The New Historical Economics is Self-Aware

What does it do? Who is it for?

Someone recently explained to me that economic history is a topics field, not a methods field. Unlike labor or econometrics, history isn’t a way of conducting research, but rather something you do.¹ As a consequence, our field tends to be a big tent. I’ve been accused of being restrictive in my definitions of what is and isn’t economic history,² which is probably fair. The first book of economic history I read—back in high school—was Joel Mokyr’s The Lever of Riches, and the second and third Greg Clark’s A Farewell to Alms and Bob Allen’s The British Industrial Revolution in Global Perspective respectively. These works tend to give one a certain impression of what economic history looks like—very much biased toward a refined cliometric history, rooted in standard economic theory but generally explicable without it.

Anyway, it’s become clear to me in the past year or two that economic history has a lot less to do with these kinds of books than I’d once thought. Of course, the questions raised by those three volumes are still some of the biggest out there, and the latter obviously remains a touchstone for much research in the British/technological history area. But the more that I read and immerse myself in the best-respected academic research of the past two decades—especially from top-5 journals—it’s become clear to me that economic history isn’t really about this anymore. And, with some reservations, that’s fine.

A few points have clarified my thinking on this head. First, I’ve been reading the Bisin and Federico-edited Handbook of Historical Economics over the past few weeks. The writing is a little rough around the edges (like my own) at times,³ but the first dozen chapters give a clear impression of what economic historians are doing today. You could answer “a lot!” and get away with it. But I’ll try to boil things down into a couple (heavily overlapping) categories.

1. Causal identification. Historians (used to) love to study causation, and economic historians more than most. Why did Europe conquer the world? Why did Britain industrialize first? etc. In chapter 12, Monnet and Velde (2021) lay out three ways that economic historians identify causality: process tracing, microeconometrics, and structural models. Process tracing lines up loosely with the old-school cliometrics that I grew up reading—it involves close reading of historical processes to establish chains of causality. An example from this morning: Napoleon III’s coup de etat demanded popular legitimacy to stay in power; the Catholic Church offered itself as a legitimating force; Napoleon sought to bolster the Church, gain a popular foreign policy success, and bolster the Ottomans by interevening against Russian Orthodoxy in the Near East; the Crimean War broke out.

2. Econometrics and Persistence. Economic history has been transformed by the Credibility Revolution in economics, which seeks to use experimental or quasi-experimental settings along with a canonical suite of econometric methods (instrumental variables, difference-in-differences, regression discontinuity) to make causal claims about economic processes. Everybody knows about AJR (2001) and their famous settler mortality instrument, but the intervening two decades has seen increasing A) creativity of instrument choices and B) use of RDDs, especially geographical. Melissa Dell’s 2010 paper on the mita region of Peru founded a cottage industry of analyses using arbitrary boundaries as policy or institutional discontinuities.⁴ Many papers in this vein are “persistence” studies aimed at studying how past variables affect contemporary outcomes (like both of the aforementioned). Chapter 9, by Hans-Joachim Voth, catalogs some of the canonical papers—on the Protestant Reformation, historical anti-Semitism, and colonialism in Africa—and provides useful discussion of when and why they fail (and no, it’s not “uniformly”). Instead, good studies convincingly explain the mechanisms of persistence and examine variables corresponding to some logical theory. He also vehemently defends himself against the “Kelly critique” on spatial autocorrelation, which I strongly recommend reading.⁵

3. Natural experiments. One of my favorite chapters in the book is Cantoni and Yuchtman’s on historical natural experiments. They outline the explosion of papers from the last decade which seek causal identification in quasi-natural experiments, when an exogenous shock is applied to one group but not a comparable other, which can be used as a control, and sort them into three categories: 1) for understanding historical processes and events (e.g. Nunn and Qian (2011) on the potato) 2) for testing economic theory (Juhasz (2018) on the Napoleonic blockade) and 3) for explaining contemporary development (persistence). Chapter 10 is devoted to unraveling the econometric logic underlying these papers, stressing that the interpretation of the coefficient of interest is contingent on the kind of heterogeneity of treatment effects. If you consider the instrument as a treatment, then your regression often recovers the effect only for those units (the “compliers”) who’ve changed their behavior as a result. The LATE-ATE distinction is inspired by the brand of labor economics pioneered by Card, Krueger, Angrist, and Imbens.

4. Models. The new historical economics may have a deeply empirical bent, but there’s still a lot of room for the use of models, which establish causality by creating a counterfactual universe from which the real world deviates by means of exogenously-injected shocks. Most of these frameworks come from macroeconomics or (more frequently) trade/urban. The canonical example is Donaldson and Hornbeck (2016) on the value of the American railway network in the 19th century. However, many primarily empirical papers are adding models—sometimes for hoop-jumping purposes, but also to either clarify the conceptual point or to make ballpark estimates of welfare consequences of the phenomena under study.

