Is There a Theory of Economic History?

Joel Mokyr
Northwestern University

April 15, 1999

One of the books I remember discussing at great length in the Economic history discussion sections that were part of Bill Parker’s course in Strathcona Hall, led by this well-known economic historian Paul Joskow, was John Hicks’s Theory of Economic History. This year is the thirtieth anniversary of the publication of this little gem. What does a theory of economic history consist of? Hicks noted that very few scholars have even ventured to answer the question. Looking at the very few who did since the heady days of Karl Marx and Oswald Spengler, we get an idea why. It is hard to do, and Hicks, who was widely read in general history -- to say nothing of economic theory -- paid little heed to what was at that time already a huge body of literature in economic history. He could not be bothered with details, because he was interested exclusively in general tendencies, trends, moments. A few observations that did not lie on his curve did not bother him, he said. Perhaps this was because he was not dealing with theory as it would be understood today: tight logical propositions formulated as theorems in which relationships between variables are derived deductively from assumptions. That kind of theorymay be useful, but will of course not provide a general theory of economic history. But Hicks was looking elsewhere.

What Hicks meant by a theory of economic history is something close to what Karl Marx did, namely some general ideas taken from economics to be applied to history, so that the pattern he thought he saw in history had some extra-historical support. A re-reading of his book suggests that in that he was at best partially successful. While the book is rich with insights and neat stories, it is not easy to find a great deal in it that could not have been worked out by a level-headed historian who had not written Value and Capital and “Mr Keynes and the Classics.” It is a sequence of clever essays on the emergence of the market, the finances of the sovereign, the importance of city states and so on, and while they are of course informed and enriched by the economic insights of one of the greatest minds of our profession, there is little evidence in these essays of formal theory. There is plenty of informal theory in his book, but perhaps informal economic theory these days is a contradiction in terms that only Douglass North can get away with. Unlike Marx, who really did construct a theory that encompasses all of history, Hicks’s book does not deliver on its promise.

In the thirty years since Hicks’s book, thousands of economic history papers have been written that applied formal some sort of economic theory to problems of economic history. Yet there have been few attempts to write a general theory of economic history (as distinct from ambitious syntheses on the “Rise of West” of which we have half a dozen a year), and the few who have tried have been widely regarded as crackpots madly in love with their own tautologies. If the project of finding a “theory of economic history” in current economic thinking has been disappointing, it is above all because in a sense economic history is bigger than economic theory, vastly bigger.

The fundamental reason for that is that economic history is anchored in facts, not in logical constructs of “stylized” social processes. When a theorist thinks of “an economy”, she thinks of a construct consisting of something like overlapping generations of representative agents with quadratic utility functions. An economic historian thinks of medieval peasants or factory workers in Imperial Germany, but immediately has to account for landlords and open fields, or Bismarckian social reform as well. This makes it “Big” and hard, and we cannot crank out easy results. To use the terminology of modern complexity theory, a lot of heterogeneity will give you a lot of interaction effects, which stand in the way of emergent properties. 

This proposition that economic history, unlike economic theory, is funda­mentally constrained by the facts of history as we know them has come under fire in recent decades in the humanities. Facts, we have been told by the postmodernist reaction, are themselves social constructs, records are written by people who have a particular notion of what they are describing and reporting, and then read and interpreted by people who are themselves conditioned by their social environment. What we kid ourselves as being “facts” are really “representations” -- reflecting power structures, prejudices, ignorance, greed, and what not. This historical nihilism has fortunately made little inroad in economic history. While economic historians have always cultivated a healthy suspicion of what the data told them and how contemporaries interpreted them, they seem immune to the notion that the facts did not exist. So we try to get close, and if there is an irreducible epsilon between us and how it really was, so be it. To pick an example, we may not know precisely who first made a mechanical clock in late 12^th century Europe but we know that somebody did. Or we may disagree what GDP was in England in 1688, but we agree that a finite and countable number of goods and services were produced in that year and that if only we could measure it and attach market prices to them, we could compute it. The body of historical fact included in economic history is vast, and growing everyday. Yes, most of these facts have a confidence interval around them, but they are not infinite.

The task of “theory” is to make sense of these facts and to help us pick and choose among them. Economic Historians are overwhelmed by data and facts, surrounded by important questions of “how” and “why.” Theory builds the connection. But there is no single theory that can possibly do that for us. The best we can do is to choose a particular issue we are interested in, and then search for a theory that seems most applicable to help us sort through the evidence and build a model. By a model I mean, quite loosely, a logical construct that connects endogenous with exogenous variables. In history, however, very little -- some would say nothing at all -- is truly exogenous. So we inevitably chop it up into manageable pieces, so that for the analysis of an event at time t, we assume that whatever was at t-1 needs no further explanation. This way of going about things is inevitable, but it also means that a truly general theory of economic history will be hard to construct. For instance, if we are to follow David Landes who has maintained in his recent book that the economic success of the West was due to a different culture, we have to allow for the exogeneity of culture. If we then link the acquisitive and aggressive culture of the West, say, to the Judeo-Christian religion in the traditions of Lynn White and Max Weber, we have to explain those. The problem was termed “the colligation problem” by my late Northwestern colleague Jonathan Hughes a year after the publication of Hicks’s book, in his Industrialization and Economic History. Perhaps something like the ubiquity of Marx’s historical materialism or Toynbee’s “Challenge and Response mechanism” (ridiculed by Bill Parker in a memorable lecture) will eventually resolve to provide a general set of tools applicable to all situation, but don’t hold your breath.

