In the previous blog, I asked why some nations are wealthy, stable, and happy, and others are not. Many theories have tried to provide an answer to this question. How do we decide which of the competing theories is true? So far, economists have not done a compelling job addressing this issue.
Let’s take Why Nations Fail, one of the best recent books by economists (or, rather, by an economist Daron Acemoglu and a political scientist James Robinson) that tackles this question. In his review in Cliodynamics Tom Currie notes that there are “two problems with A&R’s analysis as it currently stands: 1) The descriptive, case study approach adopted here makes the systematic appraisal of alternative hypotheses difficult, 2) The focus on historical contingency of the development of certain types of institutions overlooks or down-plays more general patterns about where and when these institutions have tended to develop.”
I think this criticism is fair. However, it is equally true that we simply lack appropriate data to test rival theories about the deep roots of economic development. Consider another article in the current issue of Cliodynamics, “Was Wealth Really Determined in 8000 BCE, 1000 BCE, 0 CE, or Even 1500 CE?” by William Thompson and Kentaro Sakuwa.
As these authors point out, previous empirical analyses have been plagued by a variety of problems. One is the use of modern states as geographical units, even though they may have little relation to historically appropriate units of analysis. Think of the USSR, for example – a very inconvenient unit of analysis for any historical period before 1700, when the Russian Empire emerged as a Great Power.
Another problem is how to deal with time. Some authors look at what was the situation 10,000 years ago, at the dawn of agriculture. From there they jump to 1500 BC (the Bronze Age), then to 1 AD (a pretty arbitrary date), to 1500 AD (the ‘dawn of modernity’), and finally to the present. Other analyses use different time jumps.
It’s like we have a faulty time machine. Ideally we’d like to jump back in time to the beginning of things, and then trace how they developed. So, for example, we would jump to the Fertile Crescent 10,000 years ago, and then travel forward in one century leaps, recording how everything changes. It would be like Sid Meyer’s game of Civilization, except for real. “Oh, they invented Monotheism”. “Aha, now they have Bureaucracy.”
Instead of this, eminently sensible approach, we have to endure jumps of random duration that land us in periods that may not be of critical importance to the understanding of questions we want to answer. And we pass over a lot of history between the jumps (from 8,000 BC to 1500 BC? Weren’t there a lot of interesting developments in between that we’d like to see?).
OK, enough of this analogy. We don’t have a time machine, so much the worse.
Except that we do, an imperfect and, at times, an exasperating version, but it does allow us to peek back in the past. Thousands of historians and archaeologists collectively can tell us a lot about the past – not everything, but if we could somehow put all their knowledge together, it would provide a very rich historical tapestry, which, I am sure, would allow us to reject a lot of theories – and build better, new and improved ones.
This is what’s most galling – the data that we need to test theories are there. Some of it is scattered over a multitude of published and unpublished articles. But most simply resides in the brains of historians or archaeologists specializing on particular regions and epochs. The only way to make these data useful (for a systematic testing of theories, that is) is to translate/transcribe them from human brains onto electronic, computer-readable media.
As I wrote in previous blogs (e.g., here), one of my current (and probably the most important) projects is Seshat: Global History Databank. So today I cannot answer the questions with which I started this blog. But give us a few years. Once we can trace how agricultural innovations and new ways of organizing polity, society, and economy arose and spread, we will be able to have a much better idea why some nations are rich, and others are poor.
I guess somebody has gotten this idea already. If something like it really exists it’d be nice if somebody pointed me to an online resource dealing with it.
Being that I’m currently leaning toward micro level explanations (and macro-micro integrations) what would interest me to see is a kind of map of social norms evolution. I’m not sure how it should be done, being that the experimental quantifying data in this area (trust, ultimatum etc. games) have been available for only a couple of decades (not that I think game theory data is enough, but there is much anthropological information in this area too). Social norms are, as we all know, two way streets, effects and causes. They are affected by geographical and especially historical contingencies, but they can also affect, positivelly or negativelly, the development of an area.
Just this morning I’ve heard a pop-scientific explanation of economic (under)development of Serbia, which, supposedly, suffers much from “our pessimistic outlook”, which is a consequence of unfavourable historical contingencies. I don’t know whether this “analysis” is true or not, but it does tackle the important topic of social capital … and a meshwork of other “causes & effects” on many different levels of explanation.
