According to Mandelbrot,
“pictures are undervalued in science, they are not trusted…” but “…nowadays the picture can aid, not mislead (or replace!) the scientist. It permits instant comparison, instant comprehension…” 1
Let´s forget the words “science” and “scientist” in the quotation above and just retain the “instant comparison” and” instant understanding” qualities of pictures. To that effect I will use a set of pictures on house prices that may help us have some idea of what is going on. Maybe we´ll see something which is not quite what we get from media reports and releases of aggregate house prices.
The two home price series I compare (through the third quarter of 2008) are the OFHEO House Price Index and the S&P-Case Shiller index of house prices. The first picture shown gives a bird´s eye view of house prices for the nine census regions of the US collected by the OFHEO – the “West” is subdivided in “Pacific” and “Mountain”; the “Mid-West” is classified as “West North Central” (WNC) and “East North Central” (ENC); the “South” as “West South Central” (WSC), “East South Central” (ESC) and “South Atlantic” (SA), while the “North East” is broken down into “New England” (NE) and “Middle Atlantic” (MA).
Quick observations: In general, where prices showed the steepest rise we also observe the largest drops (and the higher rates of foreclosure). That´s the case of the Pacific, Mountain and South Atlantic regions. In those regions, the shape of the figure is driven by price behavior in states such as California, Nevada, Arizona and Florida. The NE region also experienced a steep price rise but has not experienced the same drop. First “conclusion”: the differences in house price behavior both between and within census regions are significant.
The second set of figures compares the OFHEO HPI with those from S&P Case Shiller for the same metropolitan areas in the same census regions (there is no representative of the ESC region in the S&P Case Shiller Index).
1. Metropolitan areas in the Pacific, Mountain and SA regions tend to show the steepest rise and fall in prices, but even in these metro areas there are significant differences in price behavior. Compare, for example, Portland with San Diego or LA in the Pacific region, Denver with Phoenix or Las Vegas in the Mountain region and Atlanta or Charlotte with Miami and Tampa in the SA region.
2. It is hard to pin down why in some cases the quantitative information from the two price indices is so different and in other cases not so. In the case of San Francisco, for example, there are major differences on the upswing and downswing portions of the figure. In the case of Las Vegas and Detroit, for example, the two indices give out the same quantitative information at every point. In other cases, Phoenix and Miami, for example, the two indices move closely together on the rising portion of the figure but very differently on the downswing. Why do the two indices tend to be more “equal” when the price reversion is smaller as, for example, in the case of Charlotte, Portland or Denver? Why since 2002 is price behavior in Dallas so different in the two indices?
The next figure is the “great divider” of the two indices. The S&P Case Shiller national index shows that since 2000 house prices have gone up much more than indicated by the national OFHEO index and has now “dived far and deep”! While the S&P C-S shows house prices nationally falling 17% on a year ago in September, the OFHEO indicates that prices for the country as a whole has dropped a meager, by comparison, 4%. That is not an outcome I would have guessed from looking at the set of figures for the different metropolitan areas shown above (especially if we concentrate on the rising portion of the charts).
The following charts which show the C-S Composite of 15 metro areas and the Pacific census region of the OFHEO, both compared to the C-S National Index help, if not reconcile, at least make sense of the very different results obtained for the two measures of national house prices shown in the chart above. It seems that the two charts below illustrate “beasts” belonging to the same “family”. And the “family they belong to is the Pacific region house prices. This means that the information in both the C-S Composite 15 and the C-S National Index is significantly determined by the weighing methodology of the C-S indices and likely reflect less accurately what´s happening to house prices in the country as a whole, which, as I showed earlier, seems to be very diverse.
The question then becomes: why has price experience been so diverse? After all, it´s one big market, driven by the same set of fundamentals – in particular the “home ownership program” especially targeted to lower income groups devised by Congress and HUD in 1992 and implemented during the following years (which was a major factor behind the development of much of the financial innovation that has now become “toxic”) – and where construction costs cannot be very different.
A few years ago, Edward Glaser and others2 argued that much of the difference in house price behavior among regions and areas could be explained by supply factors, in particular local restrictions on new construction.
According to Glaeser et al.: “The San Francisco, San Jose, Oakland and Los Angeles markets, all in California, which is well known as the epicenter of the restrictions on new construction…” No wander prices in California have had such a distinctive behavior.
This supply restriction argument can be illustrated by comparing house supply (proxied by housing starts) in two regions: West and South. This comparison is interesting and illuminating because both regions contain a large fraction of the population targeted by the “homeownership program”. Separately or together they are home to much of the Asian, Hispanic and African American population in the US.
The figure below shows house prices (OFHEO) for the West and South and Housing Starts adjusted for population difference between the two regions. Clearly house supply was much more accommodating in the South.
From all we´ve seen, the house price problem (negative equity) exists but may not be as pervasive as generally thought. The “program” that gave rise to the problem is now dead and the proposals that are being floated to help people whose home values are under water have to take into account, in their design, the regional differences in price behavior observed as well as factors such as “legal restrictions” on supply that help explain them.
(1) Benoit Mandelbrot: The (Mis)Behavior of Markets Basic Books 2004, page 8.
(2)NBER WP No 11129, 2005 and NBER WP 10124, 2003.