Saturday, August 30, 2008

GDP ex Import Inflator: Third Straight Quarter of Decline

Due to a rise in the price of oil from less than $110 to $145+, the import portion of the GDP Deflator Inflator raised Real GDP by 4.6% (what is the GDP Deflator? click here). This is more than 2x the level seen at any other point during this decade and the highest amount EVER (we have to go back to Q1 1974 and Q1 1980 for levels even remotely close to that of this past quarter).

Had the contribution of imports on the GDP Deflator Inflator stayed the same as last quarter (which was historically large), GDP goes from the 3.3% as reported by the BEA to 0.9% for Q2. If the deflationary impact of imports were removed completely from the equation, GDP goes to -1.3% and shows GDP contraction for the third straight quarter.

Which sounds more reasonable to you?

Source: BEA


  1. Jake - an interesting approach which is smack in the middle of a lot of controversy, over which there should be some serious debate. I'm not sure you can apply those adjustments with simple algebra as the BEA constructs price indice/deflators seperately for each series in the national accounts. It's a lot of work. However a better thing to look at is Gross Domestic Purchases which takes what we make here at home and adds what we buy from abroad and subtracts what we sell. GDP YoY was slowing at 2.5% and 2.2% while GDPurchase was 1.1% and 0.4%. More importantly the GDP deflator was 2% and falling while the one allowing for expensive foreign goods was 3.5% and rising. There's never been anything like it - I just ran the numbers back to 1929 and this appears to be the biggest divergence in about 80 years. You can see the charts and detailed discussions here:

  2. The numbers I list directly impact the final GDP Deflator at the level I list. Imports removed 4.6% from the GDP Deflator, which is applied directly to Nominal GDP to "create" the Real GDP, which was listed in Q2 at 3.3%. All the numbers I use are directly from the BEA Table 1.1.8 and linked to at the bottom of the post.

  3. Well with my poor and tired brain I'm still not quite following the argument nor grasping why that works. For one thing the table you point to is the GDP price indexes which would seem to mean you can't just knock that percentage off of GDP growth but have to re-calc the real GDP stats and estimate GDP growth alternatively. Or one might look at real GDP vs Real Gross Domestic Purchases and see what that tells you. And even simpler alternative and an experiment that would be worth trying would to just subtract the sum of Exports and Imports from GDP and see what you get. I just did that with the real GDP numbers and get numbers almost identical to Real GD Purchases. Try it and see what you get. The economy once you net it of trade is very weak but not quite what your charts would seem to argue for. Be interested in your take.

  4. The chart I point to is Table 1.1.8 'Contributions to Percent Change in the Gross Domestic Product Price Index'. Imports weren't just 4.6% deflationary, they were 4.6% deflationary taking their relative size to the total GDP Deflator (the # was thus MUCH bigger).

    I did the same analysis as you detail (just removing net exports from the picture), but it still doesn't look remotely like the REal GDP figures presented. Looking at Table 1.1.2 and removing Net Exports, you get growth of ~0.1% the last 2 quarters and -1.1% in Q4 07.

    The whole point is we can't get accurate numbers because the data the government provides is manipulated to present a picture of an economy that is rebounding.

  5. This is kinda fun for a terminally wonkonerd discussion. Try table 1.1.6 real GDP - don't think one should just subtract the % changes from one another. If a composite number say 11 is 11.7 = 8.3+1.7-.4+2.0 and the %changes are calculated for each line I believe the algebra doesn't hold. Try subtracting -.4 from 11.7 to get 12.1 and look at the changes. Of course I'm looking at YoY changes instead of QtQ - perhaps that's the difference ?

  6. Agreed. A very fun / ubernerd conversation.

    I just posted a new entry which hopefully better illustrates some of my thoughts. I'm sure if anything, it will make it more unclear.

    Thanks for the great posts.

  7. Unless I have missed something, this whole piece is confused economics, and I say that as an economist.

    The GDP deflator has the sole purpose of getting rid of price effects from the value data they collect to calculate GDP: eg value of imports. .

    Since the value of imports was ballooned by the surge in oil prices, that has to be got rid of to get to volume GDP and that is why the import deflator is so large. Nothing in this column suggests there is anything wrong with the procedure they have undertaken and nothing to suggest that the 3.3% is wrong for that reason.

    Incidentally the commodity price falls in Q3 may lead to a negative import price deflator for Q3 as a quid pro quo. Will you complain then ?

