April New Home Sales – Revisited

On Tuesday, we noted that New Home Sales fell 42%. There was one small piece of the data I failed to mention earlier in the week, which is worth discussing — the March revisions:

1) Revisions: April’s (unrevised) data for new homes was 526k annualized units (+3.3). That is the identical to the number released in March — 526k units. March sales were revised to -11%
from -8.5%. So but for the revisions, March to April headline number was flat.

How did we show a 3.3% monthly gain? Tuesday’s report saw March revised DOWNWARDS from 526k units to 509k units. In other words, April did not so much showed a positive move upwards (statistical error notwithstanding) as much as the prior month comparison was revised  downwards.

2) Apples to Oranges: Why do we compare Revised versus Unrevised data? The overall trend for the past year has been mostly downward revisions. An apples-to-apples comparison would be an original release to original release (that showed flat data, not an increase in new homes sales).

Comparing the original (but soon to be revised) April data to the revised March data presents a misleading picture.

3) Cancellations: Of course, none of the new home sales data includes cancellations, which were running north of 30% — and with the recently tightened credit, it may be even worse.


UPDATE May 30, 2008 12:55pm

Over at Calculated Risk, Tanta disagrees with my assessment:

"Cancellations are not getting worse. In fact they are getting better.
For most builders, cancellation rates peaked in Q3 2007 (with the
credit crunch) and have improved significantly since then. And it’s the
change in cancellation rates that matter when analyzing the New Home

This is a key point: right now the Census Bureau is probably underestimating sales!

I do not know if cancellations are getting worse (I wrote "may be even worse"). However, we do know several things:

1. Census Department does not factor in cancellations at all. Let’s say for arguments sake that cancellations are improving — from 30+% to 15+% or better —
that only means the overstatement of sales is less, but still
overstated nonetheless.

2. Cancellations represent potential understated sales in some indeterminate

future. Past cancellations might be getting sold today — or they might not. We don’t really know. One would think if they were, however, we should see some evidence somewhere in some data — sales, homebuilder inventory, etc. 

I have yet to see that anywhere (doesn’t mean it doesn’t exist, i just haven’t seen it).

3. Right now, all of the YoY, inventory, and price data suggests that prices & sales are still falling. If and when New Home sales actually see real month over month, YoY improvements,
then CENSUS will be understating it New Sales. But thats somewhere off in the
(possibly distant) future.

If someone can show me a definitive basis for thinking New Home Sales are better than reported, I mention it on Kudlow & Co. But until then, do not count your chickens before they are hatched!


No, New Homes Sales DID NOT Rise . . .   (October 2007)

New Residential Sales 
The Census Bureau MAY 27, 2008   

Download New Res Sales_5.27.08.pdf

Cockeyed Optimists See Housing Recovery
Barron’s May 28, 2008

Print Friendly, PDF & Email

What's been said:

Discussions found on the web:
  1. Stringm commented on May 30

    Next month when they revise April’s down to 509,000 and report May as 526,000 that will be back to back monthly gains of +3.3%. We will be screaming towards recovery then!

  2. Jim Haygood commented on May 30

    “Comparing the original (but soon to be revised) April data to the revised March data presents a misleading picture.”

    Precisely. But then, a Ponzi scheme is all about CONfidence. “Releases first, data later!” as the Red Queen might have said.

    Rather than seasonally adjust the data to enable month-to-month comparisons, I’d rather have raw data which can only be compared on a year-over-year basis. Just as useful, and less scope for hanky-panky.

  3. VennData commented on May 30

    Agree with the raw data point(s). What is the problem with the unvarnished truth?

    Layout your your seasonal – and other – adjustments, skip all the narrative, give us the raw data as it comes in.

    What they did to the monthly Federal debt figures is an outstanding example; they stop it at.. guess when? 1999. Hmmm…


    .. then to get the “Bush years” you’ve got to go through this clunky process (We have no need to wonder why the Y-o-Y debt increases have no relation to the headline deficit figures they churn out, McClellan-like.)

    These .gov web sites are crack obsfucationists. We need one of the BLS guys to write a tell-all.

  4. Marcus Aurelius commented on May 30

    Other numbers recently released by our “officials”:

    -3.2%, $30,000, 1.558/1, and 52.

    If you don’t like these numbers, feel free to make-up your own.

  5. crack commented on May 30

    Good point on unrevised to unrevised. Shouldn’t this apply to year to year comparisons too?


    BR: Yes. (If we went revised to revised YoY, the -42% might be even worse).

    The most important difference between the monthly and YoY is that the monthly has been consistently below the margin of error or, “Not statistically significant” in wonk speak. The YoY changes have been “statistically significant.”

    My point is that people seem to be excited about a number that was a) not significant statistically; and b) essentially flat.

  6. Tim commented on May 30

    I recall one of your commenters already pointed this out on Tuesday. We should all pay more attention to this – there is a lot of significant contributions being made and many of your commenters have fantastic insight.

  7. Lilguy commented on May 30

    I’ve been tracking a similar phenomenon at the NAR on existing home sales. A pattern has emerged in which the preliminary sales numbers are consistently revised downward the following month.

    This allows for the announcement each month about sales to be more positive than the revised data would permit.

  8. Donny commented on May 30

    Consumer confidence fell to a 28-year low in May as inflation expectations soared

    I wonder if this represents another “bottom” indicator, from the PermaBulls. It seems like the “bottom” callers are losing, what little (if any), credibility that have left.

    The stock market is trading up on the news. LoL

    This isn’t going to end well.

  9. SR commented on May 30

    Er, Ticking Time Bomb? And how does that factor in? (or does not?)

    May 30, 2008
    Lose Homes, Pay More Tax

    Some of the biggest losers in the real estate slump are not purchasers of mansions they could not afford. They are buyers of second homes — or third ones, for that matter — who are sitting on a tax time bomb.

    Many of these people will lose their properties in foreclosure and then stagger into bankruptcy under the weight of a sizable tax bill. While Congress has granted some tax relief to people who lose their primary homes, there is no such aid for those who fall behind on payments on a getaway condo in Las Vegas, a retirement home on the Florida coast or an old house that they are renting out for income.

    Bankruptcy lawyers say they are seeing a wave of foreclosures among owners of second homes in such a position, owners who thought they had found sound advice for financial security. (cont.)


  10. Ed Miller commented on May 30


    Tanta at Calculated Risk has comments this morning about cancellations.

    “Cancellations are not getting worse. In fact they are getting better. For most builders, cancellation rates peaked in Q3 2007 (with the credit crunch) and have improved significantly since then. And it’s the change in cancellation rates that matter when analyzing the New Home data.

    This is a key point: right now the Census Bureau is probably underestimating sales!”


    In spite of this, I still think we are in a recession. I agree with your other points and appreciate the information you provide. You are usually my first morning read.

  11. KnotRP commented on May 30

    This has been going on so long (Lereah, Yun) that it needs a name….a shorthand. You should run a naming
    contest for this method of rewriting history to obtain
    an upturn in the statistics.

    Apparently, if you jump out of a plane, but face the
    sky, you are in imminent danger of floating into space,
    in the NAR world.

  12. Jmay commented on May 30

    As Scott McClellan wrote in his new book, this administration is in permanent campaign mode. So, of course statistics will always be doctored to put the best possible spin on things.

    The question is why people don’t get angrier about it. We just point it out and feel smug and self-satisfied and then go back to being ruled by shameless liars.

    We get the statistics we deserve.

  13. AGG commented on May 30

    The “national defense” expenses are a national offense. It would be nice if all this spent money actually helped us. I suspect that if added wealth to military industrial complex corporations is your definition of added security, then the shoe fits. However, for 99% of Americans, it appears we are going barefoot.

  14. Liv commented on May 30

    Credit where credit is due: it was CR himself, not Tanta.

  15. Toby commented on May 30

    Statistical noise!! Current releases are too dirty. You are still comparing a dirty rotten apple to a dirty rotten apple. Error rates in revisions are heteroskedastic. There is no consistency in the variance. Wait a month or two for the true picture. Better yet, use a three or four month moving average.

  16. Bruce commented on May 31


    Heteroskedastic?…had to look it up, didn’t take statistics in college…wonderful post…

    Also my first morning site to visit in the morning…..thanks Barry.

    Bruce in Tennessee

Posted Under