“Anything is small if you divide it by GDP.”  I’ve been quoting this line for decades, usually to point out exceptions like the gains of deregulating immigration and housing.  But who, if anyone, actually said it?  I recently decided to email Theodore Keeler, the Berkeley economist who (I believe) first shared the aphorism with me.

Hi Ted, I took IO with you about 30 years ago!  I hope all is well with you.

Question: There’s a quote you told us that I’m having trouble placing.  If memory serves, you told us that after seeing Harberger’s low estimates of the deadweight loss of monopoly, Samuelson said something like, “Anything is small if you divide it by GNP.”

Ring a bell?

Thanks either way.

Here’s Keeler’s initial response:

Hi Bryan,

It’s good to hear from you!  I certainly remember the quote, but getting the source is more difficult.  I’m going to describe my process of reading the quote in hopes that you could retrace the steps and find it.

As I probe my memory, I believe I remember reading an interchange at a conference, maybe an AEA session, from long ago, maybe the late 50’s or early ’60’s.  It may be a collected volume other than a journal; I can’t remember that for sure.  It can’t be after 1967 or 1968 that I read it, because it was early on in my time in grad school at MIT.  I believe it was indeed a session about Harberger’s estimates of the welfare loss from monopoly (there’s a small chance the loss analysis was from a follower of Harberger, but I doubt it).

I may have things mixed up here, but I believe another discussant was Dale Jorgenson, who was born in 1933, so he got his Ph. D. about 1959, and it seems unlikely to be before that.  Jorgenson had some technical objections to Harberger’s work that Samuelson thought to be naive.  He said something like “I’ll defend Harberger against Jorgenson’s quaint objections,” and I believe it was he who went on to note the quote above about GNP, and also to note that the rectangle of redistribution was much greater than the triangle of welfare loss, and might in itself be worth considering (but that’s a distinctly non-Chicago view of course, so not so much of interest to Harberger).

I wish I could come up with a more definite citation, but maybe these fragments of recollection can help.  I’m doing well.  I’ve been retired for quite a few years, though I continued to do research for some years after I stopped teaching.  At age 76, I’m enjoying full retirement now.

Again, it’s good to hear from you.  I hope my random recollections can help you, and wish you well in your research.  Please let me know if you have a follow-up question that I might possibly remember enough to answer.

With best wishes,

Ted Keeler

He soon added:

Hi Bryan,

I looked a little more into this, and I may have more information that could be helpful.  Specifically, page 28 of the following NBER pdf:


shows three references to the AER May issue that contain Harberger’s presentations and discussions of his model and results.  They are the May issues for 1954, 1959, and 1964; page numbers are shown in the reference.  I don’t believe the discussions were in 1954, so it’s probably the 1959 or 1964 issues that contain the discussion.  Back in those days, the published AEA sessions in the AER included comments by the discussants, so I think you might be able to find them there.

I also believe that Samuelson, in his graduate theory lectures, mentioned a facetious “Harberger’s Law,” which states that if you divide anything by GNP, the fraction will be very small.  I couldn’t find “Harberger’s Law” by a superficial Google search, but it could be possible.

Anyway, that’s what I can find.  It’s okay to quote me in your blog, but, if I were you, I might check the AEA sessions on the pages listed if it weren’t too difficult.  If you do quote me in your blog, please note that these are memories of mine from over 50 years ago, so I fear I can’t vouch for their veracity.

With best wishes,

Ted Keeler

I followed up on Keeler’s remarks by googling “Harberger’s Law”  – and found a pile of McCloskey references to a slightly different point:

Any economist of experience knows this. It’s enshrined in Chicago tradition as Harberger’s Law. If trade is 10 percent of national income and free trade improves the terms of trade 20 percent, then (multiplying a fraction by a fraction) the rigorously derived gain to national income is… 2 percent. That’s once-for-all, not cumulatively year-upon-year.

And for now, that’s all I know.  If you’ve got any further insight, please share in the comments.