Is Paul Krugman a Bayesian?

Or does Krugman merely want John Cochrane to be a Bayesian?

It seems to me that Cochrane’s position – he only said it was a danger, not that it would happen at any particular time, so it signifies nothing if it doesn’t happen even after four years have passed – is just untenable in its strong form. If saying that something is a danger carries no implications for the likelihood that it will actually occur, what is the point of saying it? You might as well stand up there and say “Nice day for weather” or sing “Mary had a little lamb.”

No, clearly talking about the danger of inflation was some kind of statement about probabilities – in particular, a statement that the probability of inflation is, according to the speaker’s model of the world, higher than it is in other peoples’ models of the world. And that means that actual events do or at least should matter – they may not prove that one model is wrong and another is right, but they should certainly affect your assessment of which model is more likely to be right.

In short, it’s a Bayesian thing.

Now let’s apply this to market monetarism.  Early in 2013 Krugman claimed that MM was about to be tested:

as Mike Konczal points out, we are in effect getting a test of the market monetarist view right now, with the Fed having adopted more expansionary policies even as fiscal policy tightens.

And the results aren’t looking good for the monetarists: despite the Fed’s fairly dramatic changes in both policy and policy announcements, austerity seems to be taking its toll.

The results are now almost complete.  So far both RGDP and NGDP growth, as well as job growth, has outperformed the 2012 pace.  So MM easily passed the test. And how did Krugman react?

One way to look at the US economy in 2013 is that it was, in effect, trying to begin a strong recovery, but was held back by terrible federal fiscal policy. Housing was making a comeback, state and local austerity was, if not going into reverse, at least not getting more intense, household spending was starting to revive as debt levels came down. But the feds were raising the payroll tax, slashing spending via the sequester, and more.

Incidentally, these other factors are why I don’t take seriously the claims of market monetarists that the failure of growth to collapse in 2013 somehow showed that fiscal policy doesn’t matter. US austerity, although a really bad thing, wasn’t nearly as intense as what happened in southern Europe; it was small enough that it could be, and I’d argue was, more or less offset by other stuff over the course of a single year.

Here’s more from Krugman (titled “The Year of the Weasal“):

This is, I’d argue, a significant development, because it gives us a new window into the nature of the disagreement. As late as last year you could view this as a legitimate contest between rival models. But we’ve now seen that one side of the debate not only refuses to take evidence into account, but tries to dodge personal responsibility for getting it wrong. This has gone from a test of ideas to a test of character, and a lot of people failed.

I don’t think I even need to comment.

[Update,  In case Noah is reading, maybe I do need to comment.  I mean one shouldn't attack others and then do the same thing.  Obviously I am not saying either Krugman or Cochrane lack character.]

And then this:

In Praise of Art Laffer

No, really. Business Insider contacted him about his 2009 warning that soaring inflation and interest rates were just around the corner — and he not only admits that he was wrong, he admits that his error signified something wrong with his overall economic model:

“Usually when you find the model this far off, you’ve probably got something wrong with the model, not that the world has changed,” he said. “Inflation does not appear to be monetary base driven,” he said.

This is a remarkable act of intellectual honesty — remarkable because it happens so rarely.

Finally something on which I agree with Krugman.  Yes, it happens so rarely.

PS.  I have a post over at Econlog that looks at this Krugman response from a somewhat different angle.

PPS.  Saturos directed me to a Noah Smith post on people who do not update their priors.


Tags:

 
 
 

50 Responses to “Is Paul Krugman a Bayesian?”

  1. Gravatar of Steve Roth Steve Roth
    4. January 2014 at 15:05

    Again:

    Was Krugman asserting that 2012 should be the relevant counterfactual?

    You are asserting that.

    Should we agree with that?

    If Krugman’s counterfactual is what coulda shoulda happened, MM might have failed the test.

    Ceteris not paribus and all that. Neither side can claim an “obvious” result from this “test.”

  2. Gravatar of Scott N Scott N
    4. January 2014 at 15:19

    I suppose it’s progress to hear Krugman arguing that only “intense austerity” (like southern Europe) will affect the economy. It really was too much to expect that he would admit MM was right even after saying 2013 was the perfect test. Very few people change their deeply held beliefs in the face of contrary evidence. Instead, they figure out ways to explain away the evidence in a manner that allows them to maintain their beliefs, which is exactly what Krugman did.

