آثار جانبی سرمایه اجتماعی و مرگ و میر در سوئد
|کد مقاله||سال انتشار||مقاله انگلیسی||ترجمه فارسی||تعداد کلمات|
|4094||2008||24 صفحه PDF||سفارش دهید||12342 کلمه|
هزینه ترجمه مقاله بر اساس تعداد کلمات مقاله انگلیسی محاسبه می شود.
این مقاله شامل 12342 کلمه می باشد.
نسخه انگلیسی مقاله همین الان قابل دانلود است.
هزینه ترجمه مقاله توسط مترجمان با تجربه، طبق جدول زیر محاسبه می شود:
Publisher : Elsevier - Science Direct (الزویر - ساینس دایرکت)
Journal : Economics & Human Biology, Volume 6, Issue 1, March 2008, Pages 19–42
We conceptualize social capital as an aggregate factor affecting health production and analyze the effect of community social capital (CSC) externalities on individual mortality risk in Sweden. The study was based on a random sample from the adult Swedish population of approximately 95,000 individuals who were followed up for 4–21 years. Two municipality-level variables – registered election participation rate and registered crime rate – were used to be a proxy for CSC. The impact of CSC on mortality was estimated with an extended Cox model, controlling for the initial health status and a number of individual characteristics. The results indicate that both proxies of CSC were associated with individual risk from all-cause mortality for males older than 65+ (p = 0.013 and p = 0.008) but not for females. A higher election participation rate negatively and significantly associated with the mortality risk from cancer for males (p = 0.007), and may also have exerted protective associations for cardiovascular mortality (p = 0.134) and deaths due to “suicide” (p = 0.186) or “other external causes” (p = 0.055). Similar associations were observed for the crime rate variable. The findings were robust to alternative specifications examined in the sensitivity analysis.
Social interactions in various forms and degrees are universal to all economic agents, and externalities are a fundamental feature of modern interdependent societies (Liu and Turnovsky, 2003). Individuals living in a society interact with one another, and their decisions may influence other individuals directly and/or indirectly through both market and non-market activities. In economics, the role of externalities has been widely studied in many contexts (Marshall, 1890 and Lucas, 1988). By comparing the wages of workers in US cities possessing different overall levels of human capital (education) and controlling for workers’ individual characteristics, Moretti (2004) showed that human capital spillover increases productivity over and above the direct effect of human capital on individual productivity. As a social characteristic, the effects of social capital have been hypothesized to influence various aspects of society, including health. Research into social capital has extended across various social science disciplines, such as economics and public health. However, debate is still ongoing regarding both the measurement of social capital and the mechanisms by which social capital generates health benefits for individuals. There is still dissent over whether social capital is an individual attribute or a community characteristic which operates as a public good. The post-Coleman (1990) literature commonly views social capital as an attribute of communities, and one which arises from social interaction (Glaeser et al., 2002). Anderson et al. (2004) recently found evidence to support the understanding of social capital as both a group and an individual attribute. Individual social capital (ISC) can be viewed as a person's social characteristics (Glaeser et al., 2002), and operationalized by the level of trust, membership in a network, civic participation, or participation in different community groups or activities. Community social capital (CSC) can be defined as the density of networks, groups, civic participation, or trust within a certain community (Paldam, 2000). CSC may include all of the cross-person externalities formed by the different types of ISC produced within a community (Glaeser et al., 2002). There is continuing debate concerning the sources and determinants of social capital, and the ways in which different conceptualizations of social capital should be measured. A range of potential sources and factors of social capital can be identified in the literature. For example, two major international organizations, the World Bank (1998) and the OECD (2001), identify eight sources (the family, schools, local communities, firms, civil society, the public sector, gender, and ethnicity) as being pertinent for the advancement of social capital. The Australian Institute of Family Studies (AIFS) has developed a conceptual framework of social capital where they recognize that social capital can have a range of possible determinants (similar to those identified by the World Bank and the OECD) and outcomes (e.g. individual/family wellbeing, public health, reduced crime, political quality of governance, etc.; Stone and Hughes, 2002). Social capital indicators can be classified in a number of ways, including a division into ‘proximal’ and ‘distal’ categories. ‘Proximal’ indicators of social capital are seen as consequences of social capital related to its core components of trust, networks and reciprocity, whereas ‘distal’ indicators are outcomes of social capital which are not directly, but rather indirectly, associated with its key components (e.g. interpersonal mistrust — a component of social capital — influences crime rates, and crime rates influence health) (Stone, 2001). Nonetheless, one general feature of social capital is that there may be complex feedback effects between its causes (sources) and effects, resulting in ‘virtuous circles’ of social capital creation (Productivity Commission, 2003).1 Though the practice is not beyond criticism, proximal and distal indicators of social capital are frequently used in those studies which are dependent on secondary data for social capital indicators (Stone, 2001). In a recent thorough literature review on social capital and health, Islam et al. (2006a) found that most of the studies operationalized social capital as a combination (composite indicator) of both cognitive (e.g. trust and reciprocity) and structural dimensions (e.g. informal participation or civic engagement) of social capital indicators. The authors raise the question of whether the studies are indeed comparing the same phenomenon across the countries or studies. Social participation and