Using a new dataset on e-commerce transactions in many categories of goods from Adobe Analytics, we calculated matched-model inflation and explored the importance of new products. Combining the two, the true Adobe DPI inflation rate — adjusted for new goods — was more than 3 percentage points per year lower than the CPI inflation rate for the same categories from 2014–2017.
This is the conclusion of a new NBER study by University of Chicago economist Austan Goolsbee, former chairman of the Council of Economic Advisers under President Obama and Pete Klenow, an economist at Stanford University. The study is “Internet Rising, Prices Falling: Measuring Inflation in a World of E-Commerce,” NBER Working Paper #24649, May 2018. It is, unfortunately, gated.
Note that they compared the inflation rates for the same categories, which are a subset of the items measured in the CPI. So we should be somewhat cautious in generalizing from this subset. Still, it is a good bet that the CPI overstates inflation even more than the 0.8 percentage points that Stanford economist Michael Boskin found in “Consumer Price Indexes,” in David R. Henderson, ed., The Concise Encyclopedia of Economics.
READER COMMENTS
Michael Sandifer
Jun 26 2018 at 1:48am
I’m curious as to how likely it is that you think low productivity growth is due to tight money.
David Henderson
Jun 26 2018 at 9:26am
It doesn’t relate to this post, but I consider it highly unlikely.
Jon Murphy
Jun 26 2018 at 10:19am
Wow. 3 percentage points? That’s huge.
David Henderson
Jun 26 2018 at 11:47am
Yes.
Jacob Egner
Jun 26 2018 at 12:37pm
If we were to become convinced that “true” inflation is much lower than we previously thought, does that mean we should increase our attempts to inflate? Or does deflation-via-quality-improvements deserve a different reaction than deflation caused by other things?
I also remember Scott Sumner said he was okay with ~2% long term inflation because it seemed to be enough to help combat sticky wages while not being overly distortionary. I’m not sure that deflation-via-quality-improvements makes sticky wages more of a problem.
David Henderson
Jun 26 2018 at 7:40pm
I don’t see a difference in principle between deflation by quality improvements and deflation caused by, say, productivity. But, no, I don’t think it makes sense to aim at 2% inflation IF we have an accurate indicator of inflation. I like the productivity norm that monetary economist George Selgin proposed a few years ago. See this post by Don Boudreaux for the link to Don’s idea.
Jacob Egner
Jun 27 2018 at 8:40pm
Ahh, I like this way of distinguing between inflation/deflation caused by productivity changes versus other changes. That might be the important distinction I should be making.
And this productivity-based Selgin proposal would help decrease real wages when productivity decreased, thus combating the sticky wages problem)?
Thanks for the pointer. Hopefully I’ll have time this weekend to read the relevant parts of Selgin’s “Less Than Zero”.
Mark Z
Jun 27 2018 at 10:36pm
So, we’ve been experiencing moderate deflation since 2014, according to Goolsbee? I wonder how much of the “missing inflation” predicted by so many economists during the financial crisis and the stimulus could now be explained by their use of the CPI in assessing how loose monetary policy was.
LK Beland
Jun 26 2018 at 3:27pm
I thought that most of the current non-energy inflation was driven by services. Mainly rent of primary residence (and its inputed owner’s equivalent), medical care and transportation services (such as auto insurance).
Those categories are likely not affected much by e-commerce.
Jon Murphy
Jun 26 2018 at 6:55pm
They might. Not directly, sure, but online providers of insurance could create more competition. Same with ad services like Zillow or Craigslist to get landlords’ offerings out there.
Greg Zerovnik
Jun 28 2018 at 1:15pm
Your model is so off I cannot believe it. All I know is that the prices for things I buy every day keep going up and up and up. Yes, I know my “findings” are only “anecdotal” and empirically backed by proper sampling. But I would argue the model you are using is worthless on its face.
Greg Zerovnik
Jun 28 2018 at 1:17pm
I meant to say NOT empirically backed. Sorry for the goof.
Jon
Jun 30 2018 at 3:59pm
Hmm, so an implication here is that real-rates are higher than we think.
Mark Bahner
Jul 2 2018 at 4:13pm
Hi,
For more than 15 years, I’ve thought that artificial intelligence will eventually–i.e. not necessarily this decade, nor even the next, but almost surely by the 2030s–cause nearly insane increases in the rate of global per-capita GDP growth.
http://markbahner.typepad.com/photos/uncategorized/per_capita_gdp_growth_october_2004_predi_1.JPG
But for a year or more, I’ve thought that maybe the actual outcome will be nearly insane rates of deflation. Deflation that governments might not even see, believe, or be able to react to, because the numbers will seem so strange.
Todd Kreider
Jul 4 2018 at 11:09am
Of course, a sharp rise in GDP per capita, which I also think will occur even without A.I. but with widespread use of weak A.I. in almost all jobs, and strong deflation are not mutually exclusive since it is real growth that matters.
Comments are closed.