Thursday, February 03, 2005

How far off was advanced GDP? 

The Eclectic Econoclast has a story (attributed to Phil at Market Power, who needs to be in MOB) that a computer changeover in November at Statistics Canada may have cost us half a point on GDP growth in the last quarter.

The Commerce Department three days ago said the U.S. economy cooled to a 3.1 percent annual rate last quarter, damped by a record trade deficit that included a widening gap with Canada. StatsCan, the agency responsible for Canadian trade and GDP figures, today said it mistakenly underreported November imports from the U.S. by C$1.31 billion ($1.06 billion) on Jan. 12.

Estimates for how much the error will add to U.S. GDP when corrected range from 0.1 percentage points from Morgan Stanley in New York to 0.5 points by Joseph Carson, director of economic research at Alliance Bernstein in New York.

Add this to the report of a large revision in residential construction, and I'm thinking we may end up with GDP growth in the last quarter of 2004 revised upwards to 3.8% (a full 4% is possible but not likely.)

Those unfamiliar with GDP statistics in the U.S. -- I expect that makes up most of us -- may not be aware that the Dept. of Commerce issues three estimates of GDP for a quarter. The first one is called the advance estimate, given thirty days after the end of a quarter, and it contains assumptions. The second is a preliminary (at 60 days after the end of a quarter) and the final at 75 days. Advance has a tendency to underreport relative to preliminary and final, though not by very much. Still, about one in six of the revisions to GDP numbers between advance and final are 1% or more.

It's possible this is one of those quarters. That suspicion may as well be the reason that the Fed raised the interest rate target yesterday.