Agricultural Productivity in Antiquity

[Guaporense]

Ancient agricultural productivity

A few decades ago it was typical to assume that ancient societies were all subsistence farming societies where agricultural productivity was very low, just enough for the bare survival of the average ancient peasant farming family with perhaps a small surplus to support the assumed 10% or so the of the population that was not engaged in food production. However, as historians in recent decades have applied the insights of modern sciences into ancient societies, as well as the insights from modern studies of other better documented pre-industrial economies (British Economic Growth 1270-1870 (2015) remains the gold-standard in the analysis of pre-modern economic growth) we have been able to properly appreciate how varied productivity levels could be in the pre-modern world.

For example, according to British Economic Growth, agriculture corresponded to only 45-50% of the GDP of England during the 14^th to the early 17^th centuries, which shows that manufacturing and services were already half of the economy even in England during the late-middle ages. England was not a very exceptional pre-modern economy before the 17-18^th centuries, in fact, its estimated rate of urbanization was around the European average in the 16^th century (roughly ca. 9% of the population living in towns over 5,000 inhabitants in both cases), which suggests that it was not an economy with an exceptionally small farming sector. Therefore, we can infer that it was not atypical for pre-modern farming to be productive enough to support a large population that was not directly engaged in agriculture.

In regards to ancient and medieval agricultural productivity overall, it is likely that the traditional model of mostly self-sufficient peasant households characterized farming in most ancient and medieval societies. However, there were many exceptions. For example, the evidence currently suggests that the agricultural system of the Early Roman Empire was likely very different. In its most developed regions such as Italy, farming was dominated by highly efficient large farms that functioned as business enterprises and exploited economies of scale and regional specialization, allowing for much higher levels of productivity than typical of subsistence farming.

One example is the implied levels of Roman farming productivity that can be inferred from the agricultural manuals of Varro, Columella, and Cato. Columella writing in the 1^st century AD, during the zenith of classical antiquity, wrote the following table of labor days required to farmland:

We can conclude that in those well-managed, almost industrial farming states a farm laborer by working 300 days per year could cultivate 50 iugera of barley fields (which was the cheapest ancient stable grain), using 250 modii of seed or 28 iugera of wheat fields.

To determine how much grain they could produce based on yield level, as ancient sources state, according to Goodchild (2007):

“The scant ancient references we have for ancient yields in this area range from Columella’s cereal yield of 4:1 for the whole of Italy (Rust. 3.3.4) to the exceptional yields of 100:1 for Sybaris mentioned by Varro (Rust. 1.44.1) and Pliny the Elder (NH 18.94-5). For Etruria, an area noted for its fertility, Varro (Rust. 1.44.1) gave maximum yields of 10-15:1. The remaining passage is from Cicero (Verr. 2.3.112) where he claimed a yield of between 8-10:1 in the land surrounding Leontini in Sicily.” (Goodchild page 247)

Based on a combination of modern data with ancient sources, Goodchild (2007) estimates that median crop-yield in Roman Italy was perhaps 6.5 to 1 to the quantity of seeds sown:

This means that a Roman Italian farm laborer could perhaps produce 1,625 modii of barley per year, which is about 14,000 liters of grain, or 11 metric tons of grain. In terms of wheat, it took about ¾ more time to work on the same land, which a similar yield, which means a Roman Italian farm laborer could work on 28 iugera of land working 300 days per year, seeding 140 modii of wheat and producing about 910 modii of wheat, or about 8,000 liters of wheat per year. This explains why the rich in antiquity ate wheat bread while the poor and slaves ate barley bread.

At a productivity level of 14,000 liters of grain per year, it is possible for a single Roman Italian farm worker to produce enough barley to feed 35 people (given each individual needs about 2,000 to 2,500 calories per day which can be supplied with 400 liters of barley per year) or enough wheat to feed 27 people (as the density of calories in wheat is larger so only 300 liters are required per person per year, thus the difference in calorific output from shifting from barley to wheat is smaller than in terms of liters, still barley appeared to be 20% cheaper than wheat in terms of the cost of calories). Other grains like beans and millet appear to have had labor input costs between wheat and barley.

At the median yield of 6.5 per seed, Columella’s farm of 200 iugera crewed by a workforce of 8 workers plus two pairs of oxen would be expected to produce 6,600 modii of grain in total, roughly 60,000 liters of wheat. If producing at the high fertility region of Etruria with yields around 10 to 15, according to Varro’s account from the late 1^st century BC, such a farm with 8 workers could reach an output of 50 to 75 modii per iugerum or 87,200 to 131,000 liters of wheat. One has the impression that Early Imperial Roman farms were highly productive grain factories manned by slaves rather than self-sufficient peasant farms. In fact, given estimated wheat consumption of 160 kg per person (excluding calories from other sources such as meat, wine, and olives), the population of ca. 65 million inhabitants for the Roman Empire in the 1^st century could be fed with only 120,000 of such high-performance Etrurian-wheat farms employing only ca. 1 million workers.

