Saturday, March 02, 2013

Blogger smackdown: Robert Waldmann vs. David Glasner

Somehow I missed the first few salvos in this little police action, but I'm on it now! In this corner, we have Robert Waldmann, one of the pioneers of "noise trader" finance theory. In the other corner we have David Glasner of the FTC. Today's ingredient is wild mushrooms question is whether monetary policy works.

(I have to say, it's pretty fun to see these two guys debate, since Waldmann is one of the most terse bloggers out there, and Glasner one of the most verbose.)

Ben Bernanke too has declared a policy of unlimited quantitative easing and increased inflation (new target only 2.5% but that's higher than current inflation).  The declaration (which was a surprise) had essentially no effect on prices for medium term treasuries, TIPS or the breakeven....[Y]ou have confidently asserted again and again that if only the FOMC did what it just did, expected inflation would jump and then GDP growth would increase. However, instead of noting the utter total failure of your past predictions (and the perfect confirmation of mine) you just boldly make new predictions. Face fact,  like conventional monetary policy (in the US the Federal Funds rate) forward guidance is pedal to the metal. It's long past time for you to start climbing down. 
Basically: "QE-infinity didn't budge inflation expectations, so monetary policy doesn't look like it works the way monetarists say it should work."

Glasner then jumps in to say, well, the Fed hasn't really done very much in terms of changing policy:
I mention this, because just yesterday I happened across another blog post about what Bernanke said after the FOMC meeting.  [That] post by David Altig, executive VP and research director of the Atlanta Fed, was on the macroblog. Altig points out that, despite the increase in the Fed’s inflation threshold from 2 to 2.5%, the Fed increased neither its inflation target (still 2%) nor its inflation forecast (still under 2%). All that the Fed did was to say that it won’t immediately slam on the brakes if inflation rises above 2% provided that unemployment is greater than 6.5% and inflation is less than 2.5%. That seems like a pretty marginal change in policy to me.
Waldmann then responds with two points. First, he criticizes Glasner for failing to mention the Zero Lower Bound: 
Your analysis of monetary expansion does not distinguish between the cases when the ZLB holds and when it doesn't...I think it is clear that the association between the money supply and domestic demand has been different in the USA since oh September 2008 than it was before...
Second, he says that the data show no indication of Fed policy announcements having any effect:
I claim that the null that nothing special happened the day QEIV was announced or any of the 4 plausible dates of announcement of QE2 (starting with a FOMC meeting, then Bernanke's Jackson Hole speech then 2 more) can't be rejected by the data. This is based on analysis by two SF FED economists who look at the sum of changes over three of the days (not including the Jackson Hole day when the sign was wrong) and get a change (of the sign they want) whose square is less than 6 times the variance of daily changes (of the 10 year rate IIRC).  IIRC 4.5 times.  Cherry picking and not rejecting the null one wants to reject is a sign that one's favored (alternative) hypothesis is not strongly supported by the data.
Basically: "We shouldn't assume that monetary policy works at the ZLB. We should assume it doesn't work at the ZLB, unless we have clear evidence that it does. We haven't seen any clear evidence that monetary policy works at the ZLB, so we should continue to assume it doesn't."

Glasner responds first by throwing out some of his own ideas about how monetary policy should work:
I agree that the zero lower bound is relevant to the analysis of the current situation. I prefer to couch the analysis in terms of the Fisher equation making use of the equilibrium condition that the nominal rate of interest must equal the real rate plus expected inflation. If the expected rate of deflation is greater than the real rate, equilibrium is impossible and the result is a crash of asset prices, which is what happened in 2008. But as long as the real rate of interest is negative (presumably because of pessimistic entrepreneurial expectations), the rate of inflation has to be sufficiently above real rate of interest for nominal rates to be comfortably above zero. As long as nominal rates are close to zero and real rates are negative, the economy cannot be operating in the neighborhood of full-employment equilibrium. I developed the basic theory in my paper “The Fisher Effect Under Deflationary Expectations...and provided some empirical evidence (which I am hoping to update soon) that asset prices (as reflected in the S&P 500) since 2008 have been strongly correlated with expected inflation (as approximated by the TIPS spread) even though there is no strong theoretical reason for asset prices to be correlated with expected inflation, and no evidence of correlation before 2008. Although I think that this is a better way than the Keynesian model to think about why the US economy has been underperforming so badly since 2008, I don’t think that the models are contradictory or inconsistent[.]
(Got that? There will be a quiz afterward.)

