1 What This Book Is About
This book is about a big problem in economic history research: how to study economic development in societies that lack archival records or other written sources suitable for the familiar cliometric analyses used by economic historians. Ancient economic historians have been acutely aware of this challenge for many decades. Since the 1980s it led them to look for theoretical models to bridge the gap between their shaky empirical data and macro-level realities. In the early 2000s New Institutional Economics combined with neo-Malthusian models was hailed as the new paradigm that allowed tying together the data into a meaningful explanatory narrative of growth and development. Despite growing criticism1 and the rising popularity of climate historians,2 NIE remains the dominant framework used by ancient economic historians today.
Over the same past decades, however, economic archaeology also boomed. The close collaboration between archaeologists and physical anthropologists, climate scientists, and natural scientists in general led to huge advances in abilities to identify, collect, and interpret material remains as proxies for economic interactions and performance . Ancient economic historians enthusiastically embraced the sets of proxy data produced by archaeologists but interpreting them proved excessively hard. The datasets derive from the material record but the patterns they display are heavily determined by archaeological classifications and methodologies unfamiliar to historians. Archaeologists in turn struggle to factor in human governance, structured by institutions that are rooted in shared cultural beliefs and transmitted verbally and symbolically in ways that are invisible in the material record.
The aim of this book is to stimulate the necessary collaboration between economic archaeologists and historians to overcome these difficulties by offering a theoretical and methodological framework to evaluate and integrate archaeological proxy data and data modelling in economic history research. For reasons I explain below, this is not possible by relying only on the now familiar frameworks of mainstream economicsâwhether neoclassical, Keynesian, neo-Malthusian, or neo-institutionalâeven though they continue to provide useful insights in other respects. Instead we explore and advocate the paradigm of complexity economics, developed in the 1980s and 1990s as a complement and an alternative to mainstream equilibrium economics. I will provide a more elaborate account of complexity economics below. Suffice now to say that it studies societies and economies as âcomplex adaptive systemsâ consisting of components and agents (human and non-human) that act upon and respond to real and (in the case of humans) anticipated events. These actions produce emergent patterns that in turn feedback into the agentsâ behaviour. This paradigm, we argue, provides an appealing theoretical framework to interpret archaeological proxy data because it does not interpret these a priori as a reflection of the outcomes of linear processes of supply, demand, and distribution, determined by production factors, institutions and so on. Instead the data are treated as reflecting the outcomes of non-linear network dynamics. The advantage for historians and archaeologists is that methodologically the framework calls for network analyses and agent-based modelling and thereby opens up for analyses source data that provide only anecdotal evidence or intuitive impressions when interpreted in the light of mainstream equilibrium economics.
The rest of this introduction discusses more in detail the problems involved in the interpretation of proxy data and the importance of models and theories to address them. After briefly situating the role of models in ancient history and archaeology, I discuss the current limitations of proxy data, survey previously advocated theoretical models, and discuss the advantages of âcomplexity economicsâ.
This book is situated in the tradition of reflective works on the use of theories and models in ancient economic history.3 Classicists, historians as well as archaeologists, are often wary of theories and models. To reject them in the case of economic history, however, implies accepting the assumption that all the information we need to build a reliable image of ancient economies and to explain economic developments is locked in the scanty empirical data that are available to us, and that common sense suffices to unlock this information and to bridge any gaps that might be left. This is highly improbable given the state of our sources and the imperfections of human intellect. One may not always like them but models are a necessary evil and theories are what tie them together. That being said, however, we need to resist the temptation to cherry-pick the models we like and pretend to make sense of the data we have. The social scientists who usually make these models rarely have historical questions in mind and are generally unfamiliar with the messiness of historical or archaeological sources and the technical difficulties involved in interpreting them. It is important, therefore, that we reflect on the models and theories we use. They all have limitations and pitfalls that we need to avoid.
That, however, is equally true for archaeological proxy data, on which historians have much less reflected. Interdisciplinary approaches in archaeology the past decades have massively produced new proxy data on ancient economies. Soil and climate data reveal productivity levels. Ice-core samples show air pollution caused by mining. Skeletal remains reveal differences in diets and health status. Isotopes in dental enamel show mobility of human populations and livestock. Newly identified parts and remains document the wide spread of advanced technical devices and hydraulic technology. Sedimentary budgets of rivers reveal changing agricultural methods. The list continues to grow. New digital tools make it possible to store these data in (big) digital data collections suitable for rapid information retrieval or for feeding into visualisation or modelling software. Integrating the results in economic history research has profoundly changed the way we look at and think about ancient economies.
But the process of collecting, classifying, processing, and visualising data is not in itself revealing the dynamics underneath that caused the patterns we find in the data. It is not telling us how economic processes and outcomes are connected to broader societal structures and dynamicsâsocial, political, or cultural. Millions of sherds recorded in a database, processed through graphs, plotted on maps, will not tell what was in the minds of their makers, the obstacles they faced, and how they overcame them. What does it mean if a dataset shows a diachronic shift from wheat to barley? Does the increased mining activity documented in the ice-core samples signal high economic performance, or does it merely show that the Roman state devoted massive resources to exploit silver and gold deposits in much the same predatory way as the Spanish crown did in South America more than a millennium later?4 Are the disease burdened bones from Pompeii showing us how unhealthy urban populations were, or are they indicating efficient coping strategies for diseases that would otherwise have killed their victims? What is the outcome of all this in terms of living standards and well-being? And whose living standards and well-being are we talking of? It is hard to establish how patterns in disparate and discontinuous datasets are related to each other. What is the connection, for instance, between patterns in amphorae finds and output of wine in production areas, given that barrels and wineskins have left little or no traces in the archaeological record? It is even harder to establish how they relate to patterns in non-material historical reality. Can we link urbanisation patterns to institutional changes? How many micro-level studies do we need to make statements about trends at meso- and macro-levels?
Uncertainty increases every time we posit connections between datasets or we generalise from sample sets. Nevertheless, these questions need to be addressed. Economic history is not just about documenting change. It is about understanding and explaining it. Why was the Roman Empire so good at producing and distributing not just the basic necessities of life, but so much more that constitutes âa good lifeâ or âcivilisationâ: high-quality tools, materials, and know-how to build comfortable houses, high-calorie, protein-rich food, good shoes, basic health-care, art, and entertainment? Was it markets? Administered trade? Efficient institutions? Who benefitted from Romeâs economic success? Only elites and city dwellers or rural populations as well? And why did it stop? If Roman levels of technology and logistics were so high, why did they fall back instead of break through? Roman economic performance seems to have reached sixteenth- or maybe early seventeenth-century European levels, but these were still primitively low when compared to modern economies. The Great Divergence at that time was only visible yet with hindsight.
Quantification and digital data processing become meaningful only when the data provide reliable proxies to measure or express levels of production, distribution, changes in supply and demand, and all the rest we believe determines peoplesâ welfare and well-being. There is no reliable method to convert ancient proxy data into the economic indicators used in economics and economic history.5 Attempts to guestimate economic indicators so far have used theoretical models and comparative evidence to create âmatrices of possibilitiesâ (in the words of Keith Hopkins6) that would accommodate the sparse textual evidence we have. Given the kind and quality of data available in the material record, it is unlikely that it will ever be possible wholly to dispense with this âcontrolled conjectureâ approach. Nevertheless, the impressive amount of new empirical data documenting economic phenomena and the analytical firepower provided by new software tools cannot be ignored. The potential they offer for data modelling is huge. One of the greatest challenges today, therefore, is t...