Slavery, Prosperity, and Inequality in Roman Pompeii^*

Abstract

Historians of premodern economies, in contrast to modern ones, have only infrequently contemplated the economic contribution of slavery. Here, I suggest that quantitative and statistical tools allow us to evaluate the place of slavery in an early economy, using Roman Pompeii as a case study. At the time of its destruction in 79 ce, Pompeii appears prosperous, having benefitted from the economic development thought to have characterized the Roman world. Recent discoveries, meanwhile, shed new light on the conditions of working classes and slaves throughout the city. These narratives can be seen to form two sides to the same coin, as Pompeii’s prosperity was created in large part thanks to slave labour. The connection is supported by constructing a probabilistic model, which suggests some 6 million sesterces (HS) flowed every year to Pompeii’s masters through their exploitation of slaves. Slave owning probably formed the largest single income source for the urban economy. This scale of income is shown to be consistent with recent reconstructions of wealth and income inequality in the city. The results not only speak to slavery’s profound importance to Pompeii’s prosperity, but they encourage a recentring of labour and slavery in Roman economic history.

While there is by now a substantial literature on the place of slavery in the economic history of the modern world — on slavery’s role in the making of modern capitalism, for example, or on the contribution of enslaved people to economic development in America and Europe — similar themes have received far less attention from premodern historians.¹ If the oversight is generally noteworthy considering the commonality of slavery across much of human history, it is especially so for the field of Roman history, where there is both longstanding acknowledgement of slavery’s profound importance and a healthy current interest in understanding the causes of economic development. That is, historians of the Roman Empire regularly discuss the possibility of economic growth while also acknowledging Rome’s transhistorical prominence as a slave society, but rarely contemplate any connection between the two.² However, since research on other slave societies stresses the outsized contribution of slavery to economic development, it seems fair to ask whether this was also true for the Roman Empire.

The absence to date of any attempt to confront this question cannot be ascribed simply to a lack of data. Using what evidence exists for prices or incomes, historians of the Roman economy have proven enthusiastic quantifiers, formulating grand models of GDP or wealth inequality; however, such models have rarely sought to incorporate the contributions of enslaved labour.³ What’s more, our ability to work with situations of scarce data or high uncertainty — the sort of situations typical of the study of all premodern economies — improves considerably as statistical methodologies, introduced to this journal several years ago, make increasing inroads in historical studies.⁴ It is this paper’s goal to promote the possibility of measuring the role of enslaved people in a premodern economy through a case study of Roman Pompeii on the eve of the eruption of Vesuvius in 79 ce. As a wealthy city at the core of the Empire, Pompeii can help us think about a wider Roman economy often thought to have been characterized by cities and signs of urban prosperity, not to mention slavery. Using statistical and quantitative tools and applying them to an especially well-documented example, I will argue that enslaved labour formed a primary source of Pompeii’s economic prosperity. By measuring the share of wealth generated by Pompeii’s slaves, moreover, I hope to shed light on the scale of the city’s economy and the distribution of wealth across Pompeiian society.

My inquiry draws upon Pompeii’s uniquely rich evidentiary situation, while it also aligns beneficially with several strands of ongoing research. The superintendency responsible for the site’s archaeology has followed up recent excavations of cramped productive spaces and workers’ quarters with a major exhibition on Pompeii’s underclasses.⁵ The last few years have also seen other relevant trends including considerable specialist study of the city’s economic history.⁶ More broadly, there has been rising wider scholarly interest in economic inequality in past societies which turns with increasing frequency to Pompeii to inform sweeping historical models.⁷ Still, it is my contention that there remains in all this work a disconnect between two assertions often made about the city but rarely put into any sort of relationship, mirroring the similar oversight among Roman economic historians noted above. First, it is often noted that Pompeii was relatively well off around the time of its destruction in 79 ce. This position emerges from studies of materials as diverse as human skeletal remains, environmental residues, settlement density in the surrounding landscape, consumer habits, and so forth.⁸ Positive interpretations converge to make the city in many ways emblematic for broader views of Roman economic growth and success. Second, it is widely acknowledged that Pompeii had a very large enslaved population, while slavery often forms the lens through which the city’s archaeology is interpreted.⁹ Enslaved people appear ubiquitous at Pompeii: they staffed the city’s brothels, fought in its gladiatorial shows, lived in its dingier rooms and sub-basements, served as go-betweens for its financial class, turned its flourmills, and laboured in its surrounding fields. Of course, Pompeii’s evident prosperity does not exclude the idea of a city full of enslaved and exploited workers, but the juxtaposition should remind us that Pompeiian prosperity was not shared universally, even as scholarship does not always acknowledge this fact.¹⁰ I propose to join these two narratives into a single story about how slavery created the prosperity that some (but not all) Pompeiians enjoyed.

This story is pursued here through quantitative methods and statistical tools, and I offer some immediate disclaimers. The following describes probabilities rather than certainties, as I feel this represents the best way forward for historians of earlier slave societies such as that of Rome to engage with debates about the economic contributions of slavery. To calculate their contribution to Pompeii’s economy, we will need to think about how much income enslaved people generated, what it cost to purchase and keep them, and how many laboured in the city. Certainly, Pompeii provides riches for the ancient historian looking to establish these parameters, but by comparison to the modern world the evidentiary situation still presents a high degree of uncertainty, and it seems impossible to calculate precisely the contribution of enslaved people to Pompeii’s economy, as has been done for more recent slave societies.¹¹ I thus approach the topic probabilistically by constructing a stochastic model based upon information for expenditures and incomes, including the cost of owning slaves. For all the relevant parameters, even as we cannot arrive at exact figures, we can reasonably establish plausible ranges within which the answer is likely to lie, while statistical tools can combine those ranges to report a probable level of overall slave-generated income.

As with all model-building, this will be an exercise in simplification. In particular, I focus deliberately on the economic contributions of Pompeii’s enslaved people, what economists would refer to generally as the goods and services included in broad measurements of any economy. While the approach may be conventional enough in economic literature, for historians’ purposes this means viewing a wide range of enslaved activities as essentially income-generating labour. Relatedly, I frequently turn to scholarship on American slavery not to insist on its structural resemblance to Roman slavery, but because Americanists’ focus on slave labour as opposed to other slaving strategies has developed helpful tools for my purposes.¹² Of course, Roman slavery existed for numerous reasons, and ancient slaving strategies were by no means limited to the appropriation of labour.¹³ My account does not intend to suggest otherwise, and certainly some reasons to keep slaves are more legible as productive labour than others. To give a Pompeiian example, epigraphic sources record cash incomes earned by prostituted slaves for their owners and one graffito even refers to an enslaved woman who ‘made money with her body’ (Corp. Inscr. Latin., iv. 1948, ex corpore lucrum faciebat), while the individual who wore a large, snake-shaped gold and silver bracelet inscribed ‘from her master to his slave girl’ (Ann. Epig. 2001, 803, dominus ancillae suae) was also probably enslaved for reasons of sexual gratification, although not necessarily to yield a cash income.¹⁴ I follow economic convention in understanding both these and all other activities undertaken by the city’s enslaved as resulting in goods and services, that is, in different forms of value, which were appropriated by slave owners. While the approach will not capture the full social world of slavery in Pompeii, I stress that it intends to form a response to what I perceive as Roman historians’ neglect of enslaved labour’s economic role. I also intend this move to answer a recent call for the quantification of less positive aspects of ancient economic history.¹⁵ As I noted, while Roman economic historians engage energetically in quantification, slavery is regularly left out of their calculations.¹⁶ To the extent that Pompeii and the wider Roman Empire were slave societies, we need ways to incorporate slave labour into our economic models, and simplification is undertaken here in support of that goal.

