No, New Homes Sales DID NOT Rise . . .

Look, we have to stop meeting this way.

Some half-assed piece of data comes out, the markets spasm, then you want to understand what the data really means.

OK, let’s look at New Home Sales.

U.S. Census Bureau and the Department of Housing and Urban Development releases New Home Sales each month. The data contains three elements that contextualize what they mean, and how much significance they have, beyond the headline number.

First, we see the headline number. This month, September 2007 sales of new one-family houses were up 4.8%, at annual rate of 770,000 (SA). 

Second, we look at the year-over-year data, which in this case was 23.3% below the September 2006 estimate of 1,004,000.

Third, we look at margin of error. The monthly gain of 4.8% was within the "estimated average relative standard errors" of ±10.3%. This means the data point was statistically insignificant not statistically significant

The year over year number however, at 23.3% — ±8.0% — is greater than the margin of error, and therefore is statistically significant.

Note: These aren’t my opinions; these are simple mathematical facts that the Commerce Dept. notes in the footnotes of its release. Based on this data, we know for sure that year-over-year sales decreased; What we can tell about month to month sales is that they may — or may not have — increased (we just don’t know).

Next, we look at the revisions: For the month of August 2007, the original sales report was for 795,000 new homes built (annual rate). This was adjusted downwards this month to 735,000. An apples-to-apples comparison (original release to original release) shows a decrease, not an increase in new homes sales. Comparing the original (but soon to be revised) September data to the revised August data presents a misleading picture.

Lastly, we need to consider Cancellations. The Census Bureau does not make adjustments to the new home sales figures to
account for cancellations of sales contracts. As we have seen, the Cancellation rates of Home Builders have been huge:

Firm . . . Cancellation rate for Quarter
Centex (CTX) 35%
MDC Holdings (MDC) 57%
KB Homes (KBH), 50%
Lennar Homes (LEN) 32%
D.R. Horton (DHI) 48% 
Beazer Homes (BZH) 68%
NVR (27%)

So, you can further discount the reported data by some amount relative to the above cancellation rates . . .


UPDATE:  October 25, 2007 3:15pm

Compare & contrast the NAR with the National Association of Home Builders:

"David Seiders, chief economist for the builders’ trade group, said
Thursday that the latest report has some questionable readings,
including a 38 percent rise in sales in the West, which he expects will
be revised significantly lower in subsequent months. Without that
reported increase, sales would have fallen from the already weak
revised level in August.

Seiders also pointed out that the report
does not capture cancellations by buyers who were unable to get
financing or had to pull out of sales because they couldn’t sell their

"We saw an upsurge in cancellations in August and
September, according to all the builders," he said. "The net sales, if
we could get that number, would clearly be weaker than this. It’s too
early to get hopes up on this report."

An honest trade group. Go figure . . .


Commerce Dept, SEPTEMBER 2007

Home sales: Bad and worse than they seem
Chris Isidore, October 25 2007: 12:31 PM EDT

Print Friendly, PDF & Email

What's been said:

Discussions found on the web:
  1. Stuart commented on Oct 25

    I could not agree more. Going through the figures in detail then comparing to the AP and CBS marketwatch news releases I thought I was comparing two different reports. The media headlines on this morning’s new home sales news release are so bad, they can only be construed as purposeful in their intent to mislead. It seems the David Lareah a.k.a. Lawrence Yun syndrome is spreading faster than the California fires.

  2. edhopper commented on Oct 25

    Talk to your BFFE Larry Kudlow. This means the economy is booming, so the Feds need to cut rates to help all the Financials who are NOT in any trouble. And a sign the Dow will hit 15000.

  3. Jeff commented on Oct 25

    1) “Statistically insignificant” is not the same as “not statistically significant.” Most people use the term incorrectly, as you have here. Type the term into google for some detailed explanations. [BR: Fixed above]

    2) Statistical significance is also not the same as substantive significance. There can be an important result that does not attain statistical significance because of sample size. There are also unimportant results that are statistically significant — meaning that the sample is big enough that we are confident, for example, that a change is non-zero.

    3) Comparing this month with the original release from the prior month is not “apples to apples” but rather the intentional use of data known to be incorrect. No one does this. You cover the non-farm payrolls every month. Do you think they measure the widely followed change based upon “original releases?”

    So — your headline is not supported by the data in your post or in the release. We all understand that it is a noisy series. We do not know whether sales actually increased last month because the sample is too small. But neither can you conclude that sales did NOT increase.


    BR: Jeff: I agree that, based on this data, you cannot draw any conclusion on a month to month basis. I thought I made that clear above; I’ll clarify that.