    PS The 'the import portion of the GDP Deflator Inflation' quote shows the lack of understanding of the author. By definition the GDP deflator does not contain an import price deflator as GDP does not include imports !!
    GDP measures the volume of domestic production. That is why it is gross DOMESTIC production

  8. Graham- you are both right and wrong.

    Right: GDP does exlude imports (the way the BEA does this is through the subtraction of imports from exports getting a net export figure).

    Wrong: To subtract imports from the equation, the BEA takes nominal imports and applies a GDP deflator to the nominal number creating a "real" number (THE IMPORT GDP DEFLATOR). In the latest quarter, the subtracted "nominal imports" were 18% higher (on an annualized basis) and "real imports" 7% lower (on an annualized basis) due to a 25% "deflator" (look it up). This removed 4.6% from the complete GDP Deflator.

    So, my argument is that the GDP equation itself is flawed. Your arguments implies that the U.S. economy is stronger because we paid more in aggregate for our imports, but received less units (i.e. nominal was higher, real was lower). My argument is that had we actually received more units for the nominal amount we paid, the import deflator (YES THERE IS ONE) would be lower and GDP would have been lower than the reported 3.3%.

    Again, is the U.S. economy stronger because we paid more for less? NO!!

    I have a full post explaining this which was posted last night.

    In all seriosness... thanks for the post. I do appreciate the point of view of an economist. As one, please explain why getting less for more money is beneficial besides saying that is just how it is calculated...

  9. One more thing;

    Question: Incidentally the commodity price falls in Q3 may lead to a negative import price deflator for Q3 as a quid pro quo. Will you complain then ?

    Answer: If it leads to an understated GDP all else equal. Yes.

  10. Jake,

    They start from expenditure and work from spending value to spending volume to get to real GDP : adding exports and subtracting imports on the way to get to domestic production (as exports are made at home and imports not).

    The lesser the rate of increase of import volume (ie units), which is being subtracted, the greater the pace of domestic production. Take an example:

    Example 1) If real spending ('SPVOL') rises at 1% (assume exports went up at the same rate as domestic production sold domestically for simplicity) and real imports (MVOL) rose at 1%, then GDP will rise at 1%.

    Assume actual MVOL =XVOL, again for simplicity and that MVOL= 25% of total GDP.

    Example 2) if SPVOL AND EXVOL rise at 1% and MVOL falls at -1%, GDP growth will be faster than in example 1. As the part of domestic spending on imports actually fell, the rest of SPVOL (ie spending on domestically produced goods) must have been more than 1% to make the total rise 1%. In this case , GDP will have gone up by 1.7% approx to be consistent with the fall in imports.

    Given a figure for the change in import values (MVAL%), the higher the import deflator (ie the inflation rate of imported goods and services), the lower will be the rate of change in the volume of imports and hence (see example 2) the higher will be the growth rate of GDP .

    This is not a causative real word effect but the statistics jigsaw puzzle if you start from value of spending on imports. I think this is why you are confused. You are mixing up the assembly of the elements within a simple identify with causation. A calculation of a deflator does not cause anything. It is just a reflection of what is happening and is only wrong if errors are made in data collection and labelling.

    In the real world, it is the volume of imports that happens. The deflator is just used to get to the real world number from the value date that statistician have to start with because volumes are not normally recorded by customs or at least not recorded reliably enough to be used.

    The deflator is calculated form a collection of price data . It is not an arbitrary number. Sometime parts of the deflator can be calculated from the difference between value and volume data if there is no price data and there is reliable volume data : number of vale of cruise missile exports gives an export price deflator as there is not price index for cruise missiles .

    In a very simple world where the only imports were barrels of oil, you would count the barrels of oil and use that directly ofr real GDP and only use price to get value data if that were not available to calculate the balance of payments: which needs value data

    So recently demand for oil fell and hence import fell. But the value data showed a big increase because of price increases. Hence a big deflator is necessary to get to the underlying units of oil etc imported.

    Hope that is clear. It is late at night here so I hope I have not made an error.

    There is no conspiracy or incompetence. You want to see really bad GDP data? Look at Japan. Again cock-up rather than conspiracy


  11. VERY informative.

    That does makes it clear why imports aren't necessarily hurtful to GDP on a stand alone basis. However, if I am interpreting this correctly, more real imports would hurt given a fixed amount of real consumption in a given period as that would imply less of the consumption was done by non-imports.

    Thus, if I assume imports were understated on a real basis, that same assumption implies GDP was overstated for that given period.

    The good news is we should hopefully find out if that's the case when the Q3 GDP # is released.

    Thanks for the post. I write this blog to learn and I am certainly not the expert that half my readers are.

  12. That is a correct statement Jake (ceteris paribus: ie other things unchanged)