Monday, September 1, 2008

Nominal vs. Real GDP: Why the GDP Deflator Overstates GDP

To better illustrate the impact of the Import GDP Deflator on Real GDP, the chart below shows Real vs. Nominal GDP by component.

First, a refresher... the greater the size of net exports, the greater the GDP. Net exports = exports less imports, thus an increase in imports negatively impacts GDP all else equal.

So what do we find? Nominal imports increased in Q2 by 18.8% annualized. Real imports DECREASED in Q2 by 7.5% annualized. In other words, even though we paid 18.8% more out of our pockets, we received 7.5% less units of these imports. Why? The Import GDP Deflator decreased imports by more than 25%.

Is paying more for less beneficial to our economy? No... but it is good for GDP. As detailed here, this 25% contribution by the import portion of the GDP Deflator increased Real GDP by a positive 4.6% (Imports make up roughly ~1/6 of GDP). Had we paid the same price level quarter over quarter for our imports, and thus received more given the level of nominal GDP, real GDP would have been -1.3% (I understand this WOULD NOT necessarily be a more accurate version of GDP, but it illustrates the size of the impact / just how odd this calculation is).

Ignoring the odd calculation, does this make sense to me? It doesn't. For this to make sense, it must be reasonable that we imported 39.2% less petroleum products during the quarter (in real terms) on an annualized basis. If the U.S. was able to drop its dependence by that much in just one quarter, I guess the U.S. isn't nearly as dependent on the Middle East as we thought. More important, it implies that the U.S. economy is stronger because we paid more in aggregate for our imports, but received less units as more was consumed from non-imports.

Want to hear something more likely? The impact of import inflation on the GDP Deflator is too high and GDP is overstated.


  1. Jake - much appreciate your patience. Spent some time tracking it back thru the numbers after your post and think things are getting clearer but lead to more questions. In essence what this boils down to is that import inflation was over-estimated, particularly for oil ? And as a result real imports were under-estimated, again largely oil ? And because real imports were actually much higher then subtracting them from GDP would lower real GDP estimates ? Or real GDP was estimated too high ?

    Fair enough. A couple of challenges though. For one thing oil prices did shoot thru the roof in Q2. Which deflator should have been used ?

    Aren't we glad we don't do this for a living ! :)

  2. That was the rough intention of the post and I honestly am not sure how to go about getting a better number (so many variable - i.e. should the huge decrease in oil imports put less weight to the spike in oil????)

    At the end of the day, the goal is to get a better idea as to what made up the 3.3% number. I personally care because I want to know what type of strength the economy truly has.

    In the short term, is the economy actually more productive because we used less oil (I'm sure we did, just not sure about the size of that decrease)? Or are things such as the 1.7% increase in personal consumption / 12% decrease in private investment more applicable?

    Either way, by breaking down the numbers and getting challenged by readers such as yourself, I honestly feel I am better able to form my own opinion.

  3. We're taking similar approaches for nearly identical reasons. I really tried to take apart the oil numbers and there's not a good answer - my own estimates of oil and real oil prices increases are less than the deflator but the yoy change in nominal prices and the deflator are very close. There may not be a good corrective. But the bigger picture issues are I think the important ones. Granted the price adjustment conundrums (puns intended) I think one gets a better handle on your questions with YoY looks AND by looking at Gross Domestic Purchases, i.e. GDP net of trade. I'd be happy to share those spreadsheets or - in this process - have now put up several posts trying to kick the issue to death.