Surveys and SpeculationsSocial Networks in Economic History: Opportunities and Challenges
Introduction
“From one level to another the ties thus formed – like so many chains branching out indefinitely,” wrote Marc Bloch, in his magisterial depiction of European feudal society Bloch (1962, pp. 466). Such a description immediately plants in the mind of the reader a concept of feudal society as a network, with individuals linked together in an increasingly tangled web of social obligation. The shape of that network – who one is ‘chained’ to – has consequences for individuals and social groups. Social network analysis offers a set of tools for engaging with substantive questions about the impact and structure of networks.
The centrality of networks to social life is obviously not limited to feudal society. Greif (2006) placed social networks at the center of trade relations, in his book on the networks of merchants underpinning long-distance medieval trade. Similarly, the social network of the English gentleman underscores Cain and Hopkins (1993) influential argument on the role of ‘gentlemanly capitalism’ in British imperialism. Thus foundational studies in economic history have often gestured toward social networks as pivotal in understanding historical processes, and yet have rarely offered formal analyses of the networks themselves. The appeal to network concepts is particularly prevalent in the areas of business and financial history, where correspondence networks between trading partners, suppliers and customers (Hancock, 2005) or interlocking directorates between firms (Fohlin, 1998) have been objects of extensive analysis. Nevertheless, the majority of empirical studies thus far deal with network effects by including in a regression one or more statistics that describe particular aspects of the network structure. As we discuss below, this may be inappropriate.
More recent work, however, has brought sharper methodological tools to bear on larger network datasets. As an example, van Dosselaere (2009) has returned to the work of Greif (2006) on Genoese traders armed with a dataset of “20,000 commercial relationships” describing the network of long-distance trade.1 Moreover, the use of network concepts and methods have changed the received wisdom on a number of topics. The pioneering work of Flandreau and Jobst (2009) on the currency geography prior to World War I confirmed that strategic externalities led to path dependence in international reserve currencies. However, it also dispelled the intuitive notion that persistence was strong enough to eventually lock-in one single currency at the center of the international monetary system. Their work inaugurated a new research strand on monetary equilibria with multiple reserve currencies (Eichengreen et al., 2018). Likewise, Jaremski and Wheelock (2017) confirm that national banks’ preferences were critical for the choice of reserve cities and the drawing of the boundaries of the Federal Reserve districts in the US. By using data on the pre-Fed network of banking correspondents, they were able to explain the origin of those preferences. The structure of the Fed therefore mirrored preexisting banking linkages which it was due to replace.
The growth of network empiricism in economic history is welcome, but the sources present novel challenges. Network studies in economic history force us to think carefully about the claims we make regarding the independence of our units of observation. Statistical inference in economic history conventionally rests upon the identification of discrete units which form the basis for analysis. And yet, economic theory stresses the strategic interactions between those units, as well as the externalities generated by the choices of people or institutions. Indeed, the interdependence of social outcomes lies at the heart of the concept of a social equilibrium. In general, however, empirical models in which all observations affect all other observations are not tractable. A convenient response is to pick those units one believes interact and construct a game-theoretic model of their behavior.
The promise of social network analysis from an empirical point of view is that it lies between game theory and general equilibrium (Jackson, 2016).2 Observing the ‘links’ – broadly defined – between units, as well as the units themselves, allows the researcher to pin down modes of interaction between agents that are more constrained than the case of general interference and more general than the assumption of complete independence. Recording the network of units helps in the estimation of models of individual strategic behavior, by revealing whom agents choose to interact with. Similarly, when estimating the effects of treatments at the unit level, the knowledge of the underlying network allows the researcher to model processes of social spillover that may either be of substantive theoretical interest in themselves, or at least important to account for in order to avoid biasing the estimate of the treatment effect.
