Social Capital: The Role of Social Interaction
Alrik Baltus
March 2008
Abstract
Social capital is a topic that is increasingly used by both sociologists as economists to explain why some regions and nations experience higher growth rates than others. Because this type of research is still in its infant years, the definition of social capital is fundamental to the discussion of how to quantify the stock of social capital a region or nation is endowed with. In most empirical literature trust and group membership are used as a proxy. This paper contributes to the discussion of how social capital could be related to economic growth more properly. This will be done by regarding the time spend on social interaction as the leading indicator for the regional level of social capital. Furthermore, it should be noted that the direction of causality between social capital and economic growth is still to be determined.
JEL classification: Z13, O4
Key words: Social Capital, Economic Performance, Regional Income
Acknowledgement.
1. Introduction
The concept of social capital has attained an increasing interest in understanding the differences in economic performance between countries and regions. The publication of “Making democracy work” by Putnam, Leonardi and Nanetti in 1993, in which the authors find that social capital matters in explaining regional differences in economic and institutional performance in Italy, stimulated the discussion of the role of social capital in explaining economic performance. However, there is still much uncertainty about the mechanisms that foster social capital and the effects of social capital on economic growth, mainly because the empirical studies are relatively scarce, resulting from the difficulty of measuring the level of social capital correctly. The indicators mostly used in the literature are trust and participation in social groups (Knack and Keefer, 1997; Brehm and Rahn, 1997, Krishna and Uphoff, 1999, Costa and Kahn, 2001; Beugelsdijk and Smulders, 2004).
Granovetter (1985) criticized economists as they ignored the important role of social relationships in which economic transactions are ‘embedded’. Neoclassical economic theory is heavily focused on rational individual decision making in a perfect competitive market, whereas social exchange theory is needed to deal with exchange behavior in non-Western economies and in Western economies without a perfectly competitive market (Emerson, 1976). On the one hand, sociologists see individuals as socialized and acting by social norms, rules and obligations. On the other hand, economists see the same individuals as actors that maximize utility, i.e. purely acting on self-interest (Coleman, 1988; Elster, 1989). However, economists are becoming interested in the origin and operation of social capital, as they have come to realize that it is not only human and physical capital that matters, but that markets involve a great deal of behavior that isn’t fully rational economically. But, what is meant by social capital?
Social capital includes factors as values, norms, culture, motivation and solidarity (Uphoff, 2000).1 However, there is no single accepted definition of social capital. Putnam, Leonardi and Nanetti (1993) define social capital as “the features of social organization, such as trust, social norms and networks that can improve the efficiency of society by facilitating coordinated action” (p.167). Or, as put by Bourdieu (1986); “social capital is an attribute of an individual in a social context. One can acquire social capital through purposeful actions and can transform social capital into conventional economic gains. The ability to do so,
however, depends on the nature of the social obligations, connections and networks available to you”.2 As there are many different definitions, there are also many different types of social capital used in theoretical and empirical studies. The theoretical studies consider for example open vs. closed social networks (Coleman, 1988; Beugelsdijk and Smulders, 2004), structural vs. cognitive capital (Uphoff, 2000; Hjerppe, 2003) and micro, meso vs. macro social capital (Hjerppe, 2003; Beugelsdijk, 2006).3 This study, however, focuses more on the theoretical and empirical effects of social participation, as in Putnam et al (1993), Knack and Keefer (1997), Beugelsdijk and Van Schaik (2001) and Beugelsdijk and Smulders (2004). In the empirical section of this study, trust will also be included, for it is an important measurable part of the social capital concept as a whole. Now, which determinants could be considered as being social capital?
First, levels of trust determine the degree to which agents are willing to interact on markets or within organizations. Trust is based upon the expectations of agents that partners will not exploit the vulnerabilities of the other, that exchanged information is reliable and that agreements are fulfilled (Beugelsdijk and Van Schaik, 2001).4 The problem with trust, however, is to draw conclusions from cross-sectional comparisons of trust, because of the different cross-country institutional and cultural frameworks that foster trust (Sobel, 2002). In general, trust can be seen as the dependability of individuals on others. Higher efficiency and better defined rules and sanctions provide a high-trust environment that will lead to higher economic growth. Knack and Keefer claim that “people in low-trust environments will transact more with close friends and relatives than with strangers, compared with people in high-trust environments” (1997, p.1156). This indicates that trust plays an important role in the economy and in the effectiveness and efficiency of transactions.
Second, through repeated interaction individuals can build social relationships. With these social relationships mainly the social interactions with family and friends, colleagues, sport members and members of religious organizations are meant. Coleman (1988) states that “the actor is shaped by the environment”. The values, norms, culture, motivation and solidarity stated to shape the level of social capital are all strongly influenced by the people an individual interacts with. And, expanding your network indirectly increases the social capital of both you and your fellow group members (Sobel, 2002). Social networks can be extended
2 For a discussion on the different concepts of social capital presented by Bourdieu and Putnam, see Siisiäinen (2000).
3 For more explanation of these types of social capital see Appendix A.
in various directions; individuals can choose to interact more with other individuals of various groups, for example religious groups, but also with colleagues, sports organizations, and more. However, social networks are not a natural given; the profits which accrue from group membership are the basis which make the investment in a social relationship possible (Bourdieu, 1985).5 These profits can take the form of a higher attained social utility, and beneficial outcomes for the economy, as will be made clear in the next section.
In the neo-classical model of labour supply, an individual makes labour decisions by maximizing a utility function subject to a budget constraint. In this model a trade-off takes place between leisure and paid work. This model is somewhat unsatisfying in that it only allows a limited role for the social aspect of life. The consumption-leisure problem has to be extended with another decision, which is endogenously choosing the social capital level. Both trust and social networks are part of the total level of the social capital concept. However, the fuzziness of the concept of social capital is caused by the fact that researchers from different disciplines use social capital for seemly different purposes (Beugelsdijk and Smulders, 2004). Therefore, social capital as a whole may not be the correct instrument to be used in relating the social aspect of society to economic performance. Joining and being regularly involved in social relationships has a very significant impact on individual health and well-being (Cohen, Underwood and Gottlieb, 2000; Yip, Subramanian, Mitchell, Lee, Wang and Kawachi, 2007; and more). The integration of social interactions in the household’s work/leisure decision could be an improvement of the existing neoclassical economic literature, since social capital inheres in the social relationships between individuals (Coleman, 1988). The time spend on social interactions can be seen as investing in social networks, and thus, as investments in social capital. Time spend is much more easily quantified than the total concept of social capital, and, therefore, could be implemented more easily in models. And, perhaps more importantly, empirical studies are more easily conducted, as a result of better quantification methods to study the impact of social capital indirectly. This brings us to the next research questions: Is investing in social relationships by spending time on social relationships beneficial for economic outcomes? And, if so, how can governments or organizations stimulate this investment?
