Addenda to the previous blogpost “Economic growth in ancient Greece“. I argue that certain estimates made by Ian Morris under-compute the implied growth rates in the “per capita income” of the ancient Greeks. With a proper computation Morris’s estimates simply become unbelievable. This post does a kind of reductio ad absurdum using Morris’s own assumptions.
The Calculations of Ian Morris
In the previous post, I mentioned that Ian Morris simultaneously claims a manifold increase in the population density of Greece in the period 800-300 BCE and a 50-100% increase in Greek living standards. These combined claims imply an unbelievably large increase in the output per capita of ancient Greece.
Morris 2004 suggests consumption per capita rose 0.07-0.14% per annum in those 500 years. As far as I’m concerned these are basically fabricated out of thin air, but I’ve uploaded the 3-page extract of Morris’s own description from pp 724-6 so you can judge for yourself. For now I just assume those figures are real. Morris compares these growth rates of consumption per capita with data from England and the Netherlands a millennium and a half later. He concludes:
“The performance of the Greek economy in the first millennium BC was a whole order of magnitude lower than that of post-Industrial Revolution Britain, but it bears comparison with the some of the most dynamic pre-industrial economies that we know of.”
0.07-0.14% per annum may seem modest enough to be believable, but Morris grossly underestimates the total growth in output per capita required to generate that long-term improvement in living standards.
According to Morris 2005 [pg 7-8] the population density of ancient Greece went up from 3-4 people per km2 at the end (?) of the Greek Dark Ages circa 800 BCE to 12-15 per km2 by 500 BCE. Thus, population density jumped 3- to 5-fold.
In other words, although the land was supporting many many many more people, the ancient Greek economy was not hitting a Malthusian resource constraint. Land per capita was falling, but income per capita was not. To the contrary, according to Morris, living standards were rising. But he apparently does not realise that in order to keep income per capita just constant, even before factoring in the improvement in living standards, there must have been a substantial advance in productivity, whether in yield per hectare of land or in labour productivity.
The combined changes in population density and living standards imply:
- [low estimate] 3x population density + 1.5x living standards = 450% growth in output per capita, or 0.3% per annum over 500 years
- [high estimate] 5x pop density + 2x living standards = 1000% growth in output per capita, or 0.46% per annum over 500 years
But that’s still an undercomputation. Morris assumes consumption = 90% of income in 800 BCE and 75% of income in 300 BCE. When that’s factored in, then the growth rates in “GDP per capita” implied by Morris’s own assumptions would be 540% to 1200%, or 0.34% to 0.5% per annum over 500 years.
0.34-0.5% per year for 500 years. Is that really believable ???
Consider the growth rates in Britain’s GDP per capita, as estimated (with real data, despite imperfections) by Broadberry et al. (Table 5.07) :
According to the rival Clark data, the growth rates are even lower !
So what could have possibly caused such high sustained rates of growth in ancient Greek productivity ??? Remember, there is little evidence of consequential technological progress over this period.
I can believe in the modest rise in consumption living standards. But I cannot believe that population density even tripled, let alone quintupled.
A reader asked me why, in the previous post, I barely considered “Smithian” growth as a cause of the economic expansion in ancient Greece. That represents the gains from increasing the volume of trade and the specialisation of production, especially when previously isolated or autarkic regions are integrated into a single market. As Peter Temin argued in The Roman Market Economy, the gains from market integration could last a while even if they eventually face diminishing returns. I don’t discount that might have happened in the ancient Greek case. But I can’t believe that gains from integrating the Aegean, the lower Black Sea area, and possibly the Ionian sea region could have been as great or prolonged as the Roman creation of a “Mediterranean Free Trade Zone”, which contained much of the most densely populated and advanced parts of western Eurasia. Not to mention, there was much more economically relevant technological progress under the Roman empire than in classical Greece.
Pingback: What kind of growth did Ancient Greece have anyway? | Pseudoerasmus
It seems like the only hypothesis that can be reconciled with Morris’s figures (or even a modified-down version of them) is the one you mention in the main post: that the Greek story is not one of pure “growth”, but of rebound from an artificial “dark ages” wherein both population had collapsed from previous highs and cultivation retreated from previous extent.
It’s only after following your Morris (2005) link that I see the hypothesis in the response above essentially is Morris’s theory. Do you have a problem with the “Dark Ages” hypothesis or just with Morris’s set of numbers? It is possible for a less-lurid scenario to have to occurred, with a not-so-great collapse followed by a not-so-dramatic recovery and therefore a much flatter rate of “growth”.
The idea of a great ancient Greek period of “intensive growth” depends crucially on the population density estimates. But Morris’s computations completely ignore that an increase in population density “should” have reduced income. He just says consumption per capita was X in 800 BCE and 1.5X or 2X in 300 BCE. The 150-200% change actually underestimates the required / implied growth in underlying income. Here’s an example.
