December 22, 2009

A Sprinter's Or Marathoner's Pace?

"[U.S. existing home] purchases increased 7.4 percent to a 6.54 million annual rate from a revised 6.09 million pace the prior month, the National Association of Realtors said today in Washington. The median sales price declined 4.3 percent from the same month a year earlier, the smallest decrease since November 2007.

Lower interest rates, cheaper homes and a homebuyer tax credit have resuscitated a housing market that contributed to the worst economic slump since the 1930s. A sustained recovery in housing and the economy depends on a resumption of payroll growth after employers cut 7.2 million jobs in the past two years."

Sales of [U.S.] Existing Homes Increase More Than Forecast [Bloomberg]

First Published: December 22, 2009 9:30 AM

Comments from "Plugged In" Readers

"A sustained recovery in housing and the economy depends [either] on a resumption of payroll growth [or a continued and expanding government stimulus]"

Fixed that for them.

While I'm playing bloomberg editor, anyone notice that same article mentions the second downward revision to the most recent GDP today - from 3.5 -> 2.8 -> 2.2 - the article amusingly omits the original 3.5% estimate, instead calling the 2.2 revision the "final revision".

I guess I have to stop reading bloomberg now too :-)

Posted by: dub dub at December 22, 2009 10:16 AM

They're predicting 5 this quarter.

Posted by: anonn at December 22, 2009 10:18 AM

dub dub -- oddly, they revised this article with a similar URL to mention the 3.5% original estimate: http://www.bloomberg.com/apps/news?pid=20601087&sid=aVeAMaVRygoM

I think some of the AP articles haven't necessarily mentioned that it's a second revision.

You're right about the stimulus.

Posted by: corntrollio at December 22, 2009 10:31 AM

@controlio -- thanks, they did indeed revise the article! So I will continue to read bloomberg after all :-)

Posted by: dub dub at December 22, 2009 10:40 AM

So they report a number for a single month as a annualized rate. Why the hell is that useful?

Second, I'm assuming they are reporting YoY growth and not MoM, right? So we're comparing to Nov 2008? I'm not sure anything from October 08 to March 09 is worth comparing to anything else, since just about everyone panicked, thinking the world was going to end.

Posted by: rr at December 22, 2009 10:52 AM

rr -- that's a quarterly rate for 3Q. Please read the article -- not MoM or YoY. Whether it's useful or not as an annualized number, I don't know.

Posted by: corntrollio at December 22, 2009 10:57 AM

Sorry, I made a hash of that in revising my response a couple times. That's an annualized rate for 3Q, not a quarterly rate.

Posted by: corntrollio at December 22, 2009 10:59 AM

Correct me if I'm wrong, but the sales number is November only, though the GDP number is a 3Q number:

Sales of existing U.S. homes in November rose to the highest level in almost three years as first-time buyers rushed to take advantage of a government tax credit and lower prices.

Purchases increased 7.4 percent to a 6.54 million annual rate, exceeding the highest estimate of economists surveyed by Bloomberg News, figures from the National Association of Realtors showed today in Washington. Another report showed the economy grew a less-than-forecast 2.2 percent in the third quarter as companies cut stockpiles, pointing to manufacturing gains at the start of 2010.

Posted by: rr at December 22, 2009 11:18 AM

Also, based on this article (the fourth source I looked at), the 7.4% increase is a MoM increase. Everyone seems to be speculating the obvious: the increase was due to the credit expiring on Nov 30.

Posted by: rr at December 22, 2009 11:22 AM

Sorry, yes, we're talking about 2 different things. I assumed you were responding to my response to dub dub. The home sales number is indeed reported monthly.

MoM housing numbers are tricky, because housing has pretty obvious seasonality. The seasonality this year is highly affected by stimulus, so it's a bit strange.

YoY for housing is slightly less tricky because it helps deal with the seasonality. You are right that 2008 numbers may make YoY less helpful this year, but usually YoY numbers are helpful.

Posted by: corntrollio at December 22, 2009 11:24 AM

"Everyone seems to be speculating the obvious: the increase was due to the credit expiring on Nov 30."

Indeed, it'll be interesting to see what the effect of extending the tax credit will be. My guess would be that at least some of the future demand was pulled forward (similar to Cash for Clunkers), but how much remains to be seen.

Posted by: corntrollio at December 22, 2009 11:26 AM

The economy is obviously picking up, just as predicted 9 months ago.

http://finance.yahoo.com/news/Recovery-not-as-strong-as-apf-2679594626.html?x=0&sec=topStories&pos=2&asset=&ccode=

Here is a story that gets all the info about the readjustments to GDP correct. According to it, we are growing at a yearly pace of about 5% right now. There is other evidence of strength in the economy, such as more overtime and more hours worked. This usually precedes hiring during a recovery. The bond curve is also very steep, which is another strong indicator for growth.

I expect that the recovery will be a long slow slog though as it will take a while for consumers to repair damage to their balance sheets. Most of the economic growth will have to come from corporate and government spending. Most of the longer term signs for the economy are positive, as the savings rate is up over 5%, from negative just a few years ago. The trade deficit continues to shrink, as America turns into in exporting nation. There were some worrying signs that this was starting to tick back up and bears watching, probably more than any other economic statistic.

http://www.nytimes.com/2009/12/11/business/economy/11econ.html

Posted by: NoeValleyJim at December 22, 2009 11:58 AM

"economy grew at a 2.2 percent pace in the third quarter, less than previously thought."

This is the second revision downward. Its odd that they don't mention it in the article

"According to it, we are growing at a yearly pace of about 5% right now."

I read the article as "The economy is probably growing at nearly 4 percent in the October-to-December quarter, analysts say. A few peg it closer to 5 percent."

We'll see. It would be nice but I still have my doubts. The headlines also lead with the great news and page 6 has the revisions downward. I noticed this with the cheerleading of sales growth this Christmas

Posted by: Zig at December 22, 2009 4:00 PM

Hi List,

re: GDP, here is the data so you can check for yourself:

http://www.bea.gov/national/xls/gdplev.xls

Quarter/Nominal GDP SAAR/Chained GDP SAAR

2008q3 14,546.7 13,324.6
2008q4 14,347.3 13,141.9
2009q1 14,178.0 12,925.4
2009q2 14,151.2 12,901.5
2009q3 14,242.1 12,973.0

If you want the current (annualized growth rate), then compute:

4*100*(2009Q3/2009Q2-1) = 2.57% (nominal), or 2.22% (real).

The BEA does this math for you here:
http://www.bea.gov/national/xls/gdpchg.xls

To get the annualized GDP deflator (an output-weighted price index)

Use the definition:

Deflator = 100*(Nominal GDP)/(Real GDP)

So in this case, to get the current growth rate of prices, you would compute the quarter-over-quarter change and multiply by 400 again.

400*((NominalQ3 * Real Q2)/(Nominal Q3 * Real Q3) -1) = 0.35%

And then you can verify that, roughly,

growth rate of deflator + growth rate of real GDP = growth rate of nominal GDP.

The BEA shows deflator levels here

Therefore on a quarter-over-quarter basis,

* inflation is running at 0.35% (using total output inflation, not CPI)
* real output is growing at 2.2%
* nominal output is growing at 2.6%


Merry Christmas!

Posted by: Robert at December 22, 2009 7:07 PM

Thanks Robert! Nice to see your keen data analysis back on socketsite.

You are right Zig, most economists are estimating GDP this quarter at 4% annualized growth rate.

Something that is kind of annoying is that all the GDP revisions have been downward for the last few years. There is some bug in their methodology which they still haven't worked out. It is a relatively recent thing, as they used to both under and over estimate GDP growth with the early forecasts pretty regularly. Now they are all high.

This ends up having significant policy implications because the push for another stimulus package pretty much died after the stronger than expected Q3 numbers came out. The BEA needs to get its act together here.

Posted by: NoeValleyJim at December 23, 2009 9:41 AM

Good to see Robert back -- thanks for the detail on GDP. Hope this means LMRiM and San FronziScheme will be back too.

Also, NoeValleyJim -- That's not a bug; it's a feature! (sorry, old software joke)

Posted by: corntrollio at December 23, 2009 10:09 AM

It's not a feature, it's a marketing trick. Announce rosy numbers - Quarter over quarter increases.

Then next quarter, go back and revise the previous quarter down so that your current quarter's quarter-over-quarter increase looks better, so you can announce rosy numbers again.