I slavery and roman economic development

Consensus now holds that the Imperial Mediterranean economy experienced some degree of per capita growth, something that potentially sets Rome apart in the long run of premodern economic history in the region.¹⁷ Economic growth is a matter of increases over time, while Pompeii, as a city frozen at a particular moment, does not provide the time series of data often held up as signs of economic expansion. However, the city of 79 ce continues to play an important role in such discussions, as its evident prosperity is commonly understood to reflect the longer-run growth experienced up to that date by many towns of the Empire.¹⁸ Indeed, Pompeii stands as perhaps the finest example of a Roman town in an Imperial Mediterranean that, while still predominantly agrarian, saw historically high urbanization rates thanks to rising production.¹⁹ Most scholars understand the development of Pompeii’s urban economy, as also that of the wider Roman economy, as Smithian in nature.²⁰ That is, under the pax Romana, Roman institutions provided a favourable environment for trade, encouraging specialization and leading to efficiencies in a manner famously described by Adam Smith. The most complete model of this scenario for the Roman world remains that advanced by Keith Hopkins, according to whom the extraction of taxes and aristocratic rents in the provinces in cash to be spent in Rome and the imperial core led to the rise of markets at local levels across the Empire.²¹ In Hopkins’s view, this meant that the Roman world as a whole must have experienced some level of growth. The keystone to Hopkins’s model was the acceleration of maritime commerce, a claim supported, innovatively at the time, with bar graphs of shipwrecks and coin production showing a familiar rise and fall over the Imperial period. Study of these and other proxies has become far more sophisticated over the last decades, supporting a lively discussion about the scale and development of the Roman economy; however, slavery features little in this literature.²²

Currently, the main theoretical scaffolding for understanding Roman economic growth comes from Douglass North’s new institutional economics, which sharpens focus upon transaction costs and institutions conducive to commerce.²³ Through this lens, economic development and prosperity become the almost natural response to the falling away of obstacles to trade. That much of Roman production was carried out by enslaved people matters little, even as economic historians acknowledge the centrality of slavery to Rome’s political economy.²⁴ The new institutional approach is invoked to show how reductions to the risk of slave owning encouraged this situation, but these institutions rather than slaveholding itself are seen to support economic development.²⁵ Thus, recent pronouncements of a ‘new economic history of the Roman Empire’ focus mostly on market integration.²⁶ As one recent study of slavery (not economics) concludes, ‘slavery is widely considered to be one of the most important phenomena in the ancient world, but plays little causal role in explaining many major developments in the course of antiquity’.²⁷

This state of affairs seems particular to Romanists. Ancient economic history’s turn towards institutional models has not included the accompanying interest in inequality-generating institutions visible in economics more broadly.²⁸ Considering the acknowledged importance of slavery to Greco-Roman societies, it is noteworthy that there has never been anything resembling the energetic debate over slavery’s contribution to industrial capitalism that has carried on largely continuously since the publication of Eric Williams’s Capitalism and Slavery in 1944 and especially following Robert Fogel and Stanley Engerman’s Time on the Cross.²⁹

The oversight may reveal a deeper legacy. As often noted, Hopkins’s trade-based model of Empire-wide growth responded to Moses Finley’s picture of a static, underdeveloped ancient economy. In turn, while he made key advances in the study of ancient slavery, Finley trained his seminal 1974 work on The Ancient Economy more upon the impact of slavery on markets than upon its exploitative qualities as a labour system. In doing so, Finley was transmitting the views of Karl Polanyi with whom he famously studied.³⁰ Like Karl Marx before him, Polanyi insisted that economies were structured differently across time; however, Polanyi located this distinction in systems of exchange, in contrast to Marx’s successive modes of production.³¹ Polanyi saw modern capitalism as a ‘market society’ whereas earlier societies allocated resources by non-market means, while he also saw less difference in this regard between antiquity and other pre-industrial periods. Rather than the slave mode of production, Polanyi took the Empire’s distinguishing feature to be its annonary system, the imperial provisioning of grain, which in his view was nothing more than a big redistributive system.³² This understanding of the unusualness of modernity, as well of the contrast between capitalist and non-capitalist forms of exchange with a focus on markets, was fundamental to Finley’s portrayal of the ancient economy. Subsequent work has not always shared Finley’s pessimism about Roman markets, but exchange remains at the centre of debate, as already seen in the work of Hopkins. This development is neatly encapsulated for Pompeii by the evolving thought of Finley’s student Willem Jongman, who once suggested in a book dedicated to Finley’s memory that Pompeii was a consumer city with a limited economy, but who now sees the city as ‘far more prosperous’ thanks to specialized agriculture and markets, which supported growth. The city thus forms a microcosm for broader thinking about the Roman economy.³³

In this way, debate about the Roman economy including Pompeii remains genetically predisposed to focus on commerce and trade over labour, even as its overall position on performance shifts. By contrast, Marxist analyses concentrating on the economic contribution of slavery find little mention in current scholarship. We should acknowledge the concerted efforts of, especially, Italian historians of the last century to illuminate the Roman slave mode of production, but their work has been criticized for its sometimes overly rigid view of slavery’s dominance and finds few current imitators.³⁴ By pointing this out, I am not advocating for rehabilitating Marxist understandings of class whose application to Roman society have proven controversial.³⁵ But I fear we may have thrown the proverbial baby out with the bathwater by overlooking the contribution of slave labour to Roman economic growth.

Recentring slave labour in the economic history of a paradigmatic Roman city such as Pompeii offers potential to address recent criticism that Roman historians fail to include ‘lives of working people as a major component of economic history’.³⁶ To do so, we need both to consider the finer details of the Roman mode of production and to ask how it may have contributed to broader economic development, just as similar questions are being asked about other periods.

II qualitative impressions of pompeiian slavery

Recent archaeological discoveries reveal new details about the lives of non-elites, including slaves, in Pompeii, deepening the impression that exploitation and dependency were characteristic of the city’s economy. Labour historians may find it unsurprising to learn that Pompeiian workers, too, inhabited cramped, dingy spaces; however, it is worth stressing the extent to which Romanists had until recently directed their energies exclusively to the study of Pompeii’s elite spaces and luxurious objects. New interests in ‘the other Pompeii’ thus address a significant imbalance in the specialist literature. More broadly, this evidence frames the model-building exercise in this paper in two important ways. First, it lends the subsequent mathematical approach a material basis, underscoring even if in an impressionistic way the importance of exploitation to the realities of Pompeiian labour. Second, my presentation intends to show some limitations to an archaeological approach to Pompeiian labour, especially in determining enslaved status from material evidence, or in scaling up from single pieces of evidence to a wider understanding of slavery’s role in the urban economy; these are issues that the quantitative model of the subsequent section will help address.

Pompeii and its territory already saw considerable slaveholding by the Late Republican period, as famously attested in the region by Spartacus’ great uprising in 73–71 bce, an episode possibly depicted in a fresco from the House of Amandus at Pompeii.³⁷ At the time of the eruption of Vesuvius, the city’s suburbs were dotted with productive villas.³⁸ In the city, slaves and freed slaves were very prominent, as urban social, economic, and political networks were closely informed by servile or ex-servile status. Numerous political graffiti help to demonstrate the role played in Pompeiian politics by relationships founded in slavery, naming freedmen who supported the candidacy of their former masters for municipal office.³⁹ The wax tablets from Herculaneum and Murecine show enslaved people acting as crucial agents in business affairs.⁴⁰ As epigraphy regularly attests, the enslaved, including many house-born slaves sometimes thought to have held privileged positions in their masters’ households, were in fact subject to prostitution and sexual exploitation.⁴¹

If these last examples reveal the variety of urban slaving strategies, archaeological evidence also emphasizes the importance of exploitation to Pompeii’s productive labour. We find shackles and fetters from both suburban villas and urban areas, often from underground cells, seemingly reflecting Columella’s advice to keep chained slaves below ground (De re rustica, i. 6. 3). These include stocks intended to chain up multiple individuals by their ankles.⁴² In one gruesome case, the tibia of a skeleton was found with a manacle still attached to an iron spike in the floor in the villa of the Mosaic Columns outside the Porto Ercolano, and other skeletons with ankle shackles have been discovered around the city and its environs.⁴³

The city reveals multiple spaces often associated with slaving activities. Looking at its highly controlled means of access, Elizabeth Fentress raises the idea that the large porticoed building off the forum referred to as the Eumachia after its patron functioned for slave auctions known to have taken place in the area.⁴⁴ Once purchased, enslaved people were kept throughout the city’s residences. Cato refers to separate rooms on his rural estate for slaves (De agri cultura, xiv. 2: cellae familiae). In the urban setting of Pompeii, living quarters for the enslaved have been identified with small, undecorated rooms in some more lavish houses, such as four cellae with easy access to working spaces in the House of Menander, and another set of small rooms near the kitchen in the House of the Centenary.⁴⁵ These are attractive interpretations given the presumably high presence of enslaved people in Pompeii, although we must admit that archaeology reveals nothing in such cases about the legal status of those moving through or living in these structures.

To these cases, we now add two discoveries by the superintendency of the archaeological park (the Parco Archeologico di Pompei, the institute governing archaeology in the Vesuvian region). The first (Plate 1) comes from excavation in 2020–1 of part of the suburban villa di Civita Giuliana just north of the city walls. Here, a courtyard with a chariot in its centre had a suite of three small rooms (3.6 x 3.6 metres) along its eastern side serving to stable animals, as one room contained the body of a horse. The central room housed several people. Lit only by a spare window and with undecorated walls, it was crammed with three beds of simple manufacture. At the time of the eruption, it was also used to store parts of a chariot rig and assorted amphorae and crockery. The excavators identify this as the sleeping quarters for several of the villa’s enslaved.⁴⁶ The nature of this room and its location on the estate at least suggest their lowness and their involvement in production, while the use of the bedroom to keep not only people but also farm instruments and livestock recalls Varro’s famous reference to an estate’s human labour force as its ‘talking tools’ (Res Rusticae, i. 17).⁴⁷ However, as with Pompeii’s other structures associated with slaves, we should be clear that nothing allows the identification of the legal status of the room’s inhabitants.

The second discovery brings us into the city walls where the superintendency has explored a bakery operation installed into a house in the central Regio IX.⁴⁸ The main areas of the structure, retaining the atrium plan and fine frescoes of the house, probably served retail and residential purposes. Political slogans graffitied on some walls suggest a certain volume of foot traffic. The bakery’s productive infrastructure was located along its western side in a suite of rooms separated from the structure’s main wing (Plate 2). Access to this area was highly restricted through a single door into the main courtyard, with no means of egress onto the street. Movement was further restricted, as several windows were closed with iron bars — the excavators suggest a prison-like space. In close quarters, we find a mill installation for grinding flour, basins for kneading dough, and a large bread oven. A rectangular room with a trough at the south served to stable animals for the mill. The productive spaces lack the decorative finishes of the main house, while the floor around the mill shows grooves from the constant treading of animals. The excavators compare a grim scene in Apuleius’ Golden Ass (ix. 11–13), in which the transformed Lucius is imprisoned in a flour mill with several horribly treated men, probably slaves. Fergus Millar refers to this same passage as ‘one of the most vivid presentations from antiquity of what the extraction of value from the poor really meant’.⁴⁹ Other citations can be found, as flourmills are frequently mentioned as punitive destinations for household slaves.⁵⁰ The textual evidence makes an association of the Pompeiian bakery with enslaved people attractive, although again we cannot be certain of the status of the workers. Certainly, the structure reveals the highly rationalized use of space to exploit human and animal labour for profit.

The city’s archaeological evidence thus emphasizes that Pompeii’s labour-force was exploited for productive gain in income-generating activities in both suburban villas and urban workshops. Deprivation and monitoring characterized working spaces and workers’ living quarters, as some Pompeiians denied others certain measures of well-being for their own profit. At the same time, the picture gleaned from this material remains somewhat disjointed. Even if we gather together these structures, spaces, or objects from across the city, this material can often do little more than suggest a relationship to slaves or slaving. To pursue a more complete understanding of the importance of slavery to the city’s economy, we now turn to a quantitative approach.

III a quantitative model of the contribution of slavery

If enslaved labour formed a critical part of the economic logic of production in Pompeii, how much income might slave owners expect? We can think of a Pompeiian household’s profits from slaving as a simple calculation of the income enslaved people brought less the costs of buying and keeping them. In turn, this calculation can be disaggregated into several components: the expected annual income produced by a slave, their purchase-price amortized over their years of enslavement, their annual maintenance cost, times the number of enslaved people in the urban population; the appendix provides a detailed version of the formula. This section seeks to establish these parameters in order to calculate a likely figure of the share of wealth generated every year by the city’s enslaved. That figure can then be cross-referenced with broader ideas of the overall scale and structure of wealth in the city, as the following section sets out to do. All the parameters of the calculation from slave prices to the amount of slave incomes retained by masters and so forth tend to be historically specific in every slave society and need to be reconstructed. It is therefore remarkable that Pompeii provides rare epigraphic data to inform the procedure, as we shall see. However, very much remains unknown, and my way forward, following the lead of Myles Lavan and others, is to build a statistical model that accounts for large uncertainties. For every input, rather than a point estimate or single number, I consider a plausible range within which the answer is likely to lie. In some cases, we might presume the answer lies closer to one end of the range or the other; in other cases we might think that all guesses within a range are equally probable. These sorts of a priori assumptions can inform ‘probability density functions’, essentially shaped distributions within estimated ranges. All the ranges are then put together into a computer program (see appendix), which simulates the model by inputting randomly selected variables a large number of times. The approach will not provide the resolution or accuracy offered by the archaeological and textual evidence discussed in the previous section, but it is not intended to do so. Instead, such modelling is well suited to test historical questions about unknown quantities that may be disaggregated into a number of independent parameters, as with the question here of enslaved labour’s contribution to Pompeii’s economy.

The overall contribution of the enslaved to Pompeii’s economy was in some way proportional to the number of enslaved people among the city’s general population, but the issue is not merely demographic. A key question centres upon the relative productivity of free and unfree labour and how much of a basic (free) income a Pompeiian slave could be expected to produce.⁵¹ Certainly, the archaeological data reviewed above suggest that whole or nearly whole household incomes were sometimes expected from dependent, probably enslaved, labour: at the bakery in Regio IX, workers confined to a prison-like wing undertook the entire production chain of breadmaking, from grinding grain to mixing dough and baking. However, in turning this observation into a calculation, we want to know whether Pompeiian slaves worked on average as hard as freed or free workers, and this is not something our Pompeiian evidence is prepared to tell us. We might take a cue here from historians of American slavery, who have been vexed by the question of relative slave productivity since Fogel and Engerman’s 1974 Time on the Cross. After a series of arguments for higher or lower productivity, current consensus falls somewhere in the middle, that the enslaved produced on average about as much or slightly more than free labourers. Underscoring the issue’s complexity, scholars suggest this similarity was produced by offsetting factors: labouring under coercion, American slaves worked more hours and days in a year than free labourers, but the enslaved also produced less because of go-slow resistance to forced work and because of their typically lower morale and health.⁵²

To what extent might this particular mixture of factors have applied to Pompeii’s slaves? We can only guess about the average level of Pompeiian slave health or morale, although some have seen more frequent Roman manumission as an incentive.⁵³ Meanwhile, as with the later American situation, Roman slave labour participation was probably comparatively high: juridical texts suggest that enslaved children could work from the age of five.⁵⁴ At the same time, I also note some important differences, which may suggest that Roman slave incomes were comparatively higher. Perhaps 90 per cent of American slaves worked in agriculture, generally thought to be less productive per capita than urban industries, while Roman slaves appear far more frequently in urban production. This is not only true of the primarily urban population we are considering; it was probably also characteristic of the Roman world more generally, where enslaved people were frequently used in craft industries.⁵⁵ There were also some costs to keeping slaves, such as slave insurance, that are relevant in making calculations about net productivity in the antebellum South but not in Pompeii.⁵⁶ I thus see little reason to think Pompeii’s enslaved were significantly less profitable, and some reasons to think their labour brought a slightly higher income than free hired labour.

On this basis, I use the assumption that the gross wealth produced by enslaved persons at Pompeii resembled basic free incomes, as calculated below. Once we net the cost of buying and maintaining a slave, we can arrive at a notion of the wealth made through slave owning. This in turn may be multiplied by the number of slaves to arrive at a probable total income from enslaved labour at the scale of Pompeii’s economy.

In building the model, I incorporate several assumptions worth making explicit. First, in order to capture the major share of Pompeiian slave-owning households, I seek to reconstruct non-elite incomes. These are not bare subsistence incomes, but the ideal consumption levels described here are ‘normal’, at least according to current scholarly understanding. ‘Elite’ is a somewhat imprecise term here, but surely the number of slave-owning households at Pompeii far exceeded the hundred or so members of the city’s decurional order, or other, similarly restricted reconstructions of the city’s richest households.⁵⁷ Indeed, it is often stressed that slaveholding reached far down the Roman economic spectrum, as many more households than Pompeii’s small ‘elite’ benefitted from owning slaves. For similar reasons, I reckon with one-slave families. Many Pompeiian households including those wealthiest households owned multiple enslaved people; however, I assume that increases to the scale of slaveholding brought proportional increases of income with further income coming from other sources up to a point of diminishing returns. Almost certainly, large households with multiple slaves profited from landholding, rents, and other activities related to, if not directly supported by, slaving. The banking activities of the Pompeiian L. Caecilius Iucundus, elucidated by wax tablets found at his house, help to illustrate what I am describing. Slave agents were instrumental to Iucundus’ lending activity, but a complex relationship to enslaved labour here makes it hard to conceptualize the share of profit attributable to slaving. Finally, I assume that the single enslaved person in my ideal ‘normal’ household was an adult male.⁵⁸ More expansive accounting of the productivity of Pompeiian or Roman slave labour might try to accommodate an age and sex model of the enslaved population, but the details of slave demography are very much debated. I employ the simplifying assumption that subadult or female slaves produced but also consumed less, making differences roughly proportional.⁵⁹

My approach to household earnings is to derive income levels from a handful of data for expenditures. We have essentially no direct information from Pompeii for wages themselves.⁶⁰ By contrast, the expenditure approach can draw upon the existence of several ‘shopping lists’ found in graffiti around the city. As an example, I transcribe one of the fullest and most important for my purposes, a week’s list of costs including the price of ‘bread for the slave’ (Corp. Inscr. Latin., iv. 5380).⁶¹

This and the other shopping lists are casual, everyday graffiti containing prices for groceries and other consumer goods. In working with this information, I am indebted to Kim Bowes’s investigation of the relationship between these lists and patterns of consumption.⁶²

Slave consumption levels

The calculated rationing of food is widely attested in slaveholding societies.⁶³ Cato’s description of rations of 4–4.5 modii of monthly grain to slave fieldhands and 3 to managers is frequently cited as proof of this practice in Roman society (De agri cultura, lvi–lviii). Wider Roman evidence for slave maintenance includes legal texts and several dozen papyrus contracts for ‘slave maintenance’ (trophimon doulikon). On this basis, William Westermann suggests that Greek and Roman communities possessed notional if not legally prescribed standards for daily slave rations.⁶⁴ Attested costs are extremely low, representing little more than bare subsistence.

I thus take the specification ‘bread for the slave’ (panem puero) in the shopping list Corp. Inscr. Latin., iv. 5380, to reflect the practice of controlling the dietary incomes of enslaved people. The slave ration recorded by the list is for bread only at a cost of 10 asses, equivalent to HS 2.5, for an eight-day week, or 456 asses (HS 114) per year. As Bowes points out, the specification ‘bread for the slave’ refers to quality not quantity. The same Pompeiian shopping list also includes one entry for panis cibarius, a term applied elsewhere specifically to slave bread, for the same price of 2 asses per unit.⁶⁵ Other bread costs in the same list are double or higher. Depending on how many individuals we understand the list to refer to, these costs were for single or multiple loaves of higher quality bread.

