Economists vs Historians on Economic History

Economists such as Gregory Clark and Brad DeLong like to tell people that between the dawn of time and 1800 or 1900 there was no growth in GDP per capita and very slow population growth. As an ancient historian this leaves me scratching my head. I decided to write this post after reading the 1998 version of DeLong’s ideas but similar ones appear to be common.

There is quite a bit of historical evidence against DeLong’s specific assumptions, from “people in the past were all really short” (see J. Geoffrey Kron, ‘Anthropometry, Physical Anthropology, and the Reconstruction of Ancient Health, Nutrition, and Living Standards’), to “women in the past had no control over their fertility” (see Janice Liedl or chapter 7 of Barbara Bush, Slave Women in Caribbean Society, 1650-1838) to “conditions in 18th century Europe existed everywhere from time immemorial” (I hope that I don’t need to explain the problems with this one). DeLong writes that he has changed his mind on some things, and I hope that he means that he is familiar with some of this evidence. Yet my biggest surprise is that economists consider this a worthwhile project in the first place.

First, in the ancient world only a tiny fraction of all goods and services were exchanged for money. Economists such as Diane Coyle routinely warn that GDP is a flawed metric in industrial economies because it does not account for parts of the economy which are not monetarized, and sometimes fret about the data-collecting problems which saw Nigeria’s GDP rise 89% in a year after a change in methodology, so I don’t understand how they can use this metric for preindustrial ones. (A few historians such as Walter Scheidel do attempt to estimate average incomes in terms of silver or grain in well-documented societies such as Augustan Italy or late antique Egypt, but they usually acknowledge the limits).

Second, the evidence is not sufficient to construct a history of world population before 1900. There is simply too little census data, and archaeological evidence for population is always ambiguous. Serious estimates of the population of the Roman empire under Augustus based on surviving census figures differ by a factor of three (Walter Scheidel and Geoffrey Kron are two prominent participants in this debate), while estimates of the population of the New World in 1492 based on archaeological remains and estimates by early travellers and colonists differ by a factor of twenty. Estimates of world population history such as the book by McEvedy and Jones tend to assume that population should increase exponentially, and chose numbers accordingly, so they cannot be taken as evidence that the population did in fact increase according to a smooth exponential function. That would be a circular argument. (This problem also affects estimates of ancient incomes and income distribution, since ancient historians tend to adjust their models in keeping with what economists tell them is reasonable, while economists appeal to ancient historians’ estimates to support the economists’ generalizations about ancient societies).

Third, even on purely material grounds, I am bewildered to see conditions in the Paleolithic equated with those in urbanized, literate, agrarian societies. The average person in 10,000 BCE probably lived with a few dozen close relatives, had a literal handful of possessions of stone and organic materials made by generalists from resources within a few dozen miles; his counterpart in Italy under Augustus lived with a few thousand people in a sturdy brick building and had several boxes’ worth of possessions made by specialists from materials drawn from hundreds of miles away. He might eat worse than the forager and be sick more often, but that only strengthens the impression that material conditions in those two cultures were profoundly different. It is not hard to imagine people preferring one way of life over the other, and I suspect that a bit of work would turn up examples in recent times of people who grew up in one kind of society but found the other more attractive. If estimates of GDP cannot capture this difference, so much the worse for them.

Lastly, focusing on the “global GDP” at dates thousands of years apart can obscure that the road between those two was rather bumpy. Ancient economic history was not marked by endless sameness but by an alternation of good years, famines, growing towns, and wars. Somebody born at the beginning of a period of prosperity could expect a very different life than someone born in a period of decline, and wars or revolutions or the founding of a town could have very serious consequences for good or bad. In the long run, everyone dies, but somehow we find what happens in between worth talking about.

In short, world GDP per capita is the wrong metric for the ancient world, we don’t have data to calculate it anyways, and stressing the slow long-term change on the scale of continents rather than the visible change on the scale of single countries and human lifetimes is incomplete. Economic historians are just as much the slave of economists as Keynes’ “practical man” was, but I wish that more economists would show that they take ancient evidence seriously. While I appreciate economists’ desire for generalization and abstraction, generalizations ought to fit the facts.

