Roman villas cluster in the same corner of Britain as today's richest neighbourhoods. I built the best open list of villas I could and set it against household income, to see whether the villas mark the money or just the map.
Drive west out of London along the M4, or north-west towards the Chilterns, and you pass through prosperous commuter country: the Thames Valley, the Berkshire Downs, and on to the Cotswolds. In Roman Britain the same country was dotted with villas, the farmhouses and country houses of Roman Britain's landed families, with their mosaics, bath suites and painted walls.
It is tempting to see a line running from one to the other: good land, close to a capital, chosen by the wealthy then and now. In Is Domesday Destiny? I asked whether the England of 1086 still shapes England today. Here I go back further, to Roman Britain, and ask a sharper question. Are the neighbourhoods next to Roman villas richer than others, and if so, is that about the villas, or just about lowland south-east England?
The short answer: the villa belt and the commuter belt do overlap, and a neighbourhood with a villa nearby is, on the face of it, richer. But nearly all of that gap is geography. Compare like with like, region with region and distance from London with distance from London, and the villa premium shrinks to almost nothing.
Each red dot below is a Roman villa from my combined list. The small teal dots are the 7,264 neighbourhoods (middle layer super output areas, or MSOAs) of England and Wales, shaded by mean household income in the financial year ending 2023: the darker, the richer. You can add the Roman towns and the Roman road network as well, search for a villa, or tap anything for details.
The pattern jumps out. Villas are thickest in a broad arc from Kent through Hampshire, Wiltshire and Somerset to the Cotswolds and the East Midlands. About 74% of them fall in the South East, East, South West and London regions. North and west of a line from the Severn to the Humber they almost vanish, except in south Wales and east Yorkshire. Today's money has its own shape: a dark core in and around London, fading with distance.
I call a neighbourhood a "villa MSOA" when a known villa lies within 2 km of its population-weighted centre. That is 395 of the 7,264 MSOAs, or 5.4%.
Their median household income is £57,822, against £52,315 for the rest: a gap of £5,507, or 11%. In the richest tenth of neighbourhoods, 5.5% have a villa nearby; in the poorest tenth, only 1.1%.
So far, so commuter belt. But a villa MSOA is not a random neighbourhood. It is far more likely to be in the south, in the countryside, near a Roman road, and within reach of London. All of those things also go with higher incomes today. The question is what is left once they are taken into account.
Distance from London is the obvious suspect. The chart below cuts England and Wales into rings around Charing Cross. Income falls steadily with distance: the median neighbourhood within 25 km has a household income of £72,872, and beyond 300 km it is £44,110. Villas do not follow the same curve. They are scarce in the inner ring, where London has built over most of what there was, peak at 10.5 per 1,000 km² between 75 and 100 km out, and then thin away to the north and west.
The villa belt, in other words, is the outer edge of today's commuter belt and the farmland beyond it: the downs and vales of Hampshire, Wiltshire, Oxfordshire and Gloucestershire. Both belts lean towards the lowland south-east; they are not the same belt.
To separate villas from geography I fitted a series of regressions of (log) household income on the villa dummy, adding controls one block at a time. With no controls, villa MSOAs have incomes about 8.6% higher. Comparing only within the same region cuts that to 2.2%. Adding distance to London takes it to 0.5%.
The further controls barely move it: distance to the nearest Roman road and Roman town, height above sea level, how hilly the ground is, and distance to the coast leave 0.4%; allowing for how urban the neighbourhood is leaves 0.4% (95% interval −2.0% to +2.8%). Comparing only neighbourhoods within the same local authority gives 0.3%. None of these is distinguishable from zero.
A Roman villa nearby tells you which part of the country you are in. It does not tell you which street is rich.
The result holds when I change the definitions. Using a 5 km radius, only the villas listed on Wikipedia, income after housing costs, leaving London out, or looking only at the South East, East and South West, the fully controlled premium stays between −0.4% and +1.5%.
| Version | No controls | All controls | 95% interval |
|---|
Roman forts make a useful mirror image. They sit mostly in the north and in Wales, so the 209 MSOAs with a Pleiades fort within 2 km look poorer at first: −14.5%. With the same controls that gap also all but disappears, to −0.2%. Roman sites of either kind mostly tell you where in Britain you are.
Here is a blunter test. I picked up every villa and dropped it 5 to 15 km away in a random direction (always on land), then measured the "villa premium" again for these fake villas. I did this 200 times.
If the real sites marked something special about their exact spot, the fake ones should do worse. They do not. The fake villas show a median raw premium of 11.6%, and 94% of runs beat the real villas. With all controls in place, the fake villas average 1.2%, and 80% of runs match or beat the real ones' 0.4%.
The real premium is just what you get from being in roughly the right part of England. Any spot within 15 km of a villa would do as well.
Within regions the gap goes both ways. In the South West, villa neighbourhoods have a median income 10% above the rest. That is mostly about where in the region they are: only 8 of its 170 villas lie in Devon and Cornwall, where incomes are lower, and most sit in the east of the region, around Bath, Bristol, the Cotswolds and Somerset.
In the South East the sign flips: villa MSOAs are 5% poorer than the rest. They lie a little further from London (a median of 75 km from Charing Cross, against 68 km for the rest), and they include less well-off places along the Hampshire and Sussex coast such as Havant, while the region's richest neighbourhoods are the commuter suburbs closer in.
| Region | Villas | Villa MSOAs | Median with villa | Median without | Gap |
|---|
Zooming out to the 338 local authorities of England and Wales, there is a modest positive link between villas per 100 km² and income: a rank correlation of 0.27. It survives allowing for distance to London and population density (0.29) and, more weakly, region as well (0.21). So at the scale of whole districts, villa country does lean a little richer, even after the obvious controls.
That fits the neighbourhood results rather than contradicting them. Districts are big enough to capture broad differences in land and history; the neighbourhood test shows that within a district, being next to the villa itself buys nothing. And 55% of authorities have no recorded villa at all, including most of London's richest boroughs: Richmond upon Thames, Kingston, Wandsworth and Kensington and Chelsea all score zero.
Among authorities with at least three villas, the densest are below, with their income rank out of 338 (1 is richest). Their median income rank is 101: middling, not top.
| Local authority | Villas | Per 100 km² | Mean income | Income rank |
|---|
The great show villas of Roman Britain mostly sit in comfortable, but not exceptional, neighbourhoods. The typical one is richer than 69% of MSOAs. The exception at the top is Crofton in Orpington, in the London Borough of Bromley; the exception at the bottom is Brading on the Isle of Wight.
I could not find a single open, downloadable list of Roman villas for England and Wales, so I built one from four sources and joined them:
Records from different sources were treated as the same villa when they were within 600 m (1 km where Wikipedia gives only a 1 km grid square), or within 2.5 km when one name contains the other, closest pairs first, and never joining two villas from the same source. That leaves 617 villas: 592 in England and 25 in Wales. 332 are confirmed by two or more sources and 57 by all four.
These are excavated or well recorded villas that have made it into open sources, not all the villas that existed or even all that archaeologists know about. Local historic environment records hold many more possible villa sites, and aerial photography and developer-funded excavation keep adding to them. Recording is uneven: it is richer in the south, in counties with strong local archaeology societies, and where gravel quarrying and house building have prompted digs. That bias runs towards exactly the prosperous, growing south-east this study is about, which is one more reason to treat any raw "villa premium" with suspicion.