Comparing Housing Booms and Mortgage Supply in OECD Countries

COMPARING HOUSING BOOMS AND MORTGAGE SUPPLY IN THE MAJOR OECD COUNTRIES Angus Armstrong and E Philip Davis NIESR and Brunel University London

Introduction The house price and lending boom of the 2000s is widely seen as the main and unique cause of the financial crisis that began in 2007. However, looking to the past, we find a similar boom in the late 1980s which did not lead directly to a global systemic banking crisis This raises the question whether the received wisdom is incorrect, and other factors than the housing boom caused the crisis, while macroprudential policy is overly targeted at the control of house prices and lending per se. Accordingly, in this paper we compare and contrast the cycles in house prices over 1985-94 with 2002-11.

The cycles entailed a similar rise in real house prices where booms took place, and a marked rise in the real mortgage stock along with real incomes and real financial wealth. The aftermath periods are also comparable in terms of house price changes as are econometric determinants of house prices in the two cycles since liberalisation. There remain some contrasts. Leverage rose far more in the later episode and did not contract in the aftermath. Adjustment to the long run is slower in the recent cycle. The earlier boom period showed differences with average house price behaviour which was not mirrored in the most recent boom. Despite the contrasts, on balance we suggest the evidence is contrary to the idea that the recent boom was in some way unique and hence the key cause of the crisis. Among areas for further research to capture factors underlying the recent crisis are the impact of initial levels of debt/income, or inflation, of global liquidity and monetary policy decisions, of wider securitisation, and of changing owner occupation rates and population densities. These could also be relevant for calibrating macroprudential policies.

House prices, income, wealth and debt during booms Percentage change prices Real house RPDI Real household debt Nominal house prices 1985 q1- 89q4 112 31 6 49 60 76 190 81 25 46 8 33 91 78 33 61 Real gross financial wealth 1985 q1- 89q4 61 31 37 65 27 50 95 35 46 56 22 76 57 94 80 55 1985q1 -89q4 2002q1 -06q4 1985q1 -89q4 2002q1 -06q4 1985q1 -89q4 2002q1 -06q4 2002 q1- 06q4 65 44 2002 q1- 06q4 17 33 United Kingdom United States Germany France Canada Italy Spain Austria Netherlands Belgium Denmark Ireland Finland Sweden Japan 71 12 1 28 32 32 110 68 24 32 -8 12 56 35 27 49 29 -2 64 25 20 62 -5 11 41 56 48 32 44 -17 23 17 18 14 17 17 27 21 16 17 5 16 24 10 22 10 14 5 11 19 3 17 13 -2 3 10 18 17 12 4 74 40 18 51 53 88 23 16 16 21 21 38 78 35 59 50 48 -3 42 44 40 83 26 42 29 44 145 83 45 4 9 78 35 36 90 4 21 56 67 69 35 52 -20 26 17 10 41 27 19 1 58 48 42 52 16 0 42 28 Mean Mean (boom countries) Correlation Correlation (boom countries) 35 30 18 10 42 48 72 53 60 28 40 39 18 11 47 59 0.95 0.99 0.47 0.49 0.74 0.41 0.14 0.58 0.97 0.97 0.49 0.35 0.79 0.42 0.06 0.30

The aftermath of booms Real house prices RPDI Nominal house prices 1990 q1- 94q4 -5 11 16 1 -8 45 22 13 38 28 9 14 -32 -7 -2 9 Real gross financial wealth 1990 q1- 94q4 21 16 29 17 18 15 19 21 14 -3 -1 18 -16 -22 12 10 Percentage change Real household debt 2007 q1- 11q4 2007 q1- 11q4 -6 -4 1990q1 -94q4 2007q1 -11q4 1990q1 -94q4 2007q1 -11q4 1990q1 -94q4 2007q1 -11q4 0 United Kingdom United States Germany France Canada Italy Spain Austria Netherlands Belgium Denmark Ireland Finland Sweden Japan -21 -3 -2 -8 -18 12 -7 -2 21 14 -14 -24 12 12 11 7 -1 -2 10 12 8 14 8 14 -13 11 3 6 5 3 10 19 25 -4 13 32 8 14 22 9 -19 15 -21 -18 19 -8 -9 -7 22 36 10 -1 5 18 23 12 8 20 28 -2 -17 9 7 9 3 2 3 na -1 2 -6 -23 16 -17 -14 11 -6 -2 0 0 2 2 -4 8 8 0 -16 15 -5 18 -18 -48 13 17 -13 5 9 -2 -4 6 -4 5 0 4 -9 7 -26 0 0 na -42 -26 -9 0 7 -8 9 -2 0 Mean Mean (boom countries) Correlation Correlation (boom countries) -6 -7 7 2 8 10 10 -2 9 -1 -7 -6 6 3 8 13 0.93 0.97 0.41 0.22 0.46 0.29 0.62 0.67 0.86 0.82 0.19 0.61 0.46 0.38 0.78 0.88