Glasner then says that no, since we have plausible theories about how monetary policy should work, we should assume that it does work until we have convincing evidence otherwise:
I think that the way to pick out changes in monetary policy is to look at changes in inflation expectations, and I think that you can find some correlation between changes in monetary policy, so identified, and employment, though it is probably not nearly as striking as the relationship between asset prices and inflation expectations. I also don’t think that operation twist had any positive effect, but QE3 does seem to have had some...[E]ven if Waldmann is correct about the relationship between monetary policy and employment since 2008, there are all kinds of good reasons for not rushing to reject a null hypothesis on the basis of a handful of ambiguous observations. That wouldn’t necessarily be the calm and reasonable thing to do.
Basically: "No, my hypothesis is the null!"

Waldmann fires back. First, he says, "No, your hypothesis should not be the null, mine should":
1) Get the null on your side is my motto (I admit it).  You follow this.  You suggest that your hypothesis is the hull hypothesis then abuse Neyman and Person by implying that we can draw interesting conclusions from failure to reject the null.  Basically the sentence which includes the word "null" is the assertion that we should assume you are right and I am wrong until I offer solid proof.  To be briefer, since we are working in social science, you are asking that I assume you are right.  This is not an ideal approach to debate. 
I ask you to review your sentence which contains the word "null" and reconsider if you really believe it.  The choice of the null should be harmless (it is an a priori choice without a prior).  How about we make the usual null hypothesis that an effect is zero.  Can you reject the null that monetary policy since 2009 has had no effect ? At what confidence level is the null rejected ?  Did you use a t-test ? an f-test ?  "null" is a technical term and I ask again if you would be willing to retract the sentence including the word "null".
In other words: "Macro data sucks, so whoever has the burden of proof will always lose. By putting the burden of proof on me, you want me to hand you an automatic win. But no, the burden of proof is on the people who think monetary policy does work at the ZLB."

(Update: Waldmann shows up in the comments and says: Your translation of my statement about QE# isn't exactly what I meant...The evidence that the true effect of FOMC announcements on expected inflation is small is solid, clear, and strong.")

Waldmann also continues the discussion about what the data tell us with regards to whether our monetary policy is working:
[U]sing expected inflation to identify monetary policy is only a valid statistical procedure if one is willing to assume that nothing else affects expected inflation.  If you think that say OPEC ever had any influence on expected inflation, then you can't use your identifying assumption.  In particular TIPS breakevens can be fairly well fit (not predicted because not out of sample) using lagged data other than data on what the FOMC did...I think we can detect the effect of recent monetary policy on TIPS breakevens if we agree that it (including QE) is working principally through forward guidance.  There should be quick effects on asset prices when surprising shifts are announced.  QE 4 (December 2012) was definitely a surprise.  The TIPS spread barely moved (within the range of normal fluctuations).  I think the question is settled.
Glasner retorts that all this talk of "null hypotheses" and "alternative hypotheses" is bad:
What I meant to say was that even if the evidence is not sufficient to reject the null hypothesis that monetary policy is ineffective, there may still be good reason not to reject the alternative or maintained hypothesis that monetary policy is effective. In the real world, there is ambiguity. Evidence is not necessarily conclusive, so we accept for the most part that there really are alternative ways of looking at the world and that, as a practical matter, we don’t have sufficient evidence to reject conclusively either the null or the maintained hypothesis. With the relatively small numbers of observations that we are working with, statistical tests aren’t powerful enough to reject the null with a high level of confidence, so I have trouble accepting the standard statistical model of hypothesis testing in this context.
In other words, macro data sucks, so we'll never have any firm conclusions. In the absence of good data we must rely on plausibility of theory, tempered with a little bit of data.

Glasner follows this with some musings on Copernicus, Galileo, Ptolemy, etc. He basically says "Don't toss out monetarist theory just because the data isn't solidly in its favor. Someday we may get better data." He then goes on to discuss whether QE-infinity really was a surprise and which factors we should assume are more important in driving inflation expectations.