Assuming a standard loaf of 2.5 libra, the poorer quality bread for slaves contained just below 2,000 calories.⁶⁶ For an average adult male slave, this low caloric budget ensured survival but did not grant much residual energy for work.⁶⁷ Slaves thus probably survived on more than bread. A supplement could be made as a reward or as part of the owner’s largesse, but such bonuses were discretionary and netted from a master’s disposable income.⁶⁸ What about the basic calories required to permit slaves to work? Comparative data suggests that an average adult male 1.77 metres tall and weighing 78kg needed 300 calories above ‘slave bread’ for basal metabolism.⁶⁹ Hard work in either the city or the countryside called for more food. Enslaved people working fields on big plantations in the antebellum US consumed daily diets of some 4,200 calories.⁷⁰ Cato’s fieldhands received wine, oil, fish sauce, and vinegar (De agri cultura  lvii–lviii). These cases attest to agricultural labour, but early modern British urban workers also consumed upwards of 4,000 calories a day.⁷¹ We might imagine that, like Cato, Pompeiian owners gave their slaves larger diets only when they wanted them to perform hard labour. Thus, I incorporate into the model the possibility that Pompeiian slaves received daily food above bread averaging out to 500–1,000 calories. Working from the shopping lists, those additional calories could be made up cheaply with low quality wine or oil at daily costs of 1 as, or 365 asses (= HS 91.25) per year.⁷² Rounding the numbers, I reconstruct the annual slave ration cost of HS 114 (bare subsistence) to HS 206–297 (approximately 2,500–3,000 calories).

The bulk of the cost of keeping a slave was spent to feed them, while we may assume that enslaved people shared little in other items of typical household consumption baskets. The prices of bare necessities like clothing and the enslaved’s share of fuel and shelter costs are hard to obtain directly, so I account for them by adding 10 per cent to the ration cost following historical comparisons.⁷³ Notably, the resulting low bound of HS 126, with 10 per cent added to the bare HS 114 cost of bread, closely resembles the rural slave maintenance cost of HS 140 estimated by Richard Duncan-Jones using Columella’s agricultural treatise.⁷⁴

The purchase price of a slave

Slave owning also implied the sunk cost of a slave’s purchase price. There are two variables here: slave prices, for which we have a limited number of datapoints, and the expected term of enslavement, which depended upon demography along with the customs of slave owners regarding the late-life treatment of slaves. For the first, Table 1 collects reported slave prices from the Vesuvian cities.⁷⁵

Giuseppe Camodeca has revised the text of a sale of an adult enslaved man on a wooden tablet from Herculaneum (Tab. Herc.,²  lxix). Initially thought to report a low price of HS 900, the figure misses its first digit, which he restores as HS [1]900.⁷⁸ With this revision, slave prices closely correspond to age and gender: a first level from HS 600–725 for enslaved children, a second for an adult enslaved woman, and a third for adult men, five in total, ranging from HS 1,900–4,050. Incidentally, the market favouring adult males may reveal local preference for labour over other slaving strategies.

Next, the question of how many years the enslaved worked between purchase and manumission or death. There is a great deal of guesswork here, but if masters favoured purchasing young enslaved adults who had survived the perils of childhood, a life expectancy of 31–34 from age 10–15 (e₁₅ = ~31–34) for a male slave implies a range of several decades’ survival after purchase.⁷⁹ We can deflate this figure by supposing that owners moved to sell or manumit enslaved men who reached their early 30s.⁸⁰ For my model, I amortize a purchase cost of HS 1900, 2750, 4050 (PERT distribution⁸¹) over 10–18 years (uniform distribution).

The slave owner’s income

To reconstruct a ‘normal’ income, I return to the shopping lists. The same weekly shopping list (Corp. Insc. Latin., iv. 5380) that includes ‘bread for the slave’ offers a range of other items. Foodstuffs include bread, meat, cheese, wine, and vegetables. The list also records household items such as a wick, a lamp, a cooking pot, incense, and possibly a bucket or situla. The ensemble thus resembles a weekly consumer basket.⁸² The non-food items suggest to Koenraad Verboven that the consumers were relatively well-off, but I hesitate to describe goods such as lighting or crockery as luxury items.⁸³ It is also noteworthy, as Bowes points out, that the per diem costs of this list fall within a tight range suggested by most other shopping lists. In this sense, the lists seem promisingly to record a narrow band of ‘normal’ incomes for members of Pompeiian households.

We should acknowledge, however, that this ‘normal’ Pompeiian income is high in an Empire-wide perspective, representing five or six times subsistence according to recent estimates or around twice the gross amount of base legionary pay (HS 900) in the same period.⁸⁴ As Bowes herself notes, the shopping lists form another indication that Pompeii was a wealthy city with a significant segment of the population earning incomes well above subsistence.⁸⁵ To control for the difference, however, I employ the Pompeiian data for the upper bound but extend the lower bound of the probability density function downwards to basic Roman legionary pay. Of course, this is not to insist that many Pompeiians were actually legionaries, but the figure offers an idea of a realistic standard income level more in line with Empire-wide data.

Table 2 reports the daily costs of the shopping lists, providing a range of annual expenditure-based estimates for basic incomes. For Corp. Insc. Latin., iv. 5380, I focus only on expenses of the free individuals, assuming two people for half of the eight-day week and one for the other half. Using daily bread prices as guide, I assume the other four lists record expenditures for single individuals for single days, although this is largely a guess. The highest cost forms an outlier, perhaps revealing more than one individual’s consumption, and I calculate a mean from the remaining four data points of HS 1,635.

The shopping lists describe payments on a daily or, in the case of Corp. Insc. Latin., iv. 5380, weekly schedule. Other regular expenditures are hidden from these inscriptions, either because they were made less frequently or for other reasons. These are hard to enumerate or quantify; however, we might include one known cost in an urban setting paid less frequently but nonetheless critical to household budgets, and that is rent. Following Geoffrey Kron, I use a notional figure of 10 per cent of income above other household expenditures as a low-end estimate. Caesar’s proposal for rent remission in Italy to help those who paid below HS 500 a year suggests a possible low-end normal cost.⁸⁷ However, for budgets of HS 1,635, HS 500 still represents a high percentage (30 per cent) of annual income, arguing for a lower normal figure. Thus, the model uses an annual income range of HS 900 to 1,799, adding a 10 per cent rent cost to the upper bound established as the mean of those figures reported by the shopping lists.

