Home Prices Over the Long Run

Transcription

1 Home Prices Over the Long Run Jacques Friggit, French Ministry of the Environment and Housing WEAI 84th Annual Conference, Vancouver, June 29 July 3, 2009

2 Who I am A civil servant with the French ministry in charge of (among many other things) housing Working with the internal audit department (CGEDD, formerly CGPC) We audit policies (and organizations) ex-ante and ex-post We developed audit tools, among them a long term vision We provide our long term tools and analyses to our colleagues and to the general public: since Dec 2008 on earlier, on (ADEF is a non-profit organization, independent from the government; although I published on its website and in its journal, I am not a member of ADEF) including a page and some documents in English My work on home prices in the long run began in the late 1990 s Paris and France indices published in Paris and France indices published in 2001

3 The two papers presented here «Comparing Four Secular Home Price Indices», February This working paper compares secular home price indices in four countries, one of which goes back to the 13th century. It stresses that their long term trend should be interpreted with caution. ( «Long term ( ) Investment in Gold, Bonds, Stocks and Housing in France - with Insights into the USA and the UK : a Few Regularities», January 2007 (1st version: mid-2006). This working paper comments on the evolution of asset prices for the past two centuries. Its exhibit 1 compares French and British home prices in the long term. Its exhibit 2 comments on French American cross-border investments in the past two centuries, in terms of return, volatility and diversifying power. ( I ll try here to focus more on the results than on the methodology

4 Methodology (=mostly the 1st paper) Long run home price indices are very sensitive to: their perimeter (ex: did its social status go up, down?) => a «Paris index» is not a «France» index, a «Herengracht index» is not a «Netherlands index» nor even an «Amsterdam index» their methodology (may impact an index by around 1% per year or more) Usually you don t choose the perimeter (it is determined by your database) and you try to fit the methodology to the type of data as best as you can Secular indices usually are based on repeat sales methods (least bad choice or, often, the only possibility) but repeat sales methods are not perfect and there are several variants, ex: How do you take into account changes in quality of a given building over time (obsolescence, better comfort)? What is quality? (includes or not the quality of the neighborhood) (and which measure of quality does the database provide) - A method fit to one neighborhood, one type of buildings may not be to other ones (ex: Haussmanian Paris buildings versus 13th century Paris wooden houses or versus American wooden suburbian houses)

5 Our record: d Avenel s Paris home price index but fragile: - quality effects identified but treated in an undisclosed way (not repeat sales), - transformation of the area ( center of Paris ), - «CPI», etc. although d Avenel was a very intelligent person 10 1 Herengracht Paris before adjustment for obsolescence Paris after adjustment for obsolescence Centre of Paris before adjustment for obsolescence Centre of Paris (d'avenel) Norway main cities USA before adjustment for depreciation USA after adjustment for depreciation Home price indices in constant local currency Basis 1970= ,1 0,01

6 A common feature of the four secular home price indices compared They grow at various paces, but at most like income per household, more often like the «CPI» The differences in trend growth stem in large part at least from methodogical differences

7 Results (mostly the 2nd paper) for France with comparative insights into the USA and the UK (updated charts) Home price index (Paris and France) relative to disposable income per household (France) , Q , Q ,9 0,8 0,7 0,6 0, Tunnel ,4 0,3 0,2 Home price index (France and Paris) relative to disposable income per French household Basis 1965= France Paris NB: the divider of both ratios is the disposable income per household over all of France 0,

8 The «useful» period for extracting «historical regularities» possibly pertinent to anticipate the future begins in 1965 in France 1,8 1,7 1,6 Home price index and amount of transactions relative to their long term trend (*) All of France, basis 1965= (Q1 09) 1,5 1,4 1,3 Home price index relative to disposable income per household, basis 1965=1 Auxtunnel 0 12 month cumulated total amount of existing-home transactions relative to its long term trend (*) Auxtunnel 0,9 1,2 1,1 1, ,9 0,9 Tunnel 1.07 (May 2009) 0,

9 Comparison with the USA and the UK (France seems to lag the USA and the UK by 1 to 2 years) 2 1,5 International comparison: home price index relative to disposable income per household Basis 2000=1 France USA (FHFA index on the recent past) UK (DCLG) UK (Halifax) Aux 0,9 Aux 1,1 1,1 1 0,9 0,