  3. Gravatar of Geoff Geoff
    4. January 2014 at 15:34

    Someone please conclusively refute the argument that growth would have been even better now if there was neither fiscal nor monetary stimulus.

  4. Gravatar of 123 123
    4. January 2014 at 15:35

    Obviously I don’t agree with Krugman here, but suppose we get 6% NGDP growth in 2014. Would this be a partial vindication of his views?

  5. Gravatar of TGGP TGGP
    4. January 2014 at 15:38

    I’m reminded that there was a previous case in which Friedman was predicting inflation would result from Reagan’s tax cuts, and Laffer correctly disagreed.

  6. Gravatar of HPublius HPublius
    4. January 2014 at 15:53

    The reason QE has not produced inflation is because the new money supplies provided by the Fed simply satisfied the dramatic increase in the demand for money (historical graphs here: http://tinyurl.com/q8pmmb2). Basically, the Fed prints money while the public stuffs it in mattresses (a.k.a. bank accounts).

    Also, check out the drop in money velocity since the financial collapse – more evidence of the dramatic rise in money demand. At some point, people will start spending and investing again and money velocity will begin to recover most likely prompted by the wealth effect of higher stock and real estate prices. That’s when the question of inflation will resurface. Should the demand for money begin to decline, inflation will become a problem. The problem is that the Fed does not have the right tools to fight inflation without prompting another recession.

  7. Gravatar of Al Al
    4. January 2014 at 16:23

    @Steve Roth,

    Despite being a Krugman fan, his position on monetary offset is difficult to nail down, at least if you look at his blog posts individually. In October of last year he wrote:

    “The reason austerity is so destructive is that we’re up against the zero lower bound, so that reductions in demand can’t be offset with interest rate cuts. And why are we up against the zero lower bound? The proximate cause is the financial crisis and the overhang of debt left behind by the bubble, but the reason we’re so vulnerable to getting into this position — the reason we tend to be in a liquidity trap whenever we don’t have a bubble inflating — is, or at least that’s what we’re arguing, the generally low natural real interest rate, aka secular stagnation.”

    So what does ‘destructive’ mean? If it refers to fiscal policy offsetting monetary stimulus, then why the talk on liquidity traps and the futility of [conventional] monetary policy? Does it mean a reduced level, rate of change, rate of rate of change, or output gap? His preferred metrics may have been stated in an article, but they don’t get the same play as the more ambiguous claim that austerity is bad.

  8. Gravatar of ssumner ssumner
    4. January 2014 at 16:23

    Steve, That was the claim of people like me. He’s the one that said our monetary offset was going to get tested. If you are right then there was never any test at all. In that case he was wrong back in the spring. He was either wrong then, or he is wrong now, there’s no escaping that fact.

    123, If that’s his prediction, then yes. If not, then not.

    TGGP, Good point.

  9. Gravatar of Rajat Rajat
    4. January 2014 at 16:43

    Don’t know about Bayesian. Maybe a fool wrapped up in an ego.

  10. Gravatar of benjamin cole benjamin cole
    4. January 2014 at 16:46

    Both Cochrane and Krugman seem to have been driven nuts by the fact that open-ended, results dependent QE3 has coincided with better economic growth (almost 200k monthly private-sector job growth) and good investment markets.
    Sadly, both have become shrill and dogmatic.

    But even sadder is that QE3 should have been bigger…as seen by record-low inflation…

  11. Gravatar of gofx gofx
    4. January 2014 at 16:58

    Krugman is more of a “Krayesian” than a Bayesian. He has a degenerate prior such that no amount of data can yield a posterior distribution that’s any different from the prior (that fiscal policy is preferred and monetary policy is not very effective at the ZLB). Plus, his predictive distribution has an amazing “quantum” quality in that the highest probability mass occurs in an interval about whatever the actual observed outcome is, no matter how different each actual outcome is!