    “What we can tell about month to month sales is that they may — or may not have — increased (we just don’t know).”

  4. Ross commented on Oct 25

    Lies, damn lies and STATISTICS! Quite funny actually. I love it when the media talks statistics to me.
    Looking foreward to my 2+% increase in my SS payments next year. Gotta go milk my mule..

  5. Gary commented on Oct 25

    I think you are wasting your time watching the housing market as the main cause of the next recession. It just doesn’t affect enough people to seriously damage the economy. What we need to watch is energy. Pretty much every time oil has spiked over 100% within a year a recession has followed. We started the year at $50. At $100 things will start to look a bit dicey for the US economy.

  6. tekel commented on Oct 25

    Sooooo, let me get this straight. -23% year-over-year, and -8% adjustment to last month, and effectively 0% month-to-month . . . is an increase?


  7. Stuart commented on Oct 25

    “Comparing this month with the original release from the prior month is not “apples to apples” but rather the intentional use of data known to be incorrect. No one does this. You cover the non-farm payrolls every month. Do you think they measure the widely followed change based upon “original releases?””

    What??? of course they do. Holy cow where have you been. It’s one of the common ploys. Lower the previous month figure so you can say you beat last month. Here’s a simple illustration. Say you have 12 months of collapsing home sales, each month lower than the prior month. When the current month is released, concurrent downward revision of the prior month is also reported and voila, for each month’s headline, sales were reported up compared to last month even though at the end of 12 months, sales fell each and every month. That’s the game being played. People need to see through this.

  8. Stuart commented on Oct 25

    The BS game.

    month J F M A M J J A S O N D

    1st rpt 100 99 98 97 96 95 94 93 92 91 90 89

    revised 98 97 96 95 94 93 92 91 90 89 88 87

    In reality, each month has successfuly lower sales, yet the headlines would’ve been “sales higher from previous month” for EACH month.

  9. vega commented on Oct 25

    Mortgage remits are in y’day and today. Bottom line: THEY TOTALLY AND COMPLETELY AND UTTERLY SUCK. You heard it hear first. The meltdown in the ABS/mortgage space continues. AA ABX index is down 6 points today alone. Big move. The divergence between credit and equity continues until it cannot continue. I suspect either credit improves (unlikely) or equity unravels (also unlikely — ha ha).

  10. Tech Policy commented on Oct 25

    Barry Ritholtz is Shrill Again !

    If you have been a reader of The Big Picture for as long as I have been, you would know that every now and then Barry can not take the endless spinning of data anymore. For the latest installment: No, New Homes Sales DID NOT Rise . . . or as I put it, …

  11. zao commented on Oct 25

    Can you please ask Kudlow on your next appearance how he went from “no cut – we need to save the dollar” to “shock and awe 50bp cut we need” in a week. Please.

  12. ~mikey commented on Oct 25

    Regardless of the credit mess, as financial research analyst, I ALWAYS wanted to start a report: “Some half-assed piece of data comes out, the markets spasm, then you want to understand what the data really means.” As it was I got called unprofessional enough times for what I thought was banal statistical coverage. Amen Barry! Vive la margin of error on basis points deltas. ~mmw

  13. mack macdaniel commented on Oct 25

    Important notes, all. But, these numbers from Housing Dept and Census are so shaky, confidence is beneath low. crap in-crap out.

    Stats are fine as proxies for reality, (if one trusts and understands the sample-caveat emptor) but, in this case for housing, I’m only interested in actual CLOSINGS where bank drafts are traded for title.

    Contracts written are meaningless to me, whether from Hovnanian or a local Board of Realtors.

    Fun to watch the constant struggle to read the “tea leaves”, though. At any given moment, there’s only one absolute reality.

    “Statistical significance”? Just give me the entire data set, thank you very much.

  14. Justin commented on Oct 25

    I guess, KISS – keep it simple stupid, just doesn’t work when applied to the markets or the economy? (Oh, why do that! Economist would be able to disseminate more accurate statistics, “but no!”, the markets’ Heavy-Hitters wouldn’t be able to lead the sheep asray, which would therefore be putting less money into their pockets). But wait! they can always get Ben, to lower rates. Wow, what chicanery!

  15. David Dodgen commented on Oct 25

    How many homes are in Southern California?
    What does replacing 1500 homes mean to the builders there?
    How many of those homes were in foreclosure anyway and now have a gift from heaven?
    How long will it take the insurance companies to get the funds flowing or will most of this be uninsured?
    How many people will bail on So Cal the way they did New Orleans?

