History and Institutions: It’s Complicated

Institutions are a common theme at Broadstreet. Many of us study how institutions have shaped political, economic, and social outcomes in the past. We are interested in different aspects of this problem and we use different approaches. The friction between these produces some sparks, but we needn’t get too distracted by the more adversarial aspects of ‘the relationship’ (between history and the social sciences). The approaches of these disciplines ought to be seen as complementary, rather than in conflict with one another.

As I mentioned in conversation with Scott, the questions that motivate my research are many of the same ones asked by social scientists, but I decided to study history instead. What especially appealed to me about the historical approach was the scope it offered for investigating the way society functioned as a whole; to be able to look at a society as a coherent entity and think about how the various pieces we observe in the surviving documents might have fitted together.

This approach is especially suited to research on institutions, for two reasons in particular. First, it has become evident since researchers turned their attention to institutions that the ones first identified by North et. al. – property rights and contract enforcement mechanisms – varied (and still vary) enormously from society to society. This is true even within western Europe, which is too often treated as a monolithic entity in histories of economic development. And the same goes for other institutions highlighted in the literature, including legal systems, serfdom, parliaments, guilds, towns, and communes. Second, and related to this, is the realization that institutions do many things; their functions and their effects are often intertwined with those of other institutions. They are, to borrow a term from anthropology, ‘multi-stranded’; they are connected in many ways.

The term ‘multi-stranded’ is mainly used to characterize relationships or transactions in pre-modern societies. They are portrayed in contrast to ‘single-stranded’ relationships, which characterize modern market societies: I go to the bakery, I buy my bread from the person in the shop, I leave. I don’t know the baker, the baker doesn’t know me – it’s a one-off exchange, a single strand. In a pre-modern society, the baker might sit on the village court with my husband, and his wife may be the daughter of the local miller, to whom my son is apprenticed. We all attend the same local church. What happens in the bakery is not unrelated to what happens at the mill, at the local court, or at church – the relationship has multiple, mutually-reinforcing strands.

Institutions can also be multi-stranded. Historical demography offers a good example. Demographic decisions are taken in conjunction with a range of others, including those about where to live and what kind of job to have. The simple question of why people married when they did – a subject that has caught the attention of social scientists in recent years – turns out to connect with many other questions, and ends up leading us on quite a comprehensive tour of pre-modern society and its institutional framework(s). The social science research on variation in age at first marriage has focused on more straightforward economic hypotheses, in particular, the idea that later ages at first marriage implied greater economic opportunities for women and, in this way, signaled greater wealth or levels of economic development. There is abundant evidence from modern societies to support this conjecture; it is hardly controversial. And there is no doubt concrete economic considerations were important to marriage decisions in the past. But institutional constraints also affected a person’s decision-making calculus.

There were, for instance, political constraints. In many societies throughout Europe, landlords, guilds, and communities had the power to regulate marriage and/or household size. Landlords in eastern Europe under the second serfdom could insist that peasants on their holdings marry by a certain age. Similar demographic constraints were imposed by landlords in tenurial landscapes not characterized by serfdom. Guilds, most notably in parts of central Europe, had the power to regulate marriages as well, by insisting on marriage as a requirement for advancement, and/or controlling the timing of marriage. Communities often placed constraints on marriage decisions, either by, at one extreme, requiring that permission to marry be obtained in advance, or by regulating access to land, labor, credit, mobility (requiring permission to depart or settle in a locality) and poor relief in ways that influenced demographic decisions. Inheritance practices and quality of dispute resolution mechanisms were also important factors and also varied from place to place. All of these affected the functioning of local factor markets, which mattered for decisions about whom, when, and where to marry.

There was also the social significance of marriage in traditional societies. The decision not to marry was much costlier in the past. Marriage, as I said, was often a requirement, formal or informal, for economic or social advancement. Marriage signified adulthood, i.e. full membership and rights of participation in the larger community. A superficial glance at marriage as an institution reveals its many sides: it played a powerful role in determining property rights within families and across generations, it determined custody rights to children, it determined who was economically responsible for whom, it helped to define insiders and outsiders in a society; marriage was a legal status which conferred certain privileges and obligations. While the penalties for remaining unmarried varied considerably from place to place, I can’t think of a place for which we have evidence where high social costs were not borne by the never-married population (especially women).