Hicks was the pre-eminent theorist of his generation. If he could not find in the formal models of theory much that would serve as a building block for a general theory of economic history, can we do better? My own judgment is not optimistic. If anything, the ability of economic theory to help the historian in the past thirty years had diminished. In 1969, most standard theory was still based on a world that had some reassuring properties, namely that the rationality of individual decision makers and the characteristics of the environments in which they operated promised that in most cases there was an equilibrium, that it was stable and unique, and that it usually would be a “good” (i.e., efficient) outcome. Hence, if economic history observed that British entrepreneurs were slow to adopt the basic steelmaking process or American farmers were slow in adopting the reaper, this clearly must be explained by some hitherto unobserved constraint. As Alexander Pope said, whatever was, was good, and the forces of competition assured that inefficiency could not survive in the very long run. As Bill Parker remarked, economic history shows that the market did it, and then did it again. Hence the analysis tended to be focused heavily on markets and market-like phenomena.

In the generation that passed since 1969, economics has become more flexible and in some ways more realistic. Game theory has taken over microeconomics, and what we used to call macro is now “economic dynamics” and looks suspiciously like what used to be micro. It is however an economics of multiple equilibria, of coordination failures, of unit roots and persistence, in which small perturbations can steer you off course into a variety of bad traps. A host of theoretical insights imply that very similar initial conditions can lead to vastly different outcomes, and hence we simply cannot justify historical outcomes as “logical” let alone “inevitable.” The changing emphasis in economics has been away from simple market-driven processes, in which large numbers of competitors eventually drove the system to efficiency of sorts, to game-theoretical or information-theoretical models of strategic behavior in which all bets are off. The same is true for a variety of new macro growth models with increasing returns, critical masses, various forms of learning and frequency dependence, strong complemen­tarities, externalities and other forms of positive feedback. Theory has gained in realism but has lost in neatness and predictive power. Economic History still looks at theoretical work, but the help we get from theory today seems altogether less neat, less clear-cut, more equivocal. Almost any outcome can be rationalized as a coordination failure, or a principal agent problem, or a Nash equilibrium in some kind of game. The more we study theory, the messier the world looks. Economic historians already knew the world was messy, but we looked to theory for order. We find it less today that we did a generation ago.

It is perhaps for this reason that economic history is turning outside economic theory to look for some its theoretical support. Let me illustrate this with three examples. One rather obvious example is the difficult question of the standard of living. How do we compare in some sense economic welfare between two societies that are separated by time, technology, culture and institutions? Partha Dasgupta in his An Inquiry into Well-Being and Destitution and Amartya Sen in his 1985 Tanner Lectures published as The Standard of Living both point to variables outside economics as theoretically appropriate: physical health, life expectancy, political freedom, economic security are all obvious variables that ought to be included. Looking at GDP, real wages, consum­ption per capita, or even some partial measure like sugar consumption per capita all run into serious theoretical and measurement problems. Hence the growing interest economic historians have displayed in recent years in such variables as anthropometric proxies of nutritional status,the decline in infectious diseases, changes in fertility control and infant mortality, and such. But the theoretical issues involved in that literature soon take you out of economics. Height as a measure of economic well-being is intriguing, but how much do we know about the pituitary gland that regulates the growth hormone, and how sensitive it is to nutritional deprivation as opposed to the insults of infectious disease? Economic theory will be of no help here, the economic historian needs to talk to an endocrinologist. Or consider infant mortality, widely regarded a good measure of living standards: it is by now universally viewed as proof how much better life is today than a hundred years ago. Yet it turns out that the single most important determinant of infant mortality in the nineteenth century was breast-feeding practices. Why did some women nurse babies for longer periods? How were they persuaded to change their minds? Economic historians will get part of the answer from economics, particularly from a Beckerian analysis of the opportunity costs of women’s time and possibly from the kind of household bargaining models that have been proposed by Robert Pollak and hiscolleagues. Formal models of social learning, recently analyzed by Fudenberg and Ellison can also be helpful. But you may learn a great deal more by talking to your friendly pediatrician.

Another example is institutional change, a matter of great concern. Most economic historians today would agree that almost all economic growth before the Industrial Revolution and much of it afterward depended crucially on the kind of rules by which the economic game was played. Since Hicks’s little book was published, two pioneers, Douglass North and the late Mancur Olson, have tirelessly -- and at times, some would say, tiresomely -- advocated and preached for a more explicit ana­lysis of institutional change. The logic here is that in all past societies, resources were allocated ­- very inefficiently. Market arrangements worked perhaps well when they existed, but in most environments they did not, or else they were constrained or even circumvented by political processes and vio­lence, not to mention informational limitations. There were enormous opportunities for riches to be made through the gains from trade, and even minor changes in the mobility of factors and improvements in the allocation of land could create amazing gains in income. By 1650, say, the Embarrassment of Riches of the Netherlands or England com­pared to the poverty of Andalusia or Poland was almost entirely explicable in terms of better institutions.

