Larry Summers, the Man Who Won’t Shut Up, No Matter How Wrong He’s Been

I want to thank Larry Summers for his continuing refusal to shut up about being wrong. He just got a colossal spanking from guys at J.P. Morgan.

A couple of days ago, Axios ran another story fueled by Larry. Axios is a shallower knockoff of POLITICO– a publication so gossipy and unserious that it has been nicknamed TIGER BEAT ON THE POTOMAC. It said that some Democrats feel the economy is overheating due to Biden’s bill.

This was not difficult to trace, given that nobody except Larry thinks this. Even the DINOs who thought the bill contained way too much money don’t feel the economy is overheating. Only one person does.

So Jahangir Aziz (the Head of Emerging Market Economics) and Michael Feroli (their Chief U.S. economist) decided to rebut it.


Morgan, which has a vested interest in the economy heating up, doesn’t want deficit hawks squelching anything stimulating it. So they addressed Summers’s recent claims– by pointing to a 2019 paper on secular stagnation written by… Larry Summers.

Larry defined “Secular Stagmation in a 2016 paper that was published in FOREIGN AFFAIRS:

The economies of the industrial world, in this view, suffer from an imbalance resulting from an increasing propensity to save and a decreasing propensity to invest. The result is that excessive saving acts as a drag on demand, reducing growth and inflation, and the imbalance between savings and investment pulls down real interest rates. When significant growth is achieved, meanwhile—as in the United States between 2003 and 2007—it comes from dangerous levels of borrowing that translate excess savings into unsustainable levels of investment (which in this case emerged as a housing bubble).

I’m not going to unpack any of that because it’s bullshit. Economics isn’t a science– it’s better titled “Monetary Phrenology”. Nothing that anyone says can be proven, so there are a host of competing theories about how the economy of a country works and how to fix it.

2016 Larry said the answer to this dangerous situation was stimulation– without it, a cycle of decline would deepen and we would find ourselves in an ever-deepening bear market.


In 2019, Larry signed onto a paper published by Łukasz Rachel. This looked at the economy under Trump and said things were even worse– that even more mojo might be required.

Fast forward to 2021. Joe Biden got elected– and then stiffed Larry for a job. Larry immediately published several screeds saying that Biden’s stimulus package was such an extreme overreaction to a problem that wasn’t all that serious that the bill Biden was pushing would cause hyperinflation, and we’d all need to carry money in wheelbarrows to pay for bread and milk.

The two guys at J.P. Morgan wrote that Larry’s 2021 work was disproven by the 2019 work. That work included a series of models intended for calculating what needed to be done– and what the effects would be.

They plugged the 2021 economic data into it– then used its formulas to project the impact of Team D’s bill.

The result? Not only does 2019 Larry’s model say that the impact won’t be anything close to 2021 Larry’s projections– it suggests that what is being done might not be nearly enough.


The basis for the uncertainty is this question: How much of the downturn in the last year was caused ONLY by Covid?

If you assume that all or most of the plunge can be traced entirely to the pandemic– that it will go away as the pandemic vanishes– then Larry’s 2019 model says that the bill calculated what was needed accurately and should fix things.

If you assume that Covid only made existing structural problems in the US economy worse– that things were bad even before the pandemic ramped up– then the bill isn’t nearly enough. A lot more needs to be done.

Why can’t we be sure? Because you’d need a science to be able to measure the problem precisely, and Monetary Phrenology isn’t a science. But I digress


On one level, this paper is a magic bullet that makes the problem vanish. Team Biden is now in a situation very like the defense attorney who discovers that a witness has given two very different accounts of an event.

There is no question that Larry Summers is wrong. The only issue to be resolved is “Was he wrong in 2019 or is he wrong now?” He can pick either– but he can’t pick both. And because the views are so diametrically opposed– and the timeframe is so close (August 2019 to February 2021 is barely 18 months)– there’s no way to wiggle this by saying that things have changed.

Given that the 2019 paper is 46 pages, with a few dozen formulae and oodles of tables– signed by two people and published in the National Bureau for Economic Research– while the 2021 screeds were op-eds in the media that thinks Tom Cotton is a scholar– I’d say Old Larry is more likely to be right.


The more significant point to be made– which I will make yet again– is that something like this could never possibly occur in a science.

Yes, there are differences of opinion in science, but always about topics where we don’t actually know (it because an event happened billions of years ago, or because we don’t have the tools needed to measure the particulars). There are no examples of debates about events that happened a few years ago, the details of which we can measure down to the smallest increment.

And such disputes only occur between two different experts– each proposing different hypotheses (say, Loop Quantum Gravity or String Theory) to explain a specific question.

A field of endeavor where you can hop from one extreme to the other– no proof and no consequence to your prestige– is nonsense.

What’s going on here is very simple. In 2016, Larry Summers wanted to say “Barack Obama is no good and very bad”, so he grabbed onto a theory that would say that– and invented reasons to show it fit.

In 2019, he wanted to say that Donald Trump was hateful and horrible, so he doubled down. Now he wants to punish Joe Biden for not giving him a seat on the Federal Reserve– so he’s made up new evidence

This is not to suggest that someone who has said the same thing between 2016 and 2021 is correct– they could be wrong then and wrong now (see “Moore, Stephen”). But someone who can turn on a dime– renouncing everything that has been said– is definitely not credible.

And a field that permits this sort of chicanery is quackery– and should be ignored in its entirety except for the elements that can be proven.

[CDATA[ $('input[type="radio"]
[CDATA[ $('input[type="radio"]
[CDATA[ $('input[type="radio"]
[CDATA[ $('input[type="radio"]