Social Capital, Government Expenditures and Growth Giacomo A. Ponzetto CREI, Universitat Pompeu Fabra, IPEG, Barcelona GSE and CEPR Ugo Troiano University of Michigan and NBER April 2018 Abstract This paper shows that social capital increases economic growth by raising government investment in human capital. We present a model of stochastic endogenous growth with imperfect political agency. Only some people correctly anticipate the future re- turns to current spending on public education.
Greater social di¤usion of information makes this knowledge more widespread among voters. As a result, we …nd it alleviates myopic political incentives to underinvest in human capital, and it helps the selection of politicians that ensure high productivity in public education. Through this mechanism, we show that social capital raises the equilibrium growth rate of output and reduces its volatility. We provide evidence consistent with the predictions of our model.
Individ- uals with higher social capital are more informed about their government. Countries with higher social capital spend a higher share of output on public education. Keywords: Social Capital, Education Expenditures, Economic Growth, Elections, Government Expenditures, Imperfect Information JEL classi…cation: D72, D83, H52, I22, I25, O43, Z13 We are grateful for their helpful comments to Jordi Galí, Jim Hines, Joel Slemrod and seminar partic- ipants at ESSIM, the SED Annual Meetings, the CRENoS Workshop on “Institutions, Individual Behavior and Economic Outcomes,” the University of Kentucky and the University of Michigan. Ponzetto acknowl- edges …nancial support from the Spanish Ministry of Science and Innovation (grant JCI-2010-08414), the Spanish Ministry of Economy and Competitiveness and its Severo Ochoa Programme for Centres of Ex- cellence in R&D (grants SEV-2011-0075, RYC-2013-12838, ECO-2014-59805-P and SEV-2015-0563), the Government of Catalonia and its CERCA Programme (grants 2009 SGR 1157 and 2014 SGR 830) and the BBVA Foundation through its …rst grant for Researchers, Innovators and Cultural Creators.
Troiano ac- knowledges …nancial support from the University of Michigan, the Harvard Department of Economics, the Harvard Multidisciplinary Program in Inequality and Social Policy and the Bank of Italy. We thank Chiara Ferrero and Alex Wolfe for research assistance. The opinions expressed in this project belong to the authors and the BBVA Foundation is not responsible for them. E-mail: gponzetto@crei.cat, troiano@umich.
1 Introduction How does social capital contribute to economic growth? A growing body of evidence doc- uments it does (Knack and Keefer 1997; Zak and Knack 2001; Tabellini 2009; Algan and Cahuc 2010). However, the underlying causal mechanisms remain poorly understood. The very concept of social capital popularized by Putnam (1993) and Fukuyama (1995) hinders a precise study of these mechanisms because it is too broad and too vague (Solow 1995; Durlauf and Fafchamps 2005; Jackson 2010; Guiso, Sapienza and Zingales 2011). If “social capital refers to connections among individuals— social networks and the norms of reciprocity and trustworthiness that arise from them” (Putnam 2000, p.
19), surely it must have bene…cial economic consequences through uncountable channels. Nonetheless, a rigorous analysis of particular transmission channels remains crucial to understanding the social determinants of economic development. In this paper we propose and analyze one novel mechanism through which social capi- tal causes economic growth. The greater social di¤usion of information about government activity induces greater investment, as well as higher and less volatile productivity, in the provision of public education.
We formalize this insight in a tractable model of stochastic endogenous growth with imperfect political agency. We then exploit survey evidence to show that the distinctive theoretical predictions of our model …nd empirical support at both the individual and the aggregate level. People with higher social capital are more informed about their government. Countries with higher social capital spend a higher share of income on public education.
In our model, endogenous long-run growth is supported by investments in human and physical capital. Capital accumulation cannot be undertaken by the private sector alone, but requires an indispensable government investment in public education. Politicians with heterogeneous skills set taxes and allocate spending between public education and other public services, with an eye to re-election. Voters retain or dismiss the incumbent government according to their inference of its skills, based on imperfect information.
All voters observe the provision of public services that immediately raise their utility. Instead, not everyone correctly anticipates the returns to public education in terms of future economic growth. Crucially, knowledge of the e¤ectiveness of government investment in human capital spreads through social connections. Our precise operational de…nition of social capital is the rate of social di¤usion of in- formation.
When it is low, we …nd that political career concerns induce a myopic bias in government spending. By oversupplying public services with immediate payo¤s, politicians raise their appeal among all voters. By undersupplying public education, they lower their 1 appeal only among a subset of more knowledgeable citizens. As a result, in equilibrium the government invests too little in human capital.
When social capital is higher, however, knowledge spreads more widely across the electorate. Politicians are more likely to face electoral punishment if they mismanage public education, so they raise education spending towards the …rst best. Moreover, their equilibrium selection re‡ects more accurately their skill at managing government investment in human capital. Both better political incentives and better political selection raise the long-run growth rate of output.
Better screening also reduces its short-run volatility, which stems from the endogenous evolution of stochastic government competence. Our theory accounts for several important facts, beyond our motivating evidence that so- cial capital raises economic growth. First, social capital is also associated with lower volatility of output growth (Sangnier 2013). Second, social capital improves both politicians’incen- tives and their selection (Nannicini et al.
2013; Padró i Miquel et al. Third, frictions in political agency induce underinvestment in public education, particularly in poorer coun- tries. One reason for such insu¢ cient investment is that voters underestimate its returns and thus have a distorted demand for education (Jensen 2010; Banerjee and Du‡o 2011; Bursztyn 2016).1 Better voter information raises both the amount and the productivity of government investment in human capital (Reinikka and Svensson 2004, 2005). In our empirical analysis, we also provide evidence in support of our two distinctive the- oretical predictions that had not been brought to the data before.