This paper is organized as follows. First, the literature of social capital theory is summarized and critically analyzed. Here the focus will be on trust and social interaction or group membership, whereas these two factors are the main social capital components used in
the literature so far. Then I turn to the theoretical model, which is the center of this study. After an empirical analysis, I conclude with suggestions for further research.
2. Background
The literature concerning social capital has expanded largely the past decades. Two features of social capital are worth mentioning. First, social capital varies between countries as well as regions. In some countries people are more trustworthy or socially active than in other countries (Putnam et al., 1993, Zak and Knack, 2001; Beugelsdijk and Van Schaik, 2001; Christoforou, 2003). Second, social capital can create beneficial outcomes. In developed countries social capital helps to “foster innovation, encourage investment, boost worker productivity … or stimulate the efficient use of scarce resources” (Woolcock, 1998, p. 172). Whereas in developing countries social capital creates benefits for the local society mainly through improved health and welfare conditions (Isham and Kähkönen, 2002; Narayan and Pritchett, 1999), schooling (Woolcock, 1998) and group lending (Spagnolo, 1999) and micro credits.
In this section I will look at the main factors that influence the level of social capital, after that I will show how social capital may influence the economic performance of a society. And finally, an overview of the existing empirical research will be given.
2.1 What Affects Social Capital?
Unfortunately, little research has been done on the causality between social capital and economic performance. Putnam (1995, 2000) states that in the last few decades social capital has been declining in the United States. In his book, however, a strong causal theory is not formulated (Durlauf, 2002b). The problem is the hard quantification of social capital, which makes it difficult, if not impossible, to develop a satisfactory causal theory of the relationship between social capital and economic performance.
since the improved working conditions makes employees to be dissatisfied more easily. On the other side, the economic development has positive effects. As a result of the higher incomes, people have more time available for socializing. Furthermore, the improved working conditions are beneficial for human health. However, besides economic developments, social capital can vary as a result of several factors.
First, the different moral levels that can be observed between countries, regions and generations. Also, individual characteristics, such as personal income and education, labour market status or experiences, family and social status, and national or institutional characteristics, such as income inequality and governmental quality, determine the incentive to invest in social capital (Krishna and Uphoff, 1999; Durlauf, 2002b; Swedberg, 2003; Christoforou, 2003). Social relationships die out if not maintained, whereas norms depend on regular communication at any level of social relationships. “The values, norms, attitudes and beliefs that qualify social capital are build up over time, but can be diminished and even destroyed in fairly short order” (Uphoff, 2000, p.228). Therefore, networks require the investment of time, money, information, solidarity and acceptance. Durlauf (2002b) states that the decreased decline in social pressure to join groups may be a reason for the decline in social capital. Costa and Kahn (2001) show that group loyalty in the Union Army during the American Civil War depended on socio-economic and demographic characteristics, ideology and morale. Whereas Alesina and La Ferrara (2000) show that participating in social activities is higher in more homogenous populations, both in terms of income and race or ethnicity. However, Knack (2003) states that “groups segregated by class, occupation or ethnicity may build cooperation and trust only among members, perhaps even encouraging distrust between members and nonmembers” (p.343). Education also contributes to the expansion of the stock of social capital, as years of schooling and group participation are positive related (Brehm and Rahn, 1997; Alesina and La Ferrara, 2000; Costa and Kahn, 2001). The family situation is another aspect that deserves notice. Single-parent families, for example, lack the benefit of a second at-home parent and tend to change residents more often, leading to fewer ties to other adults in the community (Portes, 1998), or as Coleman (1988, p. 105) puts it, there is a “lack of closure of the social structure”. This would bring less desirable educational and personality outcomes among single-parent children, and therefore may reduce the social capabilities of these children, and thus reduces the social capital stock of the society.
clubs, religious groups, and more. However, this same rapid technological change can also benefit the level of social capital, because it makes it easier for individuals to communicate over larger distances and with more people at the same time. Another feature of rapid technological change is that it can create a higher urge for labour mobility (Routledge and Von Amsberg, 2002). Both employers and employees try to match up the special capabilities of the worker with the particular skills needed for the job. This increased labour mobility increases the social mobility, which leads to an increasingly rapid decline in strong longstanding social and moral ties and incentive for interaction in the society (Lancaster Jones, 1969; Akerlof, 1997; Miguel, 2003). However, it has been shown that migration plays only a minor role in European regions, and the relation with per capita GDP is weak (Barro and Sala-i-Martin, 1995), and furthermore; “… just as firms, geographically, have an incentive to remain close to their current customers, individuals too have good reason not to abandon their relatives and current friends” (Akerlof, 1997, p.1010). Closer social distances would therefore increase the level of social capital. However, in the study by Miguel, Gertler and Levine (2001) no evidence is found that social capital is reduced in rapid growing communities.
Third, governments can influence the stock of social capital in their society by sound rules and fair institutions. By ensuring transparency, fairness and credibility people will have more trust and confidence in their governments. As a third-party enforcer, an effective government is able to facilitate cooperation between untrusting individuals through the creation of these institutions (Boix and Posner, 1998; Hall and Jones, 1998), whereas agents can be forced by the government to fulfill their part of the contract. This enhances the confidence in truthful interaction between agents, whereas repeated interaction can create future payoffs.6 Also, social problems as poverty and crime, racial and ethnical fragmentation, extreme income inequalities, economic underdevelopment and inefficient governments will have a negative impact on the level of social capital present in the society. For a government to create or stimulate the level of social capital in a society, it will be necessary for the government to fight these problems. Also, the stability of a society may enhance the level of social capital, while rapid and large changes do not occur, and the need for large changes in attitudes and behavioral patterns is absent (Hjerppe, 2003).
As may have become clear above, it is difficult to exactly determine the causality between social capital and economic development. Social capital is influenced by the
economic performance of a society, but as will become clear hereafter, the level of the social capital in a society certainly also influences the economic performance. Or as Portes (1998) puts it:
“… social capital is simultaneously a cause and an effect. It leads to positive outcomes, such as economic development and less crime, and its existence is inferred from the same outcomes” (p. 19).
Therefore, causality could go in both directions. Social capital is more likely to be created when income is higher, suggesting reversed causality (Ishise and Sawada, 2006). However, Knack and Keefer (1997) show that trust is more strongly correlated with per capita incomes in later years than with income in earlier years. This would mean that indeed social capital increases economic growth.
