et al., 2007Bjørnskov et al., , 2010Oishi et al., 2011;Graafland & Compen, 2015;D'Agostino et al., 2019;Piao et al., 2021). People do not only enjoy satisfaction from doing the right thing themselves but are also more happy if they feel that others can be trusted. Trust is a precondition for social order and social cohesion, without which many forms of social interaction are much more difficult (Helliwell & Putnam, 2004;Bjørnskov, 2007). People benefit from living in an environment where other people can be trusted. Helliwell and Barrington-Leigh (2010) explored the determinants of the respondents' sense of belonging to their communities and found that measures of trust explain the extent to which one feels a strong sense of belonging, which in turn increases subjective well-being. Trust reduces information and transaction costs and thus lowers the complexity and uncertainty of decision making in everyday life. Furthermore, trust is likely to increase health, because individuals will be less healthy the greater the lack of social cohesion in a country. Empirical research by Berggren and Jordahl (2006) and Jen et al. (2010) supported a positive relation between trust and health indicators. Since life satisfaction has been found to be positively associated with income and health (Helliwell et al., 2016), one would expect that trust increases life satisfaction. Furthermore, trust is often associated with other beneficial outcomes that increase life satisfaction, such as a reduction in violent crimes . This leads to our fifth and last hypothesis: Hypothesis 5 Happiness is positively related to trust.  Figure 1 can be interpreted as a mediation model. Mediation models give insight into how an independent variable (in our model: rule of law) indirectly affects the dependent variable (e.g., happiness) by changing intermediate variables, the socalled mediators. In our model, both civic virtues and trust can mediate the relationship between rule of law and happiness: rule of law may increase civic virtues, which in turn increase happiness, but also raise trust that also increases happiness. Furthermore, civic virtues may also be indirectly related to happiness through trust, as civic virtues may also increase trust (hypothesis 3), which in turn again increases happiness (Hypothesis 5). Trust thus may also mediate the relationship between the civic virtues (of other people) and happiness.

Model
To test the hypotheses, we used the following model: CV denotes civic virtue, RL rule of law, and H happiness. The index i denotes individual level, k country level, and t time period (year). For rule of law, we used 5 and 10 years lagged values (see below). X denotes (m) control variables at the country level, Y (n) control variables at the individual level (see below), and F t time dummies for each wave. As discussed above, we use individual civic virtues in Eqs. (1) and (3) and civic virtues of other people in Eq. (2). Given the limitations of the dataset, the average civic virtues in a country is used as a proxy of the civic virtues of other people as perceived by individuals in that country. To test this model, we therefore need data on the civic virtues of individuals as well as country averages. If we were to estimate the model using country averages of civic virtues, trust and happiness only, we would not be able to identify how the civic virtues of other people affect individual trust.