The other reason I’d like to see this socail norms “4D map” is because I like geographical simulations… it’d be a great screensaver 😀
Any type of description of historical evolution of norms would be interesting to see, not only a computer simulation… A book would be great. I’ve read Bowling Alone, and although it’s a great book it’s not quite what would satisfy me – I’d prefer something more macro – dealing with Europe, or Middle East, showing how the norm evolution affects groups across borders, where multiple national and religious “players” are involved.
Oh! Now I posted this I see it’s a bit off topic… but it was inspired by this post. I had no intention to hijack.
Igor – this is actually quite on topic. The goal of developing a 3D map (2 spatial dimensions and 1 temporal dimension) of the evolution of social norms, institutions, and social capital is precisely why we need that ‘imperfect time machine’ that I was talking about in the blog. Contingencies, diffusion, barriers, and so on are precisely what we want to quantify.
I read the Thompson and Sakuwa paper with interest and liked it. In fact I’ve got some criticisms of Coming, Easterly and Gong of my own that at some point I should articulate and write down. But the big elephant in the room with a lot of the papers that use Angus Maddison’s data for pre-industrial economies is that those numbers are, well, more or less made up (rather than estimates from existing records).
There’s two big problems with them. First, Maddison chose 400$ as the level of income for “primitive” societies. This is very much too low for reasons explained by Greg Clark in the relevant book review (http://hsozkult.geschichte.hu-berlin.de/rezensionen/2010-2-121#note2). By itself this wouldn’t be that big of the problem – if you want to look at growth rates you can just index everything. But for later periods Maddison switches to numbers based on actual records. This means that even if the pre-1820 data make sense on their own (at least in terms of changes in income), they are simply non-comparable to the numbers for post-1820 data.
Second, the numbers for pre-1820 are made up based on how technologically advanced a given society was according to Maddison, and the 400$ baseline scaled up accordingly. Some of this is buttressed by estimates of urbanization and such. But if you believe that the pre-Industrial world was Malthusian – and I think there’s lots of independent evidence that indeed it was – then… technology has nothing to do with income. Over the long run (and comparing 1000 BC to 1 AD is a “long run”, I think) the standard of living of Malthusian societies is independent of technology, or land, or the amount of capital. It is determined solely (*) by demography, fertility and mortality rates (to be more precise, the exogenous, non-income related components of these rates). So theoretically, this is the wrong way to go about making guesses as to what income was before the industrial revolution. And of course any tests of whether the pre-Industrial world really was Malthusian using this data will inevitably entail circular reasoning.
I think the authors encounter some of that in the construction of their index and to some extent their methodology dilutes (if not minimizes) the effect of Maddison’s numbers. Likewise, the research question they are asking is more about technological sophistication rather than income per se, so to the extent that those “GDP per capita” numbers can be thought of as Maddison’s best guesses about technology, rather than actual income, it’s potentially not a problem (I think they’re conclusions are correct, but despite the use of Maddison not because of it).
(*) the LEVEL of per capita income will be affected by the GROWTH RATE of technology but not by the LEVEL of technology (unlike modern world where growth rate will be related to growth rate). However, even these effects are likely to be relatively small.
Radek, you are quite right in your critique of Maddison’s numbers. In fact, it’s so bad that I personally think they are unusable, and that’s a strong statement for me, because I’d rather use imperfect data than none at all. This is not a fatal flaw for the Thompson/Sakuwa article, as you note, because it does not depend solely on Maddison. But we need to do better. So, rather than focusing on GDP, calculation of which presents difficulties even for today’s societies, we should go to other proxies for economic development: population densities, urbanization rates, percent of the workforce in agriculture, energy use per capita. Then there are proxies for well-being: average height, life expectancy, etc. There are complex interactions between these variables, sometimes synergistic, sometimes antagonistic, but we can tease them out at the analysis stage. So we need a richer tapestry depicting (and quantifying) human past to resolve these kinds of questions.
Bill, thanks for pointing out these papers. The data/coding are the tables 2 and 3 in the Journal of World History paper, right?
As an aside, there is actually fairly good data on wages and prices for cities of early modern Europe, collected together (from various country specific sources) by Robert Allen of Oxford (who also has an interesting book on Soviet economic growth out, though there’s more Maddison in that too). This data is based on contemporary records. While there’s probably a lot one could argue about but I think overall it’s a pretty accurate picture of early modern Europe.
