My Photo
Note: Jeff does not accept guest blog posts on A Dash of Insight.

For inquiries regarding advertising and republication, contact

Follow Jeff on Twitter!

Enter your email address:

Delivered by FeedBurner


  • Seeking Alpha
    Seeking Alpha Certified
  • AllTopSites
    Alltop, all the top stories
  • iStockAnalyst
Talk Markets
Forexpros Contributor
Copyright 2005-2014
All Rights Reserved

« Interpreting Comments | Main | Identifying Experts: Predicting Recessions »

August 27, 2007



The housing bulls were right for a very long time, until they were wrong.

Now we're seeing calls for rate cuts from the housing/market bulls, the bonus-motivated Wall Street touts, the Larry Kudlows of the right-leaning media, and even the conservative economist M. Feldstein.

With so much riding on this recovery for them to say it 'proves' Bush's tax cuts 'worked,' a drop off in general output would not be good.

Which gets to the point that many of these economists do have a bias AND... the future government action - or inaction - may have a huge impact. on housing No one even pretends to claim to know what Bush and Congress will come up with. So thoughtful economists looking at data and doing regressions may or may not get their next call right.

See Ayre's book on quant models beating experts. time and time again.


Thanks to everyone for the great comments. While we all want to weigh the substance of these arguments, my educational focus is a bit different. I am reminded of Ronald Reagan in Presidential debates, where he always seemed to know the most persuasive thing to say.

I am pondering a theme about journalists, who do great work when drawing out expert opinion. Is there a line? What happens when the journalist is thrust into the role of expert. Does the journalist offer his or her own views, or distill knowledge from others?

Email from readers tells me that they appreciated the thoughtful comments.



Josh Stern

Thanks Bill, I found it and also this link which explains the differences in the indices:

To summarize the most important points, it seems that OFHEO has more geographic coverage (good) but doesn't include many high-end, non-conforming, homes (bad) and doesn't weight the index by actual market prices (probably bad).

Bill aka NO DooDahs!


- on the OFHEO link, click in the top section under "THE LATEST ..," the "Census Divisions and US" link. Under "select regions to compare" select the United States as one of the three dropdown boxes and hit "search." The resulting table displays annual changes in the US HPI from 1975 to present, as calculated quarterly by the OFHEO using the Case-Shiller methodology, applied to all home transactions in the United States that Fannie and Freddie have data on.

There are no negative years on record with this index. There are some that are pretty close to zero, though.

The S&P C-S index uses either 10 cities, or in the broad index, 20 cities, instead of nationwide data. This explains a great deal of the discrepancy, esp. as Boston is one of those cities. It is arguable whether the data that the S&P uses is more complete for those 10 (20) cities than the OFHEO data is, because "jumbo" loans probably aren't in the OFHEO data, but the S&P data is GEOGRAPHICALLY incomplete. If you browse around their site, you'll eventually see a link to download an Excel spreadsheet with the S&P data to 1987.

In downloading that spreadsheet, which has monthly, and not quarterly, data, the Aug 1990 to Mar 1994 period shows almost exclusively negative annual appreciation. However, that is for the 10-city index, as they did not start using the 20-city index until Jan 2001. Ten cities does not a nation make. As mentioned previously, Boston is one of those 10, and Boston had local issues at the time making a huge slump. However, the OFHEO data does not show that Boston pulled the U.S. down. Also, three California cities, especially LA, had a slump, but even in the S&P city data, it shows Atlanta, Denver, Miami, Chicago, Cleveland, Portland, and Seattle as being strong in price appreciation. Were all of those in the 10-city index at that time? No. All three California cities and Boston were in, some of the others were, too, but Atlanta, Cleveland, Portland, and Seattle are not. The S&P data is biased by its sample selection, which does not reflect the nation on average.

When you are looking at data that purports to be nationwide, check to see if it is, and what the sources are. Neither of these sources is perfect, the OFHEO data is "only" Fannie and Freddie, but I would consider it far more comprehensive than the S&P data, which is limited geographically and is not certain to be more comprehensive than the OFHEO in its target areas, as I haven't pinned down their data source.

Re: the ING link you provided, do you have a *specific* link to a graph they provide that claims to be (1) nationwide, (2) home price appreciation (not starts or changes in houses under construction), and (3) showing a multi-year decline in home price appreciation? If so, I would like to see that and compare it to the other data sources ...


