Oops! The Feds Screwed Up 10 Years of Housing Data
Policy + Politics

Oops! The Feds Screwed Up 10 Years of Housing Data


Economists and analysts who have in the past decade found themselves puzzled by the U.S. Census Bureau’s monthly survey on the construction industry may have just been vindicated. In a release of its estimates for the month of November, the Census Bureau on Monday included a footnote indicating that it had restated each of its monthly estimates going back to January of 2005.

“In the November 2015 press release,” the footnote reads, “monthly and annual estimates for private residential, total private, total residential and total construction spending for January 2005 through October 2015 have been revised to correct a processing error in the tabulation of data on private residential improvement spending.”

Related: Why Oil Prices Will Rebound in 2016

While that may seem like a trivial bit of statistical housekeeping, the data in question is actually an important part of how economists measure the health of the U.S. economy, with direct consequences for the decisions made by policymakers, including members of the Federal Reserve’s Open Market Committee, which sets interest rates. (A fact noted at ZeroHedge here.)

The Census Bureau’s release doesn’t do much more than identify the specific changes that it has made to the numbers in the past decade of releases.

The average size of the corrections over the ten-year period in question has been 1.2 percent of the seasonally adjusted total value of construction “put in place.” The 1.2 percent figure is in absolute terms, but the revisions were both positive and negative at various times, and ranged from practically zero to 4.6 percent.

While the size of the revisions has waxed and waned somewhat over the years, the data shows a clear trend toward larger misstatements in the most recent data. In 2014 and 2015, all of the original reports understated the “Total Construction” value by a factor of 3.1 percent, on average. In the first ten months of 2015, that figure rose to 3.50 percent.

Related: A Bailout for Volkswagen? Congress Wants to Do Something Absolutely Crazy

All 10 of the largest monthly revisions have occurred since June 2014, and each one of them was an understatement of the value of construction work put in place during that period.

However, while the reports may have been understating the net increase in construction spending, with the revisions taken into account, they were overstating the rate of change on a month-over-month basis – a factor that economists, analysts, and Fed watchers arguably pay more attention to than the absolute figures.

In the first ten months of this year, for example the originally published numbers suggested that total construction put in place value was increasing at a month-over-month average of 1.14 percent. The adjusted figures show a considerably weaker average rate of .89 percent.

 It won’t be known for some time, if ever, how large a role these particular figures played in the Fed’s decision to begin raising interest rates from the near-zero level where they have been for years. However, to whatever extent they played a part, it was a somewhat misleading one.