HOUSE PRICE CALCULATOR METHODOLOGY

The House Price Calculator (HPC) is based upon the prices reported to the Land Registry (LR) following the sale of a property in England and Wales. LR does not hold data for properties in Scotland, Northern Ireland, Isle of Man, Guernsey or Jersey; nor does LR hold data for properties sold prior to 1995, except for all property types and all regions combined. For such properties and dates, the HPC uses various Nationwide House Price Index series. For properties in England and Wales, from 1995, LR provides prices averaged by region and, from 2000, averaged by county and London borough. We accordingly prepare separate series of prices for regions, counties and London boroughs and use these, provided that the volumes of transactions available for particular house types and localities are adequate for the calculation of a reliable current value and that recent variation in the associated index values has not been excessive. However, the value of an individual house will be affected by factors such as the number of bedrooms which is not provided in LR data. Furthermore, an HPC calculated current value reflects only the change in average values for the relevant property type and London borough, county or region in respect of which we hold and have used price data. The process is described below. Such changes may well not apply to a particular postcode area. In addition, HPC will not indicate any change in value resulting from improvements or dilapidations.

The data provided by LR make use of every transaction in England and Wales and HPC thus uses the complete dataset, as opposed to a sample. The prices used by HPC allow accurate assessment of the current value of a portfolio of properties but, for individual properties, the best that HPC can provide is a guide, for interest only, to the change in value associated with a selected property type and locality. For individual properties, the HPC value may be wide of the mark and a professional surveyor should be engaged if a reliable estimate of current value is required for financial or other important reasons. As a very rough guide, only about one third of individual valuations made by HPC are likely to lie in the range +/- 10% of the value achieved at sale. Standard deviation bands, as described below, are provided primarily for those working in the property sector.

The methodology

HPC includes a postcode look-up routine that validates the input postcode and identifies in which London borough, county or region a property is located, using the postcode without the two least significant characters. The following routines are then utilised.

London Boroughs (where relevant)

First, the HPC:

• estimates the number of sales in the year of purchase, or of the initial valuation, for the appropriate property type and borough and checks that this equals or exceeds 600 (i.e. 50 per month, on average),
• estimates the number of sales in the year of the most recent index value for the appropriate property type and borough and checks that this equals or exceeds 600,
• checks that the most recent index value for the appropriate property type and borough differs from the value for the preceding month by not more than +/- 10%.

If all the above criteria are met, the London borough data series is used. If not, the property is priced using the county data series (which includes London as a county) below, or the region data series below.

Counties

First the HPC:

• estimates the number of sales in the year of purchase, or of the initial valuation, for the appropriate property type and county and checks that this equals or exceeds 600,
• estimates the number of sales in the year of the most recent index value for the appropriate property type and county and checks that this equals or exceeds 600,
• checks that the most recent index value for the appropriate property type and county differs from the value for the preceding month by not more than +/- 10%.

If all these criteria are met the county data series is used.

Regions

Depending upon the region, the LR regional, LR UK aggregate or Nationwide data series is used as follows:

• if one or more of the criteria for use of the county data series is not met, the LR region data series is used,
• if the region is Scotland or Northern Ireland, monthly values, derived by interpolation from the Nationwide Regional Quarterly Indices (Post ’73) table up to Q1/1991, are used; thereafter, values are taken from one of the Detached (Post ’91), Flats (Post ’91), Semi Detached (Post ’91) or Terraced (Post ’91) tables (n.b. none of these tables is seasonally adjusted),
• if the region is Unknown (which will include Guernsey, Jersey and the Isle of Man), the LR average property price for all property types and for all England and Wales is used.

Dates

For a London borough or county with a month and year of purchase or initial valuation between 1/1/1995 and 31/12/1999, the region data series is used to estimate price growth up to 1/1/2000.

If the month and year of purchase or of the initial valuation are earlier than 1/1/1995, the region data series is used for the period between 1/1/1995 and 31/12/1999 and Nationwide data are used for the period up to 1/1/1995 (as for Scotland and Northern Ireland); however, values for all property types and regions are adjusted by the ratio of growth in the LR average property price for all property types and for all England and Wales to that in the UK column in the Nationwide Regional Quarterly Indices (Post ’73) table (as monthly interpolated).

Standard Deviation Bands

HPC provides bands, defined in terms of the number of estimated standard deviations, as a guide to the accuracy of valuations. In general, where a distribution is roughly normal, one standard deviation below and one above should take in two thirds of the data and two standard deviations below and above should take in about 95% of the data.

The estimates of standard deviation were derived using data from LR for 2006 and part of 2007 for properties where LR had records of two or more sales transactions for the same property. In the case of multiple transactions, only the first and last were included. A large number of ratios were constructed by dividing the first price revalued to the date of second sale by the second sale price. Because the distribution of these ratios was markedly skew, the ratios were replaced by their logs and standard deviations obtained for all the various combinations of property type and location in the three house price index data series (i.e. London borough, county and region). A mean of zero was assumed in each case and values were calculated for two standard deviations below, one standard deviation below, one standard deviation above and two standard deviations above this mean. Antilogs of all these results were calculated to provide factors that could be applied to the HPC revaluation to produce the bands that appear as part of the output.

Caution should be used in applying the standard deviation bands where the date of purchase or of initial valuation falls before 1/1/2000, since these bands may not be appropriate for the LR regional or Nationwide data series used to estimate growth before that date.