5. Culture. There are four chapters entirely about culture and/or cultural evolution; and more if you count the Becker-Rubin-Woessman entry on religion or Sara Lowes’s on field experiments. This is to be expected given that Alberto Bisin is one of the two editors, but it’s still a lot, since they’re all in the methodology section. The unifying thread appears to be an emphasis on long duree evolutionary processes, explicated through models, which lean heavily on biological analogies. Since these kinds of works are theory-guided, making only brief references to historical episodes by way of illustration, they stand in stark contrast to the empirical revolution described above. In practical terms, I think this is probably an overrepresentation based on the numbers of economic historians doing one versus the other; but maybe the theory is more influential as an inspiration for the empirical work than I know. One point (same with 6 below) is that mainstream economists seem to pay a lot of attention to this literature, even by comparison with the applied-micro papers that have made it into top 5 journals.

6. Institutions. There are also several chapters on institutions, again from a modeling perspective. Acemoglu et al. present a simplified game theoretic perspective on why institutional arrangements persist or change over time; his formula presents these decisions as a contest between interest groups with different weights jostling for position at the center of political discourse. Each faction wants to direct the trajectory of the ruling coalition toward its own policy “bliss point” and maximize the institutional gains over time, but must deal with threats of revolt, economic and technological drift, and cultural evolution which disrupt equilibria. Again, this kind of work—to my mind—isn’t something that economic historians do, but many of the ideas are testable and provocative for future research.⁶ Did the opening up of Japan to international trade force elites to grant the rule of law in exchange for defensive modernization? Did Britain extend the franchise to quell threatened revolts? Both are interesting questions that have been explored empirically.

7. Data collection. Finally, the Handbook rightly emphasizes the importance of original data collection and creation of various kinds. From digitizing Habsburg maps to OCRing old newspapers, economic historians are perhaps unique in the extent to which the sources that they bring to the table are completely original. Someone once told me that there were three ways to get your paper through: be influential, make a theoretical contribution, or find some new data. The practice of collection and creation is now taken very seriously, and it’s a notable step that several chapters offer practical advice on how to engage with oft-used sources (the Murdock map) or build your own.

As a crotchety old man-in-waiting, I’ve grumbled before about the unwillingness of economic historians to engage with the big questions, favoring empirically tractable situations with microdata and sufficiently exogenous shocks to use econometric methods. But that’s not strictly true—I mean, there are empirical papers on the origins of agriculture, the Great Divergence, and the Weber thesis—and even if it were on average, people who think seriously about this sort of thing get that and want to do better. Cantoni and Yuchtman, for example, note that historical natural experiments are especially prone to the above critique, and add that the quest for quasi-experimental exogeneity often amounts to producing an explanatory variable that is itself inexplicable—i.e. suggesting that the historical process is driven by something random, because it gives you a lottery-ish design. They and Voth completely agree with the critique that bad persistence papers have no mechanisms, pick variables willy-nilly, and ignore history.

To conclude, I think it’s well that this handbook is not called the Handbook of Economic History, or of Cliometrics, but of Historical Economics. By and large, the economic history that the book describes—heavily quantitative, practiced predominantly but not exclusively in US economics departments—is by and for economists, using historical settings and data. You have development economists, political economists, and trade economists—and then you have historical economists. They’re not the same thing as economic historians.⁷ Maybe that’s pedantic or obvious to you, but I found this very confusing. Many papers may be narrow-minded in terms of historical contribution, but economists… don’t care that much. What they are looking for is a theoretical innovation or an empirical test of said theory. Bisin and Federico don’t claim that their selection of articles is representative of all economic history, but rather of an increasingly influential brand that finally has some sway in mainstream publications. This is borne out by the self-conscious survey in Chapter 2. Field journals are not dominated by “advanced methods”—IV, RDD, and DiD make up a growing fraction, but still a minority. Conferences still have long sessions about the minute details of real wages, textile quality, and coin metallurgy. And so long as folks remember that solving economic problems is only part of the game, I think it’ll stay that way.

1. Like development, for example, and to a certain extent trade, although the latter carries some methodological baggage in terms of theoretical modeling.

2. Chalk it up to narrow-mindedness—I need to order and compartmentalize to stay sane.

3. Maybe I have a preprint.

4. The idea being that if underlying unit characteristics vary smoothly across the boundary, then being on one side or the other of the border can be viewed as a treatment, for which the control group consists of units immediately on the unaffected side.

5. The main points being that Kelly a) uses the weakest variables in each paper b) uses the first regression in each paper, which is usually OLS without geographical controls and c) uses an overly-restrictive test (regressing noise on noise).

6. Indeed, every paper that I’ve written has at least been in part motivated (positively or negatively) by some Acemoglu theory.

7. Whose main goals are studying economic questions in history, a la social or military historians.