November Construction Spending Down 0.4% After Significant Errors Found in Prior Data

With its release of November construction spending data, the Census revised all its construction data going back to January 2005, and admitted a large "processing error" that had caused all residential construction data to be misstated in the interim.  The November construction spending report (pdf) from the Census Bureau estimated that the seasonally adjusted rate of our construction spending in November would work out to $1,122.5 billion annually if extrapolated over an entire year, which was 0.4 percent (±1.5%)* below the revised estimate of construction spending at a $1,127.0 billion annual rate in October but still 10.5 percent (±1.8%) above the estimated annualized level of construction spending in November of last year.  The October spending estimate was revised from a $1,107.4 billion annual rate to a $1,127.0 billion annual rate, as were all the prior months, which will be reflected in revisions to previously published GDP figures when the annual revision to GDP is released this summer.  As a result of these revisions, October spending was less than 0.3% above that of the September, rather than the 1.0% month over month increase previously reported.  This will have the effect of reducing its positive contribution to 4th quarter GDP, even though the actual spending was larger than previously reported...

Private construction spending was at a seasonally adjusted annual rate of $828.2 billion, 0.2 percent (±0.8%)* below the revised October estimate, with residential spending rising 0.3 percent (±1.3%)* to an annual rate of $427.9 billion while private non-residential construction spending fell 0.7 percent (±0.8%)* to an annual rate of  $400.3 billion, largely due to a 4.0% decrease to $83,585 million in spending for manufacturing facilities.  Meanwhile, public construction spending was estimated to be at a rate of $294.3 billion annually, 1.0 percent (±2.5%)* below the revised October estimate, with large drops in spending for public safety, highways and streets, sewage and water supply construction only partially offset by a 5.0% increase to $71,178 million in spending for construction of educational facilities..

The previously published and revised construction data (xls) going back to January 2005 is provided by the Census Bureau as an Excel file, which Mish has transcribed 2 years of in a table at his blog, should you want to review it online.  Although it's been reported that residential construction was revised lower, from scanning the data it appears to me that most reports on it in recent years up until 2013 were revised higher, and as a result the increase from 2014 residential construction spending data to 2015 residential construction spending was smaller.  We aren't going to get into the weeds of what those revisions will mean to prior years and quarterly GDP, but we'll take a quick look at spending for October and November vis-a-vis the third quarter with an eye to how that might affect the coming 4th quarter GDP estimate...

Construction spending inputs into 3 subcomponents of GDP; investment in private non-residential structures, investment in residential structures, and into government investment outlays, for both state and local and Federal governments.  To see how this report of two month's construction spending might impact 4th quarter GDP, we have to first adjust those varied categories of spending for inflation to give us the quantity of construction in real terms.  The Census Bureau construction price indexes for new one-family houses under construction and for new multi-family homes under construction, which shows a 0.5% increase for both months, are specified as the deflator for residential investment..  However, the National Income and Product Accounts Handbook, Chapter 6 (pdf), lists a multitude of privately published deflators for the various components of non-residential investment, such as the Turner Construction building-cost indices for several types of buildings and the Engineering News Record construction cost index for utilities construction, making a true to form estimate of those adjustments too difficult to undertake manually.  However, the producer price index for final demand construction indicates that construction costs rose 1.0% in October and fell 0.3% in November, which we can use as a rough estimate of the deflator for non-residential types of construction...

Using the revised monthly annualized construction spending data for July, August and September from Table 1 of this report, we find that 3rd quarter private residential construction spending was at a seasonally adjusted annual rate of $427,642 million, and that the comparable annual rate of October and November residential spending adjusted for inflation would be at a $430,549 million rate, which would mean that real residential construction rose at a 2.7% annual rate so far in this quarter, vis a vis the 3rd quarter.  For private non-residential construction, we find that 3rd quarter non-residential construction was at a $399,249 million annual rate, while October and November non-residential spending adjusted for PPI construction inflation would give us a non-residential spending rate of $398.187 in chained third quarter dollars, indicating that real non-residential construction fell at a 1.1% annual rate.  Lastly, using just the monthly data in this report, we find that public construction averaged at a $299,967 million annual rate in the 3rd quarter, while public construction for October and November adjusted for inflation works out to a $293,291 million annual rate thus far in the 4th quarter.  Hence, real government investment spending for construction was down at a 8.6% annual rate in October November from the third quarter..

Finally, translating those changes in the rate of construction growth we have for these two months, we estimate that real residential construction growth would add 0.07 percentage points to 4th quarter GDP growth, real private non-residential growth would subtract 0.03 percentage points to 3rd quarter growth, and real public construction growth would subtract .17 percentage points to 3rd quarter GDP across the various government investment components.  With those estimates we have to add the caveat that this construction spending report does not capture everything that is included in the GDP investment categories of the same names, such as brokers’ commissions on sales and other ownership transfer costs on sales of both new and existing residential and non-residential structures, which often have a significant impact unto themselves.

Subject Meta: 

Forum Categories: 

quite a revision on Construction

Kind of an amazing error, but I feel few pay attention to these reports in the first place. Most cannot even grasp the basics of statistics no matter how much one pounds into their heads fundamentals.


Americans in general aren't so good with those number thingies. Lotteries for example are simply a tax on the mathematically illiterate.