Data Sources and Measurement
The dataset consists of 212,431 individuals from 80 countries and is taken from the World Values Survey (WVS) and European Value Survey (EVS), complemented by data from the Fraser Institute and other data sources. 2 We have data from six waves (wave 2-7) of the EVS/WVS (2021) between 1990 and 2020, surveys being carried out approximately every five years. For rule of law, we used the sub dimension "legal system and property rights" of the Economic Freedom of the World Index of the Fraser Institute (Gwartney et al., 2017), that ranges from 1 (least free) to 10 (most free). The index is a composite score of 10 policy areas: judicial independence, impartial courts, protection of property rights, no military interference in rule of law and politics, integrity of the legal system, legal enforcement of contracts, low regulatory costs of the sale of  Table 6 presents an overview of the countries included in the dataset and the number of observations per country.
real property, reliability of policy, low business costs of crime, and gender disparity adjustment. 3 The descriptives reported in Table 1 show that there is substantial variation in the index of rule of law, from a minimum of 2.41 to a maximum value of 8.93.
To measure civic virtues, we used indicators from the illegal -dishonesty domain of the Morally Debatable Behaviors Scale (Katz et al., 1994), which have also been used in the WVS/EVS. These indicators have been used in previous literature as "civic norms" (Knack & Keefer, 1997), "civic traditions" (Putnam et al., 1994), "civic culture" (Algan & Cahuc, 2009), or "civic virtues" (Frey, 2012). They reflect the civic mindedness of the respondents, i.e., responsibility towards their society. Selecting the survey questions that are available for most waves of WVS, we could use three questions from the illegal -dishonesty domain of the Morally Debatable Behaviors Scale, one on cheating on taxes, one on accepting a bribe, and one on claiming government benefits to which one is not entitled. In these questions, respondents were asked to what extent cheating on taxes, accepting a bribe in the course of one's duties, and claiming government benefits to which one is not entitled is justifiable. The scale ranges from 1 (never justifiable) up to 10 (always justifiable). 4 Correlation analysis showed that the answers to the three questions are highly correlated. To reduce multicollinearity, we used the mean response to the three questions. 5 Generalized trust is measured by the (binary) response to the question "In general, do you think most people can be trusted?" -a standard approach in literature. The answer options were "Most people can be trusted" (1) and "You can't be too careful" (0). This measure has proved to be a valid and powerful measure of social trust (Özcan & Bjørnskov, 2011).
The dependent variable, happiness, is measured by the survey question from WVS/EVS on life satisfaction ("How satisfied are you with your life?"), measured as a number on a scale from 1 (dissatisfied) to 10 (satisfied). Life satisfaction provides an evaluation of life as it is in comparison to how life should be (Veenhoven, 2000) and is a prominent standard for measuring subjective well-being.

Control Variables
As macro control variables, we included (the natural logarithm of) GDP per capita, political rights, and civil liberty (Veenhoven, 2000;Ovaska & Takashima, 2006), a dummy for South American countries and post-communist countries (Bjørnskov et al., , 2010Helliwell, 2003), and two geographical factors, latitude and landlocked (Bjørnskov et al., 2008). We also included (small) size of the government as a control variable (one of the other subdimensions of economic freedom), as Graafland and Compen (2015) and Oishi et al. (2018) have found that it significantly affects trust and life satisfaction. 6 As control variables at the individual level, we used personal income, age, marital status, religion, gender, and (un)employment status (Bjørnskov et al., 2008;Fahey & Smyth, 2004;Haller & Hadler, 2006;Yeniaras & Akarsu, 2017). Since absolute income statistics are lacking in the WVS data, we used income quintiles (Bjørnskov et al., 2008;Fahey & Smyth, 2004;Haller & Hadler, 2006). Religion is measured by the religiosity of individuals as indicated by the response to the question "How important is God in your life?" (measured on a Likert scale from 1 to 10), and by three dummies for religious affiliation (protestant Christianity, Muslim, and a-religious, with all other religions as the reference) (Bjørnskov et al., 2008;Haller & Hadler, 2006).

Estimation Technique
Since the respondents participating in the World Values Survey and European Values Survey differ per wave, we cannot use a standard panel regression model. But as rule of law is a country-level variable, we should correct standard errors (Peterson et al., 2012;Boudreaux et al., 2019). In the regression analysis we therefore used country clustered standard errors. This estimation method only requires that observations across countries are independent (Primo et al., 2007). Furthermore, we used robust standard errors to rule out biases due to heteroscedasticity.
As social trust has been found to increase economic freedom (Berggren & Bjørnskov, 2017), the results may be subject to endogeneity due to reverse causality. Endogeneity may also arise from unobserved variable bias. For example, as noted in footnote 1, trust may require the virtue of courage. As civic virtues like honesty may also demand some courage, the relationship between public virtues and trust may be due to their relationship with the unobserved virtue of courage. Unfortunately, instrumental variables that prevent simultaneity bias are lacking, as the exclusion restriction is too problematic (Alesina & Giuliano, 2015). To reduce endogeneity caused by the possible reverse effect from public virtues and trust on rule of law, we instrumented rule of law with the five and ten years lagged rule of law (Bhattacharyya & Hodler, 2014). However, this approach does not guarantee excluding all simultaneity bias and we should therefore be careful in the interpretation of the results in terms of causality.

Results
This section presents the bivariate and multivariate regression analysis. In the correlation and regression analysis we used standardized values of all variables (except the dummy variables).

Bivariate Correlation Analysis
The bivariate correlation analysis in Table 2 shows that rule of law is significantly positively related to civic virtues, trust, and life satisfaction. Furthermore, life satisfaction is positively related to civic virtues and trust. Interestingly, the correlation between civic virtues and generalized trust is rather small (0.06). This finding shows that trust and civic virtues are weakly correlated at the individual level and that research on trust cannot be extrapolated to civic virtues. Table 3 and Fig. 2 report the estimation results of the regression analysis.

Regression Analysis
The estimation results in column 1 of Table 3 show that civic virtues are positively related to rule of law. This result provides support for hypothesis 2. Furthermore, civic virtues are positively related to the age of the respondent, marital status, religiosity (particular Muslim and Protestant), and negatively related to gender (indicating that men are slightly more prepared to cheat the tax authorities or to accept bribing) and unemployment. For the macro control variables, only GDP per capita shows a small negative effect, which suggests that income has a slightly corrupting effect on the virtuousness of people.
Column 2 shows that trust is also significantly positively related to rule of law, supporting hypothesis 1. Furthermore, we find a significant positive effect of civic virtues at the country level, which provides support for hypothesis 3. Besides country civic virtues, we find a small but significant effect of the individual civic virtues. This seems to suggest that people that exhibit civic virtues themselves are more inclined to trust others, as the saying goes: "just as the innkeeper is himself, he   trusts his guests". 7 Furthermore, trust is found to be significantly positively related to income class, employment status, and latitude, whereas it is negatively related to unemployment, religiosity, political rights, and post-communist countries.
Columns 3 shows that life satisfaction is significantly positively related to trust as well as to civic virtues, which support hypotheses 4 and 5. 8 The rule of law does not have a significant positive effect. However, we do find that life satisfaction is significantly related to income per capita. Previous research by Graafland and Compen (2015) has shown that GDP per capita may mediate the positive effect of rule of law on life satisfaction. This mechanism is supported if we regress the logarithm of GDP per capita on rule of law, controlling for all control variables. The estimated coefficient of rule of law is equal to 0.62 (p < 0.001). In column 4 we therefore dropped GDP per capita as a control variable. Now, life satisfaction is significantly positively related to rule of law. For the micro control variables, we find that life satisfaction is significantly positively related to income scale, marital status, and religiosity, and negatively related to age and unemployment. Furthermore, people in post-communist countries are less happy. Table 4 presents results of robustness analysis. First, using bootstrapping with 1000 replications hardly changes the results. Only the significance of the effect of country-averaged civic virtues on trust declines (from p value is 0.042 to 0.093). Furthermore, testing if the results depend on the countries included in the sample, we found that if African countries are excluded from the sample, do the estimation results change for trust, as civic virtues at the country level becomes insignificant. Other results testing hypotheses 1-4 are highly robust.

Indirect Effects
To estimate the importance of civic virtues as a channel through which rule of law indirectly affects trust and life satisfaction, we calculated the direct, indirect, and total effects of rule of law on trust and life satisfaction. The results are reported in Table 5.
The indirect effect of rule of law on trust through civic virtues is equal to the multiplication of β 1 and β 2 (see Eqs. 1 and 2 in Section 3.1). The direct effect is equal to the estimated coefficient of rule of law in column 2 of Table 3. Although the indirect effect is significant at p < 0.05, the magnitude is relatively small (0.01). It only explains 5% of the total effect of rule of law on trust (0.20).

