Construction spending fell 1.7% in March, after construction spending for both January and February were revised much higher.  The Census Bureau's report on construction spending for March (pdf) estimated that the month's seasonally adjusted construction spending would work out to $1,284.7 billion annually if extrapolated over an entire year, which was 1.7 percent (±0.8%) below the revised annualized February estimate of $1,306.4 billion, but 3.6 percent (±1.3 percent) above the estimated annualized level of construction spending in March of last year. However, the annualized February construction spending estimate was revised 2.6% higher, from $1,273.1 billion to $1,306.4 billion, while the annual rate of construction spending for January was revised 1.7% higher, from $1,272.2 billion to $1,294.0 billion.

Details on the different subsets of construction spending are provided by the Census release summary: 

• Private Construction - Spending on private construction was at a seasonally adjusted annual rate of $987.5 billion, 2.1 percent (±0.8 percent) below the revised February estimate of $1,009.1 billion. Residential construction was at a seasonally adjusted annual rate of $536.8 billion in March, 3.5 percent (±1.3 percent) below the revised February estimate of $556.5 billion. Nonresidential construction was at a seasonally adjusted annual rate of $450.7 billion in March, 0.4 percent (±0.8 percent)* below the revised February estimate of $452.5 billion.
• Public Construction -  In March, the estimated seasonally adjusted annual rate of public construction spending was $297.2 billion, nearly the same as (±1.6 percent)* the revised February estimate of $297.3 billion. Educational construction was at a seasonally adjusted annual rate of $73.1 billion, 0.1 percent (±2.5 percent)* below the revised February estimate of $73.2 billion. Highway construction was at a seasonally adjusted annual rate of $91.0 billion, 1.2 percent (±5.4 percent)* above the revised February estimate of $89.9 billion.

As you can see from that excerpt, construction spending in this report would input 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.  There are other contributors to those investment components of GDP not included in this report, such as brokers’ commissions and other ownership transfer costs on sales of new and existing residential structures, but this report represents the lion’s share of what is reported for those components of GDP.  And with the upward revisions to the prior months, it turns out that construction spending for all three months of the 1st quarter was higher than what was reported by the BEA in their advance estimate of GDP the prior week, despite the headline 1.7% drop in March construction.

As we saw above, annualized construction spending for January was revised $21.8 billion higher, and annualized construction spending for February was revised $33.3 billion higher.  In reporting 1st quarter GDP, the BEA's key source data and assumptions (xls) indicated that they had estimated March residential construction would be $4.6 billion more (at an annual rate) than that of the previously reported February figure, that March nonresidential construction would be valued at $456.6 billion, $2.4 billion more than that of the reported February figure, and that March public construction would increase by $1.4 billion from previously reported February levels.  Thus, totaling those changes from February to March, the 1st quarter GDP report showed March construction spending at an annual rate $8.4 billion higher than previously reported February levels.  With this report showing March construction spending was down at an $21.7 billion annual rate from February figures that were revised $33.3 billion higher, that means the total annualized construction figure used for March in the GDP report was $3.2 billion too low.   Averaging the understatements in the annual rates of construction spending for the three months of the 1st quarter, that would mean that this report suggests that construction spending was underestimated by $19.43 billion (at an annual rate) in the 1st quarter GDP report.   That would mean that an aggregate upward revision to the related GDP components would have to be made at a rate that would result in an addition of about 0.47 percentage points to first quarter GDP when the 2nd estimate is released at the end of May.

(Note: the above was excerpted from my weekly economic synopsis at Marketwatch 666)