Further Reading: David Graeber’s Debt seem to have failed to make more economists take the world before 1500 seriously, although it has inspired them to educate the public about the history of Apple Corporation and the structure of the Federal Reserve. Jac J. Janssen, Commodity Prices From the Ramessid Period (E.J. Brill: Leiden, Netherlands, 1975) is an entertaining description of economic life in one ancient society; in my view anyone who wants to understand economic life before the eighteenth century should have conditions in several specific societies firmly in mind before they read their first grand theory or look at estimates of population and income. I discuss these issues at greater length on pp. 60-72 of my master’s thesis. I wrote this sometime in 2014 but decided to post it after reading another philosophical blog post by Brad DeLong (link). For other examples of ideas similar to DeLong’s see eg. William Bernstein‘s The Birth of Plenty or Gregory Clark’s Fairwell to Alms (published by Princeton University Press!).

Edit 2016-01-01: DeLong now cites Roger Fouquet and Stephen Broadberry, “Seven Centuries of European Economic Growth and Decline,” Journal of Economic Perspectives 29:4 (Fall 2014), pp. 227-44 doi=10.1257/jep.29.4.227 as showing that GDP per capita was changing noticeably in most European countries in most years between 1300 and 1800. While I am still not convinced that GDP is the right metric, I think this is a step in the right direction.

Edit 2016-01-29: See now François de Callataÿ (ed.), Quantifying the Greco-Roman Economy and Beyond. Pragmateiai, 27. Bari: Edipuglia, 2014. Pp. 260. ISBN 9788872287446. €60.00. (non vidi sed legi recensionem apud http://bmcr.brynmawr.edu/2016/2016-01-24.html)

Edit 2016-03-09: A BMCR by Kostas Vlassopoulos of the University of Crete accuses a recent book by Josiah Ober of using ideas drawn from American pop economics to tell stories about the Athenian economy without making sure that they fit the sources, particularly the sources for parts of the ancient world other than Athens. If his assessment is accurate, I still think that professional economists are much more likely to make these mistakes than professional historians.

Edit 2016-05-11: DeLong is now preparing to publish an essay which appeals to “the skeletal evidence that finds adult humans around the year 1 little more than five feet tall.” He cites an article from 1995 “Stature and the Standard of Living” by one Richard H. Stecke (link). After a quick skim, I don’t see any precise claims about heights anywhere in the world before the 18th century in that paper, and that paper is certainly overwhelmingly focused on the past 300 years and emphasizes that average heights in any given year vary from place to place.

Edit 2019-01-02: Cosma Shalizi of http://bactra.org/ reports that Oskar Morgenstern, On the Accuracy of Economic Observations (1950) argued that many of the economic statistics published in his day were at least 5-10% in error

Edit 2019-08-20: From the mouths of professors! Guido Alfani and Cormac Ó Gráda, “The timing and causes of famines in Europe” Nature Sustainability (2018) p. 287 https://doi.org/10.1038/s41893-018-0078-0 “We filled in all gaps in the estimates by assuming exponential growth rates between observations, which is the standard procedure in similar instances.” As long as their methods allowed for S curves and exponential decline, that seems reasonable, but it does show how population data is generated by assuming exponential growth, rather than that growth coming out of what passes for data from before 1800.

Edit 2019-09-01: Noel Maurer has collected research showing that GDP per capita or productivity in eastern Canada tends to be 70-95% the level of neighbouring parts of the United States from the 18th century to the present (Vincent Geloso, “Distinct within North America: Living Standards in French Canada, 1688 to 1775” https://ssrn.com/abstract=2885514). To me, this is evidence that not all low-tech economies are the same, and that we should expect that similar differences existed between say provinces in the Roman empire.

And for a thoughtful version of the position I reject, see T. Greer, “Notes on the Dynamics of Human Civilization: The Growth Revolution, Part I” (4 August 2010) http://scholars-stage.blogspot.com/2010/08/notes-on-dynamics-of-human-civilization.html {Greer presents his readers those smooth increasing figures for world population and GDP- David Henige’s ‘numbers from nowhere‘- rather than the cyclic figures based on some kind of evidence from specific places like England}

As I write, the Wikipedia page in World Population Estimates https://en.wikipedia.org/wiki/World_population_estimates is pretty good (it admits that we have no censuses for most of the world until well into the 20th century, and that estimates for the world since 1950 vary by at least 1%, and one of their alternatives to McEvedy and Jones fits its history of the population of the world into just thirteen pages (!!!)) As John C. Caldwell and Thomas Schindlmayr put it

