Revealed: huge differences in data used to compile house price indices

HM Land Registry has revealed how major house price indices are compiled - showing the huge variation in transactions behind the headline figures.

The critique, released earlier this month, gives a detailed breakdown of the different methodologies used by the main indices.

These top indices  are, according to the registry, the government’s official UK House Price Index, the LSL Acadata Index (branded as from Your Move/Reeds Rains), the indices from mortgage companies Nationwide and Halifax, and the index produced by Rightmove.

The registry analysis is the perfect Christmas reading for statistical anoraks, covering the stage in the property cycle at which the index is produced, the weighting given to the figures and the ‘adjustment methodology’ employed by each index to iron out what are considered seasonal or unrepresentative factors.

However, possibly of greatest interest to many agents is the range of transactions used to compile the indices.

HM Land Registry - perhaps aware that it is the source of the statistics behind the government’s index - says that “other things being equal, statistics based on larger, unbiased, number of observations (transactions) are likely to produce estimates closer to the true value (house price index/growth rate).”

It says Halifax and Nationwide indices are based on their own mortgage data, so naturally have a smaller number of transactions: around 12,000 to 15,000 transactions per month.

LSL Acadata uses transaction data from HM Land Registry for England and Wales and so represent all registered transactions for those parts of the UK; data from other independent sources are also used. In total this comes to around 80,000 transactions a month.

The government’s House Price Index has full UK coverage so its number of transactions is larger than LSL Acadata and is effectively a census of all housing transactions each month - some 100,000 or so.

Rightmove - which of course bases its index on asking prices - claims a similar number as reported by the HM Land Registry (100,000) but is based on advertised sales, which ultimately might not sell in that period.

The full survey is an interesting read for enthusiasts (and critics) of house price indices. You can see it here.