Why we need a Census 2026

The release of workplace data from the Census shows how little guide it provides for the next decade of policy decisions.

Blog post published on 20 December 2022 by Paul Swinney

Earlier this month saw the release of the first set of Census 2021 data to focus on work. Unsurprisingly, given that this census was done during a pandemic, the ONS has had to put a disclaimer on its outputs to ‘take care when using these data for planning and policy purposes.’ The problem is that this is the only detailed dataset we have on where people work and how they get there. If this data is used to guide investment decisions over the next decade, as has been the case in the past, then this would skew investment away from where it is needed.

Perhaps the data on this release with the largest impact on public sector investment decisions is the one for working at home. As Alastair Rae nicely illustrates below, there were very few places in 2011 where more than 25 percent of workers worked from home. Fast forward 10 years and this had totally reversed. Very few local authorities had fewer than 25 per cent of people working at home. And 13 authorities, all in or around London, had more than half of their working residents home working.

A little data nugget from this week’s (very unusual but still useful and interesting) travel to work data, using LSOAs

-map on left is areas where 25% or more worked from home in 2011

-map on right is the 2021 version

let’s wait 10 years and see what happened next pic.twitter.com/BCNM2N2oc6

— Alasdair Rae (@undertheraedar) December 10, 2022

What this map will look like in 2031 is contested. Centre for Cities set out last year why we should be cautious about extrapolating from a time of crisis and assume that working patterns in 2021 are any indication as to what they will be like in years to come. And the data since then has broadly backed this up. While TfL data for ‘City’ stations show that ridership into the office-heavy parts of the central London is around 80 per cent of pre-covid levels, it is very much higher than it was in March 2021 when the census was done.

This makes the 2021 data interesting for a snapshot in time, but pretty useless for insights into how we live our lives. And it leaves us blinded struggling to understand what commuting looks like. We’re not far off returning to pre-covid patterns. But we don’t even have a good idea of what those patterns looked like. For example, the number of jobs in Inner London increased by over 600,000 between 2011 and 2019. TfL journeys increased by 20 per cent. 

The availability of TfL data in London makes the irregularity of the Census data on our knowledge of commuting less of a problem for the capital (noting that the TfL data still only gives a partial insight). But no other city has this level of granularity on its patterns of commuting. This means that for a place like Manchester, by 2030 the last detailed read out of commuting patterns will be almost 20 years old. Given that workers are the key input into the knowledge economy, not having a good idea about how these workers supply their hours (i.e. from home, via private transport or via public transport) leaves investment decisions in transport blind to the realities of what is going on. This has implications for the quality of their public transport (which in many large cities went into Covid in a position of relative weakness), the performance of their economies, the impact on local air pollution and the volume of carbon emissions in their areas.

There has been much debate about whether we will even have another census in 2031, with the idea being that we will have good enough administrative data to provide the insights a census traditionally would do. 

Ultimately if we get to a position where the data is good enough to do this, and do it more regularly than every 10 years, then this would be much welcome.

There are three problems though. The first is whether administrative data will be able to fully replace a Census given the granularity census data provides. The second is that the last census it will be benchmarked against, which is important to check that it is giving an accurate picture, is skewed. And the third is that, whether we get a snapshot from administrative data or another census in 2031, in the meantime it leaves policy working from data that doesn’t reflect the reality of the working day throughout the 2020s.

For this reason we should have an additional census in 2026. This is not without precedent – it’s what happened in 1966 when the Government judged that there was a need for an additional count because of ‘‘rapid change and development’. What is for sure is that the amount of change between March 2021 and December 2022 is much greater than between 1961 and 1966.