Anyway, you probably get the point by now. As I see it, the take-away from this debate should be the following:

1. Macro data is really uninformative.

2. Because macro data is really uninformative, whoever gets stuck with the burden of proof in a macro debate is usually going to lose. Hence, the way not to lose a macro debate is to play "hot potato" with the burden of proof.

3. If Fed intentions and the expectation formation process are both unobservable, we'll never be able to tell how well monetary policy really works. This is a point I've made before.

At least until we are better able to understand the intentions of the Fed and how inflation expectations are formed, monetarism will involve a big leap of faith. But the pooh-poohing of monetarism will also require a big leap of faith.

(P.S. - "WADR" means "With all due respect", IIRC. AFAIK, Waldmann was asking JOOC, but IMHO that's OK, FWIW. LOL)


  1. I'm sure Scott Sumner has something interesting to say about this...

    Point 3 seems partly tautological, that is "if the true effectiveness of monetary policy is unobservable we'll never be able to tell how well monetary policy really works."

    Waldmann seems to make a lot more sense on this one. His hypothesis is the null. The null should be what a three-year old assumes about the world. Ask him if just altering the money supply ("printing money") can somehow spur growth at ZLB (okay, granted, a three-year-old hopefully won't know what that is).

    Anyway, it's irrelevant to discuss this in such scientific terms. Because say Bernanke keeps those notes coming, and coming. And NGDP doesn't get back to its old path. More and more people will start believing, hey, printing money can't save the world. And then say we entertain the idea of fiscal stimulus.

    Ten years later the monetarists will tell you in the counterfactual that we'd kept the money coming, it WOULD have returned to pre-crash path.

    Where is the science??

    1. Point 3 seems partly tautological, that is "if the true effectiveness of monetary policy is unobservable we'll never be able to tell how well monetary policy really works."

      Good point. I edited that out.

      Ten years later the monetarists will tell you in the counterfactual that we'd kept the money coming, it WOULD have returned to pre-crash path.

      Where is the science??

      Welcome to macroeconomics.

    2. So basically Glasner's argument is like Pascal's Wager?

    3. Yes, but so would be the argument of someone who says monetary policy is dangerous (e.g. Steve Williamson, all the Austrians, etc.).

      Imagine you're a medieval doctor, and your patient is sick. One assistant says "Bleed him! We have a plausible theory as to why bleeding is effective; it drains the evil humors from the body!" The other assistant says: "Actually, we have no proof that that works." The first assistant says: "OK, you have a better idea?" The second assistant says "Sure; put leeches on him. Leeches, being disgusting creatures, feed on bad humors; hence, they will preferentially suck out the bad humors while leaving the good blood in place." The first assistant responds: "OK, where're your proof that leeches suck out bad humors first?"

      Now the guy is dying. You're the doctor; the decision is yours. What do you do?

    4. Anonymous5:08 PM

      The funny part is that both bleeding and leeches are in use in modern medicine, appropriately used, of course. Humors, which both assistants believe in, bit the big one, though.

    5. I'd do neither. Because unless I'm religious my null hypothesis would be that either of them would cause harm, without any discernible benefits, and as any doctor knows: primum no nocere.

      I give Glasner more credit than either of these doctors. He supports his argument with theory. Austrians aren't because their theory sounds like superstition, but that's a different story.

      Plus we don't know that monetary policy causes harm, per se (let's ignore Rothbard for now). The reason I think Waldmann's argument comes out ahead is he's saying "hey, we don't know this works, I'm not saying don't do it, but let's try something else as well."

      So I'm the doctor. Let's say monetary policy is a magic warm towel that might work. I apply it. But this other guy comes and says, "Hey, look, you got to assume that's not going to work. Try it, but why don't you also try something else, maybe feed him."

      Waldmann is just assuming a lot less.

    6. I'd do neither. Because unless I'm religious my null hypothesis would be that either of them would cause harm, without any discernible benefits, and as any doctor knows: primum no nocere.

      OK, I see. Suppose you know he'll die if you do nothing. Would that change your calculus?

      Or, here's another scenario: What if he's already bleeding? does "do no harm" mean stopping the bleeding, or letting it continue (given that one assistant thinks bleeding will actually help)? Remember, no using your modern medical knowledge.