2.2 How Can Social Capital Affect Economic Performance?
The effect of social capital on economic performance can take two forms. Social capital affects the economic performance of a society both on the macro and the micro level. At both levels the effects of social capital can be positive and negative. At the macro level, the effects are empirically harder to prove and are less clear than on the micro level.
On the macro level, social capital especially affects governments and the functioning of markets. A favorable social infrastructure, the institutions and government policies that determine the economic environment within individuals can accumulate skills, and firms accumulate capital and produce output, “provides an environment that supports productive activities and encourages capital accumulation, skill acquisition, invention, and technology transfer” (Hall and Jones, 1998, p. 2). With more social capital in the society, governments tend to put more attention on effectively reducing unfair tax payments, bribes and violations of property rights. “Good” governments are able to take steps towards reform (Knack, 2002; Christoforou, 2003), whereas corrupt governments perform badly in effectively and/or efficiently increasing economic development and the functioning of markets,7 “free-riding”
and “rent-seeking” is less frequent in the former societies (Boix and Posner, 1998; Knack, 2002). Government officials in societies with higher trust may be perceived as more trustworthy and their policy pronouncements are therefore more credible (Brehm and Rahn, 1997; Knack and Keefer, 1997), which lowers uncertainty and therefore market transactions gain in efficiency (Shleifer, 2000; Zak and Knack, 2001).
Innovation is pushed by a higher level of human capital. Coleman (1988) shows that social capital both in and outside the family can create a higher level of attained education, because “children are strongly affected by the human capital possessed by their parents. But this human capital may be irrelevant to outcomes for children if parents are not an important part of their children’s lives” (p. 110). As families moved once or more, and this caused children to change schools, this reduces the possibility to build social strong social relationships with for example other parents. Among these children, a larger dropout rate is observed. Also, “participation in social groups may lead to the transmission of knowledge and may increase aggregate human capital” (Alesina and La Ferrara, 2000, p.849), because of positive spillover effects (Barr, 2002; Ishise and Sawada, 2006).8 Also, “actors can access heterogeneous information in a market ... because actors use network ties to search for opportunities and investments” (Uzzi, 1999, p.483), so information is shared within social relationships (Barr, 2002; Durlauf and Fafchamps, 2004), to create value for other agents for useful information in return. This information can be exchanged at golf courses, at ball games, and many more.
Companies with repeated interaction, and thus a higher level of mutual trust, may enjoy lower transaction costs, because there is less need for costly contractual agreements and monitoring of performance (Hjerppe, 2003).9 Fewer resources are spent on protection from being exploited in economic transactions, and more resources are available for investment and innovation or the firm’s profits are increased and the forgone earnings reduced. In financial markets, reducing monitoring costs is beneficial for both lenders as loaners (Eichberger and Harper, 1997).10 Also, the presence of credibility of promises and lower risks of opportunistic behavior by contracting parties allows them to rely on contracts across time and space, rather than to turn to spot market transactions, where investing in innovations and the returns on
Narayan and Pritchett (1999) in rural Tanzania show the importance of the role of social capital in governmental effectiveness and efficient markets.
8 On the effects of positive spillovers in the transmission of human capital, see Romer (1986) and Lucas (1988). 9 Some complex transactions or agreements - for example quality - are difficult to be fully ‘arranged’ in terms of contracts (Beugelsdijk and Van Schaik, 2001).
projects are considerably lower (Knack and Keefer, 1997; Keefer and Knack, 2005). However, social relationships between companies being too close reduce the need for formal monitoring and may result in cheating. There is also the danger of loss in flexibility and being locked in when relationships lead to a tendency to stick to existing linkages and networks (Beugelsdijk and Smulders, 2004). Another impact of the social component is the number of people using an innovation. Across many kinds of innovations, costs are reduced and benefits increased as more people use them (Kraut et al., 1998). This pushes the innovation level of, for example, products or more efficient communication and data base systems.
In their study, Costa and Kahn (2001) compare the army as an organization with other organizations, like companies. They studied the performance of soldiers in the military during the American Civil War. These authors do not explicitly measure social capital, but they study how socio-economic and demographic characteristics of soldiers affect group loyalty and their “productivity” within their squad. They found that the most effective military squads were those that were psychologically most homogenous. “Within heterogeneous units team production may therefore be harder because there is less social integration and less informal communication and because communication is less frequent. Team production may also be harder because social sanctions are less effective” (p.7). Heterogeneous social groups tend to be less socially integrated, and too much time and communication costs will be spent on clearing up misunderstandings and conflicts. Therefore, collective action is less likely in more heterogeneous, or socially inactive, groups, and, “everyday observation reveals that most socially interactive groups are fairly homogeneous” (Olson, 1982, p. 24). On the other hand, proponents of heterogeneous production teams claim that diversity yields creativity and effective solutions. Differences among group members are likely to stimulate discussion and exchange of information to found the different points of view that members have (see Mayo and Pastor, 2005, for more on social networks and team work effectiveness). Thus, communicating more effectively and a larger mutual tolerance both enhance the quality of exchanged information, the working atmosphere11, the cooperation between both the employer and employees and therefore the labour productivity.12
The same holds for employees, where a higher level of social capital will lead to less shirking and a higher productivity on the work floor. Efficiency of teams is increased through better sharing of information and more efficient coordination and effective leadership within
11 For more on job satisfaction, see for example Flap and Völker (2001). Co-worker solidarity seems to have positive effects on job satisfaction, and stimulates workers to “provide help to co-workers that goes beyond the
formal requirements of the job … and discouraging destructive behaviour at the workplace” (p. 299).
the team (Durlauf and Fafchamps, 2004). Furthermore, social interactions may foster workers’ effort and motivation, when interactions within the workforce are trustworthy and relaxed. Through peer monitoring, employees are more stimulated to do their best at work, and, co-workers will be more likely to sanction shirking (Dong and Dow, 1993 Bowles and Gintis, 2002; Brekke and Nyborg, 2004; Carpenter, 2005; Sabatini, 2006a). Provided that sanctioning is effective, since this can give rise to almost maximal cooperation. Strong reciprocity and norms are essential for effective sanctioning, otherwise players are not punished when defecting (Olson, 1982; Axelrod, 1986; Fehr, Fischbacher and Gächter, 2002). This increases team productivity, which may benefit all team members in the long run, through higher wages. Note that increased productivity has to be rewarded; otherwise it could seriously undermine the workers’ work morale. Also, a sound social infrastructure increases labour productivity, since rules and law make the costs of theft and diversion, such as rent-seeking, by employees too large (Hall and Jones, 1998).