But that takes us back to the colligation problem. Do we have a Theory of Institutions that explains why in some areas institutions evolved one way and in others in another? The idea of using the notion of an equilibrium in a repeated game as the definition of an institution is not new, but was applied for the first time to a specific historical situation by Avner Greif. Greif adds to the idea of using strategic behavior in a repeated game set-up the idea of consistency analysis, that is, how feedback from the game itself and its manifestations reinforce the set-up. This is a big step forward in solving the colligation problem, but to do so Greif has to rely insights from sociology and social psychology, relying on concepts such as “trust,” “values,” and “cultural beliefs.” In other words, he, too, has to look elsewhere beyond economic theory to make his theory rich enough to work for the historical issues he is interested in. After all, Greif’s institutional analysis tells us why a particular historical outcome can be sustained once it is achieved, but not why this particular one is chosen from all the possible ones except as a logical consequence of previous outcome. It remains an inductive analysis, with some extra-historical support for the way the analysis is set up, much like what Hicks had in mind.

Despite the depth with which Greif has analyzed the issues he is concerned with, and the deserved impact his work has had on economic history as well as on political science and sociology, the range of institutions he has been able to deal with is narrow, and clearly a general theory of economic institutions is still far away. How, for instance, do we apply Greif’s Historical and Comparative Institutional Analysis to such issues as corruption, despotism, representative political institutions, altruism and poor relief, trade associations, marriage contracts, families as allocation mechanisms, intergenerational contracting, feudal societies based on personal loyalty, the emergence of universal banking, and so on? Some exciting work is being done, but there seems to be no single set of multi-purpose tools that fits all cases of institutional change similar to the universality of supply and demand curves to all market analyses.

A third example has to do with technological progress and its unique function in modern economic history. Nobody would write economic history without technological change. It may not have been the driving force behind economic growth before the Industrial Revolution, but it was always a catalyst for change.. Technology au fond exists only in people’s minds. It is not, in the final analysis, a social construction although it often gets intricately mixed up in all kind of arrangements It involves an understanding of natural phenomena and regularities and its manipulation for one’s material advantage. The rest is commentary.

Well then, what kind of theory helps us deal with this kind of phenomenon? Formal, deductiveeconomics does not help us a lot with theories of knowledge which are quite distinct from information theory. What counts here is how new knowledge comes into being, how it spreads, and above all how it is believed, how it persuades households and firms to alter their behavior and use new technology in some form. The deep problem is that such knowledge is a pure public good yet it is not costless to acquire. Nosingle individual can know everything, not even Paul David, but the precise marginal cost of acquiring another piece of knowledge is not easy to specify. But even if we could, how do we specify a selection mechanism by which decision-makers decide which knowledge to act upon and which to discard? 

Modeling the emergence of useful knowledge in economic history is therefore a difficult problem on which I have wasted the last decade. Again the theoretical building blocks come mostly from outside economics proper. To be sure, here and there ingenious economists are trying to tackle this issue, such as Becker and Murphyin their 1990 QJE paper. But much can be learned here from the formal analysis carried out by a philosopher, Philip Kitcher, who tries to model formally the costs and benefits of adopting a particular piece of knowledge in a social setting and in the presence of imperfect testing procedure. A very different approach, but equally promising, is suggested by the work of evolutionary epistemologists in the traditions of Donald Campbell and David Hull and theories of cultural evolution such as Robert Boyd and Luigi Cavalli-Sforza. This approach takes me back to the Yale methodology of evolutionary economics, pioneered by Dick Nelson and Sid Winter. These models suggest for instance, a new “Darwinian” approach to modeling the growth of new knowledge, in which ideas are produced by some kind of stochastic process, and then subjected to a host of “filters” which decide what will adopted right away, what will be “stored away” for possible future reference, and what will be rejected and forgotten. Instead of worrying about “exogenous” variables “causing” knowledge to grow, we have a logical structure in which time-specific filters work on an available supply of new ideas. Much like any Darwinian model, these ideas imply that the history of knowledge is deeply contingent and indeterminate, and highly path-dependent. They also predict the possibility of big discontinuities and sudden radical paradigm shifts such as major macroinventions.

To sum up, then, economic history is too big to be amenable for a single theory. Its subject matter is all of material culture of the past. To write down a unified set of propositions that can help us sort all of that out without absurd assumptions and simplifications seems beyond reach. Some kind of impossibility theorem, in the spirit of Gödel’s Theorem, may well be formulated if not proven. Oddly enough, I find that re-assuring. Apart from Marx, nobody has had much success to persuade the rest of us that he had all the answers, and given the track record of that theory and its adherents, perhaps nobody should try. History should be analyzed in units of manageable size and no larger.