House of the Menander

Piketty in Pompeii? On measuring inequality in a Roman city

This is kind of a spin-off result from the Building Tabernae project (as it is based on the general database), but I thought it would be relevant to blog about it here anyway. Can we measure socioeconomic inequality in a Roman city? For sure, it is possible to see it: when you walk through the streets of an excavated Roman town, it is hard not to notice how houses differ in size and in wealth, and how traditional Roman-style atrium-houses were just the upper half of a domestic spectrum that includes several types of habitable units on a very small scale, such as tabernae and rental apartments. The question is whether we can translate this visible inequality into something that can be measured, and whether the measurements we can then take actually may mean something, historically. The answer, in both cases, I think, should be a qualified yes. It is possible to reconstruct Gini-coëfficients for Pompeii, and it is possible to give them a historical meaning. The latter, though, is a bit more complex than the former, and will require more work.

First, the results. It is possible to rank Pompeii’s habitable units according to the number of rooms, and according to the amount of space they occupy. If you follow the latter approach, you will get a Gini-index of 60.89; if you follow the former approach, you will get a Gini-index of 51.25. It makes sense to point out that both are indexes of access to property, and not indexes of actual income, but if reliable they do give an insight in to how wealth was divided in Pompeii – between households.


The difference between the two essentially means that people on smaller plots built houses with smaller rooms, and where thus able to squeeze relatively more domestic functions in less domestic space. They particularly shrunk the sizes of their gardens, their atrium, and their peristyles. Still, they ended up with less rooms, and less domestic comfort than most houses on larger plots: there is a clear correlation between plot size and room number.

The question is how reliable the Gini is. On the one hand, they may seem to overestimate inequality: people in smaller houses, and in tabernae and apartments made great use of upper floors, which are haphazardly preserved in the archaeological record, and thus cannot be included in any reliable calculation. Smaller households had a bit more living space than the present graph suggests, though this should not be overestimated – larger houses had upper floors as well, and these have neither been included in the calculation.

Indeed, if we want to see these indices as a starting point for discussing wealth inequality, it can be argued that both coëfficients underestimate inequality to a considerable scale. First of all, as many people may know, the tabernae and apartments that constitute the lower half of the spectrum were mostly owned by households in the richer parts of the spectrum: they were physically part of atrium houses. Second, besides owning their own house and some rental accommodation, the urban elite also is more likely than the urban poor to have owned arable land – both within the city, and outside it.  Especially the top 5% of Pompeii’s household is likely to have had more wealth than they could express in the size and embellishment of their houses – there is likely to have been a limit to the size of houses even in Roman cities.

I would suspect that, after all, the lower Gini score of 51.25 must be seen as an absolute lower limit of wealth inequality in Pompeii: in all probability, wealth inequality in real terms was higher than this. The next step is considerably more difficult: what does it mean if a Gini index of wealth inequality between households in Pompeii was probably higher than 51.25? For the Roman world, there is not yet a lot of data that helps us to contextualize this figure, but there is some. Based on papyrological evidence, scholars have reconstructed comparable Gini indices for two cities in Roman Egypt, Karanis and Philadelphia. Both are of the same order of magnitude – Karanis is a bit lower (44), Philadelphia approximately equal (52; see Scheidel and Friesen 2009: 87) – yet these are contexts so different that the comparison becomes virtually meaningless. Modern comparisons seem equally fruitless – the list of wealth Gini’s by country produced by Davies et al. includes no value lower than 54 – and refers to countries, not to individual cities.  Pompeii’s Gini would not really stand out as fundamentally different here, though it may appear a bit lower than that of most western countries nowadays (which is no surprise – see Scheidel and Friesen 2009: 87).

Indeed, our Pompeii estimate can only become historically meaningful in comparison with other estimates from the same historical context – that is: from Roman Italy in the first two centuries of our era. To judge from what I have seen in the field, I expect Pompeii will have a higher wealth Gini than most cities in the Apennines, or indeed most cities in Italy, but the only meaningful comparison that can be made is with Ostia. This  indeed is likely to become possible in the course of this project, but the comparison will be very complex as the property landscape in Ostia differs fundamentally from Pompeii.

So the graph above looks nice, and does tell some story, but at this point it does not yet mean an awful lot. It may, however, become considerably more meaningful in the near future.