Granovetter (1973, 1974), Fontaine (2004a, 2004b), and many more show that a relative large number of employees, in many professions, found their jobs through the use of their social contacts. Having the “right contacts” in the “right places” reduces regional unemployment, since more unemployed may have access to a larger number of jobs. Furthermore, since information about job opportunities and workers’ characteristics is imperfect, the process of job search takes time and effort (Stone, Gray and Hughes, 2003), thus already employed may reduce their time and costs spend on searching for a new or better job (information effect).13 If workers can reallocate to improve their efficiency, the economy as a whole benefits from the increased labour productivity. Furthermore, employers can anticipate that workers hired through networking are monitored by network members already in place, since these members are concerned about their own credibility as a “recruitment agent” (Stone, Gray and Hughes, 2003; Delattre and Sabatier, 2004). Therefore, these workers often display a higher productivity or willingness for reaching higher productivity levels (productivity effect).
Obviously, the level of social capital not only benefits the economic performance of a society, but may also have some negative effects.14 Social capital created by labour unions,
trade associations and other groups that lobby for wage demands, tariffs, subsidies or deteriorating market competition may benefit them, but at a large costs to society as a whole
13 For a discussion on the differences existing between so-called “strong” and “weak” ties, relating to jobs and wages, see Mouw (2003) and Sabatini (2006b). This indicates that social capital, as measured by the quality and quantity of workers’ social networks, may affect labour market outcomes.
(Knack, 2003; Ehrenberg and Smith, 2006). Other negative consequences of social capital are cartels, exclusion of outsiders, excess claims on group members, restrictions on individual freedom and downward leveling norms by group solidarity.15 The size of these special interest groups determine the strength and, thus, the possible negative consequences they have on society as a whole (Olson, 1982).
2.3 On The Empirics Of Social Capital
A reason for the lack of research on the causality of the relationship between social capital or social relationships and economic performance, is the insufficient amount of empirical literature regarding this topic. Few have tried to empirically prove the effects of social capital on economic growth. How to quantify social capital? The answer has not yet been given, so the quantification is mostly done by using trust and participation in social groups.
Putnam et al. (1993) study the institutional and economic performance in North- and South-Italian regions by looking at regional differences in general trust. The authors find that regions with higher levels of trust experience more rapid economic development and have better functioning institutions. This is partially confirmed by Beugelsdijk and Van Schaik (2001). They find that trust and economic growth are not related, while the European Values Surveys (EVS) data shows that being part social networks benefits regional economic growth.
In their empirical analysis, Knack and Keefer (1997), using the data of 29 countries, can not confirm the influence of group membership on economic aggregates with statistically significance. Therefore, they state that this variable cannot be used as an indicator for social capital. In their view, group membership may overestimate the stock of social capital because it includes individuals and social groups that do not take part in social activities. Knack and Keefer (1997) do find empirical evidence that per capita GDP is positively related to trust16 and civic cooperation, i.e. groups and individuals that do take part in social activities. With regard to causality, the authors show that trust is more strongly correlated with per capita incomes in later years, than with income in earlier years, suggesting that causality runs from
15 For a further discussion of these negative outcomes of social capital see Portes (1998). And see Motta (2004) for more on cartels, exclusion of outsiders and competition policy in the supply market of consumer goods. 16 Knack and Keefer also find that the correlation between trust and growth is stronger in poorer countries. They assume that this is the result of the less developed financial sectors, insecure property rights and unreliable enforceability of contracts in these countries. If “interpersonal trust is low and unlikely to improve rapidly,
trust to income.17 Furthermore, trust is correlated more strongly with recent than with past measures of school enrollment, suggesting that causality also runs from trust to the educational level. And, empirical evidence suggests that trust may improve governmental efficiency. Although the paper made a large contribution to the Social Capital Theory,18 it also has been criticized frequently. The main criticisms have to do with; a) the national data, since regions also differ in social, political and institutional characteristics (Temple, 1999), b) the small number of observations (Beugelsdijk and Van Schaik, 2001), c) the problems that plague cross-country growth regressions in general, namely the specification of the endless row of different determinants of economic growth (Durlauf, 2002a), and, d) the validity of survey data (Fetchenhauer and Van der Vegt, 2001; Beugelsdijk, 2006).
Zak and Knack (2001) describe a principle-agent model with investors as principals and brokers as agents, where the principals have to rely on the honesty of the agents. The authors show that cheating depends on social distance, the strength of formal institutions, effectiveness of social sanctions against cheating, the amount of assets invested and the wage rate of the investors. These theoretical findings have strong support by their empirical cross-country analysis. The model is tested empirically based on trust data obtained from the World Values Surveys for 29 countries. The empirics show a positive and significant relationship for trust with both economic growth and the quality of formal institutions.
The papers by Knack & Keefer (1997) and Zak & Knack (2001) have also been critically analyzed by Beugelsdijk (2006). Beugelsdijk shows that a closer look at the graph reveals three ‘clouds’ of observations.19 Furthermore, the robust results found in Zak and Knack (2001) come from adding data of 12 less developed countries to the sample of 29 countries used by Knack and Keefer (1997). In these countries a lower trust was observed, which increased the variance on the lower side of the sample, and therefore increasing the robustness. He also shows that trust is only significant in the low trust part of the sample, which are the relatively poor countries. Furthermore, the effect of trust is larger and significant for less developed countries with lower levels of economic development than the (statistically) insignificant effect of trust in rich countries. Ishise and Sawada (2006), also, show that the aggregate returns to social capital appear to be almost negligible for OECD countries, whereas the returns are much higher for developing countries. The problem,
17 In an empirical study by Claibourn and Martin (2000), however, it is shown that individuals with higher incomes are more likely to joins groups than individuals with lower incomes.
18 For a more extensive interpretation of the Social Capital Theory, see Fetchenhauer and Van der Vegt (2001). 19 See figure B.1 in Appendix B. “The graph suggests that it is important to include countries with relatively
ill-functioning institutions to obtain a significant relationship between trust and level of economic development”
however, with the study of Beugelsdijk (2006) is that he uses data from different resources and for different years. Therefore, the results he finds may not be entirely accurate.
