As part of the Census Bureau's November 2015 construction spending release, revisions to the monthly estimates of private residential construction spending, of total private, of total residential and of total construction spending for January 2005 through October 2015 were issued. According to the Census Bureau, these revisions were due to a correction in the tabulation of data on private residential improvement spending. Because BEA uses the Census Bureau's detailed construction spending data to prepare annual private fixed investment (PFI) estimates for residential structures, the corrected data will have some impact on BEA's annual estimates of residential improvements, and therefore, GDP.

Corrections to the construction spending data for residential improvements resulted in revisions to BEA's annual estimates of PFI residential improvements for 2013 through 2015 in the July 2016 annual update of GDP; estimates for earlier years will be revised as part of a future revision.

BEA uses a 3-year centered moving average of the Census Bureau's annual residential improvement spending data to prepare annual estimates of PFI residential improvements. Estimates for the most recent year reflect quarterly indicators and analyst judgment. This methodology is used to address volatility in the Census Bureau residential improvement spending data. Therefore, the revisions to annual PFI residential investment estimates are expected to be dampened relative to revisions to the Census improvements data.

BEA's quarterly residential improvements estimates, however, are not estimated using the Census Bureau's monthly residential improvement data. Instead, BEA's quarterly estimate methodology uses a composite indicator that is derived from the Census Bureau's monthly retail sales for building material and garden equipment and supplies dealers and the Bureau of Labor Statistics aggregate weekly payrolls of production and nonsupervisory employees for residential remodelers.