You can do this as often as you like. 10% growth *every quarter*, as long as you revise the previous quarter down to 1.2% every time.

Posted by: tipster at December 23, 2009 11:39 AM

The Guv has understood that psychology will make or break the recovery. From green shoots to rebound to sustained recovery, the official scenario is pretty clear there.

Now, there are some numbers that cannot be easily tweaked to your liking (unemployment), but some can be temporarily cranked up with mucho cash infusion (8K buyer credit, cash for clunkers) or simply fudged a-la-tipster (GDP). The hope is this will give time to lenders go back to solvency so that they can take over when the Guv stops buying mortgages.

Maybe the home buyers will go back to buying and shoppers will go back to shopping. For now, both are busy repairing their balance sheet just like everyone else. This recovery looks like a debt-financed dead-cat-bounce.

Posted by: wow at December 23, 2009 12:59 PM

I am not as conspiracy minded as the rest of you, but this is one situation where your theory actually makes a hint of sense. Much of economics is behavioral and improving "animal spirits" is at least as important as anything else.

If we really want to improve the employment numbers we need to institute a jobs program, like FDR did. Heaven knows that there are enough sidewalks, roads and bridges that need repair to keep millions busy for years. Too bad we don't have the political will to do it. It would be much cheaper per job than the stimulus that has been attempted so far.

Posted by: NoeValleyJim at December 23, 2009 1:30 PM

I don't think it's intentional fudging, I think the "qualitative" nature of this bailout has queered certain existing models, which is why they are so far off.

Statistical modeling is very much a garbage-in/garbage-out activity, and it's too soon to change the models because these interventions are supposed to be (ahem) temporary.

It would be nice if the models were consistently wrong in the "good" direction (positive revision), because we could at least say that the intervention (as ill-advised as it may be long term) was really working short term, unless they were fudging the revisions :-)

It's easy to manufacture tepid GDP growth (and even employment growth - stay tuned) with > 10% GDP deficits.

I think it's a little tone-deaf for the govt. to throw around the word "recovery" in a technical sense which has no practical impact on 85% of the population, at least until they start paying people to hold shovels or to work at the IRS to increase treasury revenue (or whatever).

Posted by: dub dub at December 23, 2009 2:19 PM

Re: bias in revisions

This is what you would expect. Benchmark GDP data is performed every 5 years. That is the only time you get all the data. At all other times, including annual, quarter, and monthly GDP releases, you are getting only partial data, and you fill in the gaps by assuming correlations between the data you know and the data you don't know.

To see this, consider employment. Suppose you can only measure employment at the top 100 companies each month, the top 200 companies each quarter, the top 1000 companies each year, and a comprehensive census of, say 80% of all companies is done every 5 years.

Suppose that historically, you notice that in "normal times" if employment drops by 1% in the big companies, it will also drop by 0.8% in medium sized companies, and 1.1% in small companies.

In your first release, you use these historical correlations to estimate employment. But suppose many small companies are dying off in proportion to historical trends. In that case, unemployment will be greater than you expect. And for the same reason, inventory build up (when measured from the big companies) will overestimate total inventory build up. Personal expenditure data collected from the largest companies will overstate total consumption, etc. This will happen any time the harder to measure data (e.g. home improvement expenditures, small businesses) is more sensitive to economic swings -- which is typically the case.

Unusually strong downturns *should* be accompanied by a consistent series of downward revisions, just as unusually robust expansions should be accompanied by sustained upward revisions. In the present case, both the downturn itself and the recovery have been unusually weak. Expect continued downward revisions, but from the methodology, not from any political or psychological efforts.

For my part, I doubt a real GDP growth rate in excess of 3% in the fourth quarter. I think a nominal GDP growth rate of 3% is more likely, with real GDP in the 2% range and a deflator in the 1% range, when all the revisions are out.

Posted by: Robert at December 28, 2009 6:23 PM

Post a comment


(required - will be published)


(required - will not be published, sold, or shared)


(optional - your "Posted by" name will link to this URL)

Remember Me?

(you may use HTML tags for style)


Continue Perusing SocketSite:

« The "Confidential" Infinity Short Sale Scoop: 301 Main #9E | HOME | Glass Tower To Rise Over SF Mining Exchange Building On Bush »