Acadata Blog

David Thorpe, Acadata, blog: House price indices

Each ‘house price index’ provides a unique measure. An ‘average house’ is a concept.

There are a number of ways of measuring ‘price’ and of calculating an ‘average’. Hence, one ‘average house price’ is unlikely to be comparable to another.

Plotted over a long timescale, house price indices appear to move in unison. But, from  month to month, they do so only when prices are steadily on the rise or a fall. It is best to follow a preferred index, rather than try to make sense of what are often contradictory reports.

Here are the main options.

Halifax and Nationwide use only the valuations provided by surveyors for mortgage offers to estimate the price of an ‘average house’. Data volumes are small and reflect their own lending policies. Highly valued as the long-standing incumbents. For instance, think e.g. semi-detached in the West Midlands.

Rightmove and Hometrack use estate agency asking prices. The Rightmove HPI uses more data than any other index but can prove optimistic. Hometrack employs opinions rather than house price data and is a survey rather than an index.

Land Registry searches their c.15 million price records but uses only those ‘repeat prices’, each month, for which an earlier price can be found – a low data volume similar to that used by the lender HPIs.

The Land Registry average house price uses mortgage and cash data, is lower than those of the lenders and is a geometric mean, updated monthly by a repeat sales regression methodology. Technically interesting.

ONS provides a real deal in HPIs, using more data than the lenders and mortgage completion prices akin to those transacted. And ONS estimates the average price paid for a house each month, not that of the ‘average house’ and, hence, reflects London. But (and it is a big but) ONS lacks cash data on the c.35% of properties bought with cash.

LSL Property Services plc – the LSL Acad HPI – is truly independent, prepared by the consultancy Acadata. Like ONS, LSL Acad provides the average price paid each month but employs both mortgage and cash data. Like the Land Registry, LSL Acad uses final transacted prices but employs every reported price rather than only the repeat prices – a far larger data sample than any other excluding Rightmove. Uniquely, the average price is smoothed over three months to show trends, rather than the ups and downs, each month, which result from small samples.

Downsides? Like the Land Registry HPI, LSL Acad HPI has available only the c. 38% of sales reported by month end for the Land Register but forecasts the missing 62%. So the average price is a modelled price – always closely comparable to the ONS average published for the same month.