Housing

Monthly house price index methodology

Executive Summary

Communities and Local Government launched a new, monthly house price index in September 2003.

Detailed points

  • Communities and Local Government has calculated a quarterly house price index since 1968 based on a five per cent sample of completions from a wide cross-section of the United Kingdom's mortgage lenders. It is the longest-running house price index in the UK. 49 lenders currently supply completions data through the Survey of Mortgage Lenders
  • the outgoing index was only published as a quarterly series because the relatively small sample size (only 3,000 completions per month) was insufficient for a monthly series. However, thanks to electronic data transfer, many lenders have now switched to supplying 100 per cent of their completions instead of a five per cent sample. Consequently 25,000 completions per month are included in the new monthly index
  • the new index covers purchases for owner-occupation and for buy-to-let. Re-mortgages are not included because the priceof a property represents a valuation, which is not necessarily the same as a negotiated marketprice. Purchases by sitting tenants are also omitted because they are below market level
  • Communities and Local Government has also improved the underlying index methodology (assisted by methodologists from the Office for National Statistics). The overall structure of the new index reflects the recommendations of a cross-cutting House Price Working Group (comprising representatives from Communities and Local Government, the Office for National Statistics, HM Treasury, the Bank of England and Land Registry)
  • for further information on the House Price Working Group, and its findings and recommendations, the ONS published an article in the November 2002 issue of Economic Trends entitled An improved House Price Index - an update on developments.
  • a regression model is used in the new index to produce price estimates for about 100,000 cells. The benefits of this approach are that it gives greater flexibility:
    • no need for minimum sample per cell
    • partial data can be used
    • better mix adjustment
    • more detail on location and dwelling size
    • can add new variables
  • in order to select the price estimation model, a range of alternative models were compared and tested to ensure that:
    • house prices are log-normally distributed
    • no key variables are omitted
    • the modelled means are comparable with the actual means
  • the preferred model was a main effects model including seven explanatory variables:
    • dwelling type
    • is dwelling old or new?
    • is buyer a first-time buyer?
    • number of habitable rooms
    • county (or London borough)
    • type of local authority (based on an ONS classification)
    • type of neighbourhood

and several two-way interactions between these variables

  • weights are calculated once a year (in January) based on the relative numbers of transactions during the previous three years. However, because these transaction weights are applied to average cell prices, the net effect is that the index uses expenditure weights
  • the index is an annual chain-linked Laspeyres-type index, like the Retail Prices Index
  • house price indices for the UK and the Government Office Regions are published in a monthly press release. A quarterly report will carry more in-depth analysis and a wider range of indices and other house price indicators
  • a seasonally-adjusted series will be constructed in due course but a run of between two to five years of the new data will be needed before the parameters required for seasonal adjustment can be reliably determined
  • the ultimate target is to create a house price index that reflects cash purchases as well as mortgaged purchases. Unfortunately the Land Registry is the only possible source of information on completed cash purchases - and there are two major deficiencies in the Land Registry data: (a) the excessive time lag between completion and the time details are received by Land Registry and (b) the limited range of information available for each property purchased. Land Registry's long-term plans for "e-conveyancing" and the proposed development of a National Property Databank (by the Valuation Office) may help overcome these problems
  • the Halifax and Nationwide house price indices still have an important role to play. Their indices are based on mortgage approvals - so reflect house prices at least one month ahead of completion. Moreover, since the Halifax and Nationwide use only their own in-house data they can process it immediately and do not have to await the receipt of data from other lenders. Consequently the Halifax and Nationwide indices continue to provide a much earlier indicator of house price trends than our new index
  • lenders participate in the Survey of Mortgage Lenders on a voluntary basis so we cannot insist either that they switch to 100 per cent or that they must do so by a particular point in time

Related downloads

Do you need help viewing file formats?

More affordable, more sustainable homes for the future

Find out about our plans for three million new homes by 2020.

Have your say

My favourites