Monday, December 04, 2006
RE: October Pending Sales
I think not. See Grim's blog ( now NJREReport) for a quick glance at some contrasting views on this release. The reason why the analysts expected to see a better number was partially because there's been a little rise in mortgage apps and partially because with prime traditional 30 year fixed mortgage rates running around 5.6% (and the danged average below 6.3%), one would expect a sales boom.
Instead we see continued seasonally-adjusted decreases, and actually, this month's seasonally adjusted month-to-month decrease (1.7%) was more than last month's decrease (1.1%). How is this a sign of stabilization?
It's true that the year-over-year drops are lessening slightly, but that does not indicate stabilization. It's a reflection of the reality that the peak of the market was August 2005 nationally, and that therefore we are now comparing current activity to a period of already-lessened activity, rather than a peaked market.
The mortgage market is interest rate dependent to some degree. Or it has been. The big year-over-year declines in the west suggest that creative lending is now pulling back, and if so, areas like Orange County, CA and Seattle are in for a nasty shock. The numbers show a much bigger year-over-year drop in the west, which was where the bulk of the extremely risky mortgages were being written by far, so this is a plausible theory. We'll have to wait a couple more months for verification of that, but given the really bad default stats for subprime mortgages originated in the first half of 2006, that pull-back is coming.
I was expecting a slightly better report for several reasons. The first was lower interest rates, the second was the fact that relative to last year, the base sample size should be increasing, and the third is that I thought the Purchase Money mortgage apps looked a little better than this report would suggest. Remember, Purchase Money apps are processed when prospective buyers have identified a particular property and have signed a contract, so all of the apps on these pending sales will already have been submitted by now. Perhaps more applications are being rejected, or perhaps there is an error in the seasonal correction, because the non-seasonal numbers look closer to what I was expecting.
Let me explain why I think the base sample size should be increasing for Pending Sales (implying that this number would overstate pendings compared to last year). Pendings are taken off the MLS (see NAR's explanation of the index ) and large brokers, and NAR claims that this sample reflects 20% of the total market. But last year almost no new homes up for sale were listed on the MLS, and in many cases these mortgages were being processed through special brokers, most of whom would not have been sampled. This year it's quite common in many "hot" markets for new homes to be listed in the MLS, and homebuilders have begun actively marketing to realtors, who often do refer clients to mortgage brokers. So over all, more new pending home sales should be captured in this index this year than last year, which should produce somewhat rosier numbers than a true comparison would show.
If I look at the non-seasonally adjusted numbers, they do show more of the uptick from September to October that I was expecting. But the non-seasonally adjusted numbers for September to October are quite bad, because they show a roll-back in pending sales almost to 2001. In October 2001, mortgage rates were higher than in October 2006 (I use HSH for statistics). Granted, in 2001 we were in shock from 9/11, but in 2005 we were in shock from hurricane Katrina.
Over the next two months we will get a pretty firm reading of the market at its highest post-boom real potential. The reason is that interest rates are low, some funny-money loans are still being written, most people haven't gotten worried about the general economy yet, and that very significant price reductions are drawing bubble-sitting buyers out of the woodwork. In quite a few areas, they can buy a home at 10-15% off its peak price with 20% down and at a 30 year fixed-rate under 6%. In some, they can buy at 30% off peak. 20% off peak is not all that rare. Unless these pending sales start looking better, 2007 may be an extraordinarily grim year for real estate.
After the next two months worth of releases, grimmer economic news and increased conservatism in lending is going to cut down the percentage of subprime buyers by significantly more than it will draw bubble-sitters.
Because of the oddities in this report and the aberrations in this two-month time for both the index year (2001) and last year, I am giving you both seasonally adjusted and non-seasonally adjusted for September and October. A number of 100 would mean that sampled pendings in the period were equal to sampled pendings in 2001.
Seasonally adjusted by region:I think the non-adjusted numbers give a better picture of month-to-month activity; pendings are increasing with price cuts. However, there are some likely distortions in the year-over-year comparison, so the seasonally adjusted year-over-year comparisons may be more accurate.
