GDP Deflator versus CPI
click for larger chart
chart courtesy of Eric Jantzen
>
In our exploration of how laughable the 3.3% GDP was on Thursday, we looked at the GDP Price Index versus the CPI. I was discussing this via email with Eric Jantzen of iTulip, and he showed me a variation of that, using the GDP Deflator (it should be very very similar to the price index)
Its no coincidence that the current situation resembles past ones where oil prices had spiked. Since more than half of the US Crude consumption is imported, the price and quantity go into all the GDP calculations as a negative.
This leads to a bizarre and counter-intuitive outcome: Any rise in the price of crude goes into the deflator as a NEGATIVE. This brings the total deflator down, making GDP appear better than it really is.
One of the commentators at iTulip added:
The other effect of the arithmetic is that the full effect of higher imported energy prices on the PCE happens with a lag. Once the imported oil is refined and the product moves up the GDP table from imports to Non Durable Goods consumption, the deflator for gasoline goes up and this is when it hits the PCE. The other problem this creates is that if you have period of rising import prices, the gain in period one goes into PCE in period two, but the effects of that can be masked by the further increase in import prices in period two.
what you are seeing is the arithmetic of the deflator. The CPI is showing the effect of higher refined product prices. The deflator shows the effect of higher imported oil prices.
I agree with that assessment.
And as I noted in comments yesterday, this isn’t a grand conspiracy — this is simply the way the
models are constructed. There are inherent biases built in, and this
month’s GDP data reflects that.
Yes, there is some latitude in making certain selections — I am not sure precisely how much — but I do not believe all that huge.
Understand that this is not a political issue, but rather, a quantitative/analytical one, reflecting fair-to-poor econometric modeling, one with an inherent downside bias as to the inflation data, and an upside bias when it comes to GDP and job creation . . .
Its the odd way the model raises GDP when we import inflation that is my beef.
>
GDP Price Index versus CPI
click for larger chart
>
Previously:
Is GDP (via BEA) Measuring Growth or Inflation? (August 2008)
http://bigpicture.typepad.com/comments/2008/08/are-you-measuri.html
GDP: Lowest Inflation Rate in 5 Years (August 2008)
http://bigpicture.typepad.com/comments/2008/08/gdp-gross-decep.html
Source:
Inflation in GDP Out of whack – Again
iTulip, 08-28-08, 03:58 PM
http://itulip.com/forums/showthread.php?t=3896
Oddly enough the deflator for Gross Domestic Purchases, which happen to include imports, almost exactly mirrors the CPI on a YoY basis. While YoY GDP is “growing” at 2.5% and 2.2% the last two quarters GD Purchases grew at 1.1% and 0.4%. This is literally the first time in our history that the GDP and purchase deflator’s have diverged – I’m talking about back to 1929 – like this. They’ve moved apart but never on different directions as they are now. Since GDP is designed to capture domestic activities and appears to do so quite well it’s not clear that a) the model is broken or even perverse as b) we’re chatting up the wrong model instead of the right one and c)models are built from history. So if this is the first several quarters in approx 320 quarters it would rather make sense for an anomalous situation never seen before to escape modeling. But then since another model from the same folks seems to have captured it exactly maybe nothing’s borken ? Hmmm…interesting.
This divergence is very interesting.
It seems perverse that GDP will rise even higher when imported crude rises to $200 but I guess by then lower consumption/production in all sectors will kick in to moderate this effect.
http://jeflin.net
I see you’ve spent some time reconsidering the analytical aspect that was the point of my comment and your response on your previous post.
This is not “a bizarre and counter-intuitive outcome”.
And it does not make “GDP appear better than it really is.”
And there are not “inherent biases”.
This IS the calculation of inflation for GDP.
What you are getting around to focusing on is the difference between the scope of CPI inflation and GDP inflation. You have to understand that in order to appreciate why none of these negative allegations above are true, at least with respect to this particular issue.
CPI will reflect imported oil inflation plus the inflation in any refining margin that is part of (domestic) US GDP.
The GDP deflator will include only the refining margin inflation component. It excludes the imported oil inflation, because imports are not part of GDP.
Therefore, the difference between CPI and the GDP deflator is the exclusion of a large imported oil inflation component from the GDP deflator.
So if imported oil inflation is very high relative to GDP inflation, the CPI will be pushed up above the GDP deflator but below imported oil inflation.
In general, the rising scale of [GDP deflator; CPI; import inflation] will depend on the various inflation differentials and the weighting of imports.
Imports run about $ 1.8 trillion relative to a $ 14 trillion GDP.
Import inflation runs 20 per cent or even more currently.
That’s a significant factor that shows up in CPI but not in the GDP deflator.
The model does not “raise GDP”.
It’s not a model. It’s measurement, at least on this issue.
And the measurement recognizes that GDP inflation is most certainly not the same as CPI inflation – the scope of the measurement is entirely different.