Suppose there are 100,000 hectares of land. Population is 1000 in 800 BCE. Density is 100 per hectare. The yield per hectare is $10. Income per capita is $1000.
Morris’s claim is tantamount to saying income rose to $1500 or $2000.
BUT If population rises to 5000 by 300 BCE, then without productivity growth (with no change in yield per hectare), then income per capita must fall to $200.
Therefore, in order to get the income per capita that Morris estimates, you need a 7.5 to 10 fold increase in productivity or yield per hectare.
And I’m saying that’s unbelievable, even with a mere tripling of population. I’m willing to believe there was a rise in living standards — due to repopulation after the Dark Ages and possibly emigration to colonial settlements — but I don’t believe the population density estimates for “core” Greece.
OK, I get it. Pulling some before and after Dark Age numbers from Morris (2005), we get something like:
Year BCE 1200 730 500
Areas (Hectares) 1,000,000 750,000 1,500,000
Area (km2) 100,000 75,000 150,000
Density (Persons / KM2) 10 4 15
Implied Population 1,000,000 300,000 2,250,000
Implied Annual Caloric Yield per Hectare* 620,500 248,200 930,750
Standard of Living Adjustment (730 BCE = 1) 1.25 1 2
Revised Implied Annual Yield / Hectare 775,625 248,200 1,861,500
*1700 per day per person
Certainly these are improbable numbers. and imports from colonies provide no relief here, since Morris’s 150,000 KM2 estimate for classical Greece seems to include the colonies. I’ve always heard that Sicily was the breadbasket of the Mediterranean, but this is a bit much. (However, even a lesser yield figure might explain why, during the P-war, the Athenians felt compelled to go after Syracuse.)
I can believe in a tripling of population density (this can, at least, be cross-checked with archaeological data). But I can’t take seriously the doubling of per capita consumption unless there were some enormous economies of scale.
Your point on land productivity is a very good one.
I located a book on this period, “the Economic and Social Growth of Early Greece 800-500 B.C.” (1977) by Chester Starr. The preface contains a disclaimer (p. 13) that might cause the blogger to smile: “Our inability to quantify any major series of data in early Greece stands in the way of applying the interesting approaches devised by modern economists and sociologists.”
Taking Morris’s area figures and Starr’s estimates and factorings (pp. 152-161) for rotation, barley yield wastage, and calorie requirements, one can calculate an implied maximum population (assuming constant yields) for the period. Starr does not completely rule out that arable land (not counting colonies) might have increased or that yields might have risen over the period, since rising population might have provided additional labor to work the land at critical times. He clearly doubts there was “great” growth in population in “core” Greece.
For comparison, I’ve added a “Starr table” to follow the “Morris table” and adjusted the Morris table to use Starr’s preferred daily calorie requirement of 2000/day. He translates this to 3.5 hectoliters of barley per year.
If there is any validity to Morris’s acres and Starr’s figures, it would appear that any population in excess of 1,700,000 in 500 BCE would have had to have been engaged in economic activity that both financed their subsistence AND provided for any per capita improvement in standard of living. Obviously this is an over-simplification (for one, Greece was not a barley monoculture), but it might be a starting hypothesis.
Year BCE| 1200| 730| 500|
Areas (Hectares)| 1,000,000| 750,000| 1,500,000|
Area (km2)| 100,000| 75,000| 150,000|
Density (Persons / KM2)| 10| 4| 15|
Implied Population| 1,000,000| 300,000| 2,250,000|
Implied Annual Caloric Yield per Hectare*| 620,500, 248,200| 930,750|
Standard of Living Adjustment (730 BCE = 1)| 1.25| 1| 2|
Revised Implied Annual Yield / Hectare| 775,625| 248,200| 1,861,500|
*2000 per day per person
Factor, Year BCE| 1200| 730| 500|
| Area (hectares)| 1,000,000| 750,000| 1,500,000|
50%| Arable Area | 500,000| 375,000| 750,000|
16| Barley Yield (hectoliters)| 8,000,000| 6,000,000| 12,000,00|
50%| Seed/Animal/Wastage allowance****| 4,000,000| 3,000,000| 6,000,000|
3.5| Implicit population at Factor hectoliters per person*****| 1,142,857| 857,143| 1,714,286|
I wanted to make one more comment about Chester G. Starr (who died in 1999), since this thread has taken meback into his work.
Josiah Ober, in the “Wealthy Hellas” paper mentioned by Pseudoerasmus, claims Starr’s support (in “the Economic and Social Growth of Ancient Greece, 800 – 500 B.C.  [“ESG” hereafter]) against the then-dominant view “no growth” hypothesis, epitomized by M.I. Finley.