The slave population and total slave income

The next issue to work out is how much profit in total was provided to the Pompeiian economy by slave owning. To make this calculation, we multiply per-slave profit by the number of slaves in the city. With no direct evidence for the count of enslaved people at Pompeii, the topic is best approached as a percentage of total population. Pompeiian demography is something of a hot topic following the publication of a newly discovered inscription describing the achievements of an elite Pompeiian — his name is not preserved — on his tomb outside the Porta Stabia. This text records a boast that he gave a banquet on 456 couches (tricilinia), each seating fifteen guests, to the people of Pompeii (populo Pompeiano) on the occasion of his coming of age. The populus Pompeianus thus numbers 6,840. Taking this as the city’s free adult male population, Massimo Osanna suggests Pompeii contained 30,000 people.⁸⁸

This figure is strikingly higher than previous calculations, from Giuseppe Fiorelli’s early estimate of 12,000 to the conservative 8,000 proposed by Heinrich Eschebach, and others.⁸⁹ Miko Flohr’s careful study of Pompeii’s houses leads him to suggest a range of 7,291 to 11,727 people living in 1,434 habitable units with 10,214 ground floor rooms.⁹⁰ Those who favour these lower estimates point out that some attestations of municipal banquets refer to the inclusion of women or children, and, if we understand the populus Pompeianus as similarly inclusive, we might halve Osanna’s figure, inferring about 15,000 people. However, women or children at municipal banquets seem to have been exceptional enough to note in commemorative inscriptions, as is not the case in the newly found text.⁹¹

The higher count requires us to work through some further implications. For one, the elevated figure bears upon wider Roman urbanization rates and Malthusian interpretations of the Imperial economy based on high population estimates.⁹² These much-debated topics are larger than can be addressed here, while it seems unwise to work from the even more fraught question of total Imperial population to that of a single town. Specific to Pompeii, there is the issue of urban density: where to put all those Pompeiians? The town lacked the multi-storey apartment blocks of Ostia or Rome. Gabriel Zuchtriegel has used new GIS data to raise the number of excavated habitable spaces to 12,375, with an estimated 16,100 across the entire city. Noting the room at the villa di Civita Giuliana with three beds in fewer than 16 square meters, he identifies similarly cramped living spaces within the city.

It is also fundamentally important that earlier, lower population estimates based upon the city’s housing stock sought to count people residing within Pompeii’s walls. It is not necessarily the case that the populus Pompeianus named in the inscription equated with intramural residents, nor should we confine our interests to such.⁹³ Indeed, the fuller territorial population merits special consideration because of the importance of slavery to suburban agriculture. Pompeii’s territory was densely occupied at the time of the eruption, making it possible to reach a compromise count that sees both an urban population of the order of 15,000 and a total territorial population maybe 50 per cent larger.⁹⁴ The implications of the new inscription thus might, while radical at first glance, fit wider understandings of the size of the economy, rather than the size of the walled city. This is precisely the sort of debate best represented by a probability density function rather than by any point estimate. I employ a PERT distribution of 15,000; 20,000; 30,000 (see above, n. 81). The overall range reflects the newly discovered inscription, while the mode favours lower counts in deference to previous estimates.

The next question is how many of Pompeii’s residents were enslaved people. As a wealthier region at the Empire’s core, Central Italy held a comparatively high enslaved population. Early estimates based on the prevalence of freedmen in the epigraphy of Rome and Italian cities suggested that the peninsula’s towns saw rates of enslavement as high as 30–40 per cent.⁹⁵ This range is now reinforced for Pompeii’s neighbouring town of Herculaneum by Luuk De Ligt and Peter Garnsey’s analysis of a partial list of that city’s male population inscribed on fragmentary marble plaques. This ‘Herculaneum album’ suggests exceedingly high numbers of freedmen, leading them to reconstruct enslaved populations as high as 45 per cent of the overall population.⁹⁶ These are high figures by comparison to some recent discussion of Empire-wide levels of enslavement. Egyptian census returns, for example, suggest enslaved people made up some 15 per cent of the population of Middle Egypt. It is likely that Italy at the Imperial core saw higher numbers of slaves than Egypt, and I use a lower bound of 20 per cent. For reasons discussed in the introduction, I take the entire slave population as economically contributive.⁹⁷

IV slavery and pompeiian inequality

We now have all the parameters to simulate the income available to free Pompeiians through slave owning (Figure). Following the simulation procedure detailed in the appendix, the expected value (mean) across 100,000 iterations is HS 6,208,531 with an 80 per cent confidence interval of HS 2,622,741–8,704,608. That is, according to the model, in the order of HS 6 million each year flowed to Pompeii’s masters from their ownership and exploitation of enslaved people. To know what these figures mean in historical terms, we next need to contextualize them within a broader notion of the scale of the city’s economy. Did the exploitation of the enslaved at Pompeii create the whole of the city’s annual income, or half, or some other proportion? Pompeii’s economy must have been larger than that value described by the model: there were wage-earners in the city; there were dependent labourers who were not slaves; some earned wealth through financial activities or numerous other potential sources. We would like to know how much this all adds up to, although some approaches to the question seem hazardous. For example, studies have multiplied a notional figure for subsistence consumption by total population to extrapolate the size of the Roman economy, but we have already seen in the shopping lists credible signs that many Pompeiians earned multiples of subsistence.⁹⁸ We also cannot simply equate the scale of slave-generated wealth to the share of slaves among the city’s population. The point of this study so far has been to show that many Pompeiians exploited enslaved people so that they themselves did not need to work, suggesting that both productivity and income were very unevenly distributed across the population.

Considering these issues, I approach the topic of the proportion of slave-generated income instead through the lens of economic inequality, an approach presenting three particular benefits. First, as explored below, tools recently developed by economists to study historical inequality demonstrate that the overall distribution of an economy’s wealth is constrained by its size. That is, inequality offers a way to study the overall scale of Pompeii’s economy. Second, these tools for reconstructing and studying unequal distributions of wealth seek to model precisely situations in which some Pompeiians were poor or propertyless while others were wealthy. Third, several recent studies have in fact sought to reconstruct the distribution of wealth at Pompeii. These studies universally take as their starting point the city’s housing stock, with house size understood to proxy wealth. Their results are thus based upon evidence independent from what we have so far considered, offering a way to test the model.

Measuring Pompeiian wealth inequality

House size is employed broadly to study economic inequality by archaeologists for early societies for which other data are lacking. Pompeii presents an extensive and well-documented dataset, and several studies have sought to use this data to model the city’s wealth stratification.⁹⁹The approach is not entirely unproblematic. A house’s location or decor, not merely its two-dimensional floor plan, might inform its value. It is moreover sometimes difficult to locate the precise boundaries of houses in Pompeii, as some house-owners are known to have purchased adjacent structures and incorporated them into their own, while there is the lurking problem that many houses’ upper storeys were destroyed by the eruption. Most studies assume that the number of houses at Pompeii is high enough to mitigate these issues. It is also instructive that, as Bart Danon shows, the distribution of house size in the city — that is, a plot of the number of houses of each respective size — produces a shape that looks like other historical distributions of wealth plotting more direct forms of measurement, making it plausible that the overall housing-stock of Pompeii reflects the wealth of house-owners, granting some simplification to the procedure.¹⁰⁰

Table 3 summarizes housing-based estimates of wealth inequality for Pompeii expressed as Gini coefficients.

