by Ober, Josiah
In sum, by the late classical period, Hellas was relatively densely populated. The number of Greeks who lived in urban areas was remarkably high by premodern standards. They lived in much bigger settlements, in much bigger houses, and in substantially healthier conditions than their ancestors could have dreamed of. A good part of the population of mainland Greece was fed from imported food. The classical Greek world had not fallen victim to the Malthusian trap, and its economy cannot adequately be explained by reference to subsistence agriculture alone.
EQUITABLE DISTRIBUTION OF WEALTH AND INCOME
A third measure of economic development is the distribution of wealth and income. Historically, all complex societies have been characterized by economic inequality. Yet when wealth and income are distributed extremely inequitably, such that society is bimodally segmented into a tiny elite of the very wealthy and a great mass of individuals living at subsistence, there is correspondingly little room for sustained economic growth. It is only with the emergence of a substantial and stable middle class of people living well above the level of subsistence, and therefore willing and able to purchase goods unnecessary for their mere survival, that societal consumption becomes a driver of economic growth.40
How equitably was the wealth of Hellas distributed across its relatively dense and urbanized population? Once again, house sizes can be used as an indirect proxy. Ian Morris shows that archaic/classical Greek settlements were never characterized by a few mansions and many huts. Rather, across the entire half-millennium from 800 to 300 BCE, the distribution of Greek houses tends to cluster around the median house size. The size of larger houses (the top quartile in floor plan) failed to diverge markedly from that of smaller houses (the bottom quartile). The size of larger and smaller houses grew more or less in lockstep across the period: by 300 BCE, houses in the 75th percentile of the distribution were only about one-fifth again (roughly 50 m2) as large as those in the 25th percentile.41
A comparative survey of house sizes at Olynthos (i588) and other well-preserved Greek urban areas by Geoffrey Kron of the University of Victoria confirms this general picture: unlike (e.g.) nineteenth century England, the distribution of house sizes at mid-fourth century BCE Olynthos describes a bell curve: most houses fall in the middle, rather than on the far left (tiny house) side, of the distribution. Overall inequality among house sizes at Olythos was very low.42 Not every Greek family could afford to buy a substantial house—which may have cost something in the neighborhood of 6–15 years of income.43 But many could afford to own a home: Based on recorded house costs and the Athenian census of 322 BCE in which some 9,000 citizens (of a presumed total of ca. 31,000) owned property amounting to more than 2,000 drachmas, Kron estimates at least nearly a third, and possibly as many as three-quarters, of Athenian citizen families could afford to purchase a house.44
Kron has attempted to calculate wealth distribution in late fourth century Athens more directly by reference to the standard Gini index: a coefficient (from 0 to 1) of inequality in a given population. The lower the Gini coefficient, the more equitably the good in question is distributed across the population (so 0.1 is very equal; 0.9 very unequal). The Gini coefficient may also be displayed visually by a Lorenz curve; the further below a line describing a 45-degree angle (perfect equality), the greater the level of inequality. In Kron’s calculation, the good is household wealth; we calculate the Gini index and Lorenz curve for Athenian household income below.
Athenian private wealth was certainly not distributed with anything approaching perfect equality. Based on reports of the census of 322 BCE and other sources, Kron calculates that in late fourth century Athens the richest 1% of the population owned about 30% of all private wealth; while the top 10% owned about 60% of the wealth. This yields a Gini index of 0.708. Kron compares this figure to the Gini wealth coefficients for several modern societies. The late-classical Athenian level of total-wealth inequality is roughly comparable to that of the United States in 1953–1954 (0.71). It is less equal than Canada in 1998 (0.69) but more equal than Florence in 1427 (0.788) or the United States in 1998 (0.794). It is much more equal than the United States or England in the early twentieth century (0.93 and 0.95, respectively).45
Kron’s conclusion on the comparatively equitable distribution of private wealth in late classical Athens is consistent with estimates of landholding in Athens: Two independent studies by British classical scholars concluded that about 7.5–9% of citizens owned about 30–35% of the land of Attica; some 20% owned little or no land. Excluding those at the top and bottom of the distribution, we are left with roughly 60–65% of the land being owned by about 70–75% of the citizen population.46 Ian Morris points out that the resulting range of Gini coefficients, 0.382–0.386, is strikingly low in comparison to estimated distributions of landholding for other ancient and medieval societies. Although the baseline Athenian figures do not tell us anything about some relevant factors affecting the value of land, e.g., distribution of especially productive land or financial encumbrances on landholdings, Morris is certainly right to conclude that, “the basic point is clear: landholding was unusually egalitarian in Classical Athens.”47
Economists typically assess material inequality by measuring income. although it is not possible to calculate an income Gini for all Hellas, on the basis of what I take to be a plausible model of wealth distribution for later fourth century Athens (table 4.4, with discussion below), I estimate the income Gini for the whole of Athenian society (including slaves and resident foreigners) in the later fourth century to be in the region of 0.40–0.45, based on two models (an optimistic model assuming lower inequality and a pessimistic model assuming higher inequality). The corresponding Lorenz curves are illustrated in figure 4.2. This is similar to the income Gini estimate of 0.42–0.44 suggested by the ancient economic historians, Walter Scheidel and Steven Friesen for the high Roman Empire. Yet, based on the relevant income distribution models, the shapes of the Lorenz curves for the two societies are quite different; the difference arises from the substantially larger Athenian population of people who fall in the middle range, between the richest and poorest.48
What we really want to know about income in a premodern society is how many people lived near, or well above, subsistence. Stanford’s Walter Scheidel has analyzed the real wages of unskilled workers (i.e., those at the lower end of the economic distribution) in a number of ancient and medieval communities. Scheidel’s approach is to convert daily income into a “wheat wage”—a well-established method of assessing the level of income in different currencies or in kind by reference to a single standard of liters of wheat per diem. The wheat wage can then be used to estimate the proximity of the wage earner to the base level of bare survival. Scheidel’s figures show that in most premodern societies in which daily wages can be calculated (and thus converted into liters of wheat per day), wheat wages fell in a fairly narrow “core” or “customary wage range” of 3.5–6.5 L of wheat/day with a median of 5.5 L/day.49
TABLE 4.4 Athens, Late Fourth Century BCE Income Distribution Models
NOTES: Elite = liturgical fortune, which is >3–4T (Davies 1971), and >10× subsistence. Subsistence minimum = 100 dr/year. Middling = 2.4–10 × subsistence (Scheidel and Friesen 2009).