  12. Gravatar of Saturos Saturos
    4. January 2014 at 18:39

    gofx, I believe the technical term for that is “derp”.
    http://noahpinionblog.blogspot.com.au/2013/06/what-is-derp-answer-is-technical.html

  13. Gravatar of Saturos Saturos
    4. January 2014 at 18:43

    And here is the illustrative video: http://www.southparkstudios.com/clips/103374/mr-derp

  14. Gravatar of gofx gofx
    4. January 2014 at 18:50

    Saturos,

    Thanks for the information, I didn’t know it was such a well-developed concept!

  15. Gravatar of Saturos Saturos
    4. January 2014 at 18:53

    Off topic – Other news in derp, David Brooks edition: Legalizing pot is encouraging moral failure (“nurturing a moral ecology in which it is a bit harder to be the sort of person most of us want to be”, to be precise): http://www.nytimes.com/2014/01/03/opinion/brooks-weed-been-there-done-that.html?src=recg

  16. Gravatar of Saturos Saturos
    4. January 2014 at 18:55

    gofx, the irony is that Paul Krugman HIMSELF gleefully seized upon pal Noah Smith’s blogo-rhetorical innovation in order to bash HIS opponents: http://krugman.blogs.nytimes.com/2013/07/08/urp-versus-derp/

  17. Gravatar of Ram Ram
    4. January 2014 at 19:00

    Let’s not elevate this into a debate about statistical methodology. (I say this as a statistician.) This is just a problem of using a content-free model. If one’s model is complex enough (too many degrees of freedom), one can always update the parameter estimates in the face of new data and claim to have been vindicated by events. The right way to test a model is to have it make quantitative predictions, having been fitted to a training data set, concerning a test data set (which does not overlap with the training data set). This is what gofx is referring to under the name of a posterior predictive distribution.

    This is the giveaway: “The fact that things eventually turn up is neither a refutation of Keynesian analysis…” Let’s call this out for what it is–psuedoscience. Krugman’s model is consistent with things turning up eventually. Things. Turning up. Eventually. What possible events fail to fit that description? None, as far as I can tell. Hence no data would ever cause him to reject his model. Ergo, there is no model. (We might want to call this a saturated model, but that would give it too much credit.)

    The real scandal here is that this kind of reasoning is considered legitimate at all. Krugman is certainly capable of producing a formal, quantitative model, fitting it to a historical data set, and then evaluating it against subsequent events. Why doesn’t he do it (or have an assistant do it for him)? Why doesn’t everyone demand that he do it? Maybe it’s because Krugman and his audience aren’t actually interested in doing science. Why, then, are we paying any attention to him?

  18. Gravatar of Saturos Saturos
    4. January 2014 at 19:15

    Isn’t it great that our dear derpy Brooks and Obama have the privilege of announcing in public that they used to smoke weed, without being thrown into jail or stripped of any privileges? (indeed, Brooks got paid for that column)

    Ram, I think you’re now digging into the can of worms as to what extent any of the macromodels of recent decades can really be subjected to testing. As Scott points out repeatedly many studies in recent years claim to “test” fiscal stimulus without even looking at the theoretically required assumption that monetary policy be held constant. The line we’ve heard from macroeconomists for ages is that at the end of the day the data simply isn’t sufficient to determine whether fiscal stimulus truly works as advertised.

  19. Gravatar of ssumner ssumner
    4. January 2014 at 19:26

    Saturos, Thanks, I added a link.

    Ram, My view is that 2013 does not prove monetary offset. But I think it does prove monetary offset if we buy into the methodology of internet Keynesians.

    The reason I don’t think it proves monetary offset is because I believe the methodology of internet Keynesians is flawed, as you say.

    A better argument for monetary offset is that fiscal austerity did not seem to have much impact on 2012 growth forecasts. The only reason I went out on a limb in late 2012 was that it seemed to me that markets were predicting growth in 2013 to be roughly equal to 2012. The proper test is the impact on the expected growth rate, not the actual growth rate.

    Saturos, I’d like to see Brooks go and explain to young black males in jail for drug crimes why he (Brooks) should lead a life of luxury while they rot in prison.

