Sunday, November 25, 2007

Housing Bubble Snapshot

-1990 episode was fairly concentrated, with some markets 'stalling' while others dipped.
-2007 episode, so far, has been both more uniform and mixed. More markets have dipped, but some have also leveled out this year.

Table 1. Peak-to-Trough comparison, 1990 Real Estate "Bust" to 2007 Real Estate "Adjustment". (markets that started a decline prior to December 1994).

Los AngelesJun-90Mar-9669-27%Sep-06Aug-0711-6%
San DiegoJul-90Mar-9668-17%Nov-05Aug-0721-9%
New YorkSep-88Apr-9131-15%Jun-06Aug-0714-4%
San FranciscoJun-90Feb-9444-12%May-06Aug-0715-5%
Dallas - TXOct-89Feb-9452-8%Jun-07Jul-0710%
Seattle - WAJul-94Dec-945-1%Jul-07Aug-0710%
Minneapolis - MNNov-94Dec-941-1%Sep-06Aug-0711-4%
Tampa - FLSep-94Nov-9420%Jul-06Aug-0713-11%
Cleveland - OHNov-94Dec-9410%Jul-06Apr-079-5%
Portland - ORNov-94Dec-9410%Jul-07Aug-0710%
Miamino drop--Dec-06Aug-078-9%
Phoenix - AZno drop--Jun-06Aug-0714-8%
Las Vegasno drop--Aug-06Aug-0712-8%
Denverno drop--Aug-06Mar-077-4%
Detroit - MIno drop--Dec-05Jun-0718-13%
Atlanta - GAno drop--Jul-07Aug-0710%
Charlotte - NCno drop--Aug-07Aug-0700%

*items shaded lightly indicate markets in which a downward trend started, but has stopped, either as a permanent trend reversal or as a hiccup.

Seven markets really entered into price declines in the 1990 episode. Housing starts bottomed in January, 1991 (below 1 million units). The Fed raised rates in 1994, when the other markets listed slowed.

Only the three California markets had long, slow declines. The New York and Boston markets peaked first and actually started up circa 1991.

Prices did "bounce" a bit after the falloff that started in the early 1990s, mostly. Not surprisingly, the market with the steepest per-period run-up, Los Angeles, also had the largest decline. An 18-month, post-correction bounce trims the peak-to-trough loss to 22%, however. Four of the other markets came around 12%, with the last two closer to 5%.

Annualized, all of the price declines come out to 4-6% (see next table).

Table 2. Peak-to-Peak comparisons and rise at annual rates

to now
Phoenix - AZ--Jun-061389%11%9%
Tampa - FL20%Jul-061409%11%9%
Las Vegas--Aug-061409%10%8%
Seattle - WA5-1%Jul-071518%9%9%
Minneapolis - MN1-1%Sep-061417%8%7%
San Diego68-3%Nov-0511614%7%6%
Portland - OR10%Jul-071516%7%7%
San Francisco44-4%May-0614710%7%6%
Los Angeles69-5%Sep-0612613%6%6%
New York31-6%Jun-061827%5%5%
Detroit - MI--Dec-051324%5%3%
Atlanta - GA--Jul-071514%5%5%
Charlotte - NC--Aug-071523%4%4%
Cleveland - OH10%Jul-061393%4%3%
Dallas - TX52-2%Jun-071604%2%2%

The current episode, on closer inspection, also has some concentrations, if one considers the peak-to-peak growth in prices: Miami, Pheonix, Tampa, Las Vegas. Assuming that a "normal" growth in prices might be in the range of 3-7% for most regional employment and income dynamics, the price drop need to bring returns back to the averages. Over the next three years, Miami and Tampa, for instance, might need further price drops in the 15-20% range to get back to "normal" returns. Three of the other markets are in the 10-15% range. The remainder could stay flat and returns would be be "normalized" either right now or in the next 2 years.

Altogether, without a rising rate environment and based solely on "normal" levels of return, it doesn't look like a housing crisis is in the works, except in some markets that will correct further.

Of course, a recession (weak income dynamics) could force price falls to be less gradual and have greater amplitude, but those would be compounding cyclical factors, not "bubble" factors. However, many of these factors were also present in the 1990s slump.

In fact, a similar fall in prices, peak-to-trough, comes up with a duration of the episode about equal to the prior one, passing from peak-to-trendline-growth in just about 3 years, or sometime in 2009.

Chart1. House prices on a log-level scale. The current downturn could very much look like the prior one - down somewhat, but not sharply, followed by a period of flat, if there is an underlying uptrend in prices, as shown.
Free Image Hosting at

The only way to really scare yourself is to imagine that the trendline re-based at the end of the 1990s, say, as a result of the end of the large decline in mortgage rates to a new, lower level. Such a re-basing suggests that most of the "gains" in the 2000s might have been not driven by the fundamentals.

Re-basing, however, is not supported by an economic model of the fundamentals driving the housing market, outside local supply and demand and cost-of-construction increases.

A basic, consumption-based model, without sophistication, suggests that the trend series growth analysis has some underpinnings.

Chart 2. A consumption-based model for growth in house prices
Free Image Hosting at
src: all data from Census Bureau surveys

Both the model and the trend series suggest an overall price decline of maybe 8-11%.

Most of the stress will therefore come on the payments side, the credit part of the market, as part of rate-resets or income uncertainties (cyclical unemployment). A 10% decline will wipe out the equity of those with loan-to-value less than 90%, and drag out the time it will take the markets to clear and tighter credit standards will keep such people from moving on to new housing.

data: Case-Shiller indexes, through August 2007. There are other indexes for housing prices than these and other methodologies. Some market participants, like FannieMae, do not have exposure to all segments of the markets, because they are limited by the dollar amount of so-called "conforming loans".

No comments: