Change in May New Housing Starts cannot be determined from Census Report

There were two major housing reports released this week; Robert already covered one of them, on existing Home Sales from the NAR. Today, we're going to cover New Residential Construction during May from the Census Bureau, which estimates the numbers of new building permits, new home starts and home completions during the month.  As we've mentioned before, this report is usually watched and widely covered for details on new housing starts, but the data has such a large margin of error due to the small sample it's taken from as to render the monthly reports useless for any serious analysis. To collect data for this estimate, census selects a sample of building permits from a random sample of approximately 900 permit offices nationally, to be tracked monthly and in person by field representatives, and further has field representatives conduct a road canvass in a sample of roughly 70 land areas not covered by building permits, to identify the start of new residential buildings. As inexact as that is, we'll again cover what census reported & untangle it anyway, so no one can be snookered by media hoopla over this report..

According to the Census Bureau, May housing starts were estimated at a seasonally adjusted annual rate of 914,000, 6.8 percent (±10.1%)* above the revised April estimate of 856,000 and  28.6 percent (±14.4%) above the May 2012 rate of 711,000; April's figure was revised from 853,000 starts, a five month low...the asterisk leads one to a census footnote which says "it is uncertain whether there was an increase or decrease" of housing starts in May...what the numbers mean is that census took their rough estimate from field reps of 86,200 homes started in May, a figure which has a relative standard error of ±4,000, and extrapolated that into an annual estimate based on the normal relationship of May starts to annual starts...taking the margin of error into account, what that means is that Census is 90% confident that May's housing starts, were they continued at the same pace over the year, would result in somewhere between 883,838 and 1,068,466 million homes being started over the course of a year, obviously not a figure that can be reported with any accuracy.  The year over year change, 28.6% (±14.4%), has an even wider margin of error, but at least we can reasonably say that housing starts have increased by a decent number since last year...

FRED Graph

  The data on building permits is the section of this report that is more useful, to the extent that the margin of error is small enough one can get an idea of the actual activity for permits, since about half of the permit-issuing places in the US are surveyed monthly. For May, census reports that building permits issued in May were at a seasonally adjusted annual rate of 974,000, which was 3.1 percent (±0.9%) below the revised April annual rate of 1,005,000, and 20.8 percent (±1.3%) above the estimate of a 806,000 rate from last May.  The annualized number was extrapolated for the May estimate of 93,400 permits and included 62,300 permits for single family homes, 2,300 permits for residences with 2 to 4 units, and 28,700 units in buildings with five units or more.  9,800 permits were issued in the Northeast, of which 5,400 were for single family homes; in the Midwest, there were 15,700 new permits, of which 11,000 were for single family homes; 46,100 permits were issued in the South, which included 32,100 single family homes, and 21,800 building permits were issued in the West, of which 13,800 were single family structures.. Since 98% of housing built in the US is in areas where a permit is required, this metric gives us a reasonable idea how much housing construction is intended Of course, applying for and getting a building permit does not guarantee that construction will start. Our FRED graph above shows monthly housing starts in blue and building permits in red as reported on an annualized basis; while individual monthly figures may not be dependable, there is clearly a rising trend for both housing starts and building permits issued..



housing starts cannot be determined

I think we are the only financial site which amplifies the margin of error on housing statistics and the same with the advance report on Durable Goods. Wall Street just blindly jumps on the number, which is almost always revised.

I think the reason, beyond the financial press, that traders just jump on the number with brain dead zeal is HFT.

HFT utilizes these figures and has algorithms automatically trading on them. This also explains the insane rush to obtain the number and why the release time is held down to the second.

Anyone who gets a jump on the number has a trading advantage but the thing stupid is all of this?

It becomes trading on just numbers with no comprehension on what things mean, simply because a host of HFT algos trade on those figures as do regular traders.

So, we get mind numbing nothing when we have a statistical release which has a margin of error often larger than the reported number of housing starts!

Sigh, they need to ban and regulate HFT, it's become a game of router hop delays instead of any indicator on the future profits of that business.

You must have Javascript enabled to use this form.

this has bothered me for a while

month in and month out i see article after article quoting just the census headline housing numbers as if they were gospel...and even well known bloggers, who must know better, ignore that wide margin of error...if they'd just acknowledge it once in a while, then they could go on reporting what census says and i wouldnt blame them...but in continually pretending that these reports are accurate, they lose creditablity in everything else they write...

You must have Javascript enabled to use this form.


statistics and what they mean

I've personally gone to great pains to be accurate on statistics. I've called up experts when I cannot nail something down and on productivity, you have no idea what a pain I was to the BLS in order to ascertain CPI-U-RS vs. CPI-U.

I have pointed out errors with many who ignore significant digits in calculations, especially intermediate calculations.

I hand calculate extensively, from raw data series, dig out data series, derive data and all of that original number crunching actually should be for white papers for it takes a lot of time and effort to do so.

There are others who do this, Chinn does.

So, we maybe less famous, but we sure do try on this site to be extremely accurate, but it does not "pay" in terms of readership and traffic.

Many people simply cannot read any number, their mind goes numb. Out Congress relies on that fact as do many corporations!

You must have Javascript enabled to use this form.

housing in general

i'm not an economist, an engineer... can someone who knows more than i do please explain the "housing recovery"? How can a nation whose largest employer is Wal-Mart, a nation where wages for the middle class have been stagnant since the 1970's support a housing turn around?

I just don't get it. Someone please explain.

You must have Javascript enabled to use this form.

Housing "recovery"

By limiting the supply, keeping the shadow inventory off the market and institutional investors, they cause prices to soar.

That said, home ownership is at record lows and rents are soaring for you're right, people cannot afford homes and worse, soaring rents mean people cannot afford a roof over their head.

If you are an engineer, you're probably aware 6 figures in California means you can barely make rent and are living paycheck to paycheck. Horrific!

You must have Javascript enabled to use this form.

yes $73,000/yr pre-taxes and

yes $73,000/yr pre-taxes and you can barely get by in CT. The only way i will own a home in Fairfield county CT on my income is if my parents leave me or let me buy for less than market value the home i grew up in.

And they built that home new in 1977, my mother, now 70 just a high school graduate and payroll clerk and my father had a 2 year technical degree after high school as a graphic artist.

You must have Javascript enabled to use this form.