3
On Rule of Law, Civic Virtues, Trust, and Happiness Table 4 Robustness analysis Standardized coefficients. * p < 0.05 ** p < 0.01 *** p < 0.001. The results are controlled for all control variables. See Appendix Table 6 for an overview of the countries in the full sample that belong to the regions excluded. a Bootstrapped p-values based on 1000 replications. + p < 0.  Table 3 Dependent variable Direct effect Indirect effect through: Total effect b

Civic virtues Trust
Trust 0.19*** 0.01* a 0.20*** Life satisfaction 0.11** c 0.01** 0.01*** 0.14*** For life satisfaction, the indirect effect of rule of law through civic virtues is also rather small (0.01), although significant at p < 0.01. Also, the indirect effect through trust is relatively small in magnitude (0.01). The large direct effect (0.11) can be largely explained by an indirect effect through income per capita (compare the difference in the coefficients of rule of law in columns 3 and 4 of Table 3). 9

Summary of Main Findings
This paper has studied the relationships between rule of law, civic virtues, trust, and life satisfaction. Using the World and European Values Surveys (covering a period from 1990 to 2020), we accessed a sample of 212,431 observations of individuals living in 80 countries.
This study has four main findings. First, the results show that civic virtues are positively related to rule of law, which is a key economic institution of free markets. This finding suggests some support for the doux commerce thesis that free market institutions enforce civic virtues. We do not find support for the reverse thesis, the so-called self-destruction (or parasitic liberalism) thesis that free market institutions crowd out civic virtues. A possible explanation is the focus on rule of law, as destructive effects of free markets on virtues are more likely to arise from other aspects of free markets, such as intensity of competition. Intense competition may push companies to continuously increasing sales by advertising, fostering a tendency to overconsumption and materialism (Moore, 2005). Competition may also undermine honesty, as the mobility and short-term focus that competition induces crowd out long-term commitments and lead to erosion of social bonds (Putnam, 2000;Sennet, 2000). Furthermore, Shleifer (2004) argued that competition may promote corruption, as corruption gives the briber a competitive advantage. Because of our focus on rule of law only, our analysis does not allow a more general conclusion on the relevance of the doux commerce versus self-destruction thesis. 10 Second, we find that trust is related to a mixture of formal institutions (rule of law) and informal institutions (civic virtues). This suggests that not only the liberal but also the communitarian perspective is relevant. To channel self-interest in socially beneficial directions, liberalism stresses the importance of formal institutions, such as rule of law. In liberal theory, one does not have to be a good virtuous person in the moralistic sense of the term to be a good citizen (Madison, 1998). In the communitarian perspective, virtues that enhance communal relationships are important. Social virtues sustain cooperative behavior in individuals. Our results On Rule of Law, Civic Virtues, Trust, and Happiness indicate that both perspectives are important, and that liberalism and communitarianism are not mutually exclusive (Boettke, 1998).
Third, the results support Smith's view that life satisfaction is positively related to civic virtues. The magnitude of the positive effect of civic virtues on life satisfaction is comparable to the magnitude of generalized trust. This result rejects the view of De Mandeville that private vices rather than virtues generate societal benefits. Indeed, if markets are subject to market imperfections, such as lack of full transparency, virtues become of great importance for the trust required for concluding contracts (Graafland & Nijhof, 2007). Virtuous businessmen will feel responsible for living up to the contracts they concluded. The motivation comes from within. They refrain from cheating their contract partners, even if they expect this to be financially advantageous.
Lastly, this paper shows that both civic virtues and trust are channels through which rule of law stimulates human happiness. The magnitude of the sum of these indirect effects is, however, relatively small. It only explains about 15% of the total effect of rule of law on human happiness. The reason is that there are many other factors that might mediate the positive relationship between rule of law and life satisfaction. Examples are income per capita, education, individual freedom, entrepreneurship and tolerance, as previous research has shown that each of these variables relate positively to both rule of law and happiness.

