Housing Prices: Global Correction, Local Consequences Presentation by: Prakash Loungani This presentation draws on work-in-progress with my IMF colleagues Charles Collyns, Marcello Estevao and Deniz Igan. The views presented here are those of the authors and should not be attributed to the IMF. We thank Jair Rodriguez for excellent research assistance and Hites Ahir for helpful discussions. George Washington University Research Program in Forecasting November 18, 2008

outline 1. a global boom and bust 2. how much further will house prices fall? consequences of house price declines a. adjustment of residential investment in OECD countries b. adjustment of unemployment in U.S. states

data sets selected OECD countries, quarterly data,1970 to present (shorter time series in some cases) ’51’ U.S. states, quarterly data,1975 to present

oecd countries: prices & quantities 1. global boom and bust oecd countries: prices & quantities

a global boom in OECD countries ...

... is now a global bust

magnitude depends on index: U.S. example

Residential investment has been plummeting in the U.S.

globally, a residential investment boom ...

... is turning to bust

prices and quantities: up together ... Change in Housing Prices and Residential Investment 00-06 USA GBR CHE SWE ESP NOR ITA IRE DEU FRA FIN DNK CAN -40 -20 20 40 60 80 100 120 Residential Investment Housing prices

prices and quantities: down together ... Change in Housing Prices and Residential Investment 07-08 USA GBR CHE SWE ESP NOR ITA IRE DEU FRA FIN DNK CAN -7 -6 -5 -4 -3 -2 -1 1 2 3 -35 -30 -25 -20 -15 -10 5 10 Residential Investment Housing prices

2. how much further will house prices fall? evidence from past cycles evidence from short-run model evidence from long-term relationships a more careful look for U.S. (based on work by Deniz Igan)

Evidence from past cycles OECD (2005) Claessens, Kose and Terrones (2008)

evidence from short-run model

price-income ratios are above historical values ...

... suggesting price gaps

price/rent ratios are also above historical values ...

... suggesting price gaps

U.S. price-to-rent ratio (PRR)

housing affordability, then and now

price-to-rent ratio: adjustment to benchmark

3. consequences residential investment: OECD countries unemployment: U.S. states (based on Estevao and Loungani)

3a. residential investment What explains the cross-country variation in the impact of house price busts on residential investment?

cross-country variation in residential investment impacts of housing busts: role of interest payment to GDP ratio

cross-country variation in residential investment impacts of housing busts: role of mortgage debt- to-GDP ratio

cross-country variation in residential investment impacts of housing busts: role of mortgage market characteristics

3b. unemployment What explains the cross-state and cross-region variation in the impacts of house price busts unemployment?

U.S. state unemployment rates at present

Real Housing Prices and Unemployment Rates in the U.S. Regions 1995Q1 1981Q4 1979Q1 1988Q2 2 4 6 8 10 12 14 Jan-76 Jan-79 Jan-82 Jan-85 Jan-88 Jan-91 Jan-94 Jan-97 Jan-00 Jan-03 Jan-06 80 120 160 200 240 280 320 360 Housing Bust Unemployment recession cycle Unemployment rate (LHS) Smoothed unemployment (LHS) Previous peak Trough Real house price index (RHS) (In percent) (Index) New England East North Central 1982Q4 1979Q2 Middle Atlantic Real Housing Prices and Unemployment Rates in the U.S. Regions Note: Real housing prices defined as the OFHEO home price index in a census division divided by the CPI of the larger census region. Sources: Office of Federal Housing Enterprise Oversight (OFHEO); Bureau of Labor Statistics; and Haver Analytics.

Real Housing Proces and Unemployment Rates in the U.S. Regions (cont.) 1979Q1 1989Q3 1993Q4 1990Q4 1981Q3 1995Q1 2 4 6 8 10 12 Jan-76 Jan-79 Jan-82 Jan-85 Jan-88 Jan-91 Jan-94 Jan-97 Jan-00 Jan-03 Jan-06 80 120 160 200 240 280 320 360 1976Q1 1982Q1 1985Q1 1988Q1 1991Q1 1994Q1 1997Q1 2000Q1 2003Q1 2006Q1 Unemployment: recession cycle Unemployment rate (LHS) Smoothed unemployment (LHS) Previous peak Trough Real house price index (RHS) South Atlantic (In percent) (Index) 1999Q1 2000Q2 1981Q4 14 East South Central 1988Q2 1982Q3 1979Q2 Housing Bust Unemployment recession cycle West North Central Note: Real housing prices defined as the OFHEO home price index in a census division divided by the CPI of the larger census region. Sources: Office of Federal Housing Enterprise Oversight (OFHEO); Bureau of Labor Statistics; and Haver Analytics. Real Housing Proces and Unemployment Rates in the U.S. Regions (cont.)

Real Housing Prices and Unemployment Rates in the U.S. Regions (cont.) 1983Q1 1987Q1 1990Q4 1985Q4 2 4 6 8 10 12 Jan-76 Jan-79 Jan-82 Jan-85 Jan-88 Jan-91 Jan-94 Jan-97 Jan-00 Jan-03 Jan-06 80 120 160 200 240 280 320 360 1976Q1 1979Q1 1982Q1 1985Q1 1988Q1 1991Q1 1994Q1 1997Q1 2000Q1 2003Q1 2006Q1 Unemployment: recession cycle Unemployment rate (LHS) Smoothed unemployment (LHS) Previous peak Trough Real house price index (RHS) Mountain (In percent) (Index) 1981Q1 Housing Bust Unemployment recession cycle West South Central 1982Q3 1981Q2 1995Q1 1989Q4 Pacific Note: Real housing prices defined as the OFHEO home price index in a census division divided by the CPI of the larger census region. Sources: Office of Federal Housing Enterprise Oversight (OFHEO); Bureau of Labor Statistics; and Haver Analytics. Real Housing Prices and Unemployment Rates in the U.S. Regions (cont.)

U.S. Regional Recessions 4 5 6 7 8 9 10 -12M Onset 12M 24M 36M 48M Unemployment rate (percent) Average for regional recessions without a housing bust Average for regional recessions with a housing bust U.S. Regional Recessions

Regional Housing Busts and Unemployment Rates Note: Events are centered at a peak in the real housing prices that was followed by a bust in one of the nine Census divisions. Real housing prices defined as the OFHEO home price index in a Census division divided by the CPI of the larger Census region. Sources: Office of Federal Housing Enterprise Oversight (OFHEO); Bureau of Labor Statistics; and Haver Analytics. -50 -40 -30 -20 -10 10 20 30 -16Q -12Q -8Q -4Q 0Q Beginning of cycle 4Q 8Q 12Q 16Q 20Q 24Q 28Q 32Q 36Q 40Q 44Q 5 15 25 35 40 45 50 New England: 1988q2 - 1995q1 Middle Atlantic: 1988q2 - 1995q1 East North Central: 1979q2 - 1982q4 West North Central: 1979q2 - 1982q3 West South Central: 1983q1 - 1990q4 Pacific: 1989q4 - 1995q1 Census division : Beginning to trough Real House price index ( cumulative percent change vis - à vis beginning of housing bust) 2 4 6 8 12 14 16 Unemployment rate (around housing busts) Housing busts: Census division: Beginning to trough