Support there is (just consider the title of Starr’s book), but it is support of the nuanced and non-doctrinaire sort that ought to have been taken as inspiration, not justification. One can appreciate this in several dimensions.
First, Starr was no Marxist and certainly distinguished his position from that of Finley. Most fundamentally, he sees slavery slavery as far less prevalent, and far less defining, than Finley would have, for the period under study. But Starr is equally at pains at several places in the book to emphasize, in terms that echo Finley’s that the aggrandizing landowners or aristocrats are not to be confused with modern economic actors. “Chremata”, in other words, is not exactly capital.
Second, though the Wiki devoted to Starr describes him as an opponent of the “Annales” or Braudelian approach to history, the quantitative section of the chapter on agriculture in ESG is called “The Parameters of the Possible”. With an attunement to constraints unusual for a supposed individualist, he emphatically states at several points in ESG that there was no “great growth” in agricultural productivity in the period under study.
Third, while himself a military historian (he wrote a multi-volume official history of the American 5th Army’s WWII Italian campaign) and clearly aware of the significance of hoplite-based warfare, he shows no trace of the militaristic exaggerations or oversimplified polity-warfare linkages that we see in a Victor Davis Hanson. Greek society is not defined by the military function.
So this is “Wealthy Hellas” – but with a difference. In fact it is not so easy to find the “growth” in Starr’s book, certainly not in any quantitative sense; the “growth” he was talking about (look back at the title) was just as much social as economic. It was complex and not easily summarized.
My summary is that Starr saw two stages of social change, the first from a chieftain to an aristocrat-led agricultural economy, the second from an aristocratic/largely-rural stage to an assembly or council-based urban polity, one that might be democratic or (more often) oligarchic or tyrannical in its operations. but which, overall, opened society to the participation of “middling” farmers. In this two-stage change Starr saw, first, a more efficient expropriation of surpluses leading, second, to a significant increase (“growth”) in commercial activity and external trade, an increase sufficient by 500 BCE to support a population (he guesses over 2 million) well in excess of that implied by the land’s agricultural output. Needless to say, it’s a long way these suggestions to any kind of “egalitarian male” institutional theory for Greek economic development.
Starr seems to have displayed no overriding theoretical perspective. For this reason, perhaps, he was neither a stirring writer nor an easy one; his concise books manifest humility in the face of inadequate data and a deep respect for what cannot be said with honest confidence. I wish I had known him.
Nice comments, Tom. You should blog them !
Are you citing estimated rates of change of GDP per capita to suggest that a rate of growth of productivity per hectare is implausible? I do not see how that follows, given the evidence that yields per hectare can vary quite a bit depending on the use of labour, irrigation, manuring, crop rotation, tools, and other resources (Geoffrey Kron has collected some of this, but I imagine that historians of the early modern period would be the specialists). In addition, the other issue is that having collapsed to a small-scale society of peasants and farmsteads, Dark Age Greece could pluck some “low-hanging fruit” which late medieval and early modern England with its towns and dense trade network had already eaten.
I think that almost all archaeologists would agree that the population of Greece increased substantially from 800 to 300 BCE and that the wealth per capita increased somewhat, because there are more sites from later periods and they have more and better stuff in them, but the exact numbers can certainly be debated. Morris’ seem about as precise as is warranted, and I certainly applaud him for not trying to give a number in some kind of constant dollar.
An increase in population density all by itself implies an increase either in output per hectare or output per person. If you triple population density, yet living standards increase, then the implication is that productivity has been tripled PLUS the increment in living standards. Morris apparently does not realise this. His combined claims (population density AND living standards) imply something like 0.3-0.46% per annum growth in ancient Greek productivity — an order of magnitude of growth exceeding the first half of the 18th century in England.
In pre-modern times, population density and living standards were in tension. They are not irreconcilable, but the solution requires higher yield, or higher labour productivity, or more land, or the same land with previously unexploited fallow opened up.
But my previous post was about that — the Greek economic effluorescence was probably just a combination of recovery/repopulation from the Dark Ages + colonial land expansion.
When I read Morris’ books and articles five years ago, I read very clearly that a population increase of 200-400% plus a growth of consumption per capita of 50-100% implied a five-fold to ten-fold increase in ‘GDP’. Maybe you read them differently (I got some of my understanding of his ideas in person), but I do not have a copy of all of them to hand to confirm. But see for example “Economic Growth in Ancient Greece” p. 727 “If the results in Figures 9 and 10 are reasonably accurate, then aggregate consumption in the Greek world increased fifteen- to twenty-fold between 800 and 300- ie. at an annual rate of 0.6-0.9 percent. We might compare this to aggregate growth rates of 0.5 percent for Holland from 1580 through 1820, and 2.5 percent for Britain 1820-1920.”
