Lies, damned lies and rent statistics

Let’s get the facts straight on what’s happening to rents

The use and abuse of data governs the housing debate and the entire policy response. The housing supply numbers commonly used and, until recently, the housing need numbers bandied about, have long been wrong or misleading. Given the importance of rent — the ‘price’ that tells us whether demand for housing services is outstripping the supply — using the right measure of that is particularly vital.

I suspect most followers of the housing debate in the press would agree with the statement: ‘the cost of renting a house has risen faster than general inflation and average incomes in recent years’. This is a common refrain among commentators, think tanks and campaigners. But it’s incorrect. The data being used to make the case isn’t up to the job.

In the past couple of months, separate pieces by Shelter and the IFS have fallen into this trap. This blog pinpoints the problem and shows how the conclusions are far more positive if we use the right data. It’s time we got the story straight about what’s happening to rents.

Panic!

Back in August, Shelter produced a panic-inducing analysis under the title ‘Rentquake’, showing vast swathes of a map of England coloured red to indicate ‘surging’ rents. The detail shows numerous local authorities with rents apparently rising by over 30% in just six years between 2011 and 2017, with vertiginous growth in London.

Last week the IFS released a housing paper — part of its Green Budget — showing that rents have risen by almost 38% in inflation-adjusted terms since 1996, while the incomes of 25-to-34 year olds have only risen by around 19%. Surely at this point we can all conclude that rents are rising much faster than general inflation and incomes?

Unfortunately – for the supply crisis story, at least – not. In both cases the reports use data sources that can’t support the conclusions they’re being asked to bear. The Shelter analysis uses Valuation Office Agency data on almost 500,000 rented properties, entered into the lettings information database. The IFS uses the Family Resources Survey, a national household survey that captures what rents people are paying.

Compositional change changes everything

Given the source data, both the Shelter and IFS analyses are effectively based on raw rent data that tracks simple averages of rent over time. When looking at affordability, what we really want to know is something slightly different: what has happened to the rent I would have to pay on a given type of house? Raw rent data is confounded by the changing composition of the private rented sector. And those changes have been huge. Since 1996 the private rented sector has grown by about 130%, with many more families now renting. This really matters.

Imagine I’ve rented an old terraced house since 2011, in a local authority that back then had only two rented houses in it (the top half of the diagram below). My rent, and that on the other house, has grown at an acceptable 5% over the period — below the rate of inflation.

Now imagine developers completed a luxury house in my area in 2017 and it was rented out (House 3 in the bottom half), expanding the PRS to three houses. The raw average rent in the area will jump, even though the rents on my house and the other pre-existing rented house have only changed gradually. In the example below, then, raw rents rise by a massive 27% because of the changed composition of the sector. But in terms of the affordability of rented housing, it’s only the 5% figure that matters: it makes no difference to my cost of living that a fancy new house has come onto the market down the road.

By analogy, if my Sainsbury’s Local suddenly decided to start selling gold bars (real ones that is, not the popular 90s lunchbox snack), the raw average price of the goods in their shop — the equivalent of using VOA or FRS data — would rise, but it would have no effect on the cost of my usual weekly shop. Such an approach tells us nothing about how the cost of living is changing.

Please use the ONS rent index!

None of this should be remotely controversial. Indeed, it won’t be news to anyone familiar with the construction of price indices. In fact, the VOA data even warns users directly about these issues. It says of its data “users are advised not to [use these data to] infer trends in the rental market over time.” Yet this is precisely how they’re used in the Shelter study. The same VOA publication directs users to a more appropriate data set: the ONS’s Index of Private Housing Rental Prices (IPHRP), which is based on the same data, but is explicitly designed to avoid the compositional flaws.

“IPHRP measures the changes in prices tenants face to rent residential property from private landlords. It is produced using different methodology which allows prices to be compared year-on-year”

The ONS has also done some analysis to understand how much the composition effect matters. Decomposing the growth in the VOA rent data over a five-year period, they conclude that for the rented sector in England “compositional effects are estimated to account for around 75% of the jumps in the sample average between 2010 and 2015”. They go on to explain “This is because more rental properties in more affluent areas are now included in the sample.”

Calling time on the scare stories

Unsurprisingly then, IPHRP tells a very different story to these other data sets. As I showed here, the rent index has lagged average household income growth in every region of England since the series started. And if we compare it to the FRS rents produced by the IFS, there’s a big divergence after 2005: the latter suggests real rent growth of about 7.4% by 2016, while IPHRP shows a fall of 3.5% by 2016 — the latest data shows them down over 6% on 2005 in real terms.

Unfortunately IPHRP only goes back to 2005. However, combining it with the prototype index for the UK prior to 2005— albeit based on a much smaller sample —suggests that real like-for-like rents have been pretty benign since 1996, and comfortably below average household income growth.

As ever, none of this is to say that all is fine and dandy in the private rented sector — far from it — or that affordability hasn’t deteriorated for young people, whose wages have grown slowest of all in recent years. But whenever you see a scary housing piece talking about ‘surging rents’, look to see what data it’s based on. Whatever policies we might want to see enacted in housing, it does nobody any favours to compare apples and oranges.