There is often an unwarranted circularity between the analytical approach and the major findings. All the calculations from 1750 onward are primarily anchored in modern census taking and assumptions about the most likely growth rates that led ultimately to the census numbers … Among the authors of the midcentury consensus, only Willcox (1931: 639-644) listed, and briefly discussed, sources (in six pages! – ed.) … In 1904 … the consensus on nineteenth-century populations and on the sources to be employed for earlier estimates was largely in place for further work to be done in the twentieth century. What is important to keep in mind, then, is that population census figures began to be available for America, considerable parts of Europe, and Australia from around 1800 and, for much of the rest of the world, only in the twentieth century. Earlier estimates were either informed guesses based on travellers’ observations (or, in China, imperfect counts) or, quite commonly, biblical calculations.

“Historical Population Estimates: Unraveling the Consensus,” Population and Development Review, Vol. 28, No. 2 (June 2002), pp. 183-204 https://www.jstor.org/stable/3092809

Edit 2020-03-01:And Ea and Nisaba help us, a paper published in 2020 still uses McEvedy and Jones’ population estimates: Koyama, Mark (2020) “A Review Essay of Escape From Rome.” The Journal of Economic Literature https://papers.ssrn.com/sol3/papers.cfm?abstract_id=3544313 We historians have just utterly failed to communicate what we do and whose claims to interpret historical data can survive five minutes of gentle questioning!

And Brad DeLong, gods bless him, still thinks population growth follows mechanically from nutrition when a first-year anthropology textbook will run you through all the ways paleolithic, neolithic, and gardening societies manage their fertility:

From 1680 to 1780 the population of Massachusetts doubled every forty years: that is what a nutritionally unstressed pre-industrial population does. Living standards? Perhaps 4 to 5 dollars a day as today’s development economists measure things, figure 1500 real (and 50 nominal) dollars a year per capita. At that level of prosperity the Massachusetts population was then nutritionally unstressed. The Roman imperial population was definitely nutritionally stressed: we think it was the same size in 200 as it had been in the year 1, and the population of the area that had been the year-200 Roman Empire was down by 25% come 600. Three great plagues (antonine, 165-80, St. Cyprian 249-62, and Justinian 541-2) played a role in the decline: the Roman population from 150-600 was not living below long-run subsistence. But it is very hard looking at the population history to ascribe a standard of living of more than 2.50 dollars a day per capita to the typical inhabitant of the Roman Empire.

We have utterly failed! I am serious here, I have been trying to reach people in different formats and they just do not show they understand our criticisms.

Edit 2021-12-30: DeLong claimed that pre-industrial populations will ”naturally’ double every 25 years unless restrained by hunger and disease in a chapter cut from Slouching Towards Utopia in 2021 https://braddelong.substack.com/p/outtake-the-human-economy-10000-to But in recent times, high mortality triggers high fertility! Birth rates fall when people are more confident that their children will survive to adulthood. In this post he also repeats his strange claim from 1998 about heights which is not supported by any evidence I have seen. In a later blog post he posted something which is closer to the consensus of studies of skeletons (typical adult male heights in premodern Europe and the Mediterranean around 165-170 cm not 155-160 cm), and I hope that corrected version appears in his book Slouching Towards Utopia.

Edit 2022-10-25: Once again de Long pronounces that “The typical human before 1870 would see their average son grow up stunted: 5’,3“ or 5’4”” In an email exchange I showed him evidence from the ancient Mediterranean which contradicts this, and he did not have evidence to support it (and average heights are average, there are short Dutch men and tall Vietnamese women). I am not sure whether evidence can ever penetrate this. https://braddelong.substack.com/p/a-semi-platonic-dialogue-on-slouching

Edit 2020-05-30: For wages in the 17th, 18th, and 19th centuries see now Hatcher, John / Stephenson, Judy Z. (eds.) (2018) Seven Centuries of Unreal Wages: The Unreliable Data, Sources and Methods that have been used for Measuring Standards of Living in the Past. Palgrave Studies in Economic History (Palgrave MacMillan)

Edit 2020-12-18: For a rantier version of this post focused on a paper which tried to use portraits in the British National Gallery to measure changes in trustworthiness (sigh) see Mateusz Fafinski, “Historical Data: A Portrait” https://mfafinski.github.io/Historical_data/ (his other posts are prone to the exclusive ‘you’ that does not include me or my face-to-face friends though!)

