I. INTRODUCCIÓN
1.3 Teorías relacionadas al tema
1.3.12 Arquitectura y cultura en los espacios públicos
This paper contributes to three different strands of literature. First, it discusses the claim that social capital has been declining across the US (at least up until the turn of the cen-tury). In a seminal contribution to this literature, Putnam (1995, 2000) investigates the dynamics of social capital formation and documents a persistent fall in American social capital over the course of the second postwar period. Putnam draws on a wide set of empirical measures including political participation, clubs, community associations, reli-gious organisations, professional bodies, informal groups, trust, and altruism. For each, he shows that the predominant trend is one of falling engagement in social activities and increasing isolation of the American people. Amongst the hypothesised causes of this dy-namic, Putnam enumerates increasing financial pressures and long working hours, urban sprawl, electronic entertainment and television, the changing structure of American famil-ies, female participation to the workforce, and demographic composition as the great ‘civic
who are more active in the production of social capital, their mobility will negatively affect left-behind communities. Moreover, housing markets adjustments related to inelastic downward supply faced with negative shocks exacerbate these effects as they encourage the ‘wrong’ type of migration to declining areas, attracting individuals with low levels of human and social capitals, to the extent that the two are complements (Glaeser and Gyourko,2005).
8This is consistent with a general declining trend in geographic mobility of workers documented for the US (Molloy et al.,2011,2014,2016). Molloy et al.(2016) also advances the hypothesis that this decline might be endogenously connected to lower levels of general social trust amongst Americans, further complicating the picture.
generation’ of the pre-war period is replaced by their disengaged children and grandchil-dren. Putnam (2000, p. 274) also acknowledges that with no doubt “[...] global economic transformations are having an important impact on community life across America”, yet he does not further explore this channel. Several papers have evaluated Putnam’s claim, with mixed results. Paxton (1999) examines various indicators of social capital in the US over two decades and finds a moderate decrease in a general index measure, as well as falling levels of individual trust. Yet trust in institutions and participation to associations does not seem to have declined. Paxton also emphasises that to fully understand changes in American communities, it is important to study the dynamics of dispersion in social cap-ital, not just its stocks, which is where geography can offer a valuable contribution. Costa and Kahn(2001) study changes in social capital since the 1950s, focusing on the residential dimension of this concept, that is to say, the community of family and friends formed out-side the workplace within the private sphere of one’s life. They distinguish between social capital produced within home through meetings with friends and family, and one gener-ated outside home thanks to organisational membership or volunteering activity. Their findings suggest that, once education is controlled for, women’s growing participation to the workforce predicts changes in social capital produced at home, while income inequality and community heterogeneity more in general explain the losses of social capital formed outside the home environment, in line withAlesina and La Ferrara(2000).
Second, my analysis dialogues with research on the determinants of social capital formation in general, and on the role of deindustrialisation in this process in particular. Rupasingha et al.(2006) offer a comprehensive analysis of social capital production in the US, in what is perhaps the paper most closely related to my work. The authors construct a county level index of social capital for various vintages. Their methodology provides the starting point for some of the empirical analysis in this paper. Moreover, their analysis represents one of the first nationwide studies at this geographical scale, shedding light on the spatial distri-bution of social capital stock in America, as well as on its local determinants. Their findings broadly align with the theoretical predictions outlined in Section 2.2.2, albeit with some exceptions. The authors confirm that formal education, age, and community homogeneity are strongly associated with social capital production in American counties. They also find a positive effect for community attachment and, contrary to Putnam, for female labour market participation. In their econometric model, the authors also consider the share of manufacturing workers in each county, with mixed results. However, their specification is cross-sectional or uses random effects to allow inclusion of time-invariant variables. Hence, it does not capture within-county changes in manufacturing employment that are key to
understand the impact of deindustrialisation on social capital, nor does it consider changes over time in the outcome. My aim is to provide new and more credibly identified evidence specifically on the role of manufacturing decline, and to consider changes in social cap-ital rather than its cross-section in levels. To the best of my knowledge, no other study looked at the relationship between manufacturing and social capital in a developed eco-nomy. Miguel et al.(2006) consider the case of Indonesia. The authors rely on survey data from 1985 and 1997, a period of rapid growth in the country, to examine how industrial-isation affects density of voluntary community associations, informal social networks, and trust or cooperation. Their findings suggest that districts where manufacturing expands display positive changes in various measures of social interaction, while industrialisation in nearby districts negatively correlates with associational activity and mutual cooperation in the district itself. The key hypothesised mechanisms underlying these relationships are income-growth, higher inequality, and internal migration to neighbouring booming areas.
In particular, it seems likely that the mass relocation of young Indonesians spurred by new employment opportunities in neighbouring areas may explain the disruptive effect on the social networks in origin communities. It is difficult to directly relate these findings to the American case, however.
Third, this paper also extents the literature on the societal effects of manufacturing decline by considering a novel non-economic outcome and its geography across US communities.
Several studies document the adverse consequences linked to deindustrialisation. I already discussedAutor et al.(2013a) andAutor et al. (2014), who emphasise economic outcomes, notably wages, employment, and public-transfers. Autor and Dorn (2013) consider the distributional implications of a shift towards service jobs associated with skill-biased tech-nological change and the loss of middle income job opportunities in the tradable sector.
Autor et al. (2016) examine a non-economic outcome, documenting how rising import competition from abroad increased political polarisation across exposed US congressional districts and counties. More recently,Autor et al.(2019) investigate the impact of changes in manufacturing employment on family formation, the primordial social unit, by look-ing at how labour market shocks in this sector decrease the value of young men on the marriage market. They find that localised shocks to the manufacturing sector, which dis-proportionally affect men, lead to a surge in male mortality due to abuse of alcohol and drugs, shrinking employment and wages, and ultimately a fall in marriages among young adults accompanied by lower fertility rates and higher single-parent households. My aim is to contribute to this strand of research with new evidence on the strength of community participation.
Finally, a word of caution. The above sections outline the impact of manufacturing decline on social capital. However, it is also possible that social capital is a cause, rather than a result, of changes in the economic trajectory of places. There is ample research document-ing the relevance of social capital for economic outcomes.9 Putnam et al. (1993) provide one of the earliest empirical studies linking institutions and social capital to the functioning of an economy, by comparing the performance of regions in the north and south of Italy.
Knack and Keefer(1997),Zak and Knack(2001) andBeugelsdijk et al.(2004) consider the social capital-growth nexus on a macro-level, whileIyer et al.(2005) and Tabellini(2010) examine its role in the local economic development of US and European regions. More in general, social capital is associated with several other economic or economically relev-ant outcomes such as improved educational attainments (Coleman, 1988), reduced crime (Sampson et al.,1997;Lederman et al.,2002;Akçomak and ter Weel,2012), financial de-velopment (Guiso et al., 2004), entrepreneurship (Percoco, 2012a, 2015), and innovation (Akçomak and ter Weel, 2009; Crescenzi et al., 2013a,b), to cite a few. To address the possibility of reverse causation, along with other endogeneity issues, this analysis will rely on an instrumental variable approach (further details are discussed in Section2.3.3below).