FIGURE 4.2 Athenian inequality, late fourth century BCE (Lorenz curves)
NOTE: Based on income models in table 4.4.
Scheidel suggests that 3.5 L/day defines the lower limit of the customary wage range—this level of adult male wage cannot have been far above bare subsistence (i.e., close to the edge of survival).50 If we take that level as a baseline “head of household” contribution, below which it is not possible to fall very far or for very long, if an ordinary family were to survive (3.5 L/day adult male income = subsistence baseline = 1S), we can then calculate more generous income regimes, featuring higher wheat wages, as multiples of that baseline. Thus the median of the “customary wage range,” at 5.5 L/day, is about 1.6 × bare existence (= 1.6S). This is eno
ugh to get by but is still not far enough above the level of survival to be described as comfortable, or even decent conditions of life.
Scheidel and Friesen suggest that wages in the 1–2.3S range (i.e., up to about 8 L/day) may be regarded as constituting the general category of living at the level of subsistence.51 Adult male incomes of 2.4–10S (about 8–35 L/day) are considered to define a decent, “middling” existence—families with an income earner at this level could be expected to consume some goods not necessary for bare existence. This suprasubsistence-level consumption is an important driver of economic growth if the “middling” families constitute a substantial part of the total population. Those whose incomes were greater than 10S (more than 35 L/day) are categorized as elite. This elite group was a very small part of every premodern population. Thus, the conversion of wages to wheat wages allows the possibility of estimating the distribution of the population of a given society across the broad, but analytically useful categories of subsistence, decent/middling, and elite levels (table 4.5).
Distributing populations into three income tiers (subsistence, middling, elite) is obviously artificial and reductive; it obscures meaningful differences in levels of welfare and ignores how people defined themselves relative to others. Yet it is analytically useful in assessing and comparing the economic performance of ancient societies. As noted above, if a given society is divided into a tiny wealthy elite on one end and a mass of people living at subsistence at the other, there is relatively little social surplus, and so economic performance is correspondingly low. If there is a substantial “middling” population of people living comfortably above subsistence, then there is correspondingly more demand for surplus production, making possible relatively higher economic performance.
TABLE 4.5 Per Diem Income (Real Wages Expressed as Wheat Wages)
NOTE: Based on Scheidel 2010 and Scheidel and Friesen 2009.
Scheidel and Friesen argue that Roman wages fell within the low “customary wage range” and that most Roman laborers thus remained at the Subsistence level.52 On the basis of these assumptions, they offer two simple models for the distribution of income across Roman imperial society. Their “optimistic” model is based on assumptions pointing to a relatively more egalitarian income distribution (and thus, per above, to more consumer demand and a correspondingly higher expected rate of economic growth); the “pessimistic” model employs assumptions that lead to a less egalitarian distribution (and so less demand and less growth). The goal is not to specify a single distribution (we do not have the evidence to do that) but to develop a general range into which the actual distribution of incomes can reasonably be assumed to have fallen (table 4.6).
The key point here is that even on the optimistic scenario, only a small percentage of the total Roman population fits into the middling category; most residents of the empire lived close to subsistence and thus had relatively little surplus to spend on nonessential goods. On the basis of this model, Scheidel and Friesen argue that imperial Rome, overall, failed to generate a sufficiently large social surplus to push back the Malthusian constraints that, as we have seen, limit the growth of subsistence-level economies. Once again, however, it is important to keep in mind that the Roman Empire was very large and that there must have been considerable regional economic variation.53
TABLE 4.6 Income Distribution, Roman Empire
NOTE: Based on Scheidel and Friesen 2009.