Edit 2023-01-27: For another example of Henige’s numbers from nowhere and Ober’s social-science approach to ancient history, see Branko Milanovic, Peter H. Lindert, Jeffrey G. Williamson, “Pre-Industrial Inequality.” The Economic Journal, Volume 121, Issue 551 (March 2011) pp. 255-272 https://doi.org/10.1111/j.1468-0297.2010.02403.x (purports to calcualate Gini coefficients for societies back to 8000 BCE!!!)

Edit 2023-07-22: Michael E. Smith, “Comment: Income and inequality in the Aztec Empire on the eve of the Spanish conquest” https://pubpeer.com/publications/063084EB3B247130A804B2BD326F2E#

This model of inequality in the Aztec Empire is not based on empirical data. While there is nothing wrong with hypothetical models per se, the paper is phrased as if it presents empirical findings…. One indication that the results of Alfani and Carballo may be out of line with the realities of Aztec inequality is the comparison of their Gini figures with archaeological Gini estimates (Table 1). The local contexts for which my colleagues and I have calculated wealth-based Gini measures (Smith et al. 2014) come from two Aztec provinces (Cuauhnahuac and Huaxtepec). The authors’s Gini values for these provinces are far higher than any of the archaeological or historical cases. … My basic view is that the population data are inadequate, and the social table data (that is, the assignment of income shares to the categories) are not sufficiently based on data from Aztec Mexico. There are simply not enough data available to do the kind of analysis presented in this paper. The tweaking of data and methods do not produce results that satisfy me as being reasonable estimates of the level of inequality in the Aztec Empire. Perhaps this is just an epistemological difference between our approaches to science and knowledge. Economists might look at this paper as a fine analysis, whereas archaeologists and historians will probably look at it as a study based on hypothetical data, and therefore divorced from the Aztec reality that we study.

And bonus points for explaining to them that the population estimates in McEvedy and Jones 1978 are made up and strike specialists in the ancient world outside Europe and China as implausible.

Edit 2023-07-30: a good example of logical fallacies from a pundit-economist and an academic economist struggling with the failure of the biometric evidence to meet their expectations that production of durable goods = biological standard of living

After the Fall of Rome, in what had been the western provinces of the Roman Empire trade, the division of labor, urbanization, production of conveniences and luxuries, population, and total production were at a much lower level indeed—it truly was a “Dark Age”. But thighbones tell us that the adults who lived in the Dark Age were taller and better-fed than their predecessors under the Roman Empire.

Perhaps this was because the end of the Roman Empire had seen the end of a cruel and oppressive aristocracy, and was a liberation of the people—there were many fewer slaves, and many many fewer plantation slaves worked to near-death. But it is more likely that life became nastier and brutish and more dangerous, but that depopulation did increase farm and pasture size and so produce better nutrition even though the collapse of the Roman Empire’s economic network meant lower overall average living standards—the average farmholder was distressed enough by the collapse of the Pax Romana that he was willing to give up his and his family’s free status and become a bound serf of the local landlord …

https://braddelong.substack.com/p/podcast-hexapodia-is-e-key-insight-7fd#details

Edit 2023-09-01: one economist gets it! Timothy W. Guinnane, “We Do Not Know the Population of Every Country in the World For the Past Two Thousand Years,” The Journal of Economic History, Vol. 83, No. 3 (2023) pp. ??? Preprint at https://www.repository.cam.ac.uk/items/6c7f7b04-c2c4-4e34-a3b1-bbe740ddb4df/full

Edit 2023-12-22: ACOUP post on the evidence for the three parts of the ancient Mediterranean world where we have evidence (Egypt, the Greek poleis, and Republican and Augustan Italy)

Edit 2024-02-10: https://retractionwatch.com/2024/02/05/no-data-no-problem-undisclosed-tinkering-in-excel-behind-economics-paper/ imputation

Replacing missing observations with substitute values – an operation known in statistics as imputation – is a common but controversial technique in economics that allows certain types of analyses to be carried out on incomplete data. Researchers have established methods for the practice; each comes with its own drawbacks that affect how the results are interpreted. As far as the student knew, Excel’s autofill function was not among these methods, especially not when applied in a haphazard way without clear justification. (regarding https://www.sciencedirect.com/science/article/pii/S0959652623022503 )