Census division Start 1/ Trough 2/ End 3/ Contraction 4/ Recovery 5/ Complete Cycle 6/ New England Jul-81 Jun-82 Jan-84 11 19 30 Apr-02 Apr-03 Mar-07 12 50 62 Middle Atlantic Sep-81 Jan-83 Dec-84 16 23 39 Nov-01 Dec-02 Jan-05 13 25 38 East North Central Feb-91 May-92 Apr-94 15 Oct-01 Jul-03 May-06 21 34 55 West North Central Oct-90 Mar-94 5 37 42 South Atlantic Feb-81 Dec-82 Jan-85 22 47 Nov-90 Mar-92 Dec-94 33 49 Aug-01 Sep-05 26 East South Central Jul-80 Jun-87 53 83 Jul-91 Jul-93 24 29 Sep-01 Aug-03 Sep-06 60 West South Central Mar-82 Jul-83 May-84 10 Jun-92 Jul-94 36 Jun-03 Apr-06 Mountain Dec-81 Dec-83 Jan-87 Jun-88 17 41 Sep-91 Sep-93 8 Mar-03 Dec-05 52 Pacific May-85 28 44 May-03 Feb-05 40 Average Apr-90 Nov-91 Mar-96 71 Jan-91 Aug-96 67 May-80 Nov-82 Aug-86 45 75 Sep-87 58 88 Jul-86 Mar-89 32 51 Mar-91 Nov-92 Dec-96 20 69 70 2/ Period with highest unemployment rate during the cycle. 3/ Actual unemployment rate falls below smoothed unemployment rate. 4/ Number of months from the beginning of the cycle to the trough. 5/ Number of months from the trough to the end of the cycle (recovery). 6/ Amplitude of total cycle (contraction + recovery). Length of Regional Unemployment cycles. 1/ Point where actual unemployment rate is higher than smoothed unemployment rate. Unemployment rate was smoothed using the HP filter. Unemployment cycles (excluding cycles occurring around housing busts) Unemployment cycles ocurring around housing busts (In months)

Number of observations 28 Dependent variable: (1) (2) (3) (4) (5) (6) ? rhp -1 0.52 0.19 0.33 -0.04 (0.59) (0.58) (0.47) (0.42) rin -3.69 ** -2.57 * -3.57 -2.88 (1.27) (0.92) bust -4.35 -2.93 -3.30 -1.90 (1.12) (1.22) (0.88) -0.31 0.31 0.69 1.24 (2.23) (2.07) (1.60) (1.49) Number of observations 28 Adjusted R 2 0.35 0.43 0.34 0.41 0.51 = percent change in real housing prices in the 12 months leading to the recession. = percent change in real regional income in the 12 months leading to the recession. Estimating the Effect of Past Changes in Real Housing Prices on Recession and Recovery Length using U.S. Regional Data. Recession length (number of months) Recovery length (number of months) Real housing price and regional income respectively measured as the OFHEO home price index and personal income in the division divided by the CPI of the larger census region. There are 9 Census divisions and 4 census regions. Figures in parenthesis are standard errors. * means 5 percent level of significance and ** means 1 percent level of significance.

Number of observations 28 Dependent variable: (1) (2) (3) (4) (5) (6) ? rhp -1 -0.57 -0.28 0.49 -0.39 (0.56) (0.51) (0.43) (0.38) rin -5.73 ** -4.79 -4.15 -3.24 (1.33) (1.53) (1.00) (1.13) ( ) 2 0.28 * 0.12 0.23 0.10 (0.11) (.08) (0.08) 0.62 0.50 0.26 0.13 (0.30) (0.33) (0.23) (0.25) Number of observations 28 Adjusted R 0.18 0.40 0.39 0.21 0.43 0.45 = percent change in real housing prices in the 12 months leading to the recession. = percent change in real regional income in the 12 months leading to the recession. Estimating the Effect of Past Changes in Real Housing Prices on Recession and Recovery Length using U.S. Regional Data. Recession length (number of months) Recovery length (number of months) Real housing price and regional income respectively measured as the OFHEO home price index and personal income in the division divided by the CPI of the larger census region. There are 9 Census divisions and 4 census regions. Figures in parenthesis are standard errors. * means 5 percent level of significance and ** means 1 percent level of significance.

Housing bust magnifies impact of recessions on unemployment

Shock to employment Response House Prices Response Employment Response unemployment Response labor force participation

Shock to house prices Response House Prices Response Employment Response unemployment Response labor force participation