Contributions to Literature
This paper contributes to the literature in two ways. First, this study is, to our knowledge, the first research into the relationships between free market institutions and civic virtues and between civic virtue and life satisfaction. Previous research has found a positive relationship between free market institutions and generalized trust (Knack & Keefer, 1997;Zak & Knack, 2001;Henrich et al., 2001Henrich et al., , 2004Berggren & Jordahl, 2006;Graafland & Compen, 2015) and between generalized trust and life satisfaction Oishi et al., 2011;Graafland & Compen, 2015). In this paper we focus on civic virtues from the illegal -dishonesty domain of the Morally Debatable Behaviors Scale to test the empirical relevance of the doux commerce (liberal dream) thesis. Although trust is related to civic virtues, it is not properly a civic virtue. Trust depends on the institutional context (rule of law) as well as on the civic virtues of those that are to be trusted. At the individual level, the two concepts do not necessarily relate, as persons trusting others do not necessarily excel in being trustworthy themselves in relation to others. Our bivariate correlation analysis indeed shows that the correlation between trust and civic virtues is rather low.
Whereas Graafland and Compen (2015) focused on trust, there is still a lack of micro evidence linking free market institutions and life satisfaction to civic virtues. Recently, Storr and Choi (2019) showed that civic virtues are higher for market than for non-market societies. In non-market societies, twice as many people as in market societies say that civic vices such as avoiding paying fares on public transport, cheating on taxes, stealing property, and accepting bribes are acceptable. However, these conclusions of Storr and Choi (2019) were based on a simple comparison of averages for market and non-market societies. 11 In our paper, we provide micro evidence of this relationship, controlling for macro and micro control variables. We consider microanalysis to be more reliable than macro analysis because the results are not affected by compounding effects. For example, whereas at the micro level civic virtues differs conceptually and empirically from trust, at the macro level these variables are more related, as generalized trust is informed by the civic virtues of others. Microanalysis is therefore more capable of identifying the precise nature of the relationships between rule of law, civic virtues, trust, and life satisfaction.

Scientific and Policy Implications
Research into the relationship between free market institutions and civic virtues is important, because the trust that civic virtues can build is known to have many positive effects on economic development and society. As liberalism stresses the importance of formal institutions, neoliberal (as well as neoclassical) economics has disconnected virtues from economics for a long time. Both neoliberal and neoclassical economics stress the importance of formal institutions, such as rule of law, to channel the microeconomic behavior driven by individual preferences, possibilities and beliefs, but neglect the significant role of virtues in the economic system. Economists often cite Adam Smith's Wealth of Nations (1776) as the first work to identify economics as the study of how the private self-interested actions of individuals generate public wealth as if by an invisible hand (Medema, 2010). It is common to see this thesis repeated in the introductions to economics textbooks (e.g., Mankiw, 2014) with the support of a few selected lines from the Wealth of Nations. While this may be a good representation of neoclassical economists' self-understanding of their profession, it is not a good representation of what Adam Smith said or meant, as Smith acknowledged the importance of virtues for stimulating human happiness. If market institutions affect human happiness through civic virtues, economists should take these effects into account in their economic analysis and policy recommendations. The policy implication is that governments should look for institutions that structure economic incentives in a smart way, such that they cultivate virtue (Bowles, 2016).

Limitations and Future Research
It should be noted that the empirical analysis does not allow strong conclusions about causality. For example, there might be reverse causality from trust on rule of law, as trust has been found to facilitate reforms that increase economic 1 3 On Rule of Law, Civic Virtues, Trust, and Happiness freedom (Berggren & Bjørnskov, 2017). This positive reverse effect would cause an upward simultaneity bias in our estimation results. Without the use of proper instrumental variables, we cannot exclude the possibility that the positive relationship between rule of law and trust is caused by a reverse positive effect. Similar reservations may also apply to the effect of rule of law on civic virtues and the effects of civic virtues and trust on life satisfaction. For the identification of causality in the relationship between civic virtues and life satisfaction, future waves of WVS/EVS should include additional variables that can theoretically be argued to causally drive civic virtues without affecting life satisfaction.
Another limitation of our study is our focus on civic virtues. Market institutions may also affect other types of intrinsic motivation, which are related to virtues. Other virtues that may be affected by markets institutions, and where empirical research is lacking to date, are, for example, courage, temperance, humility, industriousness, prudence, love, and hope. Although the WVS and EVS include survey questions asking about preferences on these virtues (as characteristics respondents prefer to see in their own children), these preferences do not measure characteristics of the respondents themselves. Hence, measuring these other virtues would require another extension of WVS and EVS.   Table 6  Table 7 bold: p < 0.01. Number of observations is 229,520. The number of countries is 80 Table 7 (continued)