Here’s why counting houses is hard

We literally have no idea how many houses could be in there. Image: Getty.

We may be getting better at building more houses but unfortunately we’re not very good at counting them.

In August, the housing minister was citing the latest DCLG New Build statistics as proof that the country is building again. Completions across England had apparently reached 153,000 in the year to June 2017“the highest level since 2008”. On this basis, the minister may be pleasantly surprised and slightly confused when he reads the DCLG’s Net Supply release in November and finds out that housebuilding completions had already reached 155,000 in 2014-15, and are actually much higher.

The housing minister can perhaps be forgiven some excitement over the first release of housebuilding statistics during his tenure. Based on the average tenure of previous housing ministers, he’s probably only got another three or four to look forward to. However, despite some allowance for over-excitement, it is irresponsible for the housing minister to be quoting the New Build statistics as absolute measures of housebuilding as they under-count the number of new homes actually being built.

It is particularly irresponsible because DCLG are well aware that there are issues with the New Build statistics. In the introduction to their New Build statistical release they suggest the New Build figures should only be “regarded as a leading indicator of overall housing supply”, and instead the Net Supply release “is the primary and most comprehensive measure of housing supply”.

The scale of the under-count is apparent when comparing the New Build data to the more comprehensive Net Supply release. While the Net Supply release includes conversions, changes of use, and demolitions to calculate the net change in dwellings, it also includes a more comprehensive measure of housebuilding.

The latest available Net Supply data for 2015-16 recorded 164,000 housebuilding completions across England compared to only 140,000 completions recorded in the New Build data. That suggests the New Build release is currently missing around 15 per cent of the housebuilding market.

Beyond the widespread confusion created by the publication of different housebuilding numbers, this issue has important consequences for policy makers. Our failure to accurately measure housebuilding and our limited understanding of who is doing the building make it very difficult to accurately assess the success or failure of existing policies and identify new ones that could increase new supply.

The exact reasons for the under-count are not confirmed but it appears to be linked to the falling market share of the largest provider of warranties on new homes. The National Home Building Council (NHBC) provides a substantial share of the data used to create the New Build statistics, and it’s been widely assumed that they have a market share of around 80 per cent. Based on an assumed market share, the NHBC data is grossed up to provide a measure for the whole market alongside other sources of building control inspection data.

However, recent years have seen a broader range of groups delivering new homes. Volume housebuilders still deliver the majority of new homes but there has been an increase in activity by SME housebuilders, high-density luxury developers, build-to-rent investors, and housing associations. For some of the firms and organisations in these groups, an NHBC warranty may be too expensive or not attractive compared to the alternatives. NHBC’s market share has probably fallen over this period.

A fall in NHBC’s market share is apparently confirmed by the request for a review of its market undertakings from the Competition & Markets Authority (CMA). Although most of the market share data published by the NHBC and the CMA in the review is confidential, there is an interesting finding in the CMA’s provisional decision (paragraph 4.32). Using new home data from nine warranty providers including NHBC, the CMA estimated the NHBC market share at around 70 per cent.

If, instead of grossing up the NHBC data by 80 per cent market share, we use 70 per cent then we would expect the DCLG New Supply data to be around 14 per cent higher (0.8/0.7). That difference would account for nearly all of the shortfall in the New Build completions when compared to the Net Supply housebuilding data. While there may be other factors causing the under-count, it would appear that this market share issue is the most significant factor.


It would be great if we had an accurate and regularly updated measure of housebuilding, but it turns out that counting houses is actually quite difficult. The Net Supply data is far from perfect, and it’s only released once a year with a substantial delay but it’s the best we currently have.

Meanwhile, in Ireland they’ve had the opposite problem, with an over-count of new homes. Official completions data uses electricity connections – but it turned out that the actual number of new build completions between 2011 and 2015 was 42 per cent lower than the official figures due to a large number of re-connections.

Until we see a substantial re-working of the DCLG’s New Build statistics, it appears the best option is to assess the full range of available indicators that cover both housebuilding and total supply. However, perhaps the biggest frustration is that DCLG are aware of the issues with the New Build statistics yet we still see quarterly political point scoring based on these flawed data. Given the complexities of the housing market, it is only once we move past this short-term politicking that we have any hope of solving the crisis.

Neal Hudson is an independent housing analyst, who tweets as @resi_analyst. This article originally appeared on his blog.

 
 
 
 

Which British cities have the bestest ultrafast broadband?

Oooh, fibre. Image: Getty.

The latest instalment of our series, in which we use the Centre for Cities’ data tools to crunch some of the numbers on Britain’s cities. 

Between the dark web, Breitbard News and Donald Trump's Twitter feed, it's abundantly clear that terrible things often happen on the internet. But good things happen here, too - like funny videos and kitten pictures and, though we say so ourselves, CityMetric. 

Anyway. The government clearly believes the internet is on balance a good thing, so it's investing more in improving Britain's broadband coverage. But which cities need the most work?

Luckily, those ultrafast cats at the Centre for Cities are on hand with a map of Britain's ultrafast broadband coverage, as it stood at the end of 2016. It shows the percentage of premises which have access to download speeds of 100Mbps or more. Dark green means loas, pale yellow means hardly any. Here's the map:

Some observations...

This doesn't quite fit the pattern we normally get with these exercises in which the south of England and a few other rich cities (Edinburgh, Aberdeen, York) look a lot healthier than the cities of the Midlands, South Wales and the North.

There are elements of that, sure: there are definitely more southern cities with good coverage, and more northern onse without it. But there are notable exceptions to the pattern, too. Those cities with very good coverage include Middlesbrough (88.0 per cent) and Dundee (89.4 per cent), not normally to be found near the top of anyone's rankings. 

Meanwhile, Milton Keynes - a positive boom town, on most measures - lingers right near the bottom of the chart, with just 12.9 per cent coverage. The only city with worse coverage is another city that normally ranks as rich and succesful: the Socttish oil capital Aberdeen, where coverage is just 0.13 per cent, a figure so low it rings alarm bells about the data. 

Here's a (slightly cramped) chart of the same data. 

Click to expand.

If you can spot a patten, you're a better nerd than I.

One thought I had was that perhaps there might be some correlation with population: perhaps bigger cities, being bigger markets, find it easier to get the requisite infrastructure built.

I removed London, Manchester and Birmingham from the data, purely because those three - especially the capital - are so much bgiger than the other cities that they make the graph almost unreadable. That don't, here's the result.

So, there goes that theory.

In all honesty, I'm not sure what could explain this disparity: why Sheffield and Southand should have half the broadband coverage of Middlesbrough or Brighton. But I suspect it's a tempory measure. 

All this talk of ultranfast broadband (100Mbps+), after all, superseded that of mere superfast broadband (just 24Mbps+). The figures in this dataset are 10 months old. It's possible that many of the left behind cities have caught up by now. But it's almost certain we'll be hearing about the need for, say, Hyperfast broadband before next year is out.

Jonn Elledge is the editor of CityMetric. He is on Twitter as @jonnelledge and also has a Facebook page now for some reason. 

Want more of this stuff? Follow CityMetric on Twitter or Facebook