FOOL SCHOOL
Understanding House Price Indices

May 27, 2005

Every few days, we're bombarded with new data on the housing market, saying that the average price of a three-bedroomed house in Dorset has risen by 2% or something. It's reached a stage where the Government got so concerned that it might be creating distortions in the market that they've even added their own index to the melee, which is issued by the Office of the Deputy Prime Minister.

Anyway, if we're to make any sense of the surveys, then we need to know how they're cobbled together. The basic approach is to track the price of an 'average' house, but this hides a multitude of sins and the different surveys vary according to how they actually calculate this average and where the data comes from. We're going to focus on the following surveys:

The Land Registry
Halifax
Nationwide
Hometrack
Rightmove
ODPM

Averages

Simple average

The simple option with averages, taken by the Land Registry for example, is just to take a straightforward average of everything in your data sample. This has the advantage of transparency, but the mix of properties might change between reporting periods. So you might get a load of expensive mansions selling in one period and a preponderance of cheap studio flats in the next. You'd get a falling average price, but it would be very misleading.

Simple averages tend to get more reliable as the amount of data increases, since the chance of a significant change in mix is reduced. The national figures from the Land Registry should be pretty good, while the data for individual post code areas, where there may only be a handful of sales in a particular period, will be very unreliable.

'Mix-adjusted' average

In an attempt to overcome the problem of a changing mix, some surveys, like the Nationwide, Halifax and ODPM, have developed a 'mix-adjusted' price. This is designed to represent the changing price of a 'typical' property. Each transaction in the survey is labelled with a price and a series of defining characteristics -- such as location, type of property, floor size, number of bathrooms, number of bedrooms and so on. This data is then fed into a computer that spits out the price of a house with a particular set of characteristics that are thought to make it 'typical'.

Note that the price of this typical house is not the same as the average price of a house. In fact, the average house price will tend to be higher than the price of the 'typical' house, since it's more heavily influenced by a small number of very expensive properties. If you remember your maths from school, the 'mix-adjusted', typical price is more like a median price than a mean price.

Since they use slightly different 'typical' houses, Halifax and Nationwide will come up with slightly different figures each month (and they would do even if they used identical data).

Seasonality

The spring and summer tend to produce slightly higher prices than the rest of the year because, according to Nationwide, 'more buyers are in the market and hence sellers do not have to discount their prices so heavily, in order to achieve a sale'. In fact, July is the strongest month, with 'raw' prices tending to be over 1% ahead of their 'seasonally-adjusted' level. January is weakest, with 'raw' prices tending to be some 2% below their seasonally-adjusted level.

So some surveys tweak their prices according to the time of year, to produce a seasonally-adjusted price. This should give a clearer picture if you're looking for the overall trend in prices over the last few years but happen to do so in the depths of winter. But if you're about to buy a house and want to know what prices are doing, then you'll be after actual prices even if they are a bit lower because it's the middle of winter.

Source of data

The Land Registry takes the simple approach of including all property transfers, but Nationwide and Halifax 'clean' their data to get rid of transactions that they think might cause distortions. For example, Nationwide excludes 'buy-to-let' properties and anything that's 'untypical' (for example, anything over £1m in price, anything that's not for fair market price and anything that's very large or very small).

Nationwide and Halifax source their data from the mortgages they provide on house purchases, so they are only looking part of whole market. The ODPM gets its data from a monthly survey of fifty different lenders, so it is based on a bigger sample size and so should be more representative. All three surveys however exclude remortgages and properties bought for cash (the latter of which accounts for about a quarter of home sales).

Hometrack generates its data from a list of 3,500 approved agents around the country. It then calculates a simple average price for each category of home. However, they do this from the 'average' prices provided to them by the different agents and they rely on the agents to account for things like a changing mix of properties. They probably do it quite well -- they are estate agents after all -- but it introduces a large dose of subjectivity.

Rightmove uses the asking price for all the properties placed on its website by its member estate agents (which, it estimates, represents 50% of the market at any time). So, if properties are tending to sell a little way below their asking price, as would be normal, then the price should tend to be overstated.

One further point to make is that some people suspect that the different surveys may possess a degree of regional bias depending on where they're mostly based. Halifax, for example, is thought by some to have a northern bias, while Nationwide may have a bias towards the south. They both claim that they adjust for this, but not everyone is convinced. Rightmove also makes a regional adjustment (which it says is similar to the method used by Halifax).

Timeliness of data

The final point about data is not where it comes from, but when it comes. The Land Registry, for example, gets its data from the registration of completed property transactions - which is pretty late in the day - and it only produces the data quarterly. So you can reckon that it's a few months out of date by the time it arrives. It also licences its data to house price websites such as Nethouseprices.

Hometrack's data comes from the estate agents, so it depends a bit on what they're doing. Most likely, they're looking at completions, so the data is probably a little old. Rightmove's data is probably the newest since it's based on asking prices -- before a buyer has necessarily appeared on the horizon -- but that, of course, introduces its own problems as the final sale price will very rarely be the asking price.

The ODPM, Nationwide and Halifax surveys collect their data at the point of mortgage approvals. This means it should be very up to date, including sales that haven't actually completed yet. As with Rightmove, however, the problem is that it might just be a bit too quick and some of the deals that get included might never actually go through.

Summary of the different surveys

	Land Reg	Halifax	Nationwide	Hometrack	Rightmove	ODPM
Type of average	Simple	Mix-adjusted	Mix-adjusted	Simple	Simple	Mix-adjusted
Seasonal adjmt	No	Some	Some	Some	No	No
Source	Completions	Mortgage approvals	Mortgage approvals	Estate agents	Asking prices	Mortgage approvals
Cleaning	No	Yes	Yes	Some	Yes	No
Frequency	Quarterly	Monthly	Monthly	Monthly	Monthly	Monthly
Year started	1995	1984	1973	2000	2002	2003

Conclusion

In terms of transparency and knowing what you're looking at, you can't beat the Land Registry data, but you have to be careful how you interpret it. Nationwide and Halifax, on the other hand, adopt similar scientific approaches which probably give a clearer representation of the underlying picture -- but you have to take things on trust a little. The ODPM's index should be slightly more representative of the broader market.

The best solution is not to worry too much about the short-term trends. It may be useful to get an idea of how much your house might be worth, by comparing it to the house price data, to help with long-term financial planning or to help you work out what you can afford if you're thinking of moving. It's also useful to get a very board view of whether prices are moving up, down or sideways. But trying to time house purchases and sales on the basis of monthly or even quarterly house price data is a complete waste of time. The data is pretty unreliable in itself but, in any case, the particular property you're looking at could be a long way from being 'average' -- or even 'typical'.

> For a great mortgage deal, try out Homeowning Centre!