These productivity levels are the Roman Italian ideal for the 1^st century: we are assuming that the land was optimally managed, that it was good fertile land without many rocks or trees (which as Columella stated, to deal with trees would require 3 additional laborers to cultivate 200 iugera of land), and these estimates only take into account the labor cost of producing the grain, without considering the costs of processing the grain, transporting it to retail markets, and selling it. Thus, retail prices of grain suggest a Roman Legionary soldier in the late 1^st century, who made 1,200 sesterces per year, could purchase about 400 to 600 modii of wheat (as provincial wheat prices varied from 2 to 3 sesterces per modii), roughly half of what a laborer could produce. Thus, we cannot make inferences about typical ancient income levels in terms of grain from these ancient agricultural manuals.

While this represents the upper bound in estimated productivity, the current evidence suggests that it is likely that only a small minority of the Roman workforce was directly employed in the production of grains.

Roman Italy in particular was a net importer of grains, as land was relatively expensive in Italy compared to the more peripheric provinces, so it was better to use the premium valued Italian land to produce more expensive foods such as wine and olive oil, as land yields in wine and olive oil were substantially higher (I read some estimates that it was possible to produce 4 times the calories in terms of wine compared to grain on the same amount of land).

In farms in Roman Gaul, where land was comparatively abundant given the much lower population density, the Romans focused on maximizing labor productivity to an even greater extend: they even developed the first documented case of reaping machines, besides Roman Vallus, reaping machines would only reappear again in modern times. Thus labor productivity in Gallo Roman farms was probably even higher than the ideal levels of productivity deduced from the Roman writings about Italian agriculture:

Roman Egypt was understood to attain higher levels of comparatie productivity for grain versus other agricultural goods such as meat, olive oil and wine. This is why it was the Empire’s major wheat expoting province, however, it could be the case its productivity levels were lower than Italian levels, but the difference was smaller in grains than in other agricultural goods, explanining the relative specialization of Italian farms in non-grain crops. I do not have access to any documents that allow us to estimate labor productivity levels for Roman Egypt, although in terms of wheat yields relative to cultivated land they appear to have been higher than in Italy (which were perhaps around 30-35 modii per iugerum).

Agricultural labor productivity in antiquity could be much lower than 8-14 thousand liters per worker per year. This should be understood as to be near the maximum attainable output per worker at the zenith of the Classical Civilization during the Early Imperial Period (30 BC to 180 AD), a place and time as Peter Temin has conjectured, humanity reached its highest living standards attained before the Industrial Revolution. In other ancient societies labor productivity levels in agriculture were likely lower.

For example, in the labor-abundant and land-scarce Ancient and Medieval China, as Von Glahn (2016) states, farming was done by small and mostly self sufficient family farms, it was thought to be typical for a family farm of 5 people (usually 2 adults and 3 minors) to produce only 3,000 liters of grain per year:

This implies an output of roughly 600 liters per person of the household or roughly 15 to 25 lower than the output per worker of an 1st-century laborer in Columella’s farm producing wheat or barley. However, as the family was self-sufficient they had to process the grain as besides cultivating it and naturally, labor productivity of the women and children who helped in the farm would be naturally lower than that of an adult male (which I guess is what Columella calls a laborer). Productivity levels around 600 liters per person in Chinese peasant families are consistent with estimates of Chinese economic historians of Han-period productivity levels of about 1,000 jin of grain per unit of the agricultural population, which translates to 640 liters, roughly 1.6 to 2 times minimum subsistence.

These lower productivity levels are consistent with the concept that historians often imagine of the small self-sufficient farming households that sell a small proportion of their output to the market and consume most of their output for their own subsistence. In Classical Antiquity, these small and low productivity family farms were common as well during certain places and times. For instance, in Italy the Early Republic, but as Italy became more developed and integrated into the “Mediterranean economic zone” during the peak of economic activity in the Late Republic and Early Empire small farms were displaced by larger and more efficient latifundia. Indeed, Columela’s highly efficient farm with 200 iugera of farming land, or Cato’s wine orchard of 240 iugera appears to have been around the typical Early-Imperial Italian farm size (from Goodschild (2007)):

For comparison, typical ancient and medieval Chinese family farms had around 40 to 100 mu (Von Glahn (2016), the 2 AD census measured 827 million mu of cultivated lands, or about 70 mu per rural household given China had ca. 60 million inhabitants or 12 million of their typical 5-person households, of which about 11-11.5 million were rural households ), each mu was 457 square meters, an iugerun was 2,523 square meters, which means typical Ancient Chinese peasant farms would be around 7 to 18 iugera, roughly 3.5% to 9% of the median Early Imperial Roman Italian farm size.

However, yields per unit of land and seed were likely higher in Chinese farms, as they had lots of people and little land, they had to focus their efforts on maximizing the yield of their small farms. This is why they developed technologies to not waste precious seeds like the seed drill, while Romans invented technologies to maximize labor productivity like the gallic reaper. It was perhaps an analogous dynamic to the late 19^th century USA and Germany: the USA had lots of lands so they developed farming machinery to use that land, Germany had little land, so they focused efforts to develop chemical fertilizers to maximize yield per unit of land.