What would 300,000 houses per year do to prices?

Last week the Ministry for Housing, Communities and Local Government (MHCLG) published an important document entitled ‘Analysis of the determinants of house price changes’. It gives details of the department’s house price model, including figures showing how sensitive prices are to the supply of additional housing.

Those figures have important implications for the likely effectiveness of the government’s proposed solution to the housing crisis: to build 300,000 houses per year. This blog sketches out the stark quantitative implications of those results.

Reversing the house price boom?

Average house prices in the UK have risen by about 150% in real terms since 1996. This is what most people are referring to when they talk about ‘the housing crisis’. The government has said that we need to build around 300,000 houses per year to tackle it. Thanks to last week’s publication, we now know what scale of impact the government expects that pledge to have.

The key lines from the document are:

How much do these figures suggest an extra 300,000 houses per year will improve the affordability of housing over the next 20 years? Well the housing stock would grow by 6 million, or about 21% of the stock today, pushing prices down. Meanwhile household growth, projected by MHCLG’s at around 240,000 per year, would be about 18%, pushing prices up.*

The net effect, according to the MHCLG analysis, is that house prices would be would be 6% lower in real terms by 2038, all else equal.

In other words the government’s model suggests that even achieving record levels of net housing additions, sustained over two decades, will do very little to reverse the 150% house price explosion of the past 20 years. That’s unlikely to impress many priced-out millennials.

Could building more have stopped the house price boom?

So much for the future, what about the past? Does the government’s analysis suggest significantly higher rates of building could have stopped the house price boom?

Let’s imagine we’d built 300,000 houses per year since 1996, rather than the 200,000 we actually managed. By last year we’d have had 2.1 million more houses, adding roughly 9% to the 1996 stock. Assuming unchanged household growth, and applying the MHCLG sensitivities in the way they do in their report, suggests that by 2017 prices would have been about £15,000 lower. That’s about 7% off today’s average house price.

So instead of rising by 151% in real terms, between 1996 and 2017 prices would have increased by about 134%. In 2017 money, then, average house prices would have tracked the red line in the chart below, rather than the blue.

Sources: MHCLG, Land Registry, ONS. House prices adjusted to 2017 prices by CPI.

So the overwhelming majority of the real-terms price increase of the past 20 years would have happened even in this building boom scenario. Little more than a sticking plaster for the ‘broken’ housing market. To have a real impact on house prices, the government’s model tells us that we need to be looking at other policy levers relating to the financial drivers of house price growth.

But is the model any good?

Some commentators have cast doubt on the reliability of the model, suggesting the results shouldn’t be entirely trusted. It’s certainly true that the there are some odd things in there (the results on interest rates are particularly odd — mainly because of how the model has been specified, which looks wrong to me). And relying on any one model is always a risk for policymakers.

But when it comes to the impact of supply or household numbers on house prices, there is a large degree of consensus between economists. The Ministry’s results are in line with every other similar study for the UK. There’s this one from the OBR. This one by Geoff Meen at Reading University. This one from the OECD. And this one by Oxford Economics for the Redfern Review in 2016. All of them suggest that 1% more houses would cut prices by somewhere between 1.1% and 2.1% (hence MHCLG’s use of 2% is in line with these).

Some might argue that all of these models are wrong. But they’d have to be hideously wrong before 300,000 houses per year started to look like any kind of meaningful solution to unaffordable houses.

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

*I think MHCLG’s household projections look very high, for reasons I describe here, so the impact on prices could be bigger than quantified here. Nevertheless the price impact still wouldn’t be game-changing.