First, our model predicts that social capital improves government accountability by raising voter information. We test this prediction in survey data from the American National Election Studies, which include standard proxies for both social capital and political knowledge. As predicted, respondents reporting higher trust also score higher on all standard measures of information, like their ability to name candidates and incumbents or their willingness to rate politicians’ideology. We obtain analogous results if we proxy for social capital with the propensity to discuss politics with family and friends— a less standard measure, but one particularly germane to our focus on the social di¤usion of information.
Turning from individual to aggregate outcomes, the key macroeconomic prediction of our model is that social capital raises the long-run growth rate by increasing the share of output devoted to public education. We test this prediction across countries, measuring social capital with averages responses in the World Values Survey. As predicted, government spending on education is a higher share of GDP in countries with higher social capital— whether 1 Bursztyn (2016) shows that poverty biases people against education spending, both as voters and as parents. This bias can result both from credit constraints and from misperceptions caused by the cognitive burden of poverty (Shah, Mullainathan and Sha…r 2012).
2 measured by trust or by the likelihood of obtaining information from friends and colleagues. This relationship is robust to controlling for geographic and demographic characteristics, as well as for the overall size of government. To alleviate endogeneity concerns, we follow Tabellini’s (2008) instrumental-variable strategy. Social capital is signi…cantly predicted by the grammatical structure of a country’s main language.
The positive e¤ect of social capital on public education spending is robust to the use of this instrument, which minimizes the threat of reverse causation. Our analysis focuses on the e¤ect of social capital on economic growth through higher public investment in education. The empirical importance of such an education channel is borne out by the …ndings of Gennaioli et al. Using both cross-country and cross- regional data at the subnational level, they show that social capital signi…cantly predicts economic development in a univariate setting.
However, they …nd that this association is no longer statistically signi…cant when controlling for human capital. The notion that higher investment in human capital is a key transmission channel for the causal e¤ect of social capital on economic growth is also buttressed by the leading role of human capital as a driver of economic growth in general, whether across countries (Barro 1991; Manuelli and Seshandri 2014) or sub-national regions (Gennaioli et al. 2013; Islam, Minier and Ziliak 2015). Our work is connected to several strands of literature.
Most closely, a few models formal- ize how social capital can foster economic development by facilitating market transactions among private agents. In Zak and Knack’s (2001) model, social capital alleviates agency frictions in …nancial intermediation, consistent with its empirical association with …nancial development (Guiso, Sapienza, and Zingales 2004). However, long-run growth may require a switch from transactions supported by social bonds to contracts supported by formal enforcement institutions (Routledge and von Amsberg 2003; Kumar and Matsusaka 2009; Lindner and Strulik 2015). This cautionary theoretical prediction is consistent with his- torical evidence (Greif 2006).
We propose a complementary mechanism operating through frictions in political rather than corporate agency, and we show that social capital enables higher long-run growth by improving voters’ ability to monitor their government.2 Thus, our model vindicates Bowles and Gintis’s (2002) insight that social capital and government intervention may be complements rather than substitutes. We also contribute to the broader literature in economics that provides de…nitions of social capital consistent with rigorous formal modeling. Our precise, tractable de…nition is closest to Glaeser, Laibson and Sacerdote (2002), who interpret social capital as the resources 2 Empirically, social capital correlates with higher e¢ ciency in all large organizations, whether private or public (La Porta et al. 3 that individuals can draw upon thanks to their network of interpersonal relations— a view that harks back to sociologists’original de…nition of social capital (Bourdieu 1986; Coleman 1988; Lin 2001).
Our focus on information is justi…ed by the long-standing recognition that it is among the main resources obtained through social connections (Granovetter 1973; Cole- man 1988; Lin 2001; Durlauf and Fafchamps 2005). Another product of social connections is trust, the standard empirical proxy for social capital (Glaeser et al. 2000; Valenzuela, Park and Kee 2009).3 The concept of social capital becomes ambiguous when it con‡ates social connections and shared beliefs, norms and values (Putnam 1993, 2000; Fukuyama 1995). Guiso, Sapienza and Zingales (2011) resolve this ambiguity by de…ning civic capital as the shared beliefs and values that help solve the problem of collective action.
Both theory and empirical evidence con…rm that growth-promoting cultural traits help economic devel- opment (Galor and Moav 2002; Doepke and Zilibotti 2008; Guiso, Sapienza, and Zingales 2016; Gorodnichenko and Roland 2017). Our analysis suggests that civic culture and social capital— speci…cally, the social di¤usion of information— are distinct and complementary drivers of growth. Our model also speaks to the political economy of public …nance under imperfect infor- mation. Democratic governments underprovide public goods that are shrouded from voters’ view (Eisensee and Strömberg 2007; Mani and Mukand 2007).
More opaque expenditures and taxes are also more exposed to capture by special interests (Coate and Morris 1995; Ponzetto 2011; Glaeser and Ponzetto 2014). A growing body of empirical evidence shows that policy outcomes improve when citizens are more informed because politicians become more responsive to voters’ needs (Besley and Burgess 2002; Reinikka and Svensson 2005; Snyder and Strömberg 2010). The literature has focused on the media as the main source of variation in voter information. We are the …rst to highlight theoretically and document empirically that social capital plays a similar role by acting as a knowledge multiplier.
