New Home Starts? Don’t Make Me Laugh!

I continue to be astonished by the econopundit’s lack of solid connection with reality. It’s as if Faith-based initiatives have worked their way into economic analysis and forecasting.

The reaction to last week’s New Home Sales — a surprising headline of a 4.1% rise in August — suggests to me that many people simply do not read beyond the headlines. That is a terrific way to insure you have an incorrect grasp of the data and details of any economic release.

This report was a perfect example: If you didn’t know 3 factoids about New Home Sales, you might be tempted to believe sales are stabilizing, and that the worst is over for housing. Here’s what you need to know anytime you see the New Home data:

– The data itself is reported not by a neutral observer, but by the Builders (an interested party) to the Commerce Department;
– The margin of error is typically greater than the reported number, rendering it statistically insignificant;
– Cancellations of New Home Sale contracts are omitted from both the inventory and sales data.

Let’s have a look at how strong the report was:

New Home Sales only appeared to rise due to a downward revision of sales in May, June and July. But for those downward revisions, the overall trend of slowing sales has been continuing. And, once the August data is revised, it will likely show further deterioration in sales.

Why the difference between reported sales and reality? One reason may be cancellations. As Bloomberg’s Caroline Baum noted, "rising  cancellations aren’t being captured in the aggregate statistics because of the way the survey is designed. Hence, sales are being overstated and inventories understated."


• Lennar said its cancellation rate was running at more than 30 percent. That means about 1 in 3 i of Lennar’s reported new home sales not only didn’t sell, but are back in the already bloated inventory stockpile. That’s on top of the 5 percent decrease in new slaes Lennar had in their quarter ending August 31.

• At D.R. Horton  (the 2nd largest homebuilder  in the US) "cancellation rates rose to 29 percent in the April-June quarter." They deteriorated further in July. That’s about double their historic average;

• KB Home said net orders plummeted 43 percent in the three months ended August 31. That rate is inclusive of cancellations.

As you can see from 3 of the largest US homebuilders, actual new home sales — as opposed to reported ones — did not increase over 4%. Including cancellations, I would guess that new home sales actually dropped somewhere between 5-15%.

Consider too what has been happening in some of the formerly hottest housing regions of the country:  Barron’s reported that Florida, for example, saw August existing home sales drop an amazing 34%; condo sales plummeted 41%.

Given that the homebuilder’s sentiment index (expectations of sales six months) fell to a
15-year low, perhaps the actual data is even worse (we just don’t know). 

Stabilizing? Nothing could be further from the truth.

Sales & Inventory, New & Existing Homes (Combined) 


graphic courtesy Northern Trust


Florida’s Housing Hurricane
Barron’s, October 2, 2006      

Think Housing’s Stabilized? See Cancellations
Caroline Baum
Bloomberg, Sept. 29, 2006

Print Friendly, PDF & Email

What's been said:

Discussions found on the web:
  1. zell commented on Oct 3

    Change the name of your blog to Eversharp.

  2. KL2005 commented on Oct 3

    “- The margin of error is typically greater than the reported number, rendering it statistically insignificant;”

    I have bad news for you. One of my economics professors worked on many statistics for the government before he started teaching. He said almost ALL economics statistics errors were greater than the change in the reported number.

  3. Barry Ritholtz commented on Oct 3

    I can’t speak to any others — this is one that typically has a 15-20% +/-

    What other specific BLS/BEA/Commerce dept stats are you aware of that have a greater margin of errors?

  4. TRJ commented on Oct 3

    Great post Barry.

  5. Mark commented on Oct 3

    I believe that if you look deep into the original reports they give a 90% confidence interval on this and it was somethink like +/- 11.5%.

  6. Isaac commented on Oct 3

    Good thing we have a data-dependant fed.

  7. Bob_in_ma commented on Oct 3

    “Barron’s reported that Florida…”

    What your quoting from was primarily an opinion piece by I guy who’s in a pissing match with Lennar. I can’t believe Barron’s put this in without a bio or tagline.

    Don’t get me wrong, I have puts on Lennar and agree with his thesis. But it would be a mistake to refer to this as reporting.

  8. Greg commented on Oct 3


    I think you need to look at your analysis a little on this one. Cancellation rates are not netted against sales, its a component of the order number. So the net order numbers they report already consider the higher cancellation rates.

    I agree the market is not stabilizing, but I don’t think the sales numbers are revised because of cancellations.

    I enjoy your blog.


  9. ac commented on Oct 3

    To be fair, I think the homebuilders have been much more forthright than the NAR and finance industry in the past few months. Notably the chief economist of the NAHB has openly been very bearish, and even one of the homebuilders (Lennar, I think) said August was probably the weakest month yet.

  10. wcw commented on Oct 3

    I thought the actual surprise were the pending-sale numbers yesterday. While I am as bearish as the next on residential housing and hence on homebuilders, I covered a few shorts on that number, though I kept my RUF puts. I don’t see any good news coming, but I do worry about bear-market rallies.

  11. S commented on Oct 3

    Here’s something to consider. HOV announced on its last conference call that it was discontinuing its share buyback program. At the time, the stock was hugging a multi-year low and was trading at less than 1.0x book. Are they cononcerned about liquidity in the event of a prolonged downturn?

    Note: I determined they have a big revolver, but I haven’t gone through the loan agreement yet to see how tight the covenants are. I’m not sure how much of it they can access.

  12. beechdriver commented on Oct 3

    Wasn’t it also true that the only reason that Aug was positive was that they revised the July number down? If July had been unchanged then Aug would actually have been negative?

  13. Mike commented on Oct 3


    Orders which are made and subsequently cancelled are left in the orders number. For a good explanation of how the number is contrived, see here [].

  14. Bob_in_ma commented on Oct 3

    I for one had that wrong. Greg is correct, cancellations are never accounted for in these stats. The revisions must just be due to inaccuracies in data gathering.

    S, Barron’s this week has an article on homebuilders’ debt problems.

    The main thrust of the article is to point out that many of the builders have bought land through joint ventures, because if they own less than 50% of the JV, they don’t need to show the JV’s debt on their balance sheet. There’s a very useful chart. HOV had a lot of JVs and a lot of option deposits.

    I read somewhere that debt holders were likely to make it clear to builders that spending their cash on share repurchases was probably not a good idea.

  15. rvb1977 commented on Oct 3

    The downward spiral seems to have only begun…historically housing starts fall very quickly once the rest of america – the majority of participants in the housing market that are not enomists / analysts figure out they can wait a while before buying. But then, they usually start to snap back quickly – once homes are again affordable! Like you’ve reiterated a bazillion times – looking at trends in the data is more important than one number.

    Keep up the butt-kickin’ work Barry – I read The Big Picture everyday.

  16. Lama commented on Oct 3

    Bob in MA,
    Your point is well taken. The rules can allow as much as 97% investment in the subsidiary before having to consolidate. Scary under these scenarios.
    If you want to put yourself to sleep, you can Google “variable interest entities”.

  17. Bob A commented on Oct 3

    Why isn’t CNBC highlighting the fact that “DJIA buy and holders are now back to where they were in 2000 not taking inflation into account”? Homeownership in many places may well be looking forward to similar performance for the next six years.

  18. Greg commented on Oct 3


    When order numbers are reported from the big public homebuilders, it is net of the cancellations. I have spoken to many of them about this. Average cancellation rates in orders is 18% – 20%, and that is the number that has increased significantly. The blog you are referring to is talking about sales, which is a separate data point.


  19. Mark commented on Oct 3

    Now, NOW that we have that stupid high can we get on with SERIOUS business?

    (Doubt it.)

  20. Cherry commented on Oct 3

    I hardly ever look at “intial” estimates with these types of numbers, but the revisions. They are more telling for starts and sales.

    Looks like the market is running at 2003 levels of production. Now that we have entered the RE “cold season”, will more people jump off the economic sector? Stay tuned.

  21. Lama commented on Oct 3

    Bob A,
    Have a look at the S&P, which is a broader index and more representative of where the public puts its money.

    1478 in Jan 00 to 1334 today (with the recent run-up). That’s a 10% loss without regard to almost 7 years of inflation.

    If you were a risk taker in 2000:
    NASDAQ was 5132 in Mar 00 versus 2243 today (again, think of the recent run-up). That’s a 56% loss after almost 7 years (without inflation).

  22. bob commented on Oct 3

    What’s funny – the September sales will be reported with another rise and we will see the same “surprise increase” headlines again!

    The reason is, the August data will be revised down without much of fanfare, and then September unrevised data will be above August!


  23. Sarah Reiter commented on Oct 3

    Down down down, these numbers may be a little skewed since the are not taking into the considerations that the homes might not be sold for less then value, but the incentives they are offering make the rest of the homes decrease value, because what you can get for the same price months ago is way more of a house now.

Posted Under