    I know, I know. I should go get my own information and not pester you

  16. Eric Davis commented on Oct 25

    Like the idiom says “white lies, Lies and statistics”… Market is figuring it out. There is a huge disparity between buyers and sellers right now… Hence volatility… all these bottom tests aren’t buyers, it’s short covering, and scary.
    Best Part.. No “Rate Cut” nonsense, without bad job numbers, or something… this is just a healthy sell off. Just some crying in their oatmeal. Wait till the bears suck some buyers in to get stopped out.

    Cramer started this “rate cut” nonsense, but he just meant that if the fed cut in May, we would go to 16000… Because it would be starting to help now. But they didn’t. So now somehow the market thinks that we should be at 16000 because of the fed cut, and don’t realize “an ounce of prevention…..”
    Don’t blame CNBC they are just journalists, and just feed us the lies the traders/brokers feed them. If it wasn’t for bad data they wouldn’t have any at all.

    I think it’s going to be hard to get the bears off the meat today.
    the NASDAQ is currently down more than the dow… again, I thought tech was working? I bet the whisper on Microsoft is weak.

    I could be wrong, My crystal ball says we test the august lows, slight jump on the fed, trade neutral till we find a weakening consumer and bad jobs numbers. then we test march. Then 10-11K
    Besides Higher P/E, Or trying to convince Retail investors to come in and “hold the bag”.. The market has nothing to hang it’s hat on.

    How can tech work if the chips that make them work arn’t selling..

  17. rex commented on Oct 25

    Good post Barry.

    This is how I report it each month.

    “It can take up to five months for a trend in sales to emerge. New-home sales have averaged 792,000 per month annualized over the past five months, compared with 820,000 in the five months ending in August.”

    That’s the lowest five-month average since August 1997.

  18. jds commented on Oct 25

    My take is that when homebuilders have stopped declaring bankruptcy for two straight quarters, then we have bottomed.

    From the Chicago Tribune:

    Neumann Homes, one of Chicago’s largest homebuilders, announced on Monday that it intends to file for bankruptcy.

    The company said in a press release at 5 p.m. that it had been “unable to procure adequate funding” to operate its business. Marketing director Jean Neumann said that despite other published reports, the bankruptcy had not been filed, though “it will be done shortly.”

    Too many of the numbers are massaged, spun, and processed by the media to make things look better that I prefer the actual bankruptcy filings as hard evidence.

  19. VJ commented on Oct 25


    in this case for housing, I’m only interested in actual CLOSINGS where bank drafts are traded for title.

    And what interest would the real estate bidness have in reporting THAT ?

  20. Nicolas commented on Oct 25

    To be honest, you are wasting your time talking about real estate troubles, subprime, CDOs and so on. Everyone knows about this by now and it’s already reflected in the market. That’s why brokers and banks are down sharply this year and trade at single digit PEs. You are also too focused on US economic problems, there’s a global boom under way and you’re missing it. The bearishness on this board is ridiculous.


    BR: Nicholas, if you think this is fully reflected in stock prices, than you probably believe that the Banks, Brokers, Mortgage Lenders, and Home Builders are screaming bargains.

    You should go buy a boatload of shares in all of these!

  21. Gary commented on Oct 25

    Agreed bear markets don’t start with bearish sentiment high they start with everyone only seing blue skies from now on.

  22. Justin commented on Oct 25

    Hey Nick, you don’t happen to be Santa Claus, do you? I’ll thank you for my (Teddy) Bear ahead of time!

  23. Estragon commented on Oct 25


    The housing thing has been discussed here for quite some time. “Everybody” knew housing was a problem a year ago too, with the likes of TOL and HOV substantially off their highs. Buying into the notion that it had been priced in would be a good way of getting rid of about half your investment in either of those. Likewise, we “knew” about credit issues in the spring, but CFC didn’t exactly have that knowledge fully priced in. If you think banks have it all priced in now, by all means buy the single digit p/e’s.

    BR’s also been clear on the global boom and infrastructure thing, and has repeatedly said he’s long the story.

  24. Estragon commented on Oct 25


    As much as a personal tragedy the CA fires are for those involved, in the context of an industry building ~700,000 homes in a really bad year, rebuilding 1,500 homes is a rounding error.

  25. Nicolas commented on Oct 25

    Sorry, I just don’t see how we can start a bear market with high pessimism, low public participation in stocks, stocks selling at low PEs and high dividend yields, interest rates low, central banks cutting rates and providing liquidity…

  26. Jay Weinstein commented on Oct 25

    “New House Sales Triple in October”

    According to the Census Bureau, Americans bought 3 new homes in October, up from the revised figure of one in September[down from the originally reported figure of 770,000]. Mr. Karl Bildenhausen, President of Better Homes by Bildenhausen, was responsible for all three sales. He was quoted as saying, “My dog Prince can be a very persuasive salesman.”