This was also Wesbury’s point.
anon – thanks. that’s much clearer than your last comment and very helpful. not to put too find a point on it the model is the Kuznetz definitions of national income accounting which got worked out over decades, huge data gathering and grinding and lots of simple algebra. lots of informed understanding and decisions go into things but they haven’t changed in decades in any fundamental sense.
it also strikes me that you and I have arrived at similar conclusions starting in different places; and pointing to the same basic definitional difference. CPI includes foreign inflation impacts GDP doesn’t it. reconcile the two by looking at GD Purchases.
Fair ?
“Understand that this is not a political issue, but rather, a quantitative/ analytical one, … …with an inherent downside bias as to the inflation data, and an upside bias when it comes to GDP and job creation . . .”
Certainly it is true that the party which is out of power rejoices when job creation is weak. But inflation is another matter. This relates directly to cost-of-living adjustments to social security recipients and to Federal retirees; as such that aspect of it is in a sense political.
dblwyo:
Very nearly.
Menzie Chinn’s drill down made a good start, which you picked up on as well:
a) Menzie defines:
Gross Domestic Purchases = C + I + G
This is the also the scope of CPI. It includes imports – direct or embedded content in products purchased domestically. But it excludes exports.
(If “gross domestic purchases” is accepted economic terminology, it would certainly be helpful if it were used more frequently in discussing the difference between CPI and the GDP deflator.)
b) Of course:
Gross Domestic Product = C + I + G + EX – IM
This is also the scope of the GDP deflator.
It excludes imports, but it includes exports.
Clearly, the major explanatory factor for the CPI/GDP deflator differential is the effect of import inflation on the CPI. Import inflation is the outlier in all of this, given that it’s currently running at 20 per cent plus.
But the other factor is the effect of export inflation on the GDP deflator. So far, I’ve assumed this is fairly benign in the sense that it’s likely much closer to CPI inflation than import inflation. But to be complete about relating CPI to the GDP deflator, two adjustments, not one, need to be made.
(By including exports, it includes the inflation experience of foreign countries that purchase US products. Thus, it actually includes a set of CPI components in respect of foreign countries that purchase US products.)
Brian Wesbury was merely adjusting GDP in order to ascribe it with an economic scope that matches the scope of CPI. He then deflated this ADJUSTED GDP GROWTH number with CPI in order to arrive at a net real GROWTH number. What he did was a quick fix to an approach taken by those that insist on using the wrong deflator number (CPI) in attacking the GDP growth problem. He’s right to make this correction to the methodology of those who insist in using the wrong inflation number (CPI) in analyzing real GDP growth. Obviously, you don’t get GDP when you add back imports, but you do get a comparable growth and inflation adjustment, as opposed to a completely wrong and stupid result when you attempt to deflate GDP by CPI, without thinking any further about it.
(What Wesbury probably should have done additionally was make the adjustment by not only adding back imports but also subtracting exports. But the main point he wanted to make was about the extreme inflation contribution of imports alone to the CPI, so he left the export adjustment out of it.)
thanks for the post and all the comments as well – I learn so much from reading this blog.
I don’t think that GDP growth is reflective of the consumer’s satisfaction with the economy. I would say that is more represented by Gross Domestic Purchases – Energy Expenditures. As was pointed out GD Purchases have been largely flat over the last two quarters and of course energy purchases have increased dramatically. That leaves decreasing purchasing ability for everything else.
The other point I would make is that I think we are beginning to see the pain of global economic rebalancing as exports increase and imports (ex oil) decrease. It does little for the consumer if increases in GDP go towards reducing the trade gap rather than an increase in consumption. Necessary, perhaps; pleasant, not so much. Now we are finding out what it is like to be a Chinese worker.
This is all ‘smoke and mirrors’. By combining many factors into one we are left with an indecipherable (and extremely misleading) measuring tool.
It would be far more useful to see each of these factors as ‘stand alones’.
I would say that using metrics that have the effect of willful blindness *is* a political issue.
The government and media are all playing along.
I’ve spent enough time in the research world to know how you can mis-apply legitimate methods to get the result you want.
When it comes to public policy and not just getting accolades for research results, the stakes are all the more high, and the manipulation (of the whole process) all the more unethical and damaging.
Is the CPI in this chart normalized for changes in calculation methodology over these 48 years?
CPI calculation in 1960 was a very different beast. I’d say seeing both (unadjusted as well as using today’s methodology over the full period) would be interesting.
Is the CPI in this chart normalized for changes in calculation methodology over these 48 years?
CPI calculation in 1960 was a very different beast. I’d say seeing both (unadjusted as well as using today’s methodology over the full period) would be interesting.
Is the CPI in this chart normalized for changes in calculation methodology over these 48 years?
CPI calculation in 1960 was a very different beast. I’d say seeing both (unadjusted as well as using today’s methodology over the full period) would be interesting.