How to find a housing shortage… in three misleading charts

Last week the Centre for Policy Studies produced a textbook example of the case for why a shortage of supply is behind our housing woes. Here’s why it’s wrong.

MP Chris Philp has penned a monograph for the Centre for Policy Studies on what’s behind the housing crisis. You guessed it, he concludes it’s a lack of supply. It’s a clearly-argued piece, drawing on many of the same data sources as various recent parliamentary reports and the housing white paper from earlier this year. As such it is a textbook rehearsal of the usual ‘housing supply crisis’ case.

Unfortunately the data used in each step of the argument is a mixture of the partial and the just plain wrong. If we use the right data, which I present below, the picture is reversed and the premise of the housing shortage mantra falls apart.*

First Mr Philp tells the well-worn story of how house building is way down on where it was is in the 1970s. But back then much of the new stock was replacing damaged or inadequate housing. Moreover the house building stats don’t capture all of the additions to the housing stock, since conversions and changes of use also make a significant contribution. The chart below compares Mr Philp’s ‘housing starts’ chart to one for ‘net additions’ based on DCLG’s data. If we’re interested in how quickly we’ve been increasing the supply of housing, net additions is the only relevant data source.

Rather than the doom and gloom picture painted by the new starts chart, net additions in England alone hit 217,000 last year, implying a UK rate of around 260,000 — significantly above the level of the 1970s, when net additions averaged 219,000 per year. So the appropriate headline figures don’t immediately suggest we should panic about historical underperformance.

Sources for lower chart: 1971–2014 from DCLG table 101; 1971–81 shows average annual additions over the decade; 2015 and 2016 based on growth rates for England from DCLG Table 125; March 2017 estimate based on UK net additions as a proportion of English net additions in past decade, applied 2016–17 English net additions in Table 120.

Then, the paper goes on to argue that new supply has lagged growth in household numbers in most regions of the country since 2000, leading to a “large shortage which has build up over nearly 20 years”. The trouble is the data used to back this statement is all wrong.

First, the data used to measure new supply is new build completions (DCLG Table 253), but as above, it should be the more comprehensive measure of net additions. Neal Hudson gives chapter and verse on why the former is a very misleading measure in his blog, but in short not only does it ignore conversions, changes of use and demolitions, as discussed above, but this data set even misses a chunk of the new houses being built each year.

Last year net additions were almost 50% above the new build figures used in the report. If you miss out a third of the new houses it’s easy to find a housing supply crisis. Nor is the choice of data set for these purposes even debatable: DCLG’s guidance clearly says net additional dwellings is “..the only data source to give a comprehensive picture of housing supply..”.

But the household growth figures are also wrong. The report draws on faulty forecasts of household growth from DCLG from 2014, rather than the ONS’s estimates, which come from actual household survey data. DCLG household projections have persistently overshot the actual number for reasons explained here. Indeed the forecast is now fully 800,000 households above the ONS’s view of the true number of households in the UK. (If at this point you’re itching to talk about whether rising housing costs have been increasingly suppressing household formation in recent years — or ‘endogeneity’ — read here and here).

Sources: DCLG Table 401, ONS households

Pairing data that understates supply with forecasts that overstate demand unsurprisingly paints a worrying picture. The top two charts below are the ones from the CPS report for the UK and London, showing a terrifying 40,000 annual shortfall in London recently. The bottom two charts, which I’ve put together, are built using the appropriate data. They show the opposite story: supply has outstripped household formation most of the time since 2000, both in London and the UK as a whole.

Sources for lower charts: ONS regional household estimates (3-year moving average); Change in dwelling stock taken from DCLG tables 101, 109 and 125; 2015–16 UK dwellings change estimates from change in England.

All of this culminates in Figure 3 of the paper, showing the cumulative shortfall in dwellings since 2000. London, it’s claimed, has faced an undersupply of 343,000 units! The UK’s shortfall is put at 700,000 houses. Again, the bottom chart below uses the correct data, and shows that, in fact, London added 74,000 more homes than households, and the UK as a whole 365,000 more homes than households (up to 2016).

Sources for lower chart: ONS households, DCLG tables 125 and 109

On its own terms, then, the argument made here that there has been a shortfall in supply over the past 20 years is simply wrong. With parliamentary inquiries making all the same mistakes using similar data, this type of argument continues to fuel the ubiquitous perception that a shortage of supply is responsible for driving up house prices so rapidly from their late 90s lows. Until we all start looking at the right data, a fruitful debate about the cause of high prices and what to do about them is impossible.

— — — — — — — — — — — — — — — — — — — — — — — — — — — — —

*To be clear, there are more sophisticated supply shortage cases made by people like James Gleeson and Chris Giles, the arguments forwhich I responded to here.

One clap, two clap, three clap, forty?

By clapping more or less, you can signal to us which stories really stand out.