Athens is the only pre-Hellenistic Greek community for which we have figures for daily wages.54 On the basis of the available evidence, classical Athens appears to be one of the very few known societies in the period 1800 BCE–1300 CE in which daily wages were substantially above the subsistence-level “customary wage range.”55 Construction-work wages and military wages in Athens in the later fifth century BCE averaged 1 drachma/day; the wheat price was about 6 drachmas/medimnos (1 medimnos = 52–54 L), yielding a daily wheat wage of 9 L and thus a baseline multiplier of 2.6S. This is just above the “middling” floor of 2.4S. In the 320s, unskilled laborers were paid 1.5 drachmas/day; wages for skilled laborers were up to 2.5 drachmas/day; the wheat price was 5–6 drachmas/medimnos. This yields a range of wheat wages of 13–16 L/day and a baseline multiplier of 3.7–4.6S: thus solidly within the “middling” range of 2.4–10S.56 By way of comparison, median wages in Holland, ca. 1500–1800, translate to a wheat wage ranging from 10–17 L/day, and thus a baseline multiplier of 2.9–4.9S (table 4.7).
The evidence for late fourth century Athenian wages is anecdotal, but it is consistent with what Athenians were being paid for especially important forms of public service: Citizens attending a meeting of the Athenian Assembly (which typically lasted a half-day) were paid 1dr (30 annual ordinary meetings) or 1.5 dr (10 annual principal meetings).57 The key point is that both in later fifth century BCE and, a fortiori in the later fourth century, Athenians who were engaged in unskilled as well as skilled labor (at least on construction of state-sponsored buildings) were paid wages sufficient to elevate them to a decent, middling premodern standard of living: They no longer hovered at a subsistence level perilously close to bare survival. Based on data currently available, this was rare anywhere in the world before the nineteenth and twentieth centuries.58
If we assume that the available data about Athenian wages is more or less accurate, we can make informed guesses about the distribution of the late fourth century Athenian population into the three general income categories of subsistence, middling, and elite. The figures on which the following estimates are based are detailed in table 4.4. Following the lead of Scheidel and Friesen, I posit two possible distributions: a “pessimistic” (less equitable, ergo lower consumption, lower expected growth) distribution and an “optimistic” (more equitable, higher consumption, higher expected growth) distribution. For each distribution, I assume a total population for Athens of just under a quarter million people, of which about a third were slaves, and about a tenth were resident foreigners.59
TABLE 4.7 Athens and Holland, Wheat Wages
NOTE: Based on Scheidel 2010.
Elite status in Athens can be defined by a liturgical fortune of 3–4 talents.60 Assuming a conventional annual return of 1:12, such a fortune would in fact yield a living standard of roughly 10 times bare survival. The elite population of Athens amounts to a little over 1% of the total. In the optimistic scenario, I assume that most citizens and metic (resident alien) males, and even a small number slaves (those who “dwelled apart” from their masters) would be able to make at least one drachma/day on average and so would achieve middling status. In the pessimistic scenario, I assume that only about two-thirds of citizens, a minority of metics, and no slaves received regular wages at or above the one-drachma/day level. In this simplified model, I do not take into account women’s or children’s paid labor; the middling women and children in table 4.4 are assumed to be members of “middling” families. Nor do I make any allowance for the historically exceptional absence of heavy taxes or steep rents paid by Athenian citizens below elite status.61 Both productive labor by women and children and the low-tax/rent regime may push in the direction of more optimistic scenarios than I have presented here.
The results are tabulated in table 4.8 and illustrated in figure 4.2.
Table 4.8, based in the first instance on the evidence for relatively high Athenian wages, incorporates a number of assumptions, some of which may be too pessimistic (per above). Other assumptions may be overly optimistic, notably that the wages recorded in our sources represent something approximating the market standard and that unemployment and underemployment were not rampant. On the other hand, because achieving middling status required average wages of only 1 drachma/day (rather than the reported late fourth century wage level of 1.5–2.5 drachmas/day) there is a fair amount of discounting already built in—even without taking the relative absence of exploitative taxes or rents into account. In sum, it appears likely that a substantial number of residents of fourth century Athens lived far enough above subsist
ence to enable them to live decent lives. The surplus consumption capacity of a comparatively large middling population would have been a major driver of the Athenian economy.
Assuming that my model of income distribution in later fourth century Athens is more or less correct, can we extrapolate from Athens to the wider Greek world? Were high Athenian wages at all typical of the Greek world generally? The answer to that question will depend on how we imagine Greek labor markets as operating. If (counterfactually) we assume labor markets with zero transaction costs (i.e., that there was no restriction or cost, material or psychic, to movement from one part of the Greek world to another and that people would choose to move to where wages were highest), then Athenian wages would reflect the equilibrium conditions of the Greek world and we could assume that high Athenian wages reflected Hellenic norms. The no-transaction-cost assumption is, of course, false: the value of the Athenian evidence for the rest of the Greek world depends on how high the transaction costs associated with moving from one labor market or polis to another actually were.62
