A Housing Correction or a Housing Recession? Professor William Wheaton Department of Economics Center for Real Estate MIT January, 2007 MIT, IAP
Topics: 1 Historic Housing Prices: relation to income. Cycles caused by contracting economy. 2 Why this time is different. A true housing correction. 3 What is driving the market this time? 4 Estimates of the impact of housing on the economy: MPCW, Equity extraction, GDP Could there be a true housing induced recession? 5 Sector and Area issues: homebuilding, finance, durables.
All House Price are Local: tremendous US variation between MSA in long term growth 350 1980=100 (Constant $2005) 300 Boston 250 200 Los Angeles Chicago 150 Nation 100 Dallas 50 0 1980 1985 1990 1995 2000 2005
Last 7 years has seen phenomenal growth relative to the previous 18 1980Q1-1998Q4 (76 quarters) 1999Q1-2005Q4 (28 quarters) Market W Y P W Y P Atlanta 32% 49% 9% 9% 0% 21% Boston 42% 60% 74% 14% 11% 83% Chicago 23% 34% 11% 7% 1% 42% Dallas 26% 36% -26% 14% 3% 12% Denver 22% 37% 6% 16% 9% 33% Houston 27% 25% -38% 15% 7% 19% Los Angeles 31% 13% 10% 12% 6% 123% New York 49% 48% 56% 7% 6% 89% Philadelphia 27% 44% 21% 7% 8% 60% San Diego 13% 31% 1% 10% 17% 128% San Francisco 46% 51% 43% 16% 9% 93% Washington 26% 39% 6% 9% 15% 107% Sources: Bureau of the Census, Bureau of Labor Statistics, OFHEO. Variables: W - cumulative % change in real income/employment Y - cumulative % change in real income/population P - cumulative % change in real home price index
Cyclically: a prefect Historic correlation between job recessions and Housing Market construction except for the last 5 years Year-Over-Year Change in Employment Thousands 5,000 4,000 3,000 2,000 1,000 0-1,000-2,000-3,000 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 00 01 02 3,000 2,500 2,000 1,500 1,000 500 0 Total Housing Starts Thousands Year-Over-Year Change in Employment (R) Total Housing Starts (L)
When Home production soared to new levels despite the 2002-4 Recession 2.0 Units, Mil. 1.5 1.0 0.5 0.0 1970 1975 1980 1985 1990 1995 2000 2005 Single-Family Multi-Family
House Prices too mirror Employment growth: (except for 2001-2004) 8 % Change 6 4 2 0-2 -4-6 1971 1972 1973 1974 1975 1976 1977 1978 1979 1980 1981 1982 1983 1984 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 Employment Single-Family Stock Real Home Price
Lessons: Pre 2002 corrections 1 Permits: fall in half. Most speculative building stops, custom building (40% continues). 2 National Prices drop only 10% real, nominally barely at all. No one lives in the nation. 3 But, prices drop as high as 35% real and 20% nominal in coastal markets (NYC, California..) 4 Condo s more volatile. NYC (1989-1994) Single Family prices decline 25%, Condo s 50%. 5 But this time: Severe job contraction was from Stock Market rather than the Fed tightening credit to fight inflation. The Fed loosened. Are low rates the only explanation?
Back testing the Role of all fundamentals: income, jobs, rates Estimate well respected models using data from 1976 to 1998 for house prices in each of 59 MSA. Estimate 4 models: price levels and differences with and without lagged prices and AR1 corrections. Models have correct signs and plausible coefficients with respect to the three fundamental variables. Forecast forward with the model using actual fundamentals from 1998 through 2005. In every market model under forecasts. Errors range from 100+% (Florida, Arizona, California, D.C.) to 5% (St. Louis, Memphis).
Forecasts are perfectly understandable: poor job market offsets low rates yielding just slight real price growth 5.12 5.04 4.96 4.88 4.80 4.72 4.64 4.56 4.48 RHPI FORC1 FORC4 DENVER (#14) 4.40 1975 1977 1979 1981 1983 1985 1987 1989 1991 1993 1995 1997 1999 2001 2003 2005 0.20 0.15 DEMP4 MRTG DRINCE4 Economic data 0.10 0.05 0.00-0.05 1 975 1977 1979 1981 1983 1985 1987 1989 1991 1993 1995 1997 1999 2001 2003 2005
But in some markets actual prices are just totally at odds with the forecasts.
