Arzu Uluc.
A novel housing data set with information at the Local Authority level shows that changes in the proportion of high LTI mortgages was positively correlated with the size of the house price boom (1997-2007) and bust (2007-2009). Interestingly though, changes in the proportion of high LTV mortgages had the opposite relationship. This blog sets out some stylized facts on the wild swings in house prices over this period using this new data set.
To investigate booms and busts at the local level, I compiled a novel panel dataset covering 325 Local Authorities (LAs) in England over 12 years (1997-2009). The dataset combines real house prices, real gross disposable household income (GDHI) per capita, dwelling stock per capita, and mortgage market information from numerous sources.
Stylised facts on housing booms and busts at the local authority level
1. Real house prices in England rose on average 150 percent between 1997 and 2007, and subsequently fell by 12 percent by end-2009.
A first look at the data reveals substantial differences in real house price growth across locations between and within regions (Charts 1-2).
Chart 1: House Price Growth 1997-2007
Chart 2: House Price Growth 2007-09
Sources: Land Registry, ONS and Bank calculations. Charts 1-2 display the variation in real house price growth of LAs in each region (East, East Midlands, London, North East, North West, South East, South West, West Midlands, and Yorkshire and the Humber) in England. The bottom and top of each box are the first and third quartiles, and the band inside the box is the median.
A cross-section econometric model is estimated to investigate the factors associated with larger local housing booms and busts. I modelled cumulative real house price growth as a function of growth in real GDHI and dwelling stock per capita, and mortgage characteristics such as the share of high LTI and LTV mortgages, and the share of interest-only mortgages.
It is important to note that the analysis does not account for potential reverse causality (higher house prices driving changes in mortgage characteristics) or potential third factors that might drive both income and house prices.
Due to data availability, I conducted the boom analysis across three sub-periods: 1997-2001, 2001-04 and 2005-07. The break in 2005 stems from mortgage market information coming from two different data sources: the Survey of Mortgage Lenders (SML) for the period 1997-2004 and the FCA Product Sales Database (PSD) for the period 2005-2009. In addition, the earlier period was split into two as dwelling stock at LA level is only available after 2001.
2. Changes in the proportion of high LTI mortgages were positively correlated with the local housing booms and busts.
Results presented in Table 1 indicate that house price growth was higher in areas where the share of mortgages with LTI above 3 increased most between 1997 and 2004 (Model I and II). As the median share of mortgages with LTI above 3 increased from 16 to 57 percent between 1997 and 2007 (Chart 3), I used LTI above 4 rather than 3 as a threshold of high LTI mortgages for the 2005-2009 period. The results support the conclusion that the higher the share of mortgages at high LTI, the larger the booms (Model III). In 2007-09 areas with larger decreases in the share of mortgages with LTI above 4 tended to see larger price falls (Model IV). It is worth noting that OLS estimates can be biased due to potential reverse causation. Nonetheless, the analysis provides a view on factors that were associated with the size of local booms and busts.
Table 1. Housing Booms and Busts
Notes: Robust standard errors in parenthesis. ***p<0.01, **p<0.05, *p<0.1. Observations are clustered by NUTS3 local areas. Outliers are excluded from the analysis. Since SML data consist of a sample of mortgages and I exclude areas with few observations, coverage is lower in regressions based on those data; see the Appendix for details.
Chart 3: Share of Mortgages with LTI above 3
Sources: SML (1997-2004), PSD (2005-2009), and Bank calculations. The bottom and top of each box are the first and third quartiles, and the band inside the box is the median.
3. There is a negative relationship between the changes in the proportion of high LTV mortgages and the size of the booms and busts.
SML data suggests that the median share of mortgages with LTV above 90% decreased from 44 percent to 15 percent between 1997 and 2004 (Chart 4). One reason for initial decreases in the LTV ratios is that in late 1990s lenders gave borrowers exemption from mortgage indemnity insurance (which insures lenders against mortgage default) if loan to value ratios were below 0.9. This gave borrowers a considerable incentive to reduce mortgages to below this level. LTV ratios continued to drift down: median LTV ratio decreased from 89 percent to 72 percent between 1997 and 2004 (Chart 5).
Chart 4: Share of Mortgages with LTV above 90%
Chart 5: Distribution of LTV ratios
Sources: SML (1997-2004), PSD (2005-2009), and Bank calculations. The bottom and top of each box are the first and third quartiles, and the band inside the box is the median.
The analysis yields an interesting negative correlation between changes in the share of mortgages with LTV above 90% and house price growth (Model I-III). This negative relationship could be driven by an increased share of transactions by home movers during the boom. Home movers tend to rely on lower LTV mortgages as they are able to raise equity from their existing properties to move.
Since an overall LTV ratio can mask different trends in the LTV ratios of home movers and first-time buyers, I also looked at the relation between the changes in the share of FTB mortgages with LTV above 90% and house price growth. The results show a significant and negative relation between the two for the 1997-2001 and 2001-2004 periods (estimation results not shown). When interpreting these results reverse causality should be kept in mind: high house price increases may create pressure on affordability experienced by FTBs and crowds out those with lower deposits, leading to a negative relation between house price growth and changes in the share of high LTV mortgages. Additionally, during the housing boom in the 2000s there was a big increase in the proportion of FTBs receiving support from their families, in the form of gifs, loans or bequests (Tatch, Housing Finance, 2006). So FTBs may have been less dependent on high LTV mortgages.
4. Changes in the propotion of interest-rate only mortgages were pro-cyclical, similar to the picture for high-LTI mortgages.
I also find a positive relationship between changes in the share of interest-only mortgages and larger booms in both 1997-2001 and 2005-07 (Model I and III). The share of interest-only mortgages (including endowment mortgages) actually fell during 1997-2001 – from 58 to 26 percent – so the results indicate larger house price growth in areas where the share fell less in this period. In 2007-09 areas with larger decreases in the share of interest-only mortgages tended to see larger price falls.
5. Housing booms and busts were also associated with real factors such as real income and dwelling stock growth.
The econometric analysis presented in Table 1 suggests that housing booms and busts were associated with both real factors and credit loosening. Higher real income growth were associated with larger booms in 1997-2004 period, and in 2007-09 areas with higher growth in the dwelling stock per capita tended to see larger price falls.
6. The larger the boom, the larger the subsequent bust.
Chart 6 shows that during the bust house prices declined more in areas that experienced higher house price growth during the boom. The result still holds when I control for the contemporaneous real and credit variables.
Chart 6: House Price Growth btw 1997-2007 and 2007-2009
Sources: Land Registry and Bank calculations.
Conclusion. House prices are subject to boom and bust cycles, which can threaten financial and macroeconomic stability. The analysis exploiting local variation in house prices across England to understand the factors associated with larger booms and busts give greater insight into the role that credit conditions played over the 1997-2009 housing cycle.
Arzu Uluc works in the Bank’s Macro-Financial Risks Division.
If you want to get in touch, please email us at bankunderground@bankofengland.co.uk
Bank Underground is a blog for Bank of England staff to share views that challenge – or support – prevailing policy orthodoxies. The views expressed here are those of the authors, and are not necessarily those of the Bank of England, or its policy committees.
