As UK house prices start to fall again, this blog looks back at the Global Financial Crisis and examines the differing experiences across regions and neighbourhoods. It looks beyond house price indices - which showed similar peak-to-trough falls across the country - and explores the areas in which properties were sold at a loss in later years. 'Sales at loss' were much more likely in deprived areas and in regions that recovered from the GFC more slowly. This emphasises the importance of looking beyond macro factors to consider local housing market dynamics when thinking about future property prices and has interesting implications for the design of mortgage portfolio stress-tests.
Chart 1 shows regional house price indices for England and Wales in the run-up to and aftermath of the GFC. While the peak-to-trough fall in house prices was similar across regions, house prices remained below their previous peak for much longer in some regions than others. Chart 2 looks below the regional indices to the distribution of local authority price indices for each region. It shows the time it took to 'recover' to the previous peak varied widely, from less than a year in some parts of London to over 10 years in parts of the North West and North East.
Falling house price indices can mean paper losses for homeowners and landlords, but for the vast majority, those losses aren't actually realised. We now turn to examining where the losses were realised, through the concept of 'sales at loss'. 'Sales at loss' uses Land Registry data to measure the number of properties that were sold for less than they were bought for. This is a deliberately simplistic metric, making no attempt to include broader costs associated with owning and transacting housing (stamp duty, solicitors' fees, renovations, etc).
The ONS bottom-up data sums up the LA data to get a national picture for England. It seems to fit pretty well with the ONS national data. The LRPP data should in theory match the 'Private Enterprise' completions. But it doesn't appear to. I guess this could be because not all new builds are sold? i.e. if I build a house for myself, it isn't going to show up in the LRPP data but should make it into the ONS building stats, as long as they're sourced from the planning process rather than the Land Registry.
Charts 3 and 4 look at the proportion of properties sold at a loss each year, for London and the North East. For properties sold at a loss, the colour of the bar shows the year in which they were bought. For instance, for the year 2009, around 5% of properties in London were sold at a loss, with around 1% bought in 2006 and 2% bought in 2007. For the North East, the figures are not only much higher in the immediate aftermath of the crash, but material numbers of sales at loss persist for much longer.
Charts 3 and 4 show that most of the losses in later years occurred on property originally sold between 2005-8. Chart 5 focuses in on property sold in those years and compares the regions of England and Wales. The height of the bars show the proportion of properties bought in those years that were later sold at a loss and the colours now break down the scale of the loss. It shows a far higher proportion of properties bought towards the peak of the housing boom were later sold at a loss in the North, and also that higher proportions were sold at material losses. Chart 6 looks at the distribution of sales at loss at the level of local authorities, showing strong local clustering. The pattern is even clearer at lower levels of geography.
Academic literature written on the broader economic impact of the crisis finds similar regional patterns. The resilience of the economy in London and the South East is often attributed to London’s role as a hub for knowledge-intensive, high-value added industries and global investment. Areas more dependent on lower value-added industries saw slower economic recoveries, with more deprived areas often found to have the longest-lasting downturns as they were hit by both the fall in private sector activity and the reduction in government spending.
Chart 7 shows a strong correlation between prior levels of deprivation (as measured by the ONS' Indices of Multiple Deprivation) and sales at loss. Chart 8 shows a similar relationship between sales at loss and the increase in unemployment rates after the crisis.
There is little reason to think the current downturn in house prices will follow the same patterns as the last one. But digging beneath the aggregate data to think about the underlying distribution of vulnerabilities is likely to be as important as ever. Our next blog will turn to some of those vulnerabilities. In the meantime, if you'd like to discuss any of this in more detail or would like to use this or similar analysis, just get in touch.