  20. Gravatar of Ram Ram
    4. January 2014 at 20:30

    Saturos, fair enough. Krugman’s case is pretty extreme, though. If the economy struggles, it is a vindication of his model, owing to austerity. If it does well, it is a vindication of his model (“things eventually turn up”). Anyone with any ideas at all can do better than that, Krugman especially. So we should hold him to higher standards.

    Scott, I take your point that you’re just criticizing Krugman on his own terms. A more straightforward argument for monetary offset would be some kind of demonstrated association between monetary expansions and fiscal contractions (and vice versa). A study like Romer and Romer, which identifies random variation in the stance of fiscal policy, and models the relationship with the stance of monetary policy. Of course, we’d have to think about how to define that stance. Presumably you’d want to define it in terms of market forecasts, which I get, but that bakes in a lot of other assumptions that put us in a bit of a Duhem-Quine pickle. I don’t have any ideas, but market monetarists should think harder about this problem. The seductive logic of an internally consistent theory can often make it seem unnecessary to pursue external validation, but it is useful to do so, if only to persuade our critics.

  21. Gravatar of Blue Aurora Blue Aurora
    4. January 2014 at 20:59

    Although it’s been a while since I’ve made a comment on here…I believe it would be safe to say that the answer is yes. He seems to accept the subjective Bayesian interpretation of probability as a fundamental building block. Please see the part of the video featured in this blog-post from a few years back.

    http://factsandotherstubbornthings.blogspot.com/2012/11/is-ellsberg-paradox-really-paradox.html

  22. Gravatar of W. Peden W. Peden
    4. January 2014 at 21:00

    Ram,

    Re: the Duhem-Quine problem, I think that the way to deal with that probabilistically is to say that (i) the more the assumptions one has to change to preserve the tested theory and (ii) the more probable those assumptions, the less probable it is that the tested theory is true. So you can preserve geocentrism, IF you chuck out huge swathes of physics and/or observational data, but the balance of probabilities is is against the theory.

    (Naturally this is all very messy and controversial, but such is scientific practice.)

  23. Gravatar of JSeydl JSeydl
    4. January 2014 at 22:15

    “So far both RGDP and NGDP growth, as well as job growth, has outperformed the 2012 pace.”

    This would only be relevant if you *held constant* the housing and deleveraging evolvements. I’m pretty sure this is Krugman’s point. But nice try.

  24. Gravatar of Krugman Unwittingly Confirms Scott Sumner’s Whole Point Krugman Unwittingly Confirms Scott Sumner’s Whole Point
    4. January 2014 at 22:29

    […] At his blog, Scott makes the case even more strongly, by quoting what Krugman actually wrote in April […]

  25. Gravatar of Bob Murphy Bob Murphy
    4. January 2014 at 22:37

    Great post Scott. This is just an open-and-shut case: When Krugman thought 2013 growth would be bad, he said it was a great test of MM. Then when it was fine, he said it was never really a good test. Give me break, Krugman apologists. You don’t have to endorse the Treasury View, you just have to admit Krugman is a sheister with this stuff.

  26. Gravatar of CA CA
    4. January 2014 at 22:49

    Bob, you’re spot on. And you need to push back a little more strenuously against your buddy Daniel Kuehn in the comments at your blog.

    Score one for Sumner in this round.

  27. Gravatar of W. Peden W. Peden
    4. January 2014 at 23:22

    I’m reminded of the post-1981 fiscal tightening in UK economic history. From 1979 to 1981, base money growth fell from about 13% to about 2%; Bank Rate reached 17%; and NGDP growth fell from about 17% to about 10%. By almost any measure (except broad money, which was what the government was targeting at the time and which was being greatly distorted by the removal of credit controls and exchange controls) monetary policy was extremely tight during this period.

    In 1981, the Thatcher government increased expenditure cuts and tax rises as part of a plan to bring down interest rates without boosting inflation. 364 Keynesian economists (including Mervyn King, who I imagine does not like to be reminded of this) wrote a letter to The Times, arguing that the budget would deepen the depression. It was also during this period that Frank Hahn, once the promising hope of general equilibrium theory and British Keynesianism, said that the UK would be shifted onto a permanently lower potential output trend by austerity.

    The Keynesians got it totally wrong. NGDP growth stabilised in 1982-1983, and the UK shifted onto a HIGHER potential output trend.

