Receive our newsletter - data-led analysis, original reporting and insights
Government / Local politics

The backwards-looking world of forecasting: why official housing forecasts may be too low

New household forecasts are one of those dry-sounding statistical releases that hide profound and challenging issues for all of us.

Last week’s projections from the Department for Communities & Local Government will affect both where we live, and how. And countless local planning battles over these numbers, up and down the country, will determine whether we have any chance of solving the housing crisis.

The “2012-based Sub-National Housing Projections” for England (SNHP) estimate how many households there will be in each district up to 2037. They are the starting point for councils working out how many homes are needed, which gets translated into housebuilding targets in Local Plans, which in turn decides which sites get allocated for new houses. If the figures are too low, then things are going to get pretty cramped in England’s homes.

Unfortunately, there are good reasons to think they are too low. Firstly, any demographic forecast needs to guess – and guessing is really all that we can do – what future levels of net immigration will be. The forecasts are based on assumptions that are a couple of years old, and are already now looking far too low. In fact, net immigration has been around 50 per cent higher than the model assumed for the last couple of years.

Secondly, in the backwards-looking world of forecasting, the model is heavily influenced by recent housing market turbulence, the recession and the growing backlog in new housing. It all comes down to household size.

The size of households in England has been falling for over a hundred years. That’s for a range of reasons including smaller families, higher divorce rates and longer-living pensioners. For a given number of people, smaller households mean more households, so this trend is important to the projections.

But in the 2011 Census something very unexpected happened – household size stopped falling. Now that could be because of some permanent social change, and that effectively is what the SNPP assumes – that the falling trend will resume, but never catch up with where it had been going.

 

More likely, however, is that in the middle of a recession, when the housing market had ground to a halt, people were simply unable to form new households, even though they wanted to. And after years of building too few homes, young people in “generation rent” are forced to flat share or live with their parents. If we were to build more homes then this blip might quickly reverse.

The projections “bake-in” this forced overcrowding, and project it forward, reducing the number of homes we need in 20 years times by the best part of three quarters of a million.

And this projecting forward of current trends also affects where the growth is expected. This map shows how the forecasts have changed from last time. Red areas are now expected to have faster growth in the number of households than previously thought; blue areas are now expected to have slower growth.

Click to expand.

In general there has been a marked shift of growth towards London.

In areas where very little new housing has been allowed, such as some Green Belt districts, the population will have grown very little as a result, and so a trend-based forecast will extrapolate continued low growth into the future. But that’s not a reflection of need, just an extrapolation of past under-supply.

In other words, demand for homes in some areas may be much larger than the forecasts suggest: past growth may be a poor guide to future need.

To be fair, planning guidance tells councils to take account of exactly this sort of problem, and to allow for overcrowding and forced sharing. But without guidance on how to do so, in practice the issue is being largely ignored in the plan-making process.

The fear is that these new, lower, projections will be carried through into local plans – particularly by those councils eager to find excuses to reduce politically unpopular development. That means that less land will be allocated for new homes for our growing population – and the housing crisis will continue to worsen.

Here’s one last particularly sobering map. Even if the lower forecasts are taken at face value we would still need to increase housebuilding by around 70 per cent from current levels to meet this forecast need. The map below has one dot for every 1,000 additional homes needed by 2037, according to the SNHP.

Click to expand.

Each of those dots, therefore, represents years of planning applications, and often long fights, re-submissions, appeals and even judicial reviews. What are the chances that we’re going to be able to build all those dots, and have enough space for us all to live in future?

Barney Stringer is a director of regeneration consultancy Quod, who writes about cities, economics and infrastructure. This article was originally posted on his blog here.

All images courtesy of Barney Stringer/Quod.
This article is from the CityMetric archive: some formatting and images may not be present.