New Home Starts

March 20, 2007 7:02am by

Starts_20070319New Housing Starts are due out at 8:30am.

We’ve tore these numbers apart enough in the past that it would just be going thru the motions to do so again.

For those of you new to the site, check out these prior discussions on the problems with this (and related) data series:

• New Home Starts? Don’t Make Me Laugh!

• A Closer Look at New Home Sales Data

• Dissecting New Home Sales Data

I cannot find the piece we did last year about the tendency for this data to mean revert month-to-month (Can anyone find that post? I’ll update it later on site). 

Bottom line:  Any single month’s data is very noisy, while a moving average is much more accurate. In fact, whenever one month was at any particular extreme (in either direction), the following month nearly always reverted.

So when this news comes out later today, take it with a grain of salt.

UPDATE March 20, 2007 3:26 pm

Earlier today, I mentioned the issue of new Home Starts — the data I was trying to track down was about New Home Sales.

Here’s what we found about Sales:

a) Often, the data appears to be "statistically insignificant," according to the Census Bureau;

b)
Strong historical numbers (like plus 13%) tend to be subject to
revision, but mostly stay net postive, albeit somewhat moderated;

c) Over the past 10 years, double digit months have been followed by flat to negative data the very next month (Mean Reversion).

I suspect (but have not tested) that a similar pattern exists for New Home Starts . . .

Print Friendly, PDF & Email

What's been said:

Discussions found on the web:
  1. GPS commented on Mar 20

    . . . take it with a grain of salt unless it supports your bias.

  2. Josh commented on Mar 20

    Absolutely right on this Barry. I think that July number is still changing! Originally 1,022,000 SAAR and keeps going down.

    As with any economic stat that tends to be revised, always take a running 3 month average for the true stat.

  3. Jay Weinstein commented on Mar 20

    Does anyone with a brain really think starts were up 9% in February? And if so, how is that possibly good news for the economy— we know for sure that demand stinks, inventories are rising….that’s from the builders themselves. All these starts will make it even worse, if in fact they exist.

    For those who have never read the classic book “Innumeracy” by John Paulos, I recommend it.

  4. J Hudson commented on Mar 20

    Actual (Consensus)
    —————-
    3/20 Starts
    1.525M (1.45M). Beat expectation by 5.17%. Up 8.3 % month over month

    3/20 Permits
    1.532M (N/A). Down 2.3% month over month

    2/16 Starts
    1.408M (1.600M). Missed expectation by 12%

    2/16 Permits
    1.568M (N/A)

  5. Jean C. de Mestral commented on Mar 20

    Regarding the mean reversion characteristic of the New Home starts numbers: the Hurst exponent is the right tool to measure this. Data from March 1966 to February 2006, monthly, (source: Economagic: US Total New Privately Owned Housing Units Started; Thousands; SAAR) give a result of 0.41. A number below 0.5 shows a mean-reverting time serie, and a number above 0.5 shows a trend-reinforcing time serie. The calculation clearly shows that this particular time serie is of the mean-reverting type.

Posted Under