Indicators of leverage in booms and the aftermath Debt/personal income ratio change in percentage points Debt/house prices percentage change Debt/personal income ratio change in percentage points 1990q 1- 94q4 -1 1 10 -4 7 7 -2 0 7 -2 -33 -2 -9 -12 9 -2 Debt/house prices percentage change 1985q1- 89q4 2002q1- 06q4 1985q1- 89q4 2002q1- 06q4 2007q 1- 11q4 1990q 1- 94q4 2007q 1- 11q4 United Kingdom United States Germany France Canada Italy Spain Austria Netherlands Belgium Denmark Ireland Finland Sweden Japan 25 3 -1 9 14 8 -3 -1 1 3 8 13 17 9 21 30 6 -5 12 16 14 35 8 43 11 44 84 26 21 -5 2 1 -10 -3 -8 10 24 39 22 28 5 39 17 17 16 1 -5 -19 15 35 12 31 7 25 17 18 16 42 -42 -31 -6 -9 32 23 14 15 -1 -13 15 17 13 34 28 -8 -8 65 39 20 -9 23 34 17 29 1 30 15 52 100 20 20 9 2 2 24 10 16 22 10 24 -5 0 1 25 21 7 24 Mean mean (boom countries) 8 23 9 15 8 17 28 9 27 8 16 -1 11 18

Specifications for house price determination Typically link of house prices to determinants in cointegrating framework followed by allowance for dynamics in error correction framework (inverted demand function). For example Capozza et al (2002) focus on serial correlation and mean reversion using As increases, the amplitude and persistence of the cycle will increase whilst as increases the frequency and the amplitude of the cycle will Increase. P* includes population levels, real median income levels , the long-run (5 year) population growth rate , real construction costs and the user cost of housing in US regions Other recent studies use similar variables as long run determinants of house prices, for example: Muellbauer and Murphy (2008) real disposable (non-property) income, the sum of mortgage rates and stamp duty rates, the national credit conditions index and a term which interacts the mortgage rate with the credit conditions index Barrell, Kirby and Whitworth (2011) include the real borrowing rate, the 3-month nominal interest rate, the loan-to-income ratio, the loan-to-value ratio, per capita real disposable income, the ratio of the number of households to the housing stock, and the number of households.

Adams and Fuss (2010) include economic activity, construction costs and the long term interest rate. Loungini and Igan (2012) model real house price changes as a function of changes in disposable income, working-age population, equity prices, credit, and thee level of short- and long-term interest rates. While most work has been on individual countries, recent pooled or panel studies include our own work (Davis et al 2011) for 18 OECD countries, which was focused on the possible use of macroprudential tools in housing, Capozza et al (2002) who look at US Metropolitan areas, Adams and Fuss (2010) apply panel cointegration to 15 countries using Dynamic Ordinary Least Squares, while Igan and Loungini (2012) apply pooled OLS to 22 countries. All approaches to house price determination are fraught with identification problems, which make it difficult to separate supply and demand factors, and exogenous and endogenous determinants of house prices. All work on house prices faces this challenge and there is no definitive solution, although suggestions can be made both for ECMs and VAR/VECMs.

Possible additional variables Unemployment Banking crises Value of the housing stock Mortgage spreads Gross financial wealth [Stock of mortgage debt]

Approach to estimation Panel estimation for 11 OECD countries scope to investigate the common patterns of property price movements, while at the same time controlling for heterogeneity across countries or at different stages of real estate cycles panel approach gives more informative data, more variability, less collinearity among variables, more degrees of freedom and more efficiency following Capozza et al (2002) allowing for serial correlation and mean reversion as well as sensible long run variables in inverse demand function estimated as error correction model Estimate for three sub-periods namely the pre-liberalisation period before 1982, the first post liberalisation cycle over 1982-1997 and the second broad cycle over 1998-2013 and also with leveraged coefficients. Quarterly data is used for 11 countries experiencing booms in 1980s and 2000s, namely the UK, US, France, Canada, Italy, Spain, the Netherlands, Belgium, Ireland, Finland and Sweden.