Narayan and Pritchett (1999) find that “certain ‘trust’ variables … are not affected directly by household income nor do they affect income directly, but that greater levels of trust do lead to higher village social capital” (p. 880). They further state that social capital is an exogenous determinant of income and provide results against causality running from income to greater social capital. Uphoff (2000) gives an example of an experiment in which he was actively involved in. This example makes clear that in an environment without any form of social capital, investing in social relationships increases economic growth tremendously. Gal Oya was said to be the most disorganized region in Sri Lanka. After an effective plan for engaging farmers in a joint management system was introduced, Gal Oya became one of the most efficient and cooperative management systems. The new social infrastructure made farmers to take interests of others into account, considering decision-making, resource mobilization and management, communication and conflict resolution. Similar results were found with investments in the social infrastructure in the Philippines. In addition, Hall and Jones (1998) find that “countries with a good social infrastructure have high capital intensities, high human capital per worker, and high productivity. Each of these components contributes to high output per worker” (p. 31-32).
created by interactions with fellow soldiers, which were the result of socio-economic and demographic characteristics. Certainly, group loyalty depended on other factors too, but social capital was the dominant determinant.
Beugelsdijk and Smulders (2004) model social capital as participation in two types of social networks; first, the closed networks of family and friends, and, second, open networks that considers different communities. The agents in this model prefer to have (time-consuming) social contacts, which will be traded off against material wealth. Time spend on building and maintaining social relationships, thus, comes at the expense of forgone earnings.20 However, open networks reduces the costs of nations, because individuals and
organizations will experience an increase in internal sanctioning when rent seeking and shirking occurs. Consequently, the higher level of social capital may enhance economic growth.21 The authors test their model empirically by using data of 54 European regions. They find statistically significant evidence that open networks enhance economic growth, whereas closed networks have a negative effect on economic growth. Beugelsdijk and Smulders (2004) also show that people who are less materialistic tend to invest more in social relationships that do not directly yield material advantages than more materialistic individuals.
Empirical evidence suggests that networks have an ambiguous effect on labour productivity and wages (Mouw, 2003; Delattre and Sabatier, 2004; Fontaine, 2004b; Sabatini, 2006a, 2006b). On the one hand, employees who obtain their job trough social networks have higher earnings. On the other hand, the effect of social networks on these earnings becomes insignificant once the sector of labour is controlled for. Furthermore, social networks can also induce a mismatch between job seekers and vacancies, which results in a negative relationship between networks and wages due to inefficiency (see Fontaine (2004a), p.12 for references). Bonding social capital appears to affect productivity negatively, while bridging capital seems to be irrelevant. On the contrary, linking social capital fosters labour productivity (Sabatini, 2006a). Therefore, the existing literature is far from conclusive.
The lack of sufficient empirical social capital studies seems to be practically due to the vagueness of the definition of social capital, and therefore the lack of a proper quantification of social capital, and the poorly measured data. Therefore, it is my opinion not to use the
20 Individuals endogenously choose how to allocate the time spend to closed versus open networks, depending on their preferences and the opportunity costs. Furthermore, each individual optimally chooses time spent on rent-seeking activities, on work and on investment and learning.
vague concept of ‘social capital’ in theoretical models. Alternatively, it may therefore be useful to use the time invested in the creation and maintenance of social capital. This will be done in the next section.
3. The Model
Social capital is formed through network participation and social interaction in groups. As in Beugelsdijk and Smulders (2004), I too argue that there may be a trade-off between satisfying material wants and desires for socializing. However, they assume that social interaction comes at the cost of working time, whereas I assume that the time spend comes on the cost of leisure time. Social networks may affect the labour force status through the value individuals place upon non-market time as compared to market time (Stone, Gray and Hughes, 2003), however, in this framework all social networks are assumed to be homogenous. As is shown in the previous section, participation in social networks bring positive effects in knowledge and information spillovers and reduces rent seeking, by which corruption and shirking are the most important. Therefore, social interaction is assumed to increase labour productivity, since much of the productive work in an organization is done within work groups, and communication between group members is important (Kraut et al., 1998; Ishise and Sawada, 2006). Karl Marx even stated that production could only take place if workers entered into social connections and relations (Cited in Spagnolo, 1999, p.1).
The model used is based on the shopping time model.22 Implementing the influence of group interaction, the model can be used to show how social interaction affects individual decisions. General life satisfaction is strongly positive related with interpersonal trust, and trust is positive related with group participation (Brehm and Rahn, 1997). Olson (1982) stresses the value that individuals put on social interaction with other members of the society, and Loewenstein et al. (1989) stressed the importance of people caring about the outcomes of others; social interaction, therefore, should be incorporated into the utility function of individuals. In the model, individuals derive utility from consuming goods and services (c), “consuming” leisure (l) and social interaction (s) at time t.23 Therefore, the utility function of
the representative consumer takes the form
Ut = u(ct,lt,st), uc, ul, us > 0 ucc, ull, uss < 0
22 See Walsh (2003, pp.96-100).
23 For simplicity total time available is normalized to 1, i.e. l
Where the subscribes to function symbols denote respectively the first and second order partial derivatives. Utility is increasing for higher amounts of consumption, leisure and social interaction, ceteribus paribus. However, the marginal increase in utility is decreasing for higher amounts of consumption, leisure and social interaction.
The representative consumer chooses time paths for consumption, leisure and social interaction subject to budget constraints to be specified below, with maximizing total utility
W = t u(c
t,lt,st), (1)
t = 0
Where 0 < < 1 denotes the discount rate of future utility.
To complete the specification of the model, assume that consumers can hold assets (a), in the form of past savings, net transfers from the government24, and physical capital. Physical capital produces output according to the standard neoclassical production function, yt = f(kt-1,
nt(1 + st)), with fk, fn, fs > 0 and fkk, fnn, fss < 0. The consumer allocates these resources
between consumption and gross investment in physical capital, investing in social interaction at (linear) cost b, and future assets. With being the depreciation rate of physical capital, the budget constraint of the representative consumer takes the form
f(kt-1, nt(1 + st)) + (1- )kt-1 + at = ct + kt + bst + at+1 (2)
Where output is produced with the capital stock available from the last period, for capital takes time to be installed.
The representative consumer has to choose paths for ct, kt, st and at+1 to maximize (1)
subject to (2). With dynamic programming, the value function gives the maximized value of utility the consumer can achieve by behaving optimally, given its current state.
