Shifting Population – Part II

My previous post looked at how England’s population is shifting. This one compares how homes and jobs have shifted.

The map below shows the centre-of-gravity of homes, people and jobs in England, in 2011 and 2021. All are moving southwards, but not equally.

People go where the jobs are. Many home moves are just a local change in accommodation, but when people move to a completely new area it is very often because of a new job. Relatively stronger growth in London has dragged the centre-of-gravity of employment in England several kilometres southeast. And people have followed.

However, the population is still lagging the jobs. This may be partly because they are held back by lack of housing near jobs, leading to longer commutes overall. The centre of where England’s homes are has barely moved – physical buildings take longer to change than people, particularly when we’re not building enough homes where people want them.

The chart below takes a slightly longer look, over the 20 years since the 2001 Census. Each square represents one kilometre.

In 2001 England’s population and housing was almost exactly in balance, centred just 50 metres apart, east of Clifton-upon-Dunsmore in Rugby. But as employment attracted people south-eastwards, housing has failed to keep up. The result is likely to be more overcrowding in successful areas, as well as a constraint on productivity.

---

Note on methodology and caveats

Firstly to say again that this owes a great debt to the ideas in James Gleeson’s post of 2016, of which this is simply an update.

As with my first post on population shifts I have weighted centroids of local areas, in this case by population, dwellings or jobs. I have used a mixture of 2011 and 2021 Mid Layer Super Output Areas and 2001 Census Area Statistics Wards, depending on the dataset.

For population I have taken Census usual resident population.

For homes I have used Census accommodation data for occupied homes, so this excludes vacant properties which is an important caveat – lower rates of vacancy in the south east may mean housing has actually shifted even slower than shown here.

Finally for employment I have used workplace based jobs from Census 2001 and 2011, but this is not yet available for Census 2021 so I have instead used the Business Register and Employment Survey data. Some caution is needed in interpreting this as it may not be directly comparable to the Census, however it is the best currently available source.

Shifting Population

Where are we going?

For the last century England’s population has gradually shifted southeast, but what does the latest Census tell us about where the population’s centre-of-gravity is heading?

This post is inspired by James Gleeson’s excellent 2016 post Tracking England’s Shifting Centre-of-Gravity Over Time and my curiosity to update it with the new Census.

His historical analysis showed the centre of England’s population started the 19th Century just east of Coventry, and headed steadily north for over a hundred years, up the M69 in the direction of Leicester. Around the First World Ward, this reversed, with the centre of population taking a turn southeast. At the time of the 2011 Census it was last seen racing past Rugby on the A5 (Watling Street) heading towards London.

So what happened since 2011? We have two sources, the annual Mid Year Estimates (MYEs) produced by the Office for National Statistics, and now the new 2021 Census. The map below is what these show.

So from 2011 to 2021 the centre of population continued to move southeast and has now entered West Northamptonshire, close to the Houlton housing development in Rugby. Appropriately enough, it is right in the middle of DIRFT, the country’s largest railfreight logistics interchange, just off Junction 18 of the M1. Specifically, it’s in the car park of the Sainsbury’s Daventry Distribution Centre. This area is called the logistics Golden Triangle for good reason, within 4 hours drive of 90% of Britain’s population.

The mid-year estimates for the intervening years show a slowing trend, and then an abrupt break with Census 2021 putting it noticeably west and no further south than the 2020 estimates. Of course it is normal for the Census to differ from the estimates (that’s the whole point of doing an actual Census), and in the next few months the ONS will publish rebased mid-year estimates from 2012-2020 to make them consistent.

It will be interesting to see the degree to which the ONS rebasing shows this shift in direction to be a gradual trend over the last ten years, or a sudden shift because of Covid. And if Covid has affected it, how much it is temporary or permanent? The 2021 Census in England took place during partial lockdown, with everyone asked to work remotely if they could. Many people may not have been living at their normal address on Census day.

To the extent this shows a real effect, rather than a Covid Census artifact, what might be the cause? James’s earlier post showed elegantly how population was following jobs, with housing trailing some way behind. At some point this lag – a failure to build homes where jobs are growing fastest – will inevitably constrain people’s ability to follow those jobs.

If so, you’d expect to see the results of that in high house prices, lower productivity growth, and staff shortages.

---

Methodology Note – I’ve used Mid Layer Super Output Areas (MSOA) – a census geography roughly similar in size to an electoral ward. I’ve taken the geographical centroid of each MSOA, and weighted them by population. For 2011 to 2020 I’ve used the 2011 MSOAs, and for the 2021 Census I’ve used the new 2021 MSOAs. Note this analysis is England-only because GB or UK data not available yet – Scotland took the sensible decision to delay their Census a year, so it wasn’t so badly distorted by Covid.