10 Main properties of the home price index since 1965 Mean return process around trend parallel to income per household (or equivalent) Consistent with link in space between home price and income per household (but less obvious than it seems) Average dwelling price (euros) Average dwelling price as a function of the average gross taxable income per household Paris, per "arrondissement", y = 4,4084x R 2 = 0, Average gross taxable income (euros)

11 -1,0 1,0 Main properties of the home price index since 1965 (foll.) High autocorrelation of 1 year returns Autocorrelation of home price index returns, , as a function of the time interval on which returns are calculated Autocorrélation des variations du prix des logements ,5 0,0 Pas Time de temps interval (années) (years) ,5 Monnaie Nominal courante currency Monnaie Constant constante currency In units of disposable income per household En unités de revenu disponible par ménage

12 Main properties of the home price index since 1965 (foll.) Beyond that, little link with: «Supply / demand» variables (i.e. size of the housing stock / population or number of households) Interest rates Counter to intuition but a fact Tools: - multiple regressions of returns (or residuals of 1-year autoregression), with ou without lags, on various time intervals, with respect to various aggregates - Box-Jenkins (ongoing, to be formalized) - State spaces (SAS proc ucm) (to be done, but little prospect of change in the conclusion) - Dependence with respect to «liquidity», or lenders «propensity to lend», not tested because I don t have a data series representative of these concepts Maybe explains France s 1 to 2 year lag with respect to the USA and the UK Link with the other papers presented in this session?

13 Illustration (beginner s level): no coincidence between changes in home prices and changes in interest rates 60% Taux Long d'intérêt term interest à long terme rate 1,75 Inflation Taux Long d'intérêt term à interest long terme rate net minus d'inflation inflation Prix Home des price logements index rapporté relative au to revenu disposable disponible income par ménage, per household, base 2000=1 basis (échelle 2000=1 de droite) (right scale) 50% 1,5 40% Up Down Up Down Up Down Up Hausse Baisse Hausse Baisse Hausse Baisse Hausse 1,25 30% 1 20% 0,75 10% 0,5 0% ,25

14 We also have data on the amount of real estate transactions 14% 12% 10% 1844 Total amount of property transactions as a % of gross domestic product Transactions of any type of property (residential and commercial, old and new, incl. land) subject to transaction tax, cumulated on 12 months 1990 May 2009: 10.6% 8% Total Of w hich not taxed at 0.60% Auxabscisse 1980 May 2009: 7.6% 6% % % %

15 We also have data on the value of an investment in stocks (dividends reinvested) Mar Apr Jul Mar Dec Mar Nov Nov Sep Jul Jan Value of investments in French and US stocks relative to the long term trend of US stocks. End-of-month values Jan ,9 0,8 0,7 0,6 0,5 Nov , Dec French stocks US stocks 0,5 0,4 0,6 0,7 Jun ,8 0, Mar Dec Oct Jun Jun Nov Dec May 1926 Sep Jun Sep Feb Apr Jun Jun Aug Aug Dec May 31, 2009: 0.68 May 31, 2009:

16 Investment in housing (in Paris) compared to other investments Return / volatility as proxies for return / risk couples Return / volatility couples are mostly consistent, once one factors out specific phenomena The following charts are for French investors (i.e., investors whose currency is the French one) Excluding taxes on income and capital gains but including property taxes

17 Inflation 100% 80% 5 year return volatility % 40% 20% US stocks French stocks Housing Paris 0% Gold French bonds French money market Yearly return 0% 5% 10% 15% 20% 25%

18 Inflation 100% 80% 5 year return volatility Gold % French stocks US stocks 40% Housing Paris 20% 0% French bonds French money market Yearly return 0% 5% 10% 15% 20% 25%

19 Inflation 100% 80% 5 year return volatility % Gold US stocks 40% French stocks Housing Paris 20% French bonds 0% French money market Yearly return 0% 5% 10% 15% 20% 25%

20 Inflation 00% 80% 5 year return volatility Trend (assumption: inflation 2%) 60% US stocks 40% French stocks Housing Paris 20% French Bonds 0% French money market Gold?? Yearly return 0% 5% 10% 15% 20% 25%

21 Investment in housing (in Paris) compared to other investments Not visible on charts: (physical) housing diversifies portfolios out of stocks (intuitive) and bonds (linked to absence of significant correlation with interest rates)

22 CONCLUSION Having a long term perspective helps spot risks The data series are public, available to anybody wishing to explore their properties, but be careful about their limitations (not fit for any use) be careful about the historical context (ex: the very special period in France)