Panel unit root tests (IPS) Log Real House prices (LRPH) Log RPDI (LRPDI) Real long rate (LRR) Log real liabilities (LRLIABS) Log real gross financial wealth (LRGW) Unemployment rate (U) Log real housing stock (LKH) 20-39 as a share of population (YOAD) Level -0.84 (0.2) -2.7 (0.00) -0.6 (0.27) 2.7 (0.99) 1.6 (0.94) Difference -9.45 (0.00) -14.5 (0.00) -17.7 (0.00) -8.5 (0.00) -13.8 (0.00) -2.19 (0.01) -2.44 (0.00) -6.5 (0.00) -10.7 (0.00) -1.16 (0.1) 0.79 (0.78)

Constant All Pre 1982 -0.77** 1982-1997 1998-2013 0.001 (0.1) 0.25** (2.1) 0.093 (1.2) (2.4) Log difference of RPDI Difference real long rate Log difference of house prices (-1) Log of house prices (-1) Log of RPDI(-1) Real long rate (-1) Population 20- 39 as share of total (-1) Log stock of housing (-1) Difference of unemployment rate Unemployment rate (-1) Log difference of real gross financial wealth Log of real gross financial wealth (-1) Dummy for banking crises Dummy for financial liberalisation Countries Obs Adjusted R2 SE of regression Durbin Watson Kao House prices equation for boom countries 0.17** (6.7) -0.00011 (0.2) 0.25** (3.3) 0.00099 (0.5) 0.15** (4.0) -0.00094 (1.0) 0.19** (5.5) -7.13E-05 (0.1) 0.56** (28.1) -0.0097** (4.7) -9.26E-05 (0.0) -0.0008** (4.0) 0.41** (7.1) -0.045** (2.4) 0.078 (1.5) -0.00071 (0.7) 0.53** (15.7) -0.034** (5.5) -0.0073 (0.5) -0.00073* (1.7) 0.54** (16.6) -0.013** (3.0) 0.043** (3.6) -0.00062 (0.7) 0.032 (1.4) -0.0054 (1.1) -0.61* (1.9) 0.027 (0.5) 0.12* (1.8) -0.081* (1.8) -0.031** (3.1) -0.037** (2.7) -0.0041** (3.5) -1.47E-05 (0.1) -0.0077* (1.7) -0.00052 (0.3) -0.0048** (2.9) -0.00074** -0.0027** (2.1) -0.00054* (1.8) (2.3) 0.053** (4.6) 0.052* (1.6) 0.07** (4.1) 0.038** (2.5) 0.008** (3.5) -0.0032** (2.7) -0.00089 (1.1) 0.028** (5.2) -0.0034** (2.2) -0.014** (2.2) -0.0038** (2.5) 0.00026 (0.2) 11 1612 0.5 0.16 10 275 0.38 0.02 11 687 0.53 0.16 11 650 0.6 0.011 2.13 2.09 2.09 2.11 -1.58 (0.06)* -1.85 (0.03)** -2.37 (0.01)** -2.54 (0.01)**

CoeffIcient T value Log difference of RPDI Leveraged coefficients for 1982- 1997 (in regression 1982-2013) 0.022 (0.4) Difference real long rate 0.0015 (0.8) Log difference of house prices (-1) -0.029 (0.6) Log of house prices (-1) 0.0087 (2.3)** Log of RPDI(-1) -0.0024 (0.8) Real long rate (-1) 0.0011 (1.2) Population 20-39 as share of total (-1) -0.051 (1.3) Log stock of housing (-1) 0.0031 (1.0) Difference of unemployment rate -0.0048 (2.2)** Unemployment rate (-1) 0.00041 (1.1) Log difference of real gross financial wealth 0.034 (1.4) Log of real gross financial wealth (-1) -0.00066 (0.3)

Leveraged coefficient for period 1985q1- 1989q4 and 2002q1- 2006q4 Leveraged coefficient for period 1985q1- 1989q4 0.22** Leveraged coefficient for period 2002q1- 2006q4 Extended house price equation, leveraged coefficients for booms 0.10* (1.9) -0.083 (1.0) 0.00099 (0.3) 0.015 (0.2) -0.015* (1.9) 0.0041 (1.1) -0.0023 (1.0) 0.064 (1.5) -0.0036 (0.9) 0.0032 (0.7) -0.00041 (0.5) -0.0078 (0.2) 0.00091 (0.1) Log difference of RPDI (3.3) Difference of real long rate Log difference of house prices (-1) Log of house prices (-1) 0.0036** (2.4) 0.099** (2.2) 0.0016 (0.6) 0.002 (0.8) -0.00012 (0.3) 0.048* (1.7) -0.0025 (1.4) 0.0032 (1.0) 0.00033 (1.1) 0.014 (0.6) -0.0018 (0.6) 0.0034** (2.0) 0.076 (1.5) 0.0047 (1.5) -0.00089 (0.2) -0.00048 (0.6) 0.047 (1.1) 0.00068 (0.2) -0.0016 (0.3) 0.00047 (1.3) 0.023 (0.9) -0.0072* (1.7) Log of RPDI(-1) Real long rate (-1) Population 20-39 as share of total (-1) Log stock of housing (-1) Difference of unemployment rate Unemployment rate (-1) Log difference of real gross financial wealth Log of real gross financial wealth (-1)

House prices and mortgage supply Mortgage market innovations altered terms and availability of credit, while securitisation removes the limit of capital less borrowing constraints Should mortgage stock enter house price equation clear case if rationing, less so in liberalised market as endogenous?

Lindner (2014); two alternative views of the link from asset prices to credit: Bernanke and Gertler (1989) and Kyotaki and Moore (1987) view that it is asset prices that drive credit availability via changes in the net worth of borrowers that in turn eases borrowing constraints in the presence of asymmetric information. This is consistent with the exclusion of credit from house price equations. Allen and Gale (2000) availability of credit is the more exogenous factor, with the key influence being risk shifting by lenders and borrowers in the presence of asymmetric information and limited liability, with consequent moral hazard. These may in turn be facilitated by financial deregulation. Lindner (2014) suggests that the net worth argument is most relevant to credit availability in general whereas risk shifting is appropriate for the financing of a particular asset such as housing by credit so mortgage stock is relevant. Credit spreads and conditions versus credit stock

Panel results boom countries adding debt variables Proxy for log difference of real household debt Log of real household debt(- 1) All Pre 1982 1982-1997 1998-2013 1982-2013 0.088** (9.1) 0.092** (11.0) 0.11** (4.9) 0.07** (5.5) 0.1** (6.4) -0.004* (1.7) -0.0022 (0.9) 0.0082 (0.2) -0.0047 (0.8) -0.0046 (0.8) Estimation period, 1982q1 to 2013q4 Leveraged coefficient for period 1985q1- 1989q4 and 2002q1- 2006q4 0.033933* Leveraged coefficient for period 1985q1- 1989q4 0.027821 Leveraged coefficient for period 2002q1- 2006q4 Log difference of real liabilities (proxy) Log real liabilities (- 1) -0.001399 (1.7) (1.3) (0.1) 0.000534** 0.000706** -0.000121 (3.6) (3.6) (0.7) Leveraged coefficient for periods 1990q1- 1994q4 and 2007q1- 2011q4 0.016327 Leveraged coefficient for period 2007q1- 2011q4 Leveraged coefficient for period 1990q1- 1994q4 Log difference of real liabilities (proxy) Log real liabilities (- 1) 0.061461* 0.045496 (1.3) -0.000159 (0.7) (1.8) -9.79E-05 (0.7) 0.000331* (1.6) (0.8)

Additional distinguishing factors Some structural differences between the 1980s and 2000s, common to a number of countries, that could underlie the differences and warrant further research Levels of debt and the relation to inflation. Interest rates and the impact of global liquidity. Patterns of securitisation. Changing patterns of owner occupation Population density

Conclusion Comparison of the booms of the 1980s and the 2000s shows a similar rise in real house prices where booms took place, and a marked rise in the real mortgage stock along with real incomes and real financial wealth. For the most part a similar econometric specification covers both periods. Some differences are that leverage rose far more in the later episode and did not contract in the aftermath. Also, the adjustment to the long run is slower in recent years, while the earlier boom period itself showed differences with average house price behaviour which was not mirrored in the most recent boom. The fact that the earlier cycle is in many ways comparable to the recent one despite not provoking a similar global financial crisis poses a challenge for the existing narrative claiming the housing boom was the unique and key determinant of the crisis. Further research is needed on factors potentially underlying the differences between the booms such as: the initial level of debt/income and the related impact of inflation, the impact of lower interest rates in the recent boom and global contagion via liquidity in the recent episode; the ready availability of credit from mortgage bond issuance; changing owner occupation rates. evolution of population density