What happens when you doubly seasonally adjust 2015q1 GDP? You get +1.3% instead of -0.7%.

May 29th, 2015 at 8:48 am

Running off to a meeting so must be brief, but this earlier post gives you the background you need to understand this here one. As you may have seen, real GDP fell 0.7% according to this morning’s revision of first quarter data.

But there’s a potential problem of “residual seasonality” in these Q1 data (see link above for explanation). One way to deal with this possibility to rerun the series through the same seasonal adjuster used by BEA. If you did that with the first release of 2015Q1 data, instead of getting a 0.2% annualized increase in 2015q1 over the previous quarter, you got 1.8%, a rate more in keeping with what I believe to be the underlying growth rate.

So I took that same approach using this morning’s release of the revised Q1 data, and got 1.3% versus the -0.7% in the release. I’ve also stressed the importance of looking at year/year growth rate as any bias in today’s Q1 data was likely there a year ago as well, so you’re implicitly adjusting for it. By that metric, real GDP is up 2.7% over past year.

That’s a lot of different numbers to throw at you before 9am (and I’m in a hotel room without coffee so take this all with a grain a salt), but the bottom lines are a) I’d discount the -0.7% pretty heavily, and suspect high-frequency growth is slightly below trend, and b) an important reason for that slippage is the worsening trade deficit driven by the strong dollar. Net exports shaved a big 1.9 ppts off of growth in the quarter.

Print Friendly, PDF & Email

3 comments in reply to "What happens when you doubly seasonally adjust 2015q1 GDP? You get +1.3% instead of -0.7%."

  1. rjs says:

    i would discount the -0.7% for another reason…imports jumped in March when the ships were unloaded and the trade defict increased 43%…however, those imports did not show up in March sales, inventories, or investment…BEA applied their formula to subtract imports, but the reason imports are subtracted is because they represent consumption or investment previously added to GDP that was not produced here…considering that they dont seem to have been added, they shouldn’t subtract…

  2. purple says:

    Where is the frantic readjustment of the 5% print from a few quarters ago ? Or does a search for new numbers occur only when the news is bad ?

  3. spencer says:

    It is important to remember that the seasonal adjustment problem should not impact the reported year over year change in the data. If real growth was 2.5% before you though there was a seasonally adjustment problem, you should still think that year/year growth is 2.5%.