Flawed House Price Indices, Flawed Reporting
Submitted by
CalculatedRisk on Fri, 05/02/2008 - 7:57am (61 comments)
in
Shopping at Amazon?
Wait! If you click on the Amazon banner link at the top of the page first, HCN will get a percentage of anything you purchase for 24 hours!
Open predictions
Our local book authors
User login
Who's online
There are currently 22 members and 157 guests online.
Recent Recommendations
Please don't feed the trolls.
The Case-Schiller index has been the most accurate over the past several years here. No method is perfect, though.
Freddie is out with mortgage cash-out data for Q1. The share of refis for principal amounts greater than 105% of the prior amount fell to 56% from 77% in Q1 and 83% a year earlier. The median age of refinanced loans fell to 2.2 years from a prior 3.6 years, and the median appreciation of refied property slowed to 7% from 19% in Q4 and 25% a year earlier. It all looks like a shift from refis motivated by cash-out spending to refis motivated by the need to get out of the way of resets. Between Reduced share of refis for cash-out means less money to spend, and slower debt accumulation for homeowners.
The Fox says "Of course all the chickens are still in the henhouse"
Hey, MarketWatch, I didn't catch your complaints about the Case/Shiller Index when prices where going up!!! Whats the deal bro.
Calculated Tanta!
with the inflation deflator only about half of inflation allowing gdp to not be negative 1st 1/4; and the jobs data "surprisingly" only -20,000 with over a 1/4 million jobs added by the unreleased-to-the-public methodology of the birth-death model, added to this article's counter offensive against the recent flawed reporting on the housing stats, i can only marvel at the co-ordinated push/shove of sudden positive news info into the recent market rise
if my own pr's would only do as well
Haha, Tanta seems to have hacked CR's password.
Yee haw!
"And finally, just ignore the NAR on house prices."
I think that should have your standard:
*With emphasis added
disclaimer!
OT, from HousingWire:
S&P Stops Rating Home Equity RMBS; Cites Anamalous Borrower Behavior
Standard & Poors Rating Services on Thursday did something that very few market participants expected: the agency said it would stop rating RMBS backed by so-called closed-end seconds altogether, whether prime or subprime.
Home-equity loans, as closed-end seconds are more commonly referred to outside the Street, were a huge part of the recent housing market run-up. HELs were a huge boost to consumer spending during the housing boom, and also made it possible for many homeowners to skirt private mortgage insurance by piggy-backing a second lien that covered up to 20 percent of a homes purchase price.
After reviewing and analyzing the performance data available for U.S. closed-end second-lien (CES) mortgage loans and the related residential mortgage-backed securities (RMBS), Standard & Poors Ratings Services believes that this market segment does not allow for a meaningful analysis of new issuance and securitization, the agency said in a press statement late Thursday.
No kidding. Most second lien holders finding loss severity to be at least 100 percent, if not greater. And with price declines showing no sign of letting up, S&P said that an unprecedented level of loan
performance deterioration has essentially made it impossible to rate second-lien RMBS going forward.
Apparently, there simply isnt enough credit enhancement in the world to account for losses that reach that high and while S&P said it will continue surveillance on existing CES deals, sources said that S&Ps announcement underscores just how heavy losses really are in an area of mortgage finance that was once among the industrys hottest.
How hot? Consider that S&P rated nearly $18 billion in CES deals druing 2007 alone.
Investment Dealers Digest, which interviewed S&P spokesperson Adam Tempkin, noted that S&P is seeing borrower behavior that Tempkin characterized as anomolous and unprecendented a reference to a growing number of borrowers simply walking away from their homes.
Neither Moodys Investors Service nor Fitch Ratings responded immediately to a question from HW regarding whether each competing agency intends to follow suit.
Regardless, sources told HW that the move by S&P essentially puts the nail in the coffin for much of the home equity loan marketplace, and comes as many banks are tightening the screws on outstanding HELOCs, as well.