V(at, kt-1) = max {u(ct, 1 - nt - st, st) + V(at+1, kt)} (3)
Where the maximization is subject to
f(kt-1, nt(1 + st)) + (1- )kt-1 + at = ct + kt + bst + at+1 (2)
Defining kt = f(kt-1, nt(1 + st)) + (1- )kt-1 + at - ct + bst + at+1, gives the next first order
conditions for consumption, labour supply and social interaction
uc(ct,lt,st) - Vk(at+1, kt) = 0 (4)
-ul(ct,lt,st) + Vk(at+1, kt)fn(kt-1, nt(1 + st))(1 + st) = 0 (5)
us(ct,lt,st) - ul(ct,lt,st) + Vk(at+1, kt)[fs(kt-1, nt(1 + st))nt - b] = 0 (6)
Since the consumer has no influence on at and kt-1, the envelope theorem implies
Va(at, kt-1) = Vk(at+1, kt) (7)
Vk(at, kt-1) = Vk(at+1, kt)[fk(kt-1, nt(1 + st)) + 1 - ] (8)
The first order conditions show that consumption, leisure and social interaction must yield the same marginal benefit at the optimum location. Using (5) and (7), (4) can be written as
ul(ct,lt,st) = uc(ct,lt,st)fn(kt-1, nt(1 + st)) (9)
Using (8) and (9), (6) can be written as
us(ct,lt,st) - uc(ct,lt,st)fn(kt-1, nt(1 + st))(1 + st)
+ Vk(at, kt-1)[fs(kt-1, nt(1 + st))nt - b](fk(kt-1, nt(1 + st)) + 1 - )-1 = 0 (10)
Using (7) and (8), (4) can be written as
Substituting (10) and (11), we obtain
us(ct,lt,st) = uc(ct,lt,st)[fn(kt-1, nt(1 + st))(1 + st) - fs(kt-1, nt(1 + st))nt + b]
Assuming predetermined contracts,25 this can be written as
us(ct,lt,st)/uc(ct,lt,st) = fn(kt-1, st)(1 + st) - fs(kt-1, st)nt + b (12)
The marginal rate of substitution between social interaction and consumption is set equal to the marginal productivity of labour minus the marginal productivity of social interaction plus the marginal cost of social interaction. Given the stock of physical capital, the wage rate of the representative consumer is assumed to be equal to the marginal productivity of labour, and increasing with the stock of social capital in the form of social interaction.26 In short hand notation this can be written as us/uc = wt(st)(1 + st) - fsnt + b, with wt denoting the wage rate.
Two outcomes are worth mentioning:
First, with an increasing marginal cost of social interaction, b, the marginal rate of substitution between social interaction and consumption increases. This indicates a relative substitution towards consuming goods and services, because investing in social interaction becomes more expensive.
Second, what happens when the consumer, because of a shift in personal preferences, decides to spend more time on social interaction? With an increase in s, the wage rate will increase. Furthermore, an increase in s decreases the marginal productivity of social interaction, because fss < 0. Again, the marginal rate of substitution between social interaction
and consumption increases. However, with uss < 0, here the marginal utility of consumption
declines faster than the marginal utility of social interaction. A decrease in uc indicates an
increasing level of consumption. Thus, with the labour supply, n, being constant, the economy growths with the level of social interaction. In other words, social capital stimulates the economy.
25 With predetermined contracts, consumers choose to spend their “spare time”, i.e. the total time available minus the hours spend on production, on either leisure or social interaction. The assumption on predetermined contracts is that in general employees are rather unable to negotiate on marginal changes in the amount of hours work supplied.
The Steady State.
In the steady state of the model we find that Vk(at, kt-1) = Vk(ass, kss). From (8) the solution
gives the steady-state capital stock as
fk(kss, sss) = 1/ - 1 +
In this model, the steady-state capital stock depends on the time preference parameter , the rate of depreciation , and the production function. Assuming that the marginal productivity of social interaction is strictly positive, more social interaction increases the capital stock. More social interaction increases the left hand side. To compensate this, the capital stock has to rise, whereas it is assumed that the marginal return on capital is decreasing with an increase in the capital stock.
In addition, the model is independent of the cost of social interaction, b, i.e. in the steady state consumers spend the same amount of time on social interaction. However, since the steady-state consumption is equal to f(kss, sss) - kss - bsss, an increase in the costs of social interaction reduces the consumption. In the case of increased costs of social capital, is it possible for a higher level of social capital to compensate for this loss in consumption? Assuming that social capital, acquired through time spend on social interaction, increases the efficiency of team production, and measuring increased production in growth rates of GDP, social capital stimulates domestic growth. Therefore, social capital indirectly stimulates the consumption level. However, with respect to consumption it is not possible to state that it increases with higher levels of social capital. If the additional production is unable to compensate for the increased costs of social interaction and the increased depreciation of physical capital, consumption can either increase or decrease.
4. The Empirics
The steady-state model above states that spending more time on social interaction increases production. Unfortunately, information about regional levels of labour productivity per worker is extremely scarce. Therefore, by assuming that employees earn wages related to their productivity level,27 time spend on social interaction is assumed to be positively related to the
regional income level.28 Data of individual expenditure on consumption was not available; therefore the relationship between social interaction and consumption will not be empirically studied in this section.
In order to test this model empirically, I investigate 339 regions (see table B.2 in appendix B). As is done by Beugelsdijk and Smulders (2004), by taking regions I am able to test if Putnam’s thesis on social capital can be generalized. Furthermore, as regional differences within countries are present, a cross-nation investigation, see Knack and Keefer (1997) and Zak and Knack (2001), isn’t the most appropriate empirical investigation method.
4.1 The Data
The data is taken from the World Values Surveys (WVS), which is a survey on the factors included in the social capital concept; values, norms, culture, motivation and solidarity. The WVS is a large-scale, cross-national survey including over 300,000 individuals interviewed in the period from 1999 to 2003. Unfortunately, each individual is interviewed only once, and countries have conducted the interview in only one year. Therefore, my data refers to the year 1999, whereas this year represents the largest set of regions available for empirical research. The data set comprises 28 countries, i.e. Austria (with 9 regions), Belgium (12), former Yugoslavia (19), the Czech Republic (8), Denmark (13), Estonia (5), France (8), Germany (16, former eastern regions included), Greece (12), Hungary (20), Iceland (10), Italy (20), Latvia (5), Lithuania (6), Luxembourg (1), Malta (1), the Netherlands (12), Poland (17), Portugal (5), Rumania (38), Russia (12), Slovakia (8), Slovenia (12), Spain (17), Sweden (7), Ukraine (25), the UK (11, including Scotland and excluding Northern Ireland) and the US (10).29 Regions with an extremely low number of interviewed residents are left outside the sample, whereas this could lead to a distortion of the robustness of the model.
The measurement of the dependent variable is straightforward. The WVS asked the interviewed individuals to scale their current income level, counting all wages, salaries, pensions and other incomes before taxes and other deductions, ranging from 1 to 10, with 1 being the lowest income level step, and 10 the highest, so I truncated the upper bound at 10. Regional averages are assumed to indicate the generalized average income level for these
28 For simplicity, for the rest of the paper, with the regional income level is meant; the average income of the individuals that have been interviewed in the relevant. Regional income submitted by companies or the mining of regional raw materials, for example gas or oil, have not been accounted for, since this information was unavailable.
regions. The empirical measure used to proxy trust used by the WVS is based on the question: “generally speaking, do you think that people can be trusted or that you need to be very careful in dealing with people?” The answer on this question is understood either “yes” in the former case or “no” in the latter. So-called ‘blanc’ answers are left out of the samples. Again, and also for the other variables, regional averages are assumed to be the generalized indicator for these regions. Regional income levels and trust seem to be correlated negatively in my data set. However, the correlation is not that strong with -.19 (see figure B.2 in Appendix B).
The educational level is measured as the highest level education attained, ranging from 1 ('Inadequately completed elementary education') to 8 ('University with degree/Higher education - upper-level tertiary certificate'). Unfortunately, for the time spend on social relationships, there was insufficient satisfying data available. The data obtained by the WVS results from asking individuals the question: for each activity, would you say you do them every week or nearly every week; once or twice a month; only a few times a year; or not at all? With the answer 1 being ‘weekly’, 2 being ‘once or twice a month’, 3 being ‘only a few times a year’ and 4 being ‘not at all’. Thus, a lower average level of the explaining ‘social relationship variable’ indicates that more time is spent on social interaction. This, of course, is a very poor measurement of time spend with your friends, colleagues from work, people from your religion, or with people at sport, culture and/or communal organization. However, since it is the only useful variable available for my research, I have to settle for this data on time spend with social relationships. Figures B.3a- B.3d in Appendix B show the relationships between my social relationship variables and the regional income level. Note that a negative correlation indicates a positive relationship between the regional income level and time spend on social interaction, see above. The correlations of variables friends and sport with the regional income level are -.23 and -.28 respectively, which therefore indicates that time spend on social interaction with friends or other members of sport associations are positively related to the regional income level. Interestingly, the relationship between social interaction with the regional income level appears to be rather weak, which is indicated by the weak correlation of -.096. However, even more surprisingly is the positive correlation of .16 between the regional income level and the variable religion. This suggests that spending more time on social interaction with other members of religious organizations, which is assumed to increase the norms and values of individuals, has a negative relationship with the regional income level.
seem be positively related, since education is correlated positively with time spend on all four types of social interaction. Where Brehm and Rahn (1997) find an empirical positive connection between interpersonal trust and group memberships, this connection is less straightforward when studying the connection between interpersonal trust and time spend with other members of the community. The relationship between trust and time spend with colleagues seems be insignificant, resulting in a correlation of only -.004, whereas trust is stronger correlated with time spend with friends and other people at religious organizations and sport associations. Surprisingly, the relationship between trust and time spend with other people at religious organizations seems to be positive, whereas trust seems to be related negatively to time spend with friends or other people at sport associations. Since all four social interaction variables are correlated positively, time spend on social interaction is also correlated positively for all four variables. This indicates that obtaining and remaining social relationships with people from certain networks, for example sport associations, may stimulate the time spend on maintaining the social relationships of other not-related networks and vice versa. Furthermore, it can be shown that the correlation between spending time with friends and time spend with other people at sport associations has the highest correlation, whereas the correlations between religion and friends and between religion and sport are relatively low compared to the other four mutual social interaction correlations.
Table 1. The correlations between the explaining variables.
EDUCATION TRUST COLLEAGUES FRIENDS RELIGION SPORT
EDUCATION 1.000 -0.063 -0.284 -0.168 -0.013 -0.112 TRUST 1.000 -0.004 0.318 -0.218 0.393 COLLEAGUES 1.000 0.366 0.273 0.328 FRIENDS 1.000 0.078 0.654 RELIGION 1.000 0.126 SPORT 1.000
should be able to give a qualitative better indication of the amount spend on social interaction. For example, they should be able to state that they spend 10 hours a week on socializing with friends, or that they have two working days each month which are spend on ‘team building’ with colleagues. This would make analysing the investment in social capital better justified, since the quality of the regression analysis improves and better conclusions can be drawn.
4.2 The Empirical Model
The model of regional income includes the variables of social capital, here trust and social interaction with other members of the community; friends, colleagues, religious groups, and sport groups. The variable ‘time spend with family’, unfortunately, was unavailable, so this indicator can not be incorporated in the model. Perhaps the most important explanatory variables of income are the available capital stock per worker, and educational level that the interviewed people have attained. Unfortunately, the variable regional capital stock per worker was unavailable. Although the absence of this important depending variable may lead to omitted variables30, I assume that this absence will be absorbed by the constant, c, and the error, e, terms. Therefore, the average regional educational level is the first explanatory variable that will be used in the regression analysis. However, the focus of this paper is to study the influence of social relationships, and thus social capital, on economic performance. Individual factors, including age and gender, which in reality may often account for income differences are neglected in this study.
In classical linear regression models, the independent variables are uncorrelated with the disturbances for the dependent variable. This may not the case here since group membership and trust are hypothesized to influence one another (Claibourn and Martin, 2000). However, they find no simultaneous or lagged effect of trust on the level group memberships. Only prior group membership is positively and significantly related to later group membership. Furthermore, they find no evidence of a simultaneous or lagged effect of group membership on trust. Therefore, both variables can be used in the regression analysis.
Verba and Nie (1987) stated that a higher educational level is generally attended with a higher belief of one’s ability to influence socio-political outcomes, and with a higher level of social interaction. Knack (2002) found a positive and significant association of education
with volunteering, social trust, and informal socializing. However, in my dataset, I find no clear evidence of a strong correlation between these independent variables. Therefore, using a Two Stage Leased Squares model isn’t necessary here, and a truncated Original Least Squares (OLS) model is sufficient.
Two models are tested. In the first model, the direct effects of trust and the four separate social interaction variables are tested, and in the second, the product of the four separate social interaction variables is also included in the model, as one overall variable, to study both the direct and indirect effects of social interaction.
(1) INCOME = 1*EDUCATION + 2*TRUST + 3*SOCIALi + c + e
With i = 1, …, 4.
With c being an overall constant term, and e being the error term. Since time spend with friends and time spend with other members of a sport association have a relatively high correlation, two additional regressions will be analyzed; one without the variable time spend with friends and one without the variable time spend with other members of a sport association. The high correlation of the variables friends and sport, see table 1, may result in multicollinearity in the regression. Although the correlation between the two is below the ‘required level’ of .75, the model is still tested for both including and excluding these two variables.
(2) INCOME = 1*EDUCATION + 2*TRUST + 3*SOCIALi + 4*SOC.IN. + c + e
With i = 1, …, 4.
The reason for multiplying all four social interaction variables in one overall ‘social index’, SOC.IN., is the assumption that different relationships have beneficial mutual effects. This can be shown in table 1 by the positive correlations between these four variables. Therefore, simply adding the four variables to one overall variable may not be sufficient to study the indirect effects of social interaction.
5. Results
II) Model 1 with the social variables taken separately, excluding the variable sport; III) Model 1 with the social variables taken separately, excluding the variable friends;
IV) Model 2 with both the direct and indirect effects of all four social variables, in which the direct effect is taken as one overall variable, being multiplied;
V) Model 2 with both the direct and indirect effects of the social variables sport, colleagues and religion, in which the direct effect is taken as one overall variable, being multiplied; VI) Model 2 with both the direct and indirect effects of the social variables friends, colleagues and religion, in which the direct effect is taken as one overall variable, being multiplied.
The results of these 6 regressions are given in table 2, with Soc.In.(1) being the social interaction variable where the four social variables are multiplied. Soc.In.(2) estimates the indirect effects of regression 5, with the social variables sport, colleagues and religion, whereas Soc.In.(3) estimates the indirect effects of regression 6, with the social variables friends, colleagues and religion.
Regressions 1-6 of table 2 show that education is strictly positive and significant related to the income level of regions. In regressions 1-3, an one-point increase in the regional educational level increases the regional income level with approximately .25 points, whereas in regressions 4-6 an one-point increase in the regional educational level increases the regional income level with approximately .29.
Trust and regional income appear to be negatively related, as was already show in figure 1. In 5 out of the 6 regressions, however, the effect of trust on regional income is statistically insignificant, and in the other regression, this effect is only weakly significant. Furthermore, the amount that regional income is affected by the regional level of interpersonal trust is quite small in economic terms. Therefore, I find that trust and regional income are not significantly related.
Regressions 1-3 show some interesting results. Time spend with colleagues positively affects the regional income level, however, this effect is statistically insignificant.31 However, with a qualitative better quantification of time spend, it may be possible that this variable turns statistically significant, since with a slightly larger significance level, say a 20%-level, would have been statistically significant. Another reason why the relationship between time spend with colleagues and the regional income level is insignificant, is that this time is usually spend at the cost of time that would have otherwise been used for production purposes.
Furthermore, time spend with other members of religions organizations negatively and significantly affects the regional income level. An one-point increase in time spend with other members of religions organizations decreases the regional income level with .47 points in regressions 1 and 2, and decreases the regional income level with .386 points in regression 3. This result is even more surprising, since religious organizations have large positive effects on the norms and values of individuals. Therefore, spending more time with other members of these religious organizations was supposed to increase the regional income level. One reason could be the fact that an increase in norms and values may lead to faster satisfaction of individuals. Therefore, people may be satisfied more easily, instead of pushing themselves to their limits. This means that “workers stop working once they have earned enough to satisfy their traditional needs” (Swedberg, 2003, p. 230). And, “the economic progress is held back by the negative stance of religion toward profit-making and works as the mains goals in life.” In regression 1 and 2 time spend with other members of sport associations positively and significantly affects the regional income level. However, leaving this variable out of the regression has rather significant effects on the other variables. Levels of the parameters change significantly, and especially the parameter of time spend with friends changes dramatically. The parameter turns from .87 to -.442, and even becomes statistically significant. And the parameter for trust triples and also becomes statistically significant.
Table 2. The regression results.
Regression 1 2 3 4 5 6
Dependent
variable Regional Income Regional Income Regional Income Regional Income Regional Income Regional Income Constant 4.437* (.709) 4.370* (.706) 4.038* .865 (1.586) -.948 (1.746) 1.537 (1.838) Education .242* (.068) .240* (.068) .259* (.067) .282* (.070) .295* (.069) .289* (.070) Trust -.002 (.003) -.002 (.003) -.006*** (.003) -.001 (.004) .001 (.003) -.005 (.003) Friends .087 (.219) -.442** (.179) .788** (.359) .204 (.475) Sport -.531* (.142) -.500* (.119) -.230 (.187) .275 (.261) Colleagues -.228 (.175) -.215 (.171) -.203 (.156) .247 (.260) .829** (.357) .299 (.375) Religion .470* (.121) .469* (.121) .386* (.121) .785* (.176) 1.265* (.267) .740* (.270) Soc.In.(1) -.026* (.003) Soc.In.(2) -.106* (.032) Soc.In.(3) -.085 (.058) Adj. R² .164 .163 .139 .100 .190 .145 S.E. 1.008 1.007 1.029 1.041 .992 1.027 Mean, D.V. 4.703 4.703 4.688 4.703 4.703 4.688 Sample size 330 330 333 330 330 333
* = significant at 1%-level, ** = significant at 5%-level, *** = significant at 10%-level. Even though an appropriate measurement of the concept of social capital is not yet available, it was possible to relate social capital and regional income indirectly. Spending more time on social interaction is assumed to stimulate the total level of social capital. The model presented in Section 3 finds only weak support in the data. Table 2 shows that spending time on socializing may influence the regional income level, and therefore I may also state that social capital and regional income are likely to be somewhat related.
this model does not account for profits made by companies, and other forms of regional income.
Therefore, it is necessary for upcoming studies to take these limitations into account, since the existing empirical literature has been unable to come up with unambiguous results and conclusions. Many studies, however, have shown that the social capital concept plays one of the key roles in both the society and the economic environment. Unfortunately, until a proper quantification of the level of social capital is developed, linking social capital directly to economic performance may lead to false or misleading outcomes. Sociologists and economists should consider the possibility of linking the social aspect to economic performance indirectly. Many studies use membership of different social groups (Putnam et al., 1993; Knack and Keefer, 1997; Beugelsdijk and van Schaik, 2001; Beugelsdijk and Smulders, 2004). However, this does not measure the amount of interaction that takes place, and, thus, does not accurately measure how much these individuals are really involved in the relevant memberships. The use of time spend on social interaction, makes it more accurate to relate the social capital concept to economic performance. Time spend on social interaction is much easier to be quantified properly, and, therefore, may be the lead indicator of linking social capital and economic performance indirectly.
The variables of social interaction should be quantified more properly, and, thus, not by asking individuals the question: for each activity, would you say you do them every week or nearly every week; once or twice a month; only a few times a year; or not at all? With the answer 1 being ‘weekly’, 2 being ‘once or twice a month’, 3 being ‘only a few times a year’ and 4 being ‘not at all’. The question asked should be: for each social interaction, how many hours do you spend on social interaction with other individuals of each activity? And the answer given should be an approximated average numbers of hours per week or per month. This first step can be followed by a second step, to make an attempt to value this time spend on social interaction in monetary terms.
