Few indices are watched more closely than those that track house prices. On one level that’s rational. Two thirds of households own their own home, most of them with a mortgage. Changes in house prices therefore have a huge impact on most people’s personal balance sheet – almost certainly a much greater impact than stock-market movements. But on another level, watching what is commonly thought of as ‘house prices’ is pretty irrational, because the indices are a fairly poor indicator of the value of your home.
There are many reasons why measures of house prices are flawed, but most are rooted in the truism that houses are unique. Their value is therefore much more subjective than the value of so-called ‘fungible’ (directly comparable) securities like shares or bonds. We can only be sure of a house’s value when it is actually sold.
Herein lies our first problem. All the UK house price indices are based on transaction data of one form or another, but houses are rarely traded. In the 1990s, the number of private dwellings sold only amounted to about 7 per cent of the housing stock, according to an enlightening paper on this subject by Gregory Thwaites and Rob Wood at the Bank of England.
Even indices that take account of all transaction data, like the one published by the consultancy Acadametrics, are therefore based on small samples relative to the total housing stock. And the other indices are based on small samples even relative to the 50,000 or so transactions that are currently being completed each month. Of course, small samples have the advantage of being quick to collect and process, so results can be published in time to be useful. But they may not be representative of the whole market, and results based on few data points can also be quite volatile.
A further problem is that even if all the houses on a certain street are sold in successive months, what does it actually tell us about 'house prices'? Perhaps one house had two bedrooms and another four; if one sold for twice the value of the other it does not follow that prices have doubled. Equally, comparing apparently identical houses across different areas does not work because location matters.
This sounds obvious, but it’s a headache for statisticians, who have to make so-called ‘mix adjustments’ or ‘hedonic regressions’ to arrive at an abstract 'average house' whose value their index tracks. Needless to say, these tweaks to the data can undermine their usefulness.
The only alternative is to track resale prices on houses that have already been sold. That is the approach taken by the Land Registry, but since houses are so irregularly traded, this involves ditching the majority of the data. And the same house may be in a very different condition from one decade to the next, so even resale values are not a watertight guide to house-price inflation.
Of course, these problems are immediately obvious to anyone who follows the indices, because they invariably contradict each other on a month by month basis, and often beyond. Take the best-known indices, produced by the mortgage providers. Nationwide found the price of a ‘typical home’ sold in the three months to March was 0.9 per cent lower than a year earlier, while Halifax found it was 2.9 per cent lower.
It was in response to a 6.4 percentage-point divergence between the Nationwide and Halifax indices that Mervyn King, then the Bank of England’s chief economist, called for greater clarity on house prices in a speech to the Building Societies Association in 1998. He attributed the divergence then to the lenders' different mix-adjustment techniques, but admitted this was only a “preferred explanation” of a phenomenon that remained “largely a mystery”.
A number of alternative indices have been set up since, in response to the strong housing market as much as to Mr King’s plea. They add to the picture, but also to the confusion. The best house-price watchers can do is simply to look at all the indices together whilst understanding what underpins them – I’ve summarised the key points in the table. At least the main indices do tend to agree on the direction and strength of the market, if not on its growth rate.
A final caveat, however: even if you understand these indices intimately, you will not be much the wiser about the value of your own home. That’s because housing is a local market, and although most indices split prices down by region, and some by County Council or London Borough, even the most granular subdivisions include many postcodes with distinct characters and housing markets. “There’s an overriding issue of generality,” complains Neil Young of the Young Group, a South London landlord and property-management firm.
This problem has been alleviated by the advent of internet-based search engines like Zoopla, which publish the Land Registry’s data on sold house prices. It is now very easy to find out what houses on your street have fetched on the market, albeit about three months after completion. If you know the houses in question, this can be a useful guide. But remember, the only sure way to find out what your house is worth is to sell it.
The indices - all you need to know:
The Nationwide and Halifax indices are both based on the lenders’ proprietary mortgage books. Their great merits are timeliness – the Nationwide figures for March were published on 29th March and the Halifax ones on 4th October – and longevity – Nationwide has data going back to 1952.
Their main drawback is small sample size. They do not include cash purchases, which now account for about a quarter of transactions. And Nationwide and Halifax don’t control the mortgage market as they use to: Nationwide’s share is now only 9 per cent and Halifax’s is 14 per cent, reckons Ray Boulger at mortgage broker John Charcol.
Both indices adjust the data for mix of properties by size and region, as well as seasonality, resulting in a seasonally-adjusted “average house” whose price the index tracks. But the results will still reflect the market position and lending strategy of the lender on whose mortgages they are based. One explanation for the relative robustness of the figures published by Nationwide is the building society’s traditional strength in the more resilient South, for example.
The Land Registry(LR), official guardian of land rights, owns the most complete and accurate data set on house prices, with 100 per cent of transactions in England and Wales recorded. But it does not cover Scotland or Northern Ireland and only goes back to 1995. And to construct its index the LR only uses houses which have been sold more than once since 1995 – about 35 per cent of transactions. A further problem is timeliness – the LR index for March was only published on 1st May, and will only include about a third of the full March data.
The LSL Acadametrics index (formerly FT Acadametrics) also draws on the LR data, but is not based on repeat sales like the LR index itself. Instead, Acadametrics takes all available data and adjusts for seasonality and mix. It also tries to overcome the delays to which LR data is subject by forecasting a month ahead, based on the mortgage providers’ data, and then gradually aligning its forecasts with the actual numbers as they trickle in from the LR. This is complex, but thorough, and for this reason it is Mr Boulger’s preferred index. The company has also overcome the geographical limitations of LR data by launching a similar index based on data from Registers of Scotland.
The house-price index published by the government’s Communities and Local Governmentdepartment is based on mortgage data, but a much wider range than either Halifax or Nationwide. That makes it more accurate – it covers about half the mortgage market – but less timely as it has to collect data from roughly 50 lenders. It won’t publish its October results until December.
Estate-agent aggregator Rightmove also publishes an index based on asking prices cited on its website. This has the advantage of being bang up to date and covering 90 per cent of the market, but asking prices are an inaccurate guide to sales prices. It is the only index currently showing positive annual growth – 3.4 per cent for the year to Apr 2012 – which suggests vendors are still pricing properties unrealistically.
Property consultancy Hometrack publishes an index based on a monthly survey of estate agents. It asks agents in every postcode what the achievable selling price would be on a given date for four standard property types. It then aggregates the results on a volume-weighted basis to come up with an “average price”.