trust have been suggested in the literature as representing different aspects of social capital, and both these aspects of social capital are thought to mutually enhance each other (Putnam, 1993); however, in reality this is not always the case (Fukuyama, 1999). To capture the multidimensional aspect of social capital, some researchers have employed the procedure of factor analysis to measure the underlying latent variables (Onyx and Bullen, 2000 and Bjørnskov and Svendsen, 2003). However, uncertainties in identifying the underlying factors and in determining the total number of factors, and the possibility of the presence of error in the original data, may limit the chance of obtaining reliable or valid factor scores (Hair et al., 1998), and undermine the rationale for using the factor analysis procedure.2 Methods for obtaining a precise picture of the sources and determinants of social capital, and constructing a valid and reliable social capital index, are still in the early stages of development. In this paper, we use two variables as proxy measures of CSC: voting participation in the municipalities’ political elections, and crime rates in the municipalities. Although these are not perfect choices, they are motivated by a number of factors. Firstly, theorists of social capital have hypothesized that political efficacy is a consequence of social capital (e.g. the “civic voluntarism model” of Verba et al., 1995), and also acknowledged that voting and political participation may be regarded as both measures of civic engagement in general and as proxy measures of CSC (Putnam et al., 1993). It has also been predicted that high levels of interpersonal mistrust will be associated with various criminal activities, namely homicide, assault, robbery, and burglary. Other forms of social capital, in particular shared values and norms, may have an impact on crime rate (distal indicator) and thus on health status. It has been argued that the crime rate can be used as a proxy for social capital and indicators of collective wellbeing (Wilkinson, 1996, Kawachi et al., 1999 and Putnam, 2001). Secondly, a recent Swedish Government Bill on National Goals for Public Health, within the goal area Social power and Participation, proposed to the parliament that voting participation should be used as a measure of social capital (Government Bill, 2002).3 Thirdly, imprecision in the CSC measures or its proxies used in various studies allows the possibility of measurement error in the explanatory variable, which leads to biases and inconsistency (due to attenuation bias) in all estimators within the model.4 Our proxy variables of municipality-level social capital are based on registered data, and thus also comprise an objective proxy measure of CSC; use of such variables may reduce the risk of attenuation bias in the estimated effects. Finally, given that our proxy variables are not constructed from individual responses (i.e. not based on a sample), but rather are based on registered data, the endogeneity problem is less likely to be an issue; that is, the reversed causality from mortality to CSC. There is conflicting evidence regarding the association between social capital and mortality. Studies from the USA have generally found that living in communities with a higher CSC is beneficial for good health or lower all-cause and cause-specific mortality and individual wellbeing in general (Kawachi et al., 1999, Subramanian et al., 2002, Wilkinson et al., 1998, Putnam, 2000 and Folland, 2006a). Milyo and Mellor (2003), however, found no significant associations between age-adjusted mortality and social capital in the USA. Veenstra (2002) found social capital to be weakly associated with age-adjusted mortality rates in Canada. Lynch et al. (2001) found weak and inconsistent associations between a country's level of social capital and its age-specific and cause-specific mortality rates, in 16 OECD countries. Mohan et al. (2005) also found little support for CSC as a contextual influence on the probability of survival in England. Moreover, the effect of CSC on the mortality of the elderly has not been conclusively documented, and the evidence in this regard is also rather conflicting (Anderson et al., 1997, Waitzman and Smith, 1998a, Waitzman and Smith, 1998b and Wen et al., 2005). In the related literature on income inequality and mortality, it is debatable whether results from countries such as the USA are generalizable to other countries which are more egalitarian, such as Sweden. It could be the case that different public social institutions within welfare-states, and the more equitable distribution of income and wealth in these countries, may modify the mortality impact of CSC. For example, in contrast to the conclusions of most research conducted in the USA on area-level income inequality and mortality, Gerdtham and Johannesson (2004) found that municipality-level income inequality had no effect on mortality in the Swedish context. Moreover, the existing literature dealing with CSC and mortality is mostly ecological and usually based on US state-level data. While some studies use individual-level data, they include only a limited number of control variables at the individual level, and use combined mortality analysis for all age groups (Mellor and Milyo, 2005). This study aims to examine the effect of CSC externalities on individual all-cause and cause-specific mortality risks in Swedish municipalities.5 Municipality-level local self-government is very strong in Sweden, in comparison to most other countries, and there is great freedom for each municipality to make decisions about its own activities. Local democracy in the municipalities is an important part of Swedish governmental institutions. Recent research in economics recognizes that community human capital has spillover effects on individual productivity (Moretti, 2004). By considering similar mechanisms and using as a basis the ideas of Becker and Murphy (2000) and Grossman (1972), we explore a conceptual model and estimate the effects of CSC externalities on health (mortality) in an innovative way. We make use of a large data set of approximately 95,000 individuals followed for up to 21 years, and based our analysis on high-quality register data concerning mortality—an objective indicator of health, and of social capital. We also performed a sensitivity analysis of alternative models, to evaluate the robustness of our results.