While Olick certainly has better TV presence, she left a less-sanguine impression because of her statement about uncertainty due to the mortgage resets. I thought I noticed Blinder nodding in response as the clip closed. With regards to the housing market, blogger bulls don't seem to have a good record to date.

Josh Stern

Bill, I didn't find data covering the period 1989 to 1994 at those links. Did I miss it somewhere?

This link has some graphs of various housing related indicators that show multi-year slumps from time to time:,b2530qgh

Bill aka NO DooDahs!

In relation to BS from Yale on home prices going down for 5 years straight, no such even happened as measured by the price index, unless he is talking about a local event.

If BS from Yale is talking about the S&P CS index, , then he is certainly not talking about a national event, as that index is comprised of 10 (or in the broader index, 20) cities, and is FAR from nationwide.

Bill aka NO DooDahs!

Check out this story, which makes one wonder how many sub-prime foreclosures are related to out-and-out fraud, and I really doubt that those foreclosures, in particular, will impact consumer spending! LOL. This one fraud ring accounts for over 100 subprime foreclosures ...

Josh Stern

Following up, here's Bob Shiller (Economist from Yale) mentioning how home prices went down for 5 years straight between '89 and '94:

Josh Stern

I'm not a CNBC watcher so I've got no opinion about what Dianna Olick normally says or her style. But I watched the linked video clip and don't really see what you are driving at. In my mind all three interviewees said reasonable things about housing and didn't directly contradict each other at the level of details. Olick's point that a lot more ARM resetting is ahead of us than behind us is simply true, and I'd be shocked if there are not economists out there who are worried about that. When the other two guys say we are probably closer to the trough than the peak in housing prices, it's not clear whether that view is based on some well validated model or is just their hunch (economist's make personal guesses too).

There shouldn't have been any new info in what any of them said for someone who follows the U.S. economy and the stock market. One thing I would like to see and have not is where the interest payments from those ARM's are currently accounted for. How much is in future earnings estimates for banks? How much has been borrowed against by hedge funds? If the future payment stream from those mortgages actually stayed kind of constant (because of a mixture of deliquent and non-deliquent loans), which business entities would fall the most from their present value? It occurs to me that pessimists may be right about the probable quantity of defaults but still be double or triple counting the economic shortfalls that result. Can both the banks and the consumers be left with less money because of this? Only to the extent that the lenders have already spent the money.

Bill aka NO DooDahs!

Comments in chronological order of the points they address.

IMO the approach taken is more important than qualifications of the analyst, as those with both qualifications and flawed approaches are legion. Expertise matters, but one should be careful of the metrics one uses to measure expertise.

For the effect on sheeple, er, well, yes, sheeple, emotional content and plays to inductive thinking are far more effective than structured analytical approaches. This effect is magnified in mass media presentations like a television debate segment. Give a hypothetical Diana Olick or any of her ilk 30-60 minutes of preparation and she will "clean the clocks" of the experts who present data in the way that experts like to see data presented. Analytical thinking is not a skill held by the population at large.

Stereotypes make an interesting discussion. In the vernacular, a stereotype is constructed from inductive observation of a large sample size, and is representative of a modal behavior, or perhaps the behavior of a plurality or majority of the group the sample is chosen from. As such, stereotypes constructed this way are perfectly appropriate for discussion when dealing with large samples, but fail when dealing with small samples or with individuals. I think the point you may be driving at is that the "Olicks" of the world are attempting to improperly construct a stereotype from an insufficient sample size.

Taken more literally, this is a misuse of inductive reasoning, drawing conclusions from a sample size that is either non-random, too small to be significant, or otherwise biased, and then applying those improper conclusions deductively to other groups.

The major arguments I have rec'd from the housing bears boil down to two points: (1) look at these specific egregious examples! and (2) look at this flawed biased data!

(1) is self-evident, anyone pointing to anecdotes rather than distributions or histograms of impacts is missing the boat.

(2) is harder to spot.

When folks like "Mess that Greeny Made" start drawing conclusions based on the S&P C-S 10-market composite, one needs to point out that those 10 metro areas are not necessarily an unbiased estimate of the entire U.S. housing stock.

The median price of units that sold last month is also not an unbiased estimate of the entire U.S. housing stock. There is quite possibly a different distribution of units offered (desired) than of units in existence.

Anyone with the chutzpah to attack the statistical significance of numbers he doesn't like and don't support his agenda should apply the same criterion to the numbers he DOES like, or that support his agenda.

The comments to this entry are closed.