In the larger political economic context, this often played out in the following way. A local ruler granted monopoly privileges to certain corporate entities (noble landlords, guilds, towns, communes) in order to obtain rents or services from these groups. As part of this arrangement, the privileged groups could impose rules about marriage and household arrangements, about mobility and rights of settlement, about access to particular markets, about access to local resources, and about rights to welfare provision. Those rules were, in some cases, mutually reinforcing. In other cases, they exerted pressure in opposing directions. The people living in these societies were navigating all of these at once.

But wait, don’t the “fancy research designs” of social science make it possible to deal with these complexities? Statistical techniques can cope with many correlated variables. The influence of some variables can be controlled for in order to isolate the effects of others. Various sophisticated interventions can be employed, right? Yes – so long as the researcher creating the model has an understanding of how the pre-modern society in question worked, which effects she will need to control for, how they might have varied from place to place, and what good data sources/proxies for those would be. Many assumptions are made when designing a model; I think historians worry even more about those assumptions than about the origins of the source that generated the data. It worries us when we see one strand being plucked out of this web and identified as causal. Not the exercise in principle – social scientists do have powerful tools for investigating causal relationships, and these have been used to shed much light on economies and political systems. Nor is it causality in itself that’s the problem. It’s that institutions, at the local level, tend to work together, and it can be difficult to extricate the one that is having the observed effect. And if it’s the wrong one, the wrong conclusions may be drawn. For instance, “Serfdom was abolished, but peasants stayed poor. Maybe institutions didn’t matter to rural development.” Or maybe there were less visible institutions that dovetailed with landlords’ privileges – communal land tenure, barriers to entering certain markets, settlement restrictions in towns, the segregated legal status of peasants – and these hardly changed at all.

I chose the demographic example, because it’s fairly straightforward to sketch out the multiple “strands” that connect a demographic outcome to a range of different underlying institutions. But when the variables involved are complex and unwieldy like “property rights” or “legal systems” or abstract like “individualism” or “patriarchal values”,  there is even more scope for arbitrariness in assumptions made and proxies chosen. And bigger concerns arise if these are being applied across a number of institutionally heterogeneous societies.

But this is where the complementarity of the different approaches comes in. The different toolkits of historians and social scientists enable them to approach such problems from different, complementary angles. Social scientists use statistical methods to get at questions of causality. Historians use their sources (textual and quantitative) to situate their questions in a larger social (and/or economic and/or institutional) ecology. Understanding how things fit together often is the historian’s question. Historical research can provide social scientists with the lay of the land: how the society worked in practice, all the things that might have influenced a particular variable, and what kinds of assumptions can plausibly be made for this time and place. And social scientists can use this to formulate more precise questions and zoom in on particular features. The findings generated by their analyses can feed back into future historical research in HPE-related fields. If quantitative research, for instance, indicates that feudal arrears could not have been driving peasant rebellions, a historian might want to figure out why so many people at the time were writing about the crisis in feudal arrears. And so on, in mutual reinforcement. A productive division of labor.

In future posts, I intend to further explore this notion of multi-stranded institutions, using concrete historical contexts and specific problems that interest researchers in HPE. At the same time, I hope to draw attention to the considerable variation in the way institutions functioned from place to place, as revealed by micro-level historical studies, and discuss the implications of this variation for various research programs in HPE.


  • Tracy Dennison

    I am Professor of Social Science History at the California Institute of Technology. I am interested primarily in the character and role of institutions in premodern societies, and how they affected the decisions ordinary people made in their everyday lives. To date, my work has focused on the role of entities like states, landlords, communities, and households in central and Eastern Europe before 1900. My current research project examines the relationship between state capacity and the rise and decline of serfdom in Europe, in particular the cases of Prussia and Russia.

3 thoughts on “History and Institutions: It’s Complicated

Leave a Reply