نتیجه گیری انگلیسی
We have used the theoretical framework proposed by Becker and Murphy (2000) and Grossman (1972) to present a mechanism and explore the effects of community social capital (CSC) externalities on individual all-cause and cause-specific mortality risks. Using this as a basis, we made estimates employing extended Cox models controlling for initial health status and individual social capital across 275 municipalities in Sweden. We found, firstly, that higher municipal voter turnout and mortality are negatively associated, whereas higher crime rates and mortality are positively associated for the elderly (65+) male population living in a given municipality. Secondly, that higher municipal election participation rates significantly associated with lower risk of cancer mortality; in contrast, higher cancer mortality significantly associated with higher crime rates. Our results were found to be robust to alternative models based on different distributions and specification testing in the sensitivity analysis for elderly males. Our hypothesis was that CSC works as an aggregate factor affecting health production, and that these impacts may manifest themselves through four different pathways. CSC might affect health by having an effect on psychosocial stress (a reduction in such stress), health-related behaviors through norms and values (an increase in “healthy” behaviors), access to health care and amenities (an increase in access), and community social disorders (a reduction, particularly in violent crime and homicide). The pathway which would be most relevant concerning the association between (low) CSC and cancer would be first and foremost health-related behaviors but also, secondly, access to health care and amenities. As conceptualized by Becker and Murphy (2000), complementarities between social capital and health behaviors, particularly smoking, might be one of the main mechanisms which in turn affect cancer mortality. In the municipalities with low turnout or high crime (i.e. low CSC), there may be strong peer effects (e.g. most of the friends or neighbors smoke) which influence an individual's starting to smoke or continuing to smoke or the amount smoked. CSC had slightly beneficial effects on cardiovascular mortality, “suicide”, and mortality risks from “other external causes”, though none of these was found to be significant. One likely explanation could be that in old age, when the background level of disease is so high, the ceiling effect makes it difficult to detect any marginal excess risk associated with a risk factor (e.g. low CSC) (Kaplan et al., 1999). Moreover, age-related co-morbidity, along with uncertainties in detection and diagnosis of disease in the elderly, may also create noise in the estimations. Furthermore, different forms of CSC (or their proxies) may have different implications for different causes of mortality. For example, our crime variable was based on all sorts of reported crimes, but homicide or violent crime rates could be more relevant proxies of community social disorders that could be strongly associated with cardiovascular mortality via stress (Kawachi et al., 1999 and Wilkinson, 1996). In addition, it is also possible that low statistical power meant we were unable to detect significant associations of municipal voter turnout or crime on the deaths due to “suicide” or “other external causes”. The low prevalence of mortality for the younger age group could be an explanation for our finding no significant associations between our proxies of CSC and all-cause mortality. The next issue one might raise is the question of why the impact of social capital should differ for different age groups and genders. “Community context” may be relatively important for the elderly (Diez-Roux, 2002b and Robert and Li, 2001), a proposition that is supported by our findings. Gender differences in social capital and their effect on health status are also discussed in earlier literature (McCulloch, 2001). In a Swedish study, Islam et al. (2006b) found that individual health-related quality of life scores were significantly associated with linking CSC for males but not for females. The finding that our proxies for linking CSC, particularly voting turnout, had no significant association with female mortality can be viewed in the light of both individual and contextual data. Individual data from Sweden shows that when females were first given general suffrage in the national parliament elections in 1921, the female voting participation was lower than that for men—and it remained lower until the 1960s, although the sex difference in voting participation declined. Finally, beginning in the 1970s and 1980s, female voting participation became somewhat higher than for men. It is highly plausible that the current lower overall voting participation for men entails more pronounced differences between different socio-economic groups and between electoral districts/areas with low as opposed to high voting participation. That is, men of high socio-economic status tend to vote to the same extent or even more than women of high socio-economic status; while, on the other hand, men of low socio-economic status tend to vote less than women of low socio-economic status. This would explain the differences between men and women regarding significant as opposed to non-significant associations between voting participation and mortality. Contextual level data concerning the ten city quarters in Malmö in southern Sweden has revealed that male life expectancy has a range of more than 6 years