Back testing the Role of all fundamentals: What explains forecast errors? Errors range from nearly 100+% (Florida, Arizona, California, D.C.) to 5% (St. Louis, Memphis). Examine 2005 forecast errors and find two factors explain 65% of model error: 2 nd or Investment home buying. Greater Subprime mortgage lending activity.
Since 1998, total new housing supply exceeds total household formation by a record 2.5 Millions 2.0 1.5 1.0 0.5 1964 1968 1972 1976 1980 1984 1988 1992 1996 2000 2004 Source: Economy.com, Torto Wheaton Research. New Households Housing Starts
As Individuals Discover Real Estate and Gobble up the Excess Supply as Investment and 2 nd Homes (Condos excluded) Investment and 2nd Home Loans as Share of New Loans, % 0 10 20 30 40 50 Nation San Diego Sacramento Riverside Miami Tampa 1999 2005 Phoenix Orlando Las Vegas Fort Myers Atlantic City Source: Loan Performance, Torto Wheaton Research
Also Recent unprecedented and huge shift into home ownership from renting 70 Homeownership Rate, % Renter Households, Mil. 40 68 36 66 32 64 28 62 24 60 20 1965 1970 1975 1980 1985 1990 1995 2000 2005 Homeownership Rate, % (L) Age-Expected Homeownership Rate, % (L) Renter Households, Mil. (R)
Caused by explosive growth of the sub-prime Lending Market: loans with bad credit, high LTV, subordination 700 $ Percen 28 600 24 500 20 400 16 300 12 200 8 100 4 0 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 Subprime Loan Origiations Subprime as % of Total Originations Subprime Originations as % Total Mortgage Debt Outstanding 0
Without these 2 Wildcard factors, fundamentals outlook should yield a mild correction: +5% to -15% real 5.5 5.4 5.3 5.2 5.1 5.0 4.9 4.8 4.7 RHPI FORC1 FORC2 FORC4 DENVER (#14) 4.6 1975 1978 1981 1984 1987 1990 1993 1996 1999 2002 2005 2008 2011 2014 0.20 0.15 DEMP4 MRTG DRINCE4 Economic data 0.10 0.05 0.00-0.05 1975 1978 1981 1984 1987 1990 1993 1996 1999 2002 2005 2008 2011 2014
The UK experience. Rates rose (only 200bps) just enough to cool Housing. But the UK is not US. (variable/fixed, supply constraints, conservative lending) Halifax house price index, % change y/y 35 30 25 England s Housing Prices % Chg YOY Forecast 20 15 10 5 0 00 01 02 03 04 05 06 Source: Halifax PLC, Torto Wheaton Research
Assessing the impact of the 2 Wild Cards #1. Investor/2 nd homes represent net supply. A small amount of selling/buying has a huge impact on vacancy and prices - much more so that selling by primary home owners (churn) Will investor s head for the door at once (deep correction)? If they have high reservations = slow long drag on the market? #2. Recent first time home buyers are especially fragile with respect to interest rate or income shocks. What if 5 million recent buyers were foreclosed upon and were forced back into the rental market?