Flohr’s study employs the largest dataset of 1,168 houses containing 10,294 rooms. Globally comparative studies by Thomas Kohler et al., by Mattia Fochesato, Amy Bogaard and Samuel Bowles, and by Elizabeth Stone all employ a sampling from two excavated regions of the city, representing perhaps 20 per cent of Pompeii’s houses; considering this more limited dataset, the closeness of their results to those of Flohr is encouraging. Notably, Kohler et al. suggest Pompeii featured one of the highest levels of inequality of the sixty-three pre-industrial sites they examine. An impression of historically high inequality at Pompeii emerges from other quarters. Based on a sample of 366 houses, Danon argues that, even as we can identify no senator of Pompeiian origins during the Empire, Pompeii contained not only households of decurional wealth, but several of senatorial-level wealth with assets valued at HS 1 million or more.

The highest degree of wealth inequality at Pompeii is suggested by Samuli Simelius, whose upper-end Gini coefficient of 0.92 would imply a tiny elite controlling the vast majority of the city’s wealth, something that seems unhistorical in comparative terms. The basis for Simelius’s inflated figures is his observation that Flohr’s calculations do not consider the large number of Pompeiians who did not own houses. Simelius charts non-owner Pompeiians as holding zero property, creating an enormous left tail to his graph, resulting in the very steep Lorenz curves that produce high Gini coefficients. Simelius’s point that housing-based models do not account for non-owners is valuable. However, his method of incorporating these people is flawed. First, he assumes that a full Pompeiian population of up to 30,000 people resided within the city walls, whereas we have already found reason to question this claim. Second, he assumes that Pompeiians who did not own houses also did not have any wealth. This probably applies to most slaves, at least technically, but free non-owners at a minimum had the means to pay for rent and subsistence.

The issue of how to include the propertyless into models built upon housing stock or other elite datasets such as funerary assemblages is taken up independently by Mattia Fochesato, Amy Bogaard and Samuel Bowles, who explore the problem across multiple premodern societies. The authors develop a mathematical approach based on a theoretical three-class population with a portion (⁠$n$⁠) holding little wealth (⁠$s$⁠), another portion with zero wealth (⁠$u$⁠), and a wealthy class constituting the fraction $1−u−n$ who own the rest of the wealth in that society. Following discussion of this hypothetical society’s wealth distribution, they find that, if G is the Gini coefficient for $1−u$ percentage of a population, in other words, only for those with wealth, then a modified Gini coefficient (⁠$G$⁠) incorporating u can be calculated theoretically as follows:¹⁰¹

They then test their theory against Gini coefficients calculated from thirty-two datasets from early modern Florence to Hohokam and Columbia River societies where evidence allows for the reconstruction of shares of wealthy, poor, and those without any wealth. The mathematical method reports results which closely resemble the historical ones. As this procedure thus looks like a reasonable way to adjust for propertyless members of a society, we may apply it to Flohr’s figures with a conservative share of 30 per cent slaves at Pompeii, inflating his Gini coefficients to 0.67–0.73. These figures remain historically high, although they are lower than those suggested by Simelius.

Income and the inequality possibility frontier

These studies of Pompeiian inequality based on housing focus on wealth, not income as my model has done. This is not strictly a problem, as the correspondence of trends in wealth and income inequality in pre-industrial societies is well established in economic literature.¹⁰² However, there may be specific reasons to speak about income rather than wealth inequality for Pompeii. Bowes’s study of coin hoards in the city suggests a very low savings level, as Pompeiians do not appear to have been amassing cash savings despite high rates of monetization. The limitations of our evidence make it hard to say if this situation was typical, or if it formed a particularity of Pompeii’s archaeology: whether it reflected, for example, the flight of cash out of the city before its destruction. However, the low visibility of liquid savings in the city’s archaeology at least complicates the usefulness of wealth, rather than income, as a metric for thinking about the Pompeiian economy.¹⁰³ Perhaps more importantly, Roman slaves were largely deprived of property by law, and it was manipulation of their income, not their wealth, that generated wealth for others.

Shifting discussion of inequality from wealth to income also offers a means to think more closely about the weight of slave-based income within Pompeii’s wider economy. Over a series of studies, Branko Milanovic has developed what he calls the inequality possibility frontier (IPF).¹⁰⁴ The metric depends on the idea that income inequality is not limitless in a given economy but depends upon that economy’s size. The proposition is logical: because all members of a society must receive at least enough to survive, the surplus above subsistence minimum is what is available for richer classes to capture without causing mass starvation. In an economy that produces little above surplus, less separation is possible between wealthier and poorer members. In a larger economy, more income can be concentrated at the top. Milanovic observes that, for a hypothetical economy whose elite approach zero as wealth is increasingly concentrated in fewer and fewer individuals, the maximum possible Gini coefficient (G_max) is expressible as a function of mean income (μ), where α is the multiple of subsistence income (⁠$s)$⁠, or $μ=αs$⁠, assuming $≥1$⁠:¹⁰⁵

It is worth stressing that the IPF represents an upper threshold rather than a realistic figure, as most if not all societies did not display the concentration of almost all available wealth upon a very small elite. In the case of Pompeii, we possess the ingredients to calculate this threshold IPF above the city’s real Gini coefficient. Because Pompeiians’ mean income (depending on economy size and population size) relies on several debated parameters, I use probability density functions rather than point estimates. For minimum subsistence income, I employ the low-end estimate for slave rations of HS 114. The resulting estimated value for the IPF of Pompeii using only the size of income generated by slaves is a Gini coefficient of 0.56.

Bearing in mind that housing-based calculations of Pompeiian inequality look at wealth, not income, we can still observe that this IPF based on slave-generated income alone already resembles the lower range of most estimates of Pompeiian inequality, leading us to think that the total economy was not very much larger. The exception is Simelius’s upper estimate, but this may simply be a flaw to his model: the Pompeiian economy was not large enough to create such enormous levels of inequality.

As this already suggests, the IPF can help to address the question of what proportion of the total economy slave production comprised. As the real size of Pompeii’s economy was undoubtedly larger than the share produced by slaves, the IPF will also have been higher, and we may explore several hypothetical scenarios (Table 4).

Consider the Gini range of 0.67–0.73 for Pompeiian inequality, corresponding to the adjusted figures from Flohr. To produce a distribution consistent with this range, we must think that slaves contributed between 50 and 75 per cent of the total economy. If the overall economy were much larger relative to slave-generated income, it would allow for an IPF far higher than the levels of inequality otherwise reconstructed from the city’s housing stock. Thus, it is consistent with what we see in the city’s housing stock to think that slaves were responsible for half or even more than half of Pompeii’s economy.