    When the 1981 budget has been brought up in recent debates, I’ve noticed that Keynesians have argued that monetary policy offset may apply when interest rates are well above zero, as they were in the early 1980s, but not under current circumstances. In other words, fiscal policy isn’t as important as Keynesians used to think, but at least it sometimes matters.

    2013 is interesting because it was an example of monetary policy offset at the so-called “zero-lower bound”. The Keynesian fiscal-stimulus-of-the-gaps now has to get moved into an even smaller gap…

  28. Gravatar of SS SS
    5. January 2014 at 02:24

    http://digitaljournal.com/article/362353

  29. Gravatar of Kevin Donoghue Kevin Donoghue
    5. January 2014 at 02:47

    Andy Harless on Twitter:

    There isn’t a market monetarist textbook, is there? Somebody should write one.

    I’ll form a view on Market Monetarism when I’m told what it is.

  30. Gravatar of 123 123
    5. January 2014 at 03:06

    Scott:
    ” If that’s his prediction, then yes. If not, then not.”

    Yes, he’s very optimistic about 2014:
    “There’s an alarming amount of optimism out there about US economic prospects for 2014. Let me make the situation even more alarming by saying that I basically share that optimism.

    Why? Basically because of the Three Stooges effect: if you’ve been banging your head against a wall for no good reason, you’ll feel a lot better when you stop.”

    But he is not optimistic in a Bayesian sense though:
    “US austerity, although a really bad thing, wasn’t nearly as intense as what happened in southern Europe; it was small enough that it could be, and I’d argue was, more or less offset by other stuff over the course of a single year.”

    So heads Krugman wins, tails you lose.

  31. Gravatar of dtoh dtoh
    5. January 2014 at 03:21

    OMP doesn’t do anything if the Fed’s ultimate counter-party is just other banks. If you want an effective test, the Fed needs to hold ER constant (or at least limit the rate of ER growth to a rate below the rate of increase in the base.)

  32. Gravatar of SG SG
    5. January 2014 at 04:17

    Krugman’s Keynesian framework is again leading him astray. He talks about things like an improving housing situation and reduced state/local austerity as things that boost the economy regardless of what the Fed does. Of course, monetary policy is basically setting a ceiling on economy-wide growth, and the composition of that growth is irrelevant. Krugman fails to recognize that less-tight policy is a necessary condition for those positive economic developments to exist in the first place (or, to be more precise, for them to exist in the absence of spending reductions in other sectors of the economy).

  33. Gravatar of ssumner ssumner
    5. January 2014 at 06:19

    Blue, He sees the theoretical appeal of Bayesian thinking, but not when it applies to him.

    JSeydl, So you are saying that Krugman was lying when he said 2013 would be a test of market monetarism? That’s not very polite.

    Bob, That’s so obvious, but lots of people don’t get it, like the commenter before you. They act like I’m trying to put words into Krugman’s mouth. They act like I’m the one that claims it was a test of MM. No, that wasn’t my claim, it was Krugman’s claim. But he won’t tell us the result of the test.

    W. Peden. Good point.

    Kevin, I did a post “What is Market Monetarism?” I’ll try to find it.

    123, The quote you provide says that Krugman expects a good year, not that he expects NGDP growth to be 6%. As far as I can tell (we don’t have GDP futures markets) the market expects a good year, albeit more in the 4.5% range. (Maybe 2.8% growth and 1.7% inflation). Interestingly, Keynesians are now complaining that the removal of extended UI will slow growth, while supply-siders like Mulligan suggest it will speed up growth. At this point we don’t know if the extended UI will be added back in by the new Congress. In a post I’ve written but not posted, I give unemployment forecasts with and without extended UI.

    Again, if Krugman wants credit for a prediction, he needs to be specific. I don’t see 6% in the quote you provide.

    SG, Yup, and recall the Fed almost tapered in September. It was a very close call. A bit more growth elsewhere and they would have.

