A precautionary tale for all scientists. First, some prelude.

I’ve always been grateful for Paul Krugman’s public opposition to the Iraq War. When the rest of the establishment lined up in favor of the war, Krugman was one of the very few public figures who had a long reach and the willingness to oppose that idiotic folly. So I like the man.

But if Krugman has a weakness, it is that he can accept at face value the pronouncements of other prestigious economists without much examination. He did this with Obamacare where, according to Jon Gruber, Obamacare/Romneycare would be awesome for everyone, even though those of us who lived with it in Massachusetts realized that, for a subset of people (who weren’t wealthy), it would make things worse. People would have to pay for insurance they couldn’t afford, as well as a significant fraction of people would forgo needed medical care because they couldn’t afford the out-of-pocket costs. Yet Krugman kept talking about how affordable it was. This exact scenario was not only predictable, but predicted years ago. It wasn’t until about a month ago, in no small part due to Sen. Bernie Sanders’ constant call for single payer healthcare, that Krugman grudgingly admitted this problem (which makes claims of being empirically driven unlike Sanders’ supporters somewhat galling).

Enough prelude. Recently, economist Gerald Friedman–a Clinton supporter–released a pretty optimistic take on Democratic presidential candidate Bernie Sanders’ economic plan. As the kids used to say, you’ll never guess what happened next (boldface mine):

Gerald Friedman, an economics professor at the University of Massachusetts at Amherst, produced an analysis of Bernie Sanders’ economic plan predicting eye-popping benefits from the candidate’s program: 4.5 percent real GDP growth between 2016 and 2026, at which time median income would be $82,151 — about $23,000 above the Congressional Budget Office baseline.

Reaction from the economics establishment was swift and vicious. Democratic Party heavy hitters — Alan Krueger of Princeton, Austan Goolsbee of the University of Chicago, plus Christina Romer and Laura D’Andrea Tyson of Berkeley, all four former chairs of the Council of Economic Advisers [CEA] — put out an ex cathedra declaration that Friedman’s paper was utterly beyond the pale of serious analysis.

Paul Krugman joined the dogpile, writing three consecutive posts (“Worried Wonks,” “What Has the Wonks Worried,” “Wonkery Has a Well-Known Liberal Bias,” — noticing a theme?) on how Friedman’s paper was utterly preposterous, and demanding Sanders immediately denounce it. Brad DeLong was kinder, but still insisted that Friedman was enabling right-wing economic derp.

Ironically, in the frenzy to destroy Friedman’s reputation, nobody actually explained in detail what the problems were with his paper. The CEA pronouncement had no data or economic argument at all — it was 100 percent political handwringing. Krugman gave a very brief gloss suggesting that Sanders couldn’t possibly get labor force participation back up to 1990s levels due to aging, and trying to do so would cause inflation. Kevin Drum gave a similar incredulous stare argument about worker productivity and GDP growth, pronouncing it “insane,” worse than Republican “magic asterisks.” [note that Kevin Drum has since backtracked on this]

Krugman and the former CEA chiefs (BAND NAME!) apparently pissed off James Galbraith, former executive director of the Joint Economic Committee, the Congressional equivalent of the CEA. Galbraith’s style at times could be called acerbic, and he really brings the heat on this (pdf; boldface mine):

You write that you have applied rigor to your analyses of economic proposals by Democrats and Republicans. On reading this sentence I looked to the bottom of the page, to find a reference or link to your rigorous review of Professor Friedman’s study. I found nothing there.

You go on to state that Professor Friedman makes “extreme claims” that “cannot be supported by the economic evidence.” You object to the projection of “huge beneficial impacts on growth rates, income and employment that exceed even the most grandiose predictions by Republicans about the impact of their tax cut proposals.”

…So, let’s first ask whether an economic growth rate, as projected, of 5.3 percent per year is, as you claim, “grandiose.” There are not many ambitious experiments in economic policy with which to compare it, so let’s go back to the Reagan years. What was the actual average real growth rate in 1983, 1984, and 1985, following the enactment of the Reagan tax cuts in 1981? Just under 5.4 percent. That’s a point of history, like it or not.

You write that “no credible economic research supports economic impacts of these magnitudes.” But how did Professor Friedman make his estimates? The answer is in his paper. What Professor Friedman did, was to use the standard impact assumptions and forecasting methods of the mainstream economists and institutions. For example, Professor Friedman starts with a fiscal multiplier of 1.25, and shades it down to the range of 0.8 by the mid 2020s. Is this “not credible”? If that’s your claim, it’s an indictment of the methods of (for instance) the CBO, the OMB, and the CEA.

…In the specific case of this paper, one can quibble with the out-year multipliers, or with the productivity assumptions, or with the presumed impact of a higher minimum wage. One can invoke the trade deficit or the exchange rate. Professor Friedman makes all of these points himself. But those issues are well within mainstream norms.

There is no “magic asterisk,” no strange theory involved here. And the main effect of adjusting the assumptions, which would be a perfectly reasonable thing to do, would be to curtail the growth rate after a few years – not at the beginning, when it would matter most.

It is not fair or honest to claim that Professor Friedman’s methods are extreme. On the contrary, with respect to forecasting method, they are largely mainstream. Nor is it fair or honest to imply that you have given Professor Friedman’s paper a rigorous review. You have not.

What you have done, is to light a fire under Paul Krugman, who is now using his high perch to airily dismiss the Friedman paper as “nonsense.” Paul is an immensely powerful figure, and many people rely on him for careful assessments. It seems clear that he has made no such assessment in this case.

Instead, Paul relies on you to impugn an economist with far less reach, whose work is far more careful, in point of fact, than your casual dismissal of it. He and you also imply that Professor Friedman did his work for an unprofessional motive. But let me point out, in case you missed it, that Professor Friedman is a political supporter of Secretary Clinton. His motives are, on the face of it, not political.

What does the Friedman paper really show? The answer is quite simple, and the exercise is – while not perfect – almost entirely ordinary.

What the Friedman paper shows, is that under conventional assumptions, the projected impact of Senator Sanders’ proposals stems from their scale and ambition. When you dare to do big things, big results should be expected. The Sanders program is big, and when you run it through a standard model, you get a big result.

That, by the way, is the lesson of the Reagan era – like it or not. It is a lesson that, among today’s political leaders, only Senator Sanders has learned.

Don’t make Galbraith angry. You won’t like him when he’s angry.

To review, we have Very Prominent Scientists proclaiming something that turns out to be incorrect–but, thanks to their prominence, is passed far and wide as accepted wisdom, all the while stomping all over a less prestigious researcher who used acceptable standards for the field. I guess economics is a real science after all. But I digress.

I don’t think Krugman was ratfucking–his political biases just enabled him to be duped. But the CEA chiefs probably knew exactly what they were doing. Though, given their own forecasting histories, maybe I’m wrong about that:

As David Dayen notes:

Let’s remember that the CEA has a full staff and the weight of all the data-gathering resources of the U.S. government behind it, compared to one economist in Massachusetts playing with hypothetical models. And yet the CEA still gets it wrong routinely. Which is fine—history tells us we should not expect such precision. But let’s not allow one subset of Democratic economists to take the high road of “evidence-based” mathematics when they’re all throwing darts at a board.

Scientists shouldn’t interact with the public like this at all. And I think the CEA chiefs are too smart to not realize what they were doing. From a standpoint of professionalism, this puts the p-hacking and ‘irreproducibility’ crises to shame.

With behavior like this, maybe economics really is too important to be left to the economists.