More generally there is the Global Price and Income History Group at UC Davis, headed by Peter Lindert (http://gpih.ucdavis.edu/) which brings all the historical data of that nature together (http://gpih.ucdavis.edu/Datafilelist.htm) from across the world (rice prices from the 10th century China! Income distribution in 15th century Serbia!)
Of course these series are much later than the indicators in Bill’s papers. But just their existence does suggest that if you dig hard enough you can find something – there’s still more data out there.
Eventually we will have much better tools for finding such data – I envision a giant vacuum-cleaner sucking relevant data from anything an the web (which soon will be everything as all books are digitized and all new articles posted).
Thanks for the comments on Thompson/Sakuwa. I understand the frustration with the Maddison data. They are, of course, fabrications. The problem is coming up with data that encompass some 10 millennia. Peter initially balked at using these data by themselves and that forced my coauthor to come up with alternatives. At first, I did not think it was possible but soon enough realized that we could use city size data. Note that the city size data which are less constructed yield roughly similar results to what Maddison’s data suggest. Sometime down the road someone else will look at the same question with other indicators and the current findings will either be supported or not. So it goes. The very stark choices are a) do nothing because the data are not satisfactory or b) do something with the data that we have or can obtain within a reasonable time period. I will go with b every time with the expectation that someone else will be indignant enough to construct better indicators if that is the perceived problem or a better argument with the same data if that is the reaction. I have tried, incidentally, the systematic coding of ancient historians. See Andre Gunder Frank and William R. Thompson, “Bronze Age Economic Expansion and Contraction Revisited.” Journal of World History 16,2 (2005): 115-172 and same authors, “Early Iron Age Economic Expansion and Contraction Revisited,” in Barry K. Gills and W.R. Thompson, eds., Globalization and Global History (London: Routledge, 2006). The idea was to code a number of Eurasian zones through several millennia to see whether they were experiencing prosperity or depression more or less at the same time. The overarching question was Frank’s argument about the extent of an ancient world system (the assumption was that similar economic conditions might suggest systemic interactions. Frankly, while I spent a couple of years collecting these data, I don’t know that they are any better than Maddison’s constructions despite relying on a large number of ancient history sources. Since I have never seen any reaction to these articles, I can only infer that others are not too taken with the quality of the data (or maybe its the question) either. So, good luck with creating better data inventories. I hope more folks try their hand at this undertaking because it simply will not happen unless people invest their time in creating better data sets. At the same time, the data for ancient times are never going to be as good as we would like – time machine or no time machine.
Extremely timely article.
I myself have been examining the hunter /gatherer: agriculture change which clearly occurred at different locations, periods and specialisations throughout prehistory. There is a way at present of understanding what was going on by examining all the data collected by the specialists and synergising but only if we start by rigorously re-examining previous received theories as to the role of monuments and artefacts. Unfortunately, this is going to meet a lot of hostility from archaeologists and historians who seem to think that eveything had a religious purpose or that the ‘ big man theory’ accounts for social change.
This is especially true when one is outside the academic world and have insights in different fields such as mine; food economics and agricultural policy. Personally I have far more respect for our ancestors who had the same intelligence as us but had to deal with a very much ‘wilder world’ not ‘blitzed ‘by technology.
I feel that food supply and preservation had a far deeper meaning to our ancestors than any great warrior or king. Obvious but fascinating. Just what were pre agricultural societies eating?
Our nearest example of the sort of society that may have been common during the hunter /gatherer: agriculture interface were the Native American tribes of New England and I would welcome any insights others in this area of study could give.
PS. Being a bit of a maverick, I also think that mega monuments such as the Pyramids etc were work creation schemes to keep a standing army happy, fit and preventing conspiracies. Warfare needs engineers.
“I feel that food supply and preservation had a far deeper meaning to our ancestors than any great warrior or king.” That’s true, until an invading army rides in and kills and enslaves you and your family…
-“So today I cannot answer the questions with which I started this blog. But give us a few years.”
I am from spain. I need this information soon. Could you make one sketch?
Wow, fantastic weblog structure! How long have you been running a blog for?
you make running a blog look easy. The entire look of your site is fantastic, let alone the content
material!