  34. Gravatar of ssumner ssumner
    5. January 2014 at 06:37

    Kevin, Here’s a description of MM.

    http://www.themoneyillusion.com/?p=13353

  35. Gravatar of 123 123
    5. January 2014 at 07:30

    David Beckworth: “Paul Krugman Plugs Market Monetarism”

    http://macromarketmusings.blogspot.com/2014/01/paul-krugman-plugs-market-monetarism.html

  36. Gravatar of Patrick R. Sullivan Patrick R. Sullivan
    5. January 2014 at 09:44

    ‘I’m reminded that there was a previous case in which Friedman was predicting inflation would result from Reagan’s tax cuts….’

    The guy who was famous for saying that, ‘Inflation is always and everywhere a monetary phenomenon.’?

    Are you confusing Friedman with Jimmy Carter?

  37. Gravatar of Dustin Dustin
    5. January 2014 at 09:57

    I pretty much disagree with Krugman on this.

    Vehicle crash fatalities are a danger and despite their rare occurrence, I would bet Dr Krugman wears his seatbelt when driving.

    Risk management is a basic strategic planning function – identify risks, prioritize them as a function of likelihood and impact, and develop / implement mitigating controls.

    Dr Cochrane simply identified what he felt was the most likely risk associated with the policy, saying nothing of its probability. Dr Krugman did not even acknowledge the impact and fails to provide his arbitrary threshold of probability that merits risk acknowledgment.

  38. Gravatar of Kevin Donoghue Kevin Donoghue
    5. January 2014 at 12:06

    Scott, thanks for that explanatory link. I’m inclined to think the differences between (say) Nick Rowe and Simon Wren-Lewis are a bit too subtle to thrash out in a comment thread. It wouldn’t be easy to pin down why Nick R self-identifies as a MM and Simon W-L doesn’t.

  39. Gravatar of 123 123
    5. January 2014 at 15:49

    Delong reacts, mischaracterizes market monetarist views: http://delong.typepad.com/sdj/2014/01/the-relative-efficacy-of-fiscal-and-monetary-policy-at-the-zero-lower-bound-where-are-the-goalposts-anyway-the-honest-bro.html

  40. Gravatar of Noah Smith Noah Smith
    5. January 2014 at 16:57

    For what it’s worth, my prior – that neither fiscal nor monetary policy has a big effect – has been strengthened by events in the U.S., though Abenomics is causing me to slightly question that.

  41. Gravatar of ssumner ssumner
    5. January 2014 at 17:45

    Noah, Unless you tell me whether “no effect” means no nominal effect or no real effect I don’t know how to interpret your comment. Money is neutral in the long run, so presumably you don’t mean no nominal effect. But then are you saying that nominal shocks don’t have real effects?

  42. Gravatar of dtoh dtoh
    5. January 2014 at 18:39

    Noah, Scott,

    I think you need to look at the nominal effect in the context of MO – ER. Monetary policy which is simply the exchange of assets for deposits between private banks and the Fed have no more effect on the economy than an exchange of assets for deposits between the St. Louis Fed and the Minneapolis Fed.

    In 2013, we’re actually looking at slight decline in the rate of growth (of MO-ER) so while policy is still expansionary in 2013, it was less so than in 2012.

  43. Gravatar of Jim Glass Jim Glass
    5. January 2014 at 21:55

    Vehicle crash fatalities are a danger and despite their rare occurrence, I would bet Dr Krugman wears his seatbelt when driving…

    Dr Cochrane simply identified what he felt was the most likely risk associated with the policy, saying nothing of its probability. Dr Krugman did not even acknowledge the impact and fails to provide his arbitrary threshold of probability that merits risk acknowledgment.

    Compare this to Krugman’s columns on the Democrats’ (failed) carbon bill, where he said the risk in the long tail of global warming probabilities is so devastating that persons opposed to the bill were committing “treason”.

    This is just how the guy’s mind works, if he thinks a given line of logic supports his position, those who don’t accept it are “betraying the planet” — if he thinks the same line of logic goes the other way, it is all ‘huh?’

    Those of us who can remember before 2000, when Clinton ruled and before PK became so obsessed with Bush, will recall how he used to do this to his ‘friends’ on the left. (William Greider, a frequent victim, remembers not liking being called a Luddite.)