between the city quarters with the highest socio-economic status population and the city quarters with the lowest socio-economic status population. The corresponding range in life expectancy for women is only 3 years. It might be that voting participation is highly associated with area socio-economic status (in our data, we also observed that municipality mean income and turnout rates were positively and significantly correlated). The wider range of life expectancy and thus also mortality for men would, then, explain the significant association between voting participation and mortality for men but not for women. It could also be the case that different forms and dimensions of social capital may operate differently for different genders. If this could be the case, then one may also raise the question whether voting participation and crime are good proxies of CSC for women and if in general CSC effects are not similar for both gender, this may further lift the debate whether CSC is a property of a place or individual. That CSC is owned by individuals not by places is one of the major critiques by sociologists of how many social epidemiologists seem to misuse the construct of social capital (Astone et al., 1999). Given the limitations of our data, we were unable to see the distributions and relations of different forms of CSC by gender and to test this hypothesis directly, and so leave further exploration of this issue for future work. Our study is not without its limitations. First, we used only two proxy variables of CSC—election participation and crime rates; in reality, there are many other potential proxies. Besides, due to unavailability of municipality-level crime data over the study periods, we considered average of crime rates for three available time points, however, the constructed crime variable might not reflect the overall state of municipal social disorders over time. Second, although we assumed that CSC may be one of the driving forces behind better health services, however, the opposite may occur where health authorities invest more heavily in health services for communities that are the most troubled and in need. We also assumed that individual personal characteristics may reflect ISC, there may still remain other aspects of ISC (e.g. individual trust) which might not be explained by these individual attributes. Due to lack of data, other community and individual attributes were not considered in our analysis. Third, residential mobility should be another concern; however, a lack of data meant that we were unable to adjust for this factor and hence did not raise this issue in the analysis. It could also be a problem that many individuals spend much of their time (e.g. for work) in municipalities other than the ones in which they live. Although working conditions and workplace environment may influence individual mortality risks, again the lack of data meant that this issue was not addressed in our analysis. However, Sundquist et al. (2004) note that in Sweden, 75% of individuals had lived at their current home addresses for at least 8 years preceding the time of survey (based on information collected by Statistics Sweden, 1986–1993). Moreover, our analysis mainly focused on the elderly, who are most likely comparatively less mobile than younger people, and so these residential mobility or working conditions may be less important for this age group. What are the policy implications of our findings? In a civil society, higher voter turnout may reflect trust in civil organizations, politics, and politicians, and may indicate a society with good governance and less corruption and crime. The development economics literature identifies a number of ways in which good governance and social capital may directly facilitate increased wellbeing (Wallis et al., 2004). In comparison to the insignificant association between income inequality and mortality in Sweden, one may argue that marginal differences in municipal-level social capital (a country with a high stock of social capital) may have little explanatory power regarding mortality risk. However, we have observed that CSC maybe important for the Swedish population, particularly for elderly males, so one may not simply ignore CSC externalities in Swedish municipalities. Thus, if higher CSC is important for mortality in Sweden (a comparatively egalitarian and homogenous country), this may have important implications for other societies. Investing in civic participation and controlling abnormal social behavior in local communities maybe one way to improve the health of communities. To conclude, we have observed that CSC has spillover effects on mortality risks, over and above the individual personal characteristics of the aged male population across Swedish municipalities. Before drawing definite conclusions as to the effects of CSC externalities on all-cause or cause-specific deaths by age group or gender, further careful scrutiny is warranted. We hope that the conceptualization and findings presented in this study will stimulate further research. To consider CSC as a determinant of health, and to explain the underlying mechanisms by which CSC may generate social wellbeing, future research should make use of rich data pertaining to social capital measures, such as different social capital proxies and longitudinal data, while applying appropriate econometric methods, such as, where applicable, the use of the instrumental variables approach to account for potential endogeneity, particularly reverse causality from health status to ISC/CSC.