A second metric to consider is what studies of income inequality refer to as the ‘inequality extraction ratio’ (IER). This is the ratio between the observed Gini coefficient and the maximum possible Gini coefficient.¹⁰⁶ This IER represents an index of the ability of a society’s wealthy to concentrate income upon themselves. As the IER approaches 100, a smaller and smaller number of individuals capture available surplus. When the IER goes above 100, the situation becomes unsustainable, as some segment of society falls below subsistence income. Study of twenty-eight pre-industrial economies by Milanovic and his co-authors suggests a typical mean IER between 75.5 and 76.8. This tells us that, in the majority of historical cases, the highest income groups were able to ‘extract’ about three quarters of possible inequality.¹⁰⁷ Scheidel suggests that the Roman Empire in the mid second century ce saw an IER of 80, slightly higher than this global average.¹⁰⁸ For Pompeii, imagine a hypothetical economy twice as large as the income generated by slaves with an IPF of 79 (see Table 4). Dividing this figure by Gini coefficients of 0.67–0.73 produces an IER ranging from 85 to 92, comparatively high figures. A highly extractive Pompeiian elite would not be entirely out of line for a Campanian city with more slaves than other parts of the Empire; however, the high range perhaps favours a somewhat larger overall economy relative to slave-generated income with a higher IPF and lower IER.

All these calculations show again how the model of slave-generated income fits generally with other ways of measuring Pompeii’s economy. Even if we cannot say precisely how large Pompeii’s economy was in 79 ce, it seems likely that enslaved people accounted for about half of the overall economy. If we imagine a highly extractive Pompeiian elite, we might reconstruct a higher share of income based upon slaving, or vice versa. Whatever the case, the income produced simply by denying the enslaved full share of their production was undeniably substantial within the context of the wider economy. Probably, it represented the largest single source of income. In fact, we might go one step further: if the city’s economy was significantly larger and Pompeiians possessed sources of income larger than slaveholding, we might expect to see a very different city with an even steeper distribution of urban houses.

These calculations take on greater interest because we are here in a position to move beyond a somewhat anodyne discussion of inequality and extraction to show what these concepts mean in human terms. After all, we are not merely talking about ‘elite extraction’, but discussing a world of masters and enslaved people. A primary way for prosperous Pompeiians to gain wealth was by appropriating the labour of slaves. The prison-like wing of the bakery in Regio IX, stocks and fetters found around the city, and the cramped sleeping quarters in the villa di Civita Giuliana all add detail to the reality of concepts of inequality and extraction.

V conclusion

This paper has intended to show how quantitative methods can serve to strengthen an impression of slavery’s importance to Pompeii’s economy, as a case study for evaluating the place of slavery in a premodern economy. A relatively full archaeological record speaks to the exploitation of labour in productive settings in this Roman city, where the material evidence of working lives sometimes resembles those bleaker pictures painted by ancient sources. A statistical model helps overcome uncertainties inherent in this evidence, placing us in a position to suggest with a high level of certainty that the prosperity of Pompeii was built in large part upon enslaved labour, which is likely to have formed the largest single source of income for the city’s economy. More generally, my hope is that these calculations compel us to think more seriously not only about slavery as an engine of Pompeii’s economy, but also about how we might recentre labour in Roman economic history. Thanks to its unusual evidentiary situation, Pompeii presents opportunities rarely encountered elsewhere in the Roman world. However, the city is often held up as a not atypical case of urban life and economic development at the centre of the Empire. Scholars of the Roman economy should bear in mind that Rome was a slave society, and one suspects therefore that enslaved labour was economically critical across the Empire just as at it was at Pompeii. This is not to exclude the role of trade and commerce upon which previous models of Roman economic development have focused. However, in the end, someone must have risen to meet the opportunities provided by novel institutions or expanding markets, and in the Roman world that someone was very often a slave owner. This suggests that we go beyond simply recognizing the place of slavery in the Roman economy, and ask whether this form of labour was in fact causal to economic development. This paper’s procedure has intended to argue that we now possess the analytical tools to pursue this question.

A reorientation of Roman economic history around labour will require us to explore a number of topics anew, and I close by returning to the subject of economic inequality and its measurement. As noted, calculations of inequality take on particular interest in a situation not of generic ‘elite extraction’ but in which a widespread legal institution, slavery, allowed some Pompeiians to exploit others for their own gain. It is interesting in this light to consider the link between an economy’s scale and the IPF, the inequality possibility frontier. By implication of this metric, the greater the size of Pompeii’s economy relative to that portion generated by enslaved people, the greater the capacity for Pompeiian inequality: as the share of slave-generated wealth goes down, the potential for inequality goes up. This deduction may come as something of a surprise in a slave society: can a city less dependent upon unfree labour be considered more unequal? Perhaps the question starts to reveal some unaccounted-for complexity to the ways in which we assess Roman inequality. At Pompeii, one important thing that slavery accomplished was to generate income for others, as owning slaves offered masters the possibility of not working. This would seem to present an unequal distribution of labour as much as of wealth, something we might now seek to measure, while one suspects there is considerable relevance to Roman concepts of otium and leisure, or elite fascinations with watching others work.¹⁰⁹ At the macroeconomic level, some historians evaluating the Roman economy now stress the difference between endogenous economic development and the appropriation of wealth made by the knowledge or labour of others.¹¹⁰ I leave for another time the question of whether the Roman Empire as a whole was more successful at creating wealth or simply better at appropriating it from others. I merely note with encouragement that, while we appear increasingly well positioned to measure the sources and structure of income and wealth in a historical economy such as Pompeii, there remains work to be done in grasping everything that the combination of ancient slavery, prosperity, and inequality implied.

Footnotes

APPENDIX

A MODEL OF INCOME GENERATED BY SLAVEHOLDING AT POMPEII

Minimum total income (K) earned each year by Pompeii’s masters through slave owning can be understood as basic income net the costs of slaveholding (maintenance plus amortized purchase price) multiplied by the number of enslaved people in Pompeii. Reckoning all parameters on an annual basis and all monetary sums in sesterces (HS), let i_f equal basic income, i_s slave income or cost of slave-maintenance, p the slave’s purchase price, and a years of enslavement. Then u is the proportion of slaves in the total population d.

Where a uniform distribution is assumed, I give minimum and maximum values; for PERT distributions, I provide modal values.

Using these parameters, a Monte Carlo simulation of K is run 100,000 times in RStudio. The code is as follows:

# required packages

require(HDInterval)

require(mc2d)

# number of iterations

n <- 100000

# DEFINE INPUTS #

If <- runif(n,900,1799)

Is <- rpert(n,125,226,326)

P <- rpert(n,1900,2750,4050)

A <- runif(n,10,18)

U <- runif(n,.20,.45)

D <- rpert(n,15000,20000,30000)

# CALCULATE TOTAL INCOME STEP ONE #

SlavCost <- Is + (P / A)

SlavPop <- U * D

# CALCULATE TOTAL INCOME STEP TWO #

NetIncPC <- If - SlavCost

# CALCULATE TOTAL INCOME STEP THREE #

K <- NetIncPC * SlavPop

# CALCULATE EXPECTED VALUE #

EVK <- round(mean(K))