2 nd homes contribute to the greater volatility of condos relative to Single Family Homes: NYC 300 Real Condo Real Single Family Multi-Housing Permits (5+ Units) 30 250 25 200 20 150 15 100 10 50 5 0 1980 1985 1990 1995 2000 2005 0
Lesson from historic Florida condo boom (1970s) 5.5 5.4 5.3 RHPI FORC1 FORC4 MIAMI (#32) 5.2 5.1 5.0 4.9 4.8 4.7 Condo Boom No Construction, Strong Economy 4.6 1975 1977 1979 1981 1983 1985 1987 1989 1991 1993 1995 1997 1999 2001 2003 2005 0.20 0.16 DEMP4 MRTG DRINCE4 Economic data 0.12 0.08 0.04 0.00-0.04-0.08 1975 1977 1979 1981 1983 1985 1987 1989 1991 1993 1995 1997 1999 2001 2003 2005
Historic recessions: Sales - 30%, Inventory +60%, months supply nearly doubles. Current correction just beginning - but already there Months of supply on the market 9.0 7.0 5.0 3.0 Existing single-family (R) Existing condominiums (R) Sources: FRB, NAR
Recent Fannie Mae Study: Only conforming, conventional loans are insulated from Rate Rise! 80 Cumulative Share of Loans With Rate Resets, % 60 40 20 0 2002 2004 2006 2008 2010 2012 2014 Subprime Jumbo Alt-A PCC
Subprime Delinquency is beginning to look Ominous Mortgage Delinquency 16.0 14.0 12.0 10.0 Percent 8.0 Prime conventional loans Subprime conventional loans 6.0 4.0 2.0 0.0 1990 1995 1998 1999 2000 2001 2002 2003 2004 2005 Year
Determining the Marginal Propensity to consumer wealth (MPCW) Theory: MPCW = [interest rate] if wealth preserver. >< if borrower or saver. Change in wealth generates a permanent shift in consumption, then consumption growth resumes as before. Empirical work. Aggregate time series useless cannot disentangle correlated effects. (1996) Micro study (PSID). Housing gains: MPCW=.01, but housing losses MPCW=.30 (!) (2001) Panel study. US States, MPCW =.03 -.09, International (14 countries) MPCW =.10-.14
Take 2: MPCW = spending rate out of cash out borrowing. Puzzle: if consumption rate is constant, owners did not spend what they borrowed! ( last 5 Q?) Consumption rate, % share of disposable income 102 Existing single family house price, % change year ago 14 13 101 12 100 11 10 99 9 8 98 7 97 6 5 96 1999Q1 2000Q1 2001Q1 2002Q1 2003Q1 2004Q1 2005Q1 2006Q1 4 Source: OFHEO, BEA Consumption Rate (L) House Prices (R)
The Puzzle again: 2000-2006 Personal consumption rose from 7 to 9.2 trillion. Cumulative increase was 6.2 trillion This was very close to the increase in Personal Disposable income. Just 270 billion or so more (last 5 quarters). Mortgage debt increased 4.3 Trillion (from 5.0 to 9.3). Home equity increased 3.5 trillion (from 7.0 to 10.5 trillion, its share declined from 58 to 54%) Hence 4.3 Trillion in borrowing did NOT go into PCE! Only 270 billion did. = a MPCW of 270/7800=.030. Can Academics be right?
The Puzzle again: 2000-2006 4.0 Trillion of the 4.3 increase in debt must have gone back into Investment = asset purchases: back into bigger more expensive houses, major home expansions, 2 nd homes, real estate investments, Suppose we had an unprecedented 20% future drop in national Housing values (40% drop in housing wealth). Generates a 3% decline in consumption which we grow out of at the end of that year. GDP stagnates. If spread out over 2-3 years just a slowdown (GDP growth remains positive).
Sector Impacts: a pure housing correction 1). Home building/construction/materials - Likely to be more severe than traditional housing contractions since price decline will be greater. - If much of the wealth gain went back into real estate, much of the loss could go out! 2). Durable goods. - Likely to be less severe than traditional housing contractions since income and jobs not declining at the same time, this time. - Durable/nondurable ratio related to Housing? - Was the 270 billion cumulative increase in PCE targeted? Growth in home/electronics sales?
Which Markets are most exposed to a more serious Housing Slowdown? Location quotient (construction, real estate brokerage & real estate services), U.S.=1 HARD? West Palm Beach 1.81 Dayton 0.63 Fort Lauderdale 1.57 Philadelphia 0.69 Orange County 1.55 Memphis 0.72 San Francisco 1.53 Tulsa 0.72 Orlando 1.44 Cleveland 0.73 San Diego 1.44 Oklahoma City 0.74 Las Vegas 1.41 Detroit 0.75 Houston 1.39 Louisville 0.75 Oakland 1.37 El Paso 0.76 Sacramento 1.37 Greensboro 0.78 Jacksonville 1.31 Columbus 0.79 Washington 1.29 Greenville 0.84 Phoenix 1.27 Pittsburgh 0.84 Portland 1.27 Kansas City 0.85 Norfolk 1.22 Nashville 0.85 Seattle 1.22 Fort Worth 0.86 Riverside 1.21 New York 0.86 Miami 1.17 San Jose 0.88 Honolulu 1.16 Northern NJ 0.89 Albuquerque 1.14 Chicago 0.90 SOFT? Source: TWR, Economy.com, BLS