The home ATM is officially out of cash, and has been for some time, said one source, a bond analyst who asked not to be identified by name.
Well Said, although I would have enjoyed if you ripped apart his points.
I am surprised anyone took the MarketWatch article seriously. CR's last sentence summed it up perfectly. The MarketWatch columnist was giving space for Lawrence Yun to rant.
I liked the blue line better.
Ha! Nades, after I went back to read the post, I copied the exact same words:
"And finally, just ignore the NAR on house prices."
Best advice anyone has ever given.
Gary - you read my mind. lol.
Way to layeth the smacketh downeth, CR.
CR,
There is another source of inaccuracy I'm wondering about.
The volume and value of housing sales is not spread out evenly over the country.
Suppose Modesto, CA, has lots of sales of depressed properties that were bought in 2006. That would depress the average sales price in Modesto, and also nationwide if you took a simplistic average of all sales.
BUT it would not be relevant for property values in areas with little distress.
In other words: do these indexes accurately represent the aggregate value of all houses in the country, without skew towards areas with lots of distress and turnover?
(by the way, thanks for your wonderful website)
OT - Pure play corp spending bellweather...
"Meanwhile, server giant Sun Microsystems saw its shares drop more than 19% after the company swung to a surprise loss for the March quarter, citing "significant challenges" from a slowing U.S. economy. "
I just got back from Vegas- I was attending the APA- American Planning Association convention. Scared folks.
They should have had some segments on dealing with broke developers and see through buildings, bk negotiations, and other useful topics.
Too much optimism by the shills, and the beginnings of real fear in a lot of the attendants.
That Feb 08 pic from Case Shiller is beyond any reality that we have lived in- so now we have entered into the realm of collapsing house prices, collapsing mortgage instruments, a massive increase in the TAF, and Wall Street is standing there looking like this is no problem at all.
This is not just a problem, it is a crisis of a magnitude not seen since the 1920-30s, and yet equity thinks it is something other than dead in the water.
Obviously, we have too much hot money chasing yield, and not enough scared money delevering yet.
Leverage will provide a lot more damage to come, in a lot more unexpected places.
This is most likely the end of the second inning.
Someday this war's gonna end....
As I stated in the article, the Case-Shiller index might overstate the national price declines. A reasonable approach might be to take 70% of Case-Shiller and 30% of OFHEO for national prices.
That sounds like a good idea -- but there might be reasons to give more weight to the OFHEO.
The S&P/Case-Shiller index, which Tuesday posted a 12.7% decline for February, is skewed for two reasons of its own -- it tracks just 20 major markets, many among the hardest hit, and its "repeat sales" survey by design pulls in individual homes both bought and sold in the last few years.
If Case-Shiller over weights for repeat sales, that could be a problem.
A substantial portion of the foreclosures and REO sales are properties that were refi'd or purchased in the past 3 years with ARMs -- since ARMs are designed to be refi'd before the reset, if values do not continue to rise, high LTV homeowners are SOL -- which in turn leads to foreclosure and resale at a lower price.
In a normal market, resales of the same property within a few years can be a good indicator of value trends; in the current market, it can give a false low indicator of declining values. It's not that the declines aren't real, but REOs will be discounted until they sell, thus driving down the average sale price.
If the percentage of REOs in a sub area increases, it can cause an exaggerated decline in the average sale price, even if the non REO market is stable. Some buyers won't even look at REOs, and they continue to buy top of the market homes.
Raw data for sales and resales (when the resale is an REO) can also be flawed, in that the decline in property condition may be a significant factor in the discounted REO sale price.
A more accurate paired sale analysis would be to look at current sales, with terms and condition similar to the sales several years ago.
This is not to say this is a flaw in the Case-Schiller methodology, but it has that potential.
^^^^^^^^^^^^^^^^^
Developing indices of ANY range of accuracy has brought large paychecks to all. . . Case-Schiller, Zillow, Moodys, Realty trac, etc.
All the price indexes show home prices are plunging. Arguing about which one is right is like arguing over whether the fire that's engulfing your house is 500 degrees or 400 degrees.
Happy Homeownership Month!
"During National Homeownership Month and throughout the year, we applaud the men and women who work to achieve the dream of homeownership, and we are grateful for those who provide counseling, lending, real estate, construction, and other services to these individuals."
--GW Bush, 2006
404 Page Not Found | The White House
will t: interesting post.
so can HELOCs etc be bundled into the secondary market if the ratings agencies won't rate?
this could be big, no?
-ck-
In a "normal" market? That sounds awfully subjective to the Shnapster.
There is only one market: the current one at any given point in time.
Suggesting that it's 'abnormal' because there are more REOs than there used to be doesn't do any good. Nor does it mean that home price indices (which attempt to reflect the market prices) should be adjusted to estimate what things "might be like" if all those pesky REOs weren't corrupting what "prices should be."
Your paired-sale idea would actually be less accurate.
EJ_San_Fran writes:
...do these indexes accurately represent the aggregate value of all houses in the country, without skew towards areas with lots of distress and turnover?
S&P/CS 20 is a weighted index, weighted by aggregate value of single-family housing stock for each reagion - e.g NY 19% of index, LA 15%, Phoenix 3%, Las Vegas 1% (see p15 of the doc CR mentions.)
Also note CS the repeat sales used must have "two or more recorded arms length sale transactions", so the first sale of a house after its first construction is excluded, so probably excluding some proportion of tract housing sales - possibly overstating the index.
Y2L - that decision applied to closed-end loans(HEL) not revolvers (HELOCs).
I don't think anyone is still securtitizing either anyway.
I shouldn' have said anyone, there might be a trickle of activity.
Just wanted to say that before bacon dreamz posts some ABS 08 issue just for the sake of nahnahneeboobooness.
As if the NAR has any credibility left whatsoever.
BTW, is anybody as irritated by me about those ever-present NAR commercials about homeownership as a primary vehicle of wealth building? GEEZ, they don't even have a disclaimer (e.g., past performance not indicative of future returns).
NYC and LA make up almost 35% of the Composite 20. I wonder what the composite would look like excluding NY and LA?
Kinda like CPI ex Food and Energy.
Bravo, CR, definitely the best piece on the subject.
Proving once again CR is the anti-blog: high on analysis, low on ego.
EJ_San_Fran writes:
...do these indexes accurately represent the aggregate value of all houses in the country, without skew towards areas with lots of distress and turnover?
oas smoas writes:
S&P/CS 20 is a weighted index, weighted by aggregate value of single-family housing stock for each reagion - e.g NY 19% of index, LA 15%, Phoenix 3%, Las Vegas 1% (see p15 of the doc CR mentions.)
Okay, fair enough.
BUT: you can still have distortion, I think.
Suppose Modesto has 2 parts of town: the 'bad' part is a new development constructed in 2006 that has now lots of foreclosures and auctions.
The 'good' part of town is an old treelined neighborhood where almost nobody is selling now.
The indexes record a big decline for Modesto, but is it relevant for the 'good' part of town?
Can somebody in the 'good' part who sells today still get a good price?
"with the inflation deflator only about half of inflation allowing gdp to not be negative 1st 1/4; and the jobs data "surprisingly" only -20,000 with over a 1/4 million jobs added by the unreleased-to-the-public methodology of the birth-death model, added to this article's counter offensive against the recent flawed reporting on the housing stats, i can only marvel at the co-ordinated push/shove of sudden positive news info into the recent market rise" --Adan Lerma
Don't forget the stimulus package calculated to be 1.5% of GDP and benefiting Q2. If we get an "official" 2 quarter recession, it won't be until Q3 and Q4, and by then Bush at least won't care.
A longish interview with Robert Shiller at the following link:
Interview With Robert Shiller - The World According To - Portfolio.com
CR, You lose all credibilty when you don't give the NAR credibility. I demand an apology. -- Lawrence Yu
Schnaps --
What's the old saying? There are lies, damned lies, and statistics.
It's not that statistical data is invalid, but how it is weighted and how it is used (mean, median, mode) can definitely affect what the end product looks like.
In Denver, it was reported that the Case-Schiller index showed a 5% year to year decline. On average, that's probably about right -- but I've recently done 2 appraisals in neighborhoods that have declined 20-30% in the past 18 months, while other neighborhoods are stable or increasing in value. What does a 5% average decline tell us? Not much.
Likewise, within many neighborhoods, REOs have caused a decline in the average sales price, but the top of the market is relatively unaffected; this is market stratification within a neighborhood, and doesn't show up in the raw data.
Unless you do a more detailed analysis, the aggregate data can be misleading.
Paired sales analysis is the heart of appraisal theory (principle of substitution) -- properties of like kind in size, style, location, amenities, age, and condition are used as comps, and adjusted against the subject property.
My point was that if resales are given excessive weight in the CS index, it could create a misleading downward indicator when the resale is in inferior condition, or is an otherwise stigmatized and discounted REO.
^^^^^^^^^^^^^^^^^
EJ_San_Fran said: "...Suppose Modesto has 2 parts of town: the 'bad' part is a new development constructed in 2006 that has now lots of foreclosures and auctions.
The 'good' part of town is an old treelined neighborhood where almost nobody is selling now.
The indexes record a big decline for Modesto, but is it relevant for the 'good' part of town?
Can somebody in the 'good' part who sells today still get a good price?"
Maybe...by comparison with similar homes in the "good" part of town. As a homeowner, if someone approached me with an offer to buy my house based on the offering price of an REO in a new development 10 miles outside of town, I'd respectfully decline.
You and poster -ck- both bring up important points. When lumping together all the housing figures you get one point of view, but when breaking them down into components things can look very different.
I think (no, I know it) that the Case-Shiller numbers are heavily skewed to reflect the bubble extremes of the housing market in places like California, with the OFHEO numbers reflecting a much more-reasonable median.
Sebastia
Where the official NAR graph? You know, the one that only goes up?
This is just another ploy to lure suckers into the housing market. The spring selling season is bust, and putting out silly ads touting people to buy an overpriced shack because "buying a home is not just a finacial decision, and now is a great time to buy!" is not working. So, they turn to the next tactic - smearing the people who are reporting on the popping of the Housing Bubble. Typical!
Sebastian,
I think that home prices are falling alot. Go ask my neighbors. They'll tell you the same thing. I don't live in a bubble state (although it is high growth and has a high quality of life), but prices are aleady down 20% in my desirable neighborhood. I don't think the Shiller or OFHEO is getting it right. Prices are down much more.
Elvis said: "Go ask my neighbors."
Ask mine. You'll get a very different answer.
In fact, if somebody showed up on my doorstep today with a duffle-bag full of cash and offered me 15% over the market for my house I'd couldn't accept the offer, because rent would cost me as much or more than my current monthly mortgage payment.
Sebastia
Oh, yes. That man with the duffel-bag full of cash came to my house yesterday, but, I couldn't justify it either. The bag was heavy and I was tired and I didn't feel like going to the bank, so I told him to try your house. But, he said, homes in NC are overvalued and prices are about to get hammered, so why would he want to buy a house there? I said good point. Then I went back inside and took a nap.
Tried following that link to the Case shiller data and got an error message. Probably a problem on my end, but just in case.....
Sebastian: "rent would cost me as much or more than my current monthly mortgage payment." (after someone paid you 115% of market value for your home)
I'm probably lazy and have not inferred your current mortgage situation from your posts.
You are implying that the rent vs. buy-right-now calculation in your neighborhood strongly favors buy?
Sebastian wrote: "I think (no, I know it) that the Case-Shiller numbers are heavily skewed to reflect the bubble extremes of the housing market in places like California, with the OFHEO numbers reflecting a much more-reasonable median."
But this ignores the fact that the OFHEO numbers are also heavily skewed as they ignore the bubble extremes since they only include GSE loans. That means that expensive areas in California (SD, OC, LA, Bay Area, Silicon Valley, etc.) and elsewhere, which saw huge price increases and now are seeing large price decreases are not included (at least, not much) in the OFHEO numbers because so many of those homes had jumbo loans.
As CR noted, none of the indices are perfect, and some mix of C-S and OFHEO might be more accurate. But you can't just cherry-pick one index based on what you see in your neighborhood, as that shows the same bias that you claim others are showing by picking another index.
To follow up one more time:
Suppose I want to dream up an 'ideal' index that represents the total value of all homes in the country, how would I do it?
The most accurate way to do it: ask for an appraisal of each and every house in the country and add up all the values.
Sounds impractical, right? And so the indexes use other methods. Each method has distortions. The simpler the method, the bigger the distortion. For instance, it doesn't take much effort to recognize that the method of listing the median sales price nationwide as used by NAR introduces a huge distortion. OFHEO and Case-Shiller are better, but not perfect, as evidenced by the remarkable difference between these two.
How many 'ordinary' people are ultimately going to be under water? I hesitate to rely on Case-Shiller for the answer.
Callous -
I think Sebastien is omitting the fact he probably has a house with a very small mortage.
To fully analyse the situation would be to compare (annual price of rent - investment return of proceeds from the sale of his house) versus mortgage cost + upkeep + taxes.
The imponderable is always quality of life from either locations. Proximity from work/school/transport, square footage, greenery, view, pollution, etc...
That is a more intelligent comparison.
Sebastian should know better, as he usually backs up his comments with data.
-ck-
I don't see how your proposed methodology changes anything in the aggregate. It just makes it harder to find data.
Even if you could somehow get all the necessary data -- your suggested approach wouldn't circumvent the fact that IN THE MARKET, homes are declining in condition as well as price. So, again, that is just the reality of the market. Let's say a given home you want for your index because it was sold with freshly manicured topiary shrubs out front and the scent of fresh-baked cookies in the kitchen -- that doesn't mean you get to conveniently ignore and omit from your index the 2 other sales of homes with the knee-high grass whose kitchen aroma is something closer to 'Eau de Bandeaux".
It's also flawed to think there is a 'c' in Shnaps.
"If Case-Shiller over weights for repeat sales, that could be a problem."
That sounds like saying that stocks that trade a lot should receive less weight in the S&P.
C-S may have geo difficulties, but tracking multiple trades of the same abode as a proxy for a market in lots of the same thing seems like a good idea.
stuck_in_reverse - I agree. However, Sebastian's an interesting fellow, living in a completely different region than me, and I wanted to get the details on what seems to be a somewhat incredible statement.
I certainly don't expect him to fall for "my monthly payments are higher/lower!" Nor should we.
callous said: "stuck_in_reverse - I agree. However, Sebastian's an interesting fellow, living in a completely different region than me, and I wanted to get the details on what seems to be a somewhat incredible statement."
Well, check out the mortgage/rent ratios at this link (I'm in Raleigh, NC):
Housing Tracker
I keep trying to explain that California has some of the most extreme housing-related stats in America and simply can't be used as a reasonable example for the rest of the country, but I can't get any traction on it. The MSM and the bearish housing blogs have larger megaphones and more of them.
Sebastia
It's also flawed to think there is a 'c' in Shnaps.
And yet it constantly re-appears. Perhaps it is an optical illusion.
Maybe a CR Housing index would be better, take Cris CobraDriver in south FLA, Average Joe in Chula Vista, Anne in Seattle, and David in CT, and average them, after all, 99.9% of the US population is between them.
Ooops, I don't think Anne bought yet, how about PNW,...
EJ_San_Fran said: "How many 'ordinary' people are ultimately going to be under water? I hesitate to rely on Case-Shiller for the answer."
Another point I keep hammering at (with little success) is that the state of the economy is a critical factor in answering this question, it can't be answered with housing data alone.
During a normal economic expansion, most "ordinary" people aren't much hurt by housing ups and downs. But when recession hits, hundreds of thousands (millions) of homeowners suddenly are out of a job and are literally forced to put their homes on the market. When your income suddenly goes to zero, you have far fewer options than a homeowner who simply has his ARM adjusted up. That is when a lot of "ordinary" people get hurt.
Case-Shiller doesn't tell us anything about the economy or how many people are unemployed, so a critical piece of the puzzle is missing.
Sebastia
EJ_San_Fran writes:
Suppose I want to dream up an 'ideal' index that represents the total value of all homes in the country, how would I do it?
The most accurate way to do it: ask for an appraisal of each and every house in the country and add up all the values.
Come to think of it, this is actually quite doable with statistical sampling.
As follows: Take a sample of 1000 houses statistically representative of all houses in the country. Then get an appraisal for each. Aggregate. Done.
The cost would about a $1M. Quite doable.
Why isn't our government doing something like this?
Because appraisals are as reliable as NAR spokesmen.
Because appraisals are as reliable as NAR spokesmen.
Okay, let's just ask Countrywide! Or how about we just let the fact that the banks are blanketing the country with HELOC reduction notices serve as a proxy?
Sure, appraisals can be flawed.
However, Case-Shiller only looks at transaction prices.
An appraisal would look at transaction prices in the neighborhood and make an intelligent assessment of the property considering all of its circumstances.
Potentially more accurate, I think.
Sebastian:I think (no, I know it) that the Case-Shiller numbers are heavily skewed to reflect the bubble extremes of the housing market in places like California, with the OFHEO numbers reflecting a much more-reasonable median.
The only houses in most of California that could have been included in OFHEO formula(= < 417k) have been dog houses. How is that an accurate measurement?
Sebastian has a first and a second on his house (he took out the second to pay down credit card debt, if i remember correctly) and still pays less than renting? i call bs.
Also Sebastian: how can you exclude CA from your estimate of housing in this country, it has 38 MILLION people (12% of US population). Go ahead and exclude WY or SD or ND, but not CA.
Sebastian: Thank you. In both relative and absolute terms, home ownership is in a much better fiscal situation in your neck of the woods. I am most familiar with Bay Area, Portland, and Minneapolis.
A far more serious issue has been brought to light by this, however: The rather unbelievable assertion that there exist places outside of the Bubble Zones!
Sebatian writes: Well, check out the mortgage/rent ratios at this link (I'm in Raleigh, NC):
404 Not Found
Actually according to that index Raleigh has a rent/mortgage ratio of 1, so you aren't better off, just even. Worse, that index assumes down payments are free! If you assume the down payment can be invested for a similar return to the mortgage rate, then that index should be increased by 25% and Raleigh is very much on the "buying is bad" side with a 1.25 index. If you're right about stocks being a great investment, then people should be able to get better than 6% with a good stock portfolio and it's even worse. So, if you're consistent and believe your own claims, you should sell your house, rent, and put the equity into the stock market.
In any case, you're definitely not better off buying in Raleigh.
ck:
Unless I misunderstand your definition of "repeat sales" I think you need to read the C-S methodology. The index is comprised of only repeat sales of single family homes. As such it provides a good picture of the market for presumably non-speculative transactions in a given market or aggregation of markets. The index intentionally leaves out new construction, flips, and properties that sold at an unusually low price given prior sales and the market.
The median American lives in an MSA of 1.3mm (Jacksonville, FL) so the country is fairly urban, though more city than metro.