    Anyone know that humor reference without resorting to Google? LOL

    Keep up the war on bad statistics.

  27. Frankie commented on Oct 25

    “Comparing this month with the original release from the prior month is not “apples to apples” but rather the intentional use of data known to be incorrect.”

    The quote is: “An apples-to-apples comparison (original release to original release)”

    If prior month original release compared to this month original release is not an apple-to-apple comparison, then pray tell what would be one. Maybe it is kiwi-to-kiwi? Mango-to-mango?

    “the intentional use of data known to be incorrect”

    Intentional? WTF are you talking about?
    Has it crossed your mind that it could’ve been an honest mistake? (Assuming there IS a mistake to begin with)

    Is you a mind-reader? Can you decipher intent based on a short sample of written text?

    The FBI is hiring people with those talents. Six-figures starter.

    Don’t wait.


  28. L’Emmerdeur commented on Oct 25

    So if the average cancellation rate is say 40-50% for September (according tot he rates quoted in the post), does this mean the September number might be revised down to, say, 400k?

    That would be… spicy.

  29. Justin commented on Oct 25

    Nick, first goes the consumer, then goes the earnings and then its the P/E ratios. Participation rate is low because they all just got done playing the real estate market, and are maxed out. Employment numbers were skewed from change in the school year of several big states. Lack of unemployment is due to illegals from Mexico working as 1099’ers, or for cash…

  30. me commented on Oct 25

    “So if the average cancellation rate is say 40-50% for September (according tot he rates quoted in the post), does this mean the September number might be revised down to, say, 400k?

    That would be… spicy.”

    Apparently Jeff say you can’t say that.

    He could be right though, this from the report itself:
    “These statistics are estimated from sample surveys. They are subject to sampling variability as well as nonsampling error including bias and variance from response,
    nonreporting, and undercoverage. Estimated average relative standard errors of the preliminary data are shown in the tables. Whenever a statement such as “2.5
    percent (±3.2%) above” appears in the text, this indicates the range (-0.7 to +5.7 percent) in which the actual percent change is likely to have occurred. All ranges
    given for percent changes are 90-percent confidence intervals and account only for sampling variability. If a range does not contain zero, the change is statistically
    significant. If it does contain zero, the change is not statistically significant; that is, it is uncertain whether there was an increase or decrease. The same policies apply
    to the confidence intervals for percent changes shown in the tables.”

  31. hpov2000 commented on Oct 25


    Aug Original 795,000
    Sept Original 770,000

    So “apple to apple” comparision will shows that rate of new home sales declined from Aug to Sept.

    Only when you use Aug (downward revised) number of 735000, you get increase of 4.8% as mentioned in the headline. But in that case it is not apple to apple comparison. You need final (revised) number for sept.

  32. David commented on Oct 25

    Ah, good, I’m glad I found you again. I knew there was one other person on the planet who both followed these silly announcements and had taken Statistics and Data Analysis 101, but I’d lost your blog.

  33. Jeff commented on Oct 25

    To all of those with the fruit and mango stands:

    Government data gets revised as information gets better (more people respond to the survey, for example). The final data series represents the revised data. This is important to those doing economic research, who need to know WHEN effects occurred.

    Those interested in studying the market must be careful to relate market effects to the data known to the participants at the time, not the final revised version.

    The right choice depends upon your purpose.

    When the government is trying to show a monthly change, as they are in this series, it is important to use the most accurate data they have. That is the latest revision. As I noted in an earlier comment, to do otherwise would be misleading, so they do not do it.

    There is no reason to expect that revisions will consistently be in a single direction, as Stuart suggests in his example. There is nothing that makes the unrevised data “an apple” except that it is less accurate than the later version.

    I invite anyone to point out a professional time series with revisions that does not follow the process used by the BLS.

    If anyone is actually interested in learning about this, you can look at the raw data at various government sites.

    Let’s all try to do our best in interpreting data. Last month was worse than we thought at the time. This is not like a football game or something where we do something to this month to even the score.

  34. The Big Picture commented on Nov 30

    New Home Sales Rebound! (Sales drop 23.5%; Prices fell 13%)

    Yesterday morning, Commerce released the New Residential Sales data. Encouragingly, the media has caught on to the monthly nonsense built into the way numbers get reported, as evidenced by this WSJ headline: New-Home Sales Post Modest Gain Because Prio…

Posted Under