    My favorite from back then was a 1996 speech (since edited for politeness and reprinted in “Accidental Theorist”) in which he savaged those urging the Clinton Administration to aim for 4% annual growth in light of the productivity boom.

    He proudly described himself as the “obnoxious economist” who had to teach reality to “the four percenters” — instructing them that “4% growth for 5 years would mean an eventual unemployment rate of something like 1.5%”, which with NAIRU being around 5.5% would be wildly inflationary, which the Fed never allow, and thus was plainly simply impossible by arithmetic, QED.

    “So how can smart people think that a 4% growth target is feasible? Now you may think that what I am saying is that these guys are dumb … What’s wrong with the kind of economics [they] practice is that they have failed to understand principle. They think that you do economics the way a lawyer prepares a brief for a client — first you decide on your opinion, then you marshall as many plausible arguments as you can in your support.”

    No, they would first have to think through objective, impartial scientific models, as he did, before their thinking would become “coherent”.

    Of course, 5+ years of average real GDP growth of 4.5%, all years 4%+, promptly followed, with unemployment briefly bottoming at 3.9%. An economist I know who attended a later AEA convention said PK was asked about why he was wrong about this and replied, wrong? the fundamental relationships changed (labor force participation shot up), how was he supposed to foresee that? so how was he wrong? (Though he had explicitly mocked the 4%ers for saying that fundamental relationships were changing.)

    Of course the current howlingly big example, every time he dismisses as foolishness the notion that rising debt poses any real risk to future national welfare, while dismissing the shoddy ne’er-do-well Rogoff-Reinhart analysis, is how he claimed to be “terrified” by the US govt debt back in the days of the Bush deficits, and of the “looming threat to the federal government’s solvency” via the surging interest rates and inflation that must follow QED from such fiscal recklessness.

    When I mentioned this in another forum recently, a PK defender told me “But you must give him credit for admitting he was wrong about that back then”. Well, I dunno, it seems pretty convenient to admit having been wrong in the past in ways that free you to endorse the current political policy of your own side that you personally want. While never being wrong the other way. Why it actually smacks of cherry picking convenient arguments on behalf of the political client of the moment! (E.g., In light of his past experience of making that serious error, does PK express any caution that his current opinion might also prove wrong — and if it does, what the cost could be?)

    Speaking of which, Krugman has no idea how we lawyers work, either — another mistake on his part. We do not work by compiling a list of attractive arguments for our client. (Another slur!) That’s how marketers and PR people work.

    We start by taking very seriously the arguments of the other side, and addressing them on the merits. Because when it comes to a competent impartial judge of the matter, trying to dismiss the other side’s arguments with name calling and ridicule instead of addressing them seriously on their own terms is to *lose*. (Though sometimes you can get away with it with a jury of the uninformed by manipulating their emotions that way.)

    In fact, because we are *forced* to deal with the arguments of the other side seriously and with respect, we probably are less prone to all the errors of “motivated reasoning” than are any other human beings. I hesitate to claim any general moral superiority for lawyers as a result of this, but…

    Krugman would totally fail as a litigator. “Your honor, how could you have just ruled for the other side when their arguments are self-evidently so patently amateurishly stupid and evil that I didn’t even have to take them seriously? What does that make you??” (Though his technique is a lot less ineffective when sold to the choir who buy his books which have Republicans with Hitler mustaches on the cover.)

    Anyhow, from what I’ve seen, that’s a sampling of how PK works with his priors. :-)

  44. Gravatar of Noah Smith Noah Smith
    6. January 2014 at 10:49

    I meant no real effect, but I wonder about nominal effects too…

  45. Gravatar of J Mann J Mann
    6. January 2014 at 14:13

    1) Krugman’s responding in December to a July blog by Cochrane – is that typical for PK or the internet generally?

    2) I’m an imperfect Bayesian, but I would think a Bayesian restatement of Cochrane’s response would be:

    “Obviously in hindsight, the probablility of runaway inflation and a catastrophic collapse in confidence in US governent assets occuring during the prior 4 years can now be seen to be 0, but I stand by the assessment I made then based on what I knew then, and I continue to see it as a significant risk going forward, because the underlying risk factors haven’t substantially changed.”