Sept 2005: 126.3; Sept 2006: 109.1; Monthly: -1.1%; YOY: -13.6%
Oct 2005: 123.5; Oct 2006: 107.2; Monthly: -1.7%; YOY: -13.2%
Sept 2005: 106.9; Sept 2006: 89.9; Monthly: -5.9%; YOY: -15.9%
Oct 2005: 101.7; Oct 2006: 88.0; Monthly: -2.1%; YOY: -13.5%
Sept 2005: 118.1; Sept 2006: 96.4; Monthly: +2.1%; YOY: -18.4%
Oct 2005: 113.2; Oct 2006: 95.8; Monthly: -.06%; YOY: -15.4%
Sept 2005: 137.3; Sept 2006: 125.0; Monthly: -1.3%; YOY: -9.0%
Oct 2006: 135.5; Oct 2006: 122.9; Monthly: -1.7%; YOY: -9.3%
Sept 2005: 132.7; Sept 2006: 112.5; Monthly: -0.4%; YOY: -15.2%
Oct 2005: 132.6; Oct 2006: 109.5; Monthly: -2.7%; YOY: -17.4%
Unadjusted by region, including existing home sales for October:
Sept 2005: 120.8; Sept 2006: 100.8; Monthly: -16.1%; YOY: -16.6%
Oct 2005: 117.1; Oct 2006: 102.7; Monthly: +1.9%; YOY: -12.3%
Sept 2005: 105.0; Sept 2006: 84.4; Monthly: -17.0%; YOY: -19.6%
Oct 2005: 98.8; Oct 2006: 88.4; Monthly: +4.7%; YOY: -10.5%
Sept 2005: 115.9; Sept 2006: 91.2; Monthly: -13.8%; YOY: -21.3%%
Oct 2005: 105.6; Oct 2006: 91.3; Monthly: +0.1%; YOY: -13.5%
Sept 2005: 126.7; Sept 2006: 113.6; Monthly: -15.7%; YOY: -10.3%
Oct 2006: 127.1; Oct 2006: 115.0; Monthly: +1.2%; YOY: -9.5%
Sept 2005: 129.2; Sept 2006: 103.7; Monthly: -18.0%; YOY: -19.7%
Oct 2005: 127.9; Oct 2006: 106.7; Monthly: +2.9%; YOY: -16.6%
We all should keep in mind that pendings have been dropping out of escrow at an increasingly higher rate all this year. In order to correct the trend, in some markets price adjustments while the property is in escrow are becoming almost usual, so that may be lessening. However, the pendings this year are likely sampled higher due to marketing changes for new homes year-over-year, which would distort the year-over-year comparison favorably for 2006. Total sales of new homes run about 1/7 to 1/8 of all home sales, so the likely correction is somewhere around 1/15th downwards (-4%).
Existing home sales for October showed an unadjusted monthly decline (nationally) of 2.1%. Pending sales run about 6 weeks ahead of existing home sales in this market, so the big question in the existing home sales report for November will be whether national sales volume increased or decreased. If it shows a small increase, then the pending home sales overstatement is not as high as I suspect. The best correlation for existing home sales and pending home sales recently has been in the unadjusted numbers with a two month lag.
If this holds true, then seasonally adjusted home sales have probably peaked this fall already. From August to September, unadjusted national pendings fell from 120.1 to 100.8, compared to July to August's increase of 116.6 to 120.1. The relative increase showed up slightly lower in the October existing home sales report, which would seem to indicate that November's existing home sales will come in well below 500,000 units. I'm waiting with some anticipation for the NAR holiday spin on that one.
Either way, it's becoming extremely difficult to look at these numbers and figure out how inventory is going to be cleared.
They've apparently decided to use 'stable' in the medical connotation, as in "the victim is in critical but stable condition at Regional Trauma Center."
Links to this post: