Detail from Antique Pipe Rack, kelvinko.com
For US home prices, there was greed, then fear, and now … the loathing.
The ‘greed’ era ran from 2001 through 2006; followed by a brief bout of ‘fear’ in 2007 and 2008. Now it is 2009, and time for the ‘loathing.’
For about 15 of the past 22 years (1987 – 2008), home prices followed income. This relationship broke down in 2001, and we are now living through a correction of a surge that peaked in 2006.
This paper will look at national home price behavior (using the RPX 25 MSA Composite, available at radarlogic.com) and briefly:
- Review recent home prices and their relationship to income;
- Examine alternative descriptive cycles of home price corrections; and
- Estimate when the current cycle will give back all of the post 2001 gains in real home prices.
I will then develop an informal “Greed, Fear, and Loathing” model of national post-2001 home prices, using three factors:
- Greed: The extent to which home prices exceeded those consistent with relatively conservative mortgage underwriting practices;
- Fear: The unemployment rate; and
- Loathing: Estimates of national foreclosure activity.
The model will then be used to estimate what would be required to halt the current home price decline.
The national data used for the above is generally available on the web. I believe that the approach could also be applied regionally at either a state or MSA level, using proprietary sources/subscription services. If this interests you … well, I’m available.
Recent Home Price History, and Relationship To Income
Below is a chart of three relatively broad home price indexes for the last 20 years or so:
- S&P Case Shiller 10 MSA [1987 – 2008];
- RPX/Radarlogic 25 MSA [2000 – 2008, backfilled with S&P CS 10 MSA prior to 2000]; and
- S&P Case Shiller National [1987 – 2008].
Figure 1: Home Price Indices, 1987 – 2008
Until the 21st century, US home prices could largely be “explained” by simple fundamental factors, such as household income.
As indicated by the top chart in Figure 2, below, between 1987 and 2001, home prices and income generally moved together, in relatively long seven-year cycles. Beginning in 2001, however, things changed.
Figure 2: Home Prices and Income, 1987 – 2008, Nominal [top] and Real [bottom]
As the bottom chart of inflation-adjusted home prices and income demonstrates, even as real household income meandered along a relatively flat path, home prices exploded after 2001. This was due, in part, to:
- A steep yield curve;
- The widespread use of ARMS;
- Flexible mortgage underwriting standards; and
- Mortgage product innovations (subprime, Alt-A, Option ARMs).
All of the above encouraged home ownership. By 2006, prices had peaked, and began to correct.
Corrections in Real Home Prices
Figure 3: Inflation-Adjusted RPX Composite Home Prices – Up Cycle
The real peak in home prices occurred (using the RPX composite) by mid-year 2006, as noted by Figure 3, above.
If you looked at US 1980 - 2005 data (that is, data that excluded the correction beginning in 2006), you might conclude that once inflation-adjusted home prices reached their peak, they would give up about 32% of the increase that occurred in the five years prior to the peak.
If you looked at similar international home price performance, you might suggest that real home prices would give up 70% of the increase that occurred in the five years prior to the peak.
Figure 4: Alternative Real Home Price Corrections, Rules of Thumb
E.L. Glaeser & J. Gyourko, NBER, Housing Dynamics, WP12787, Dec 2006.
Morgan Kelly, UCDublin - On the Likely Extent of Falls in Irish House Prices, WP07/01, Feb 2007.
Figure 4, above, summarizes these rules of thumb, and provides the names of the economists who developed these “rules of thumb”. I have also included similar figures for a “100% give back” – or complete reversion – in home prices.
Figure 5: Alternative Real (RPX Composite) Home Price Declines & Benchmarks
If we apply each of these cycles to real RPX prices, for the cycle that peaked in mid-2006, we find (see Figure 5, above) that the:
- 32% correction (Glaeser-Gyourko) occurred in Nov 2007;
- 70% correction (Kelly) occurred in Oct 2008; and
- 100% correction will occur once real home prices drop about 10% from their year-end 2008 value.
If we assume that real home prices, and income, continue to fall at the same pace as they declined in 2008, then the 100% give back (of the 2001 – 2006 increase) will be reached (as indicated in Figure 6, below) by mid-year 2009.
This would also be, more-or-less, about the same time that real home prices again “converge” with real household income. Since 1987, for what it’s worth, real home prices and household income have converged every six-to-eight years or so. What’s different this time is the magnitude of the last cyclical increase – NOT the timing.
Figure 6: Estimated Timing of 100% Home Price Correction & Relationship To Household Income
It’s unrealistic to hope that real home prices will begin to recover as soon as home prices achieve “income parity” by mid-year 2009.
As Figure 6 indicates, during the post-1990 real home price correction, home prices lagged income for six-to-eight years. Something similar will probably occur this cycle as well.
Simple Model of National Home Prices: Greed, Fear, and Loathing
In order to see what might prompt a national home price recovery, I estimated a simple model of national home prices using monthly 2001 – 2008 data, based on:
- The difference between actual prices and those consistent with conservative underwriting [greed, in excess of affordability];
- Unemployment [fear]; and
- Foreclosure activity [loathing].
Below is a brief description of each of these factors.
Greed, In Excess of Affordability
Every month, the Federal Home Finance Board (www.fhfb.gov) releases its “Monthly Interest Rate Survey.”
This survey provides a reasonably comprehensive national snapshot of the terms (average rate, LTV, original term) on conventional mortgages used to purchase single-family residential homes. The survey also reports the average purchase price of the homes financed by these mortgages.
I combine the FHFB data with household income data available from the BLS to produce the price that would be consistent for a conservatively underwritten mortgage loan. After a couple of adjustments, I then compare this “conservatively underwritten” price with the ACTUAL purchase price reported by the FHFB.
My “Greed, In Excess of Affordability” factor equals the difference between the “conservatively underwritten” and the actual price:
- If actual prices exceed the “conservatively underwritten” price, then my “greed factor” is negative – meaning prices “should” fall.
- If actual prices are less than the “conservatively underwritten” price, then my “greed factor” is positive – meaning prices “should” rise.
- The factor is denominated, in thousands of dollars, so if actual prices exceed the conservatively underwritten price by $50,000; then the factor is –50.
Below is a chart (Figure 7) showing the relationship between my “greed” (i.e., affordability) factor and the annual change in the RPX composite.
A close examination of the graph reveals that the factor is currently “neutral” – meaning that prices and rates have fallen so low that a simple “affordability analysis” would suggest that home prices were “fair.” Equivalently (see Figure 6, above), national home prices are – more or less – just slightly above that consistent with household income.
This suggests that – to the extent that home prices continue to fall – this is due to a factor other than an absence of greed (i.e., homes ARE affordable). Must mean that it’s time for fear.
Figure 7: Home Prices and Greed (Affordability) Factor
In order to capture the “fear” effect, I will use the unemployment rate. Specifically, the “total unemployment rate” released by the BLS that covers people:
- Unemployed and currently looking for full time employment; plus
- Those employed in part time work who would prefer to work full time; plus
- “Discouraged” workers, who are out-of-work and have given up looking for a new job.
Naturally, the “total unemployment” is higher, and about twice the size of, the more widely reported “headline” unemployment rate.
- Note: For a great read on the most recent BLS release, see the PaperEconomy post of April 9, 2009 - On The Margin: Total Unemployment March 2009.
Below is a chart ( Figure 8 ) that depicts the relationship between the “total unemployment” unemployment rate and home prices.
Figure 8: Home Prices and Fear Factor (Unemployment)
As Mark Hanson of the Field Check Group (www.fieldcheckgroup.com) notes in his detailed and precise analysis of California foreclosures:
- The …massive pool of properties in the foreclosure process most influence reported median house prices. Therefore, [real-time] tracking … of properties throughout all stages of the foreclosure process should be the best indicator of future reported house prices near and long-term.
Field Check’s early April report carefully peers into the future:
- But coming down the pipe is a massive wave of [foreclosure activity] … that dwarfs the highs we saw when things began to get really out of control in 2008Q1. In the month of March new loan defaults hit a record high…
- There is a …foreclosure wave bearing down … [that] will result in ... [a massive increase in] foreclosures [that will be] … reported in May.
If one looks at the broadly defined nationwide foreclosure data from the MBA, one can see (Figure 9, below) the relationship between nationwide home prices and foreclosures, noted above by Field Check:
Figure 9: Home Prices and The Loathing Factor (Foreclosures)
- Note: The year-end dip in MBA foreclosure starts (see Figure 9, above) does NOT represent any sort of fundamental improvement in the macroeconomy, it is an artifact of the legislative and executive jawboning that temporarily convinced servicers to delay new foreclosures, as the following chart (see Figure 10, below) indicates.
- Believe that other proprietary data sources, such as RadarLogic and Altos Research, could be used to get a realistic measure of distress, and motivated sales, on both a national, and MSA-specific basis.
Figure 10: National 90-Day Delinquency Rates and Foreclosures
Home Price Model: Greed, Fear, and Loathing
The factors can be combined to create a simple monthly model that relates the annual change in RPX home prices for 2001 – 2008 to each of the three factors. All of the factors (greed/affordability, fear/unemployment, and loathing/foreclosures) are statistically significant and the coefficients have the expected signs.
Figure 11: Actual and Predicted Change in Home Prices (RPX Composite)
While popular attention is now turning towards the possibility that banks might actually make money, believe that a more important – and fundamental - question would be the following:
- If the relationships that prevailed between 2001 and 2008 persist, what must happen to affordability, unemployment, and foreclosures to halt the home price decline?
Assuming that all three factors adjust proportionately, each factor would have to drop by roughly half from their current levels.
In terms of the most familiar “headline unemployment” figure (8.5% in March, according to the 3 Apr BLS release), this means that unemployment would need to drop to about 4.3%, a level last seen in 2001.
Don’t think this will happen anytime soon? Then hold off on your celebration for the housing market recovery.