Bath may be located in the South West, Oxford in the South East, Cambridge in the East of England and Aberdeen in Scotland but trends in house price growth suggest that's not completely true in terms of housing markets. The correlation between regional and local level price growth suggests that many local markets actually have a stronger correlation with a region other than the one in which they are physically located.

We have created our own version of the Land Registry house price indices but based on local authorities rather than unitary authorities/counties. That allows us, for example, to look at the performance of Cambridge city rather than just the wider Cambridgeshire market. Using these indices, we have created a map to show the former Government Office Region with which each local authority has the strongest correlation based on house price growth.

Amongst other trends, the correlation map highlights London's spread beyond its official boundary with red areas in the Home Counties, Cambridge and Oxford. It also shows some of the limitations in using the former Government Office Regions to assess housing markets. It is perhaps more important to understand how cities interact with their surrounding areas.

These price growth correlations have important implications for future local market housing activity. Following a period of very strong growth, we are predicting London house price growth to be lower than all other regions over the next five years. For local markets closely linked with London, the question will be whether they see continued strong growth or follow lower growth London?

London house prices are 33 per cent above their 2007 peak while prices in the North of England are 18 per cent below according to the Land Registry index. That's just below their 2009 trough (-17 per cent) but above 2013 levels (-23 per cent).

Create your own house price index

The official Land Registry index is calculated using repeat sales regression. That means it looks at the price change for properties that have sold multiple times since 1995 and allows it to compare like-for-like price growth.

The Nationwide and Halifax, with more information on property characteristics, use hedonic regression for their indices. This method attempts to attribute a value to each component characteristic of a property, such as location, type of neighbourhood and size of property, and then aggregate the component values back up to an ‘average’ property value.

All indices have their strengths and weaknesses. Nationwide and Halifax are useful for their timeliness but limited by their sample sizes and dependence on mortgage approvals. The Land Registry is more comprehensive and provides more robust local level results but lags behind the news cycle and is revised over time. The Office for National Statistics recently consulted on combining their existing and Land Registry indices into an ‘official’ index but our preference will remain monitoring all available indices while understanding the reasons for any differences.

Thanks to great work by my colleague Andrew Smith, we are now able to create bespoke repeat sales indices for custom areas provided there is a sufficient sample size. This work has raised an important issue with the basic repeat sales method: it fails to account for any change to the property. It assumes that a property remains the same over time with no allowances for any depreciation, improvements or additions such as a new conservatory. The Land Registry accounts for this effect by adjusting their indices using data on home improvements from the English Housing Survey. The unadjusted indices are still useful as there may actually be cases where any improvements in the quality of stock are an important consideration such as in areas experiencing regeneration.

The official Land Registry house price index has been adjusted downwards by 94% since January 1995. That implies we’ve improved our homes by 1.5% of their value per year (on average).

Taken from
Housing Market Notes