With the colonies, keep in mind that mainland Greece paid for what it brought from them, so production in mainland Greece had to increase in order to produce goods to trade for that Egyptian and Scythian grain.
Did you read the previous post ? I had cited Morris’s “Economic Growth in Ancient Greece” there.
The issue is not with aggregate growth but with per capita growth rates.
Morris apparently does not realise that if you increase population density, then per capita GDP falls, absent an increase in productivity of land or labour. Therefore, he under-computes the per capita growth necessary to achieve the increase in living standards.
Precisely. I’ve said that and more in the previous post. I don’t see how that helps with Morris’s position.
Look, it’s simple. Morris says “GDP per capita” grew 0.07-0.14% per annum over 500 years. I’m saying that’s an undercomputation GIVEN the increase in population density. The population density estimates plus Morris’s consumption data imply that “GDP per capita” grew 0.3-0.46% per annum. That beats England in 1700-50 and rivals it in 1750-1800, and who could possibly believe that.
Even the “corrected” rate of 0.3-0.46% is lower than the “true” correction would be, because I’m ignoring Morris’s assumption that consumption was 90% of “GDP” in 800 BCE and 75% in 300 BCE.
Technically even THAT is an underestimate because Morris assumes consumption = 90% of income in 800BCE and 75% of income in 300 BCE. I did not bother with factoring that in because, apparently, some people are already confused. But if I did factor that in, then the implied growth rates in GDP/capita based on Morris’s own estimates/assumptions would be 0.34% to 0.5% per annum over 500 years. Absolutely unbelievable.
I do not understand. Morris 2004 says that Aegean Greece in the period 800 to 300 BCE saw a modest increase in production per capita and a large increase in population, therefore an even larger increase in total production. He tries to quantify this, waving his hands a lot but getting numbers which most specialists would say are within the bounds defensible from the evidence. He then estimates a rate of change in production per capita (around 0.1%/year) and a rate of change in production/’GDP’/’the size of the economy’ (around 0.6-0.9%/year). Only the total ‘size of the economy’ had to increase at 0.6-0.9%/year, not ‘GDP per capita’ (to use two bad metaphors).
Now of course that population growth would have put pressure on the standard of living, but maybe that is why the changes in standard of living were more modest than the changes in population? (And why so many Greeks gathered into cities). Between 800 and 300 BCE the societies of mainland Greece underwent many changes which we would expect to have improved the standard of living.
I don’t know how many times or how many different ways I can explain. I am doing a kind of reductio ad absurdum with Morris’s own assumptions/arguments.
Your 1st paragraph has no relevance to what I’m arguing. As for your 2nd paragraph,
Sure, but then those urban Greeks had to be producing something to pay/exchange for food. They obviously did. And usually, urban productivity is higher. Whether the food consumed by city dwellers came from within or from abroad does not change the fact that it had to be paid for by their production.
But all that is implicitly covered in the following statement : if population density is rising, then the output per capita of the Greek economy as a whole (combined farm & urban sectors) had to rise by X in order to keep living standards merely constant. That X is equal to the rise in population density itself.
This is pretty elementary.
Morris’s 2 key assumptions/observations are : (1) population density rose 3-5 times. He does not understand that this ALONE requires output per capita to increase 3-5 times in order to keep income per capita CONSTANT. Therefore, he has a 300-500% under-computation for per capita income just right there. (2) Consumption per capita, he says, rose by 150-200%. This actually implies output per capita rose by 180% to 240% because he tells us he assumes consumption is 90% and 75% of output, respectively, for 800BCE and 300BCE.
I’m willing to believe in the rise of living standards based on house size & age at death from skeletal samples & perhaps height of skeletons.
That suggests that there is something funky about the population density estimates.
The initial (ca. 800 BCE) density estimates were indeed funky. Land area cannot be 75,000 km2, population cannot have fallen “by half to two thirds” from a Mycenaean million, and population density have been 3-4 persons per km2 all at the same time. At 4 persons/km2 the maximum population for 75,000 km2 would have been 300,000.
But the population estimate, even at 333,000 or 500,000, could be low. Assume that subsistence is about 3.5 hectoliters of barley per person per year. If half the land were arable, if barley yield was 16 hectoliters per hectare, and if only 60% of the harvest could be eaten after accounting for seed corn, wastage, etc., the “surplus” would have been larger than the subsistence requirement. If the archaeology forbids a higher figure, then we certainly the situation set up for explosive growth – and that is what Morris thinks happened.
I’m working on a post that will calculated all the inter-related growth rates that are implied by Morris’s numbers and assumptions. I’ll notify when the post is up.