    Is that reasonably Bayesian, and would it make Krugman any happier?

    3) Clicking through to Cochrane’s original 09 statement, he said that in his opinion, the structural factors aren’t in place to cause deflation or a depression, and that if anything, the possibility of inflation or a catastrophic run on US debt is more concerning at this time.

    His reasoning in the piece Krugman maligns is:

    a) Cochrane feels that the only long run ways out of the current debt and other government obligations are (i) substantially increased growth; (ii) inflation; or (iii) default. (He doesn’t see sufficient spending cuts and/or tax increases as practicable).

    b) Given that, he thinks there’s a substantial risk of future inflation or default, and a consequent run on US debt.

    c) His preferred policy response is structural reforms to encourage growth.

    As far as I understand Krugman, he seems to be saying that because we haven’t seen much inflation in the last four years, a good Bayesian would increase his internal likelihood that we might never see inflation in the forseeable future. Of course, that’s nonsensical, so I assume Krugman and DeLong have private understandings of what Krugman means.

    But if that’s what he means, that’s not what Bayes says. If I inspect a coin, observe it to be of typical weight and size and to have a head and a tail, I am going to predict a 50% chance that if I flip it, it will come up tails. And if I flip it 4 times and it comes up heads, I am not going to discount my prediction by much.

    If my prediction doesn’t have a stated probability, like Cochrane’s didn’t, that discount might not even be relevant.

    If I see you load a gun and play Russian roulette to pay the mortgage, I will say “Your greater danger isn’t homelessness, it’s a bullet to the head.” And if you pull the trigger a few times and don’t get the bullet, I am not going to discount my probability by much regarding the next shot. It’s POSSIBLE that there is some slight of hand going on, and each missed shot increases my assessment of that possibility marginally, but not enough to render my initial statement inaccurate.

  46. Gravatar of dtoh dtoh
    7. January 2014 at 05:24

    Noah,
    Do you really mean

    1. No effect at all, in which case the Fed can buy all government debt (and we can eliminate all taxes) and the Fed can give us all 100 year zero interest fixed rate home, auto, and credit card loans.

    Or do you mean,

    2. The effect is a step function. (If this is what you mean, would you mind letting us know where the step occurs?)

  47. Gravatar of Macro Experiments are not Controlled | askblog Macro Experiments are not Controlled | askblog
    7. January 2014 at 05:27

    […] Read the whole thing, especially if you do not know the austerity-test controversy to which Alex is referring. And if you want more, you can read Mark Thoma or Scott Sumner either at EconLog or at MoneyIllusion. […]

  48. Gravatar of K Simpson K Simpson
    7. January 2014 at 09:28

    Summary:

    1) Krugman is a self-described “obnoxious economist”. Few will argue with the “obnoxious”. But, calling yourself an “economist” implies that you are a scientist, which implies that you change your posteriors when evidence refutes your priors. Krugman never adjusts his priors; He only continues to obnoxiously herp derp. Ergo, Krugman can only be described as “obnoxious”. He is not, at his core, an economist, but a leftist, big-government hat rack (where Keynes-exploders can hang their hats in promoting more vote-buying handouts .

    2) As long as counterfactuals exist, Krugman can continue to surf his wave of derp to fortune, fame, and high acclaim. Perhaps even some shiny European medal.

    3) Counterfactuals will always exist.

  49. Gravatar of Gregor Bush Gregor Bush
    7. January 2014 at 13:57

    Noah,
    “I meant no real effect, but I wonder about nominal effects too…”

    Tell me, if there are no nominal effects from monetary policy how is it that central banks able to successfully target inflation? (Canada, New Zealand and Australia have all had average inflation rates that are EXACTLY in line with their targets for more than 20 years). Also, if there are no nominal effects, why not just have the Fed monetize the entire stock of US debt? Why are you paying taxes to service a huge mountain of debt that you could simply replace with freshly printed cash without any inflationary consequences?

    Monetary policy has no nominal effects? Really?

  50. Gravatar of ssumner ssumner
    8. January 2014 at 06:16

    Gregor, That’s a point I’ve also been emphasizing recently. I think his wiggle room is “big impact.” Surely he thinks there is some impact.

Leave a Reply