The increasing availability of networked data – such as data from online social networks – has accelerated the trend of network studies, and in the last decade nearly every social science has reflected on the rising use of networks within their field.3 The large scale of the literature on networks thus forces us to be selective in our presentation of methodological material. We chose to make two substantial omissions. First, our emphasis will be on how to run a regression to estimate a causal effect when the data forms a network; we will not review the models to explain the formation of such a network in the first place. In certain circumstances, economic historians can use archival evidence to understand network formation without having to resort to structural models to explain it. Second, because this review deals with empirical network models, we do not cover the large literature on ‘games on networks’, although these models will appear in some of the historical papers we survey. An additional omission concerns empirical work in spatial econometrics, given that our focus is on social networks. Despite the similarities in statistical methods across the two fields, spatial methods typically take the network as given (or exogenous).4 Finally, so as to avoid unnecessary overlap with the numerous existing surveys, we focus on networks in non-strictly social settings, such as trading in markets or political conflict.5
In the next section, we examine the statistical methods to quantify network impact. Given that a network exists, how do the characteristics of some units affect the outcomes of their peers? We begin this section with a survey of commonly observed characteristics of social networks, and then turn to a discussion of network regressions. In Section 3 we survey a large number of papers that have addressed the role of networks in different areas of economic history. Section 4 concludes by suggesting some research areas in economic history ripe for the application of network methods.
Section snippets
Network impact
In this section we review the core statistical framework for estimating how units’ characteristics can affect the outcomes of their peers across a social network. We begin, however, with a brief excursion into some of the stylized facts about empirical social networks, as a way of introducing some helpful concepts, many of which are summarized in Table 1.
Networks in economic history
The preceding sections have surveyed how a researcher might estimate the propagation of effects across social networks, and some of the difficulties associated with studying network data. Despite these difficulties, economic historians have been engaging fruitfully with networks, in both causal and descriptive studies. In what follows, we survey economic history research where scholars identified networks as central to the phenomenon they studied. The studies reviewed in this section present a
Looking ahead
In this paper we have attempted to do two things. First, we surveyed the main methods to estimate the economic impact of networks. We took the perspective of the applied empirical researcher and focused on empirical methods, while leaving out structural models and the literature on strategic network formation (games on networks). Secondly, we reviewed the burgeoning literature in economic history that applies network methods, organized along four main themes: markets, financial intermediation,
Acknowledgments
We thank the editor Kris Mitchener and two anonymous referees for very helpful suggestions. All remaining errors are ours. The research leading to these results has received funding from the People Programme (Marie Curie Actions) of the European Unions Seventh Framework Programme FP7/2007-2013/ under REA grant agreement no 608129. In addition, Gabriel Geisler Mesevage gratefully acknowledges IAST funding from the French National Research Agency (ANR) under the Investments for the Future
References (119)
The perils of peer effects
Labour Econ.
(2014)- et al.
Identification of peer effects through social networks
J. Econ.
(2009) Inventing social capital: evidence from African American inventors, 1843–1930
Explor. Econ. Hist.
(2011)- et al.
Mostly pointless spatial econometrics?
J. Reg. Sci.
(2012) - et al.
Visual analytics of an eighteenth-century business network
Enterp. Soc.
(2015) - et al.
A generalized spatial two-Stage least squares procedure for estimating a spatial autoregressive model with autoregressive disturbances
J. Real Estate Financ. Econ.
(1998) - et al.
Introduction to Spatial Econometrics
(2009) Two-step estimation of network-formation models with incomplete information
J. Econ.
(2015)- et al.
The Great Divergence: A Network Approach
CESifo Working Paper
(2015) - et al.
The Anatomy of Scottish Capital: Scottish Companies and Scottish Capital, 1900–1979
(1980)
The network origins of aggregate fluctuations
Econometrica
System risk and stability in financial networks
Am. Econ. Rev.
The Social Dynamics of Collective Action: Evidence From the Captain Swing Riots, 1830–31
Cambridge Working Paper in Economics
Entrepreneurs, formalization of social ties, and trustbuilding in Europe (fourteenth to twentieth centuries)
Econ. Hist. Rev.
Feudal Society
Linear social interaction models
J. Polit. Econ.
Agglomeration economies in classical music
Pap. Reg. Sci.
Some challenges in the empirics of the effects of networks
Financial contagion in networks
Exclusion Bias in the Estimation of Peer Effects
NBER Working Paper Series
British Imperialism: Innovation and Expansion, 1688–1914
Liquidity Risk, Bank Networks, and the Value of Joining the Federal Reserve System
Federal Reserve Bank of Richmond Working Paper
Substitutes for legal protection: corporate governance and dividends in victorian britain
Econ. Hist. Rev.
Power-law distributions in empirical data
SIAM Rev.
Social capital and economic performance: trust and distrust in eighteenth-Century gold shipments from Brazil
Eur. Rev. Econ. Hist.
Systemic Risk and the Great Depression
Working Paper
Early modern social networks: antecedents, opportunities, and challenges
Am. Hist. Rev.
Dealer Networks in the World of Art
Working Paper Series
Did J.P. Morgan’s Men Add Value?: An Economist’s Perspective on Financial Capitalism
Contract enforcement, institutions, and social capital: the Maghribi traders reappraised
Econ. Hist. Rev.
How Global Currencies Work: Past, Present and Future
Applied spatial econometrics: raising the bar
Spat. Econ. Anal.
Between Monopoly and Free Trade: The English East India Company, 1600–1757
Malfeasance and the foundations for global trade: the structure of english trade in the East Indies, 1601–1833
Am. J. Sociol.
Social networks and macrosocial change
Annu. Rev. Sociol.
Social networks and port traffic in early modern overseas trade
Soc. Sci. Hist.
Networks, institutions, and uncertainty: information exchange in early-Modern markets
J. Econ. Hist.
The Square and the Tower: Networks, Hierarchies and the Struggle for Global Power
Monetary geography before the industrial revolution
Camb. J. Reg. Econ. Soc.
The empirics of international currencies: network externalities, history and persistence
Econ. J.
Relationship banking, liquidity, and investment in the german industrialization
J. Financ.
Finance Capitalism and Germany’s Rise to Industrial Power
Can collapsing business networks explain economic downturns?
Econ. Model.
Trade and Institutions in the Medieval Mediterranean: The Geniza Merchants and their Business World
Cited by (13)
A modified connectivity link addition strategy to improve the resilience of multiplex networks against attacks
2022, Reliability Engineering and System SafetyCitation Excerpt :Multiplex networks have multiple interacting layers of networks, and the same set of nodes are connected via more than one type of link, which is usually used to simulate social and infrastructure networks. In social networks, individuals are networking through numerous social relationships, such as friendship, kinship, and co-partnership, and via a multitude of communication channels, including online and offline contact [2]. Regarding infrastructure networks, critical infrastructure provides essential support for the functioning of modern society through concerted operations of multiple interlinked and interrelated networks such as energy production and supply, telecommunication, and transportation networks [3–5].
Religion in economic history: A survey
2021, The Handbook of Historical EconomicsCollider bias in economic history research
2020, Explorations in Economic HistoryCitation Excerpt :A final type of collider bias may occur when the researcher conditions on a pre-treatment collider. Elwert and Winship (2014) highlight homophily bias in social network analysis as an example, and this could certainly apply to economic history research (Esteves and Geisler Mesevage, 2019). However, another example arises when the researcher only observes individuals (or firms) after they have made a choice about where to live (or locate).
The Feudal Origins of Manorial Prosperity: Social Interactions in Eleventh-Century England
2023, Journal of Economic HistoryMarket Access and Information Technology Adoption: Historical Evidence from the Telephone in Bavaria
2023, Journal of Economic HistoryApplication of Social Network Analysis in the Economic Connection of Urban Agglomerations Based on Nighttime Lights Remote Sensing: A Case Study in the New Western Land-Sea Corridor, China
2022, ISPRS International Journal of Geo-Information