    7. Well, I'd let him continue to bleed.

      Anyway. sure I'd let him continue to bleed. Especially if one assistant thinks it will help and the other is doubtful it will help, and expresses no opinions on whether it will hurt.

      That's my point though. One guy is sitting and cheering "bleed, bleed, bleed!" and pointing to his theory (maybe, right, we don't know) that it works. The other guy is saying, "I don't know that it doesn't work, I'm not saying don't bleed him, why don't we do something else, as well?"

      Re: your first point, 'suppose he'll die anyway'. Depends. If I act independently and have no opinions of my own, I would do nothing, because I have no reason to believe bleeding him is better than doing nothing. But if I consider these assistants to be experts and another expert doesn't tell me bleeding causes harm then yes, I'd bleed him.

      Back to the previous question... Why rely like religious ducks to see if the Fed works? Sure, maybe it does, and work towards that assumption. In the mean time, also do other things like, oh I don't know, not slashing spending?

    8. Though to be fair I think bleeding (or leeches) are an unfair comparison to Glasner's argument. Even the ancient doctors knew that bleeding or leeches hurt without knowing (as in justified true belief) that it helped.

      We don't know increasing the money supply is painful in the same sense that bleeding is. That comparison implies some sort of sacrifice.

    9. At any rate, hopefully you see my point policy usually requires a leap of faith in one direction or another... ;-)

    10. Absolutely, never doubted that! (Or expected this whole thing to be some scientific calculatory endeavor).

      Would you agree, though, that it's a bigger leap to believe something works than to believe something may work, go through with it, but also do something else? I know there are lots of people that think expansionary fiscal and monetary policy at the same time is a bad idea, but limited to the frame of argument provided in your post, I can't see the argument for not doing both, which is why Glasner seems to be the more blindly faithful one!

      (see the fun of things like this is idiots like me can have an opinion and debate... Where can I debate , oh I don't know, whether the Copenhagen interpretation is appropriate w/o being an expert!)

    11. Wouldn't the null hypothesis always be "change nothing from the present"? So if Mr. 2nd Medieval Doctor just said, "We have no proof that bleeding does anything - we should do nothing at all", he'd be in the "null" situation.

    12. Brett,
      this whole discussion has got off target. We are not looking for a "null" policy. We are trying to discern what works and what doesn't. One guy is saying "there is no conclusive evidence that my policy doesn't work". The other is saying "there is no evidence that it does work".

  2. The Blorch5:28 PM

    Monetary policy doesn't make sense because it is quaint. It was meant to control the business cycle in a post Malthusian world where technological breakthroughs in productivity occur at a rate faster than population growth causing growth in both real and nominal per capita income, as well as inflation.

    But now look what happened. Globalization has socked the developed world a Malthusian, Hawaiian punch. Suddenly and with great discontinuity the population of people not employed in the industries of the developed world swelled as the developed world grew to include these populations in the new global political economy. Real per capita incomes are now stagnant once more so monetary policy is redundant.

    1. Except population in the developed world isn't growing geometrically. And there are 2x as many calories being produced as needed for a real diet.

      Doesn't sound Malthusian to me.

      Monetary policy has to exist, right, even if that is the ultimate negative of no state alteration of the money supply at all (basically the world of Ludwig von Mises).

    2. The Blorch3:55 PM

      I was referring to the Malthusian trap which, ironically I guess, isn't very Malthusian.

      The only monetary policy now is trivial. Its just set it and forget it.

    3. What makes you think that monetary policy is trivial in a stationary world. There are still trade deficits. There is still new credit being issued and old credit being retired. The world is still active under the hood. It is like looking at the atmosphere from Jupiter and staying it is in stasis. Hey man, there are hurricanes down there.

    4. P.S. Yes there is a famous quote from Keynes which is quite apposite here.

  3. "WADR" was genuinely mysterious to me. One might suspect that this is due to the fact that the concept of "due respect" is alien to me, but, in any case, I didn't guess the phrase for which it is the acronym (your mileage does vary, but I am 52 and just can't handle the way the kids text these days).

    What does "JOOC" mean ? I am honestly asking for information. I suspect it means something like "feigning ignorance" with the implication that I was implying that I would never have guessed that WADR means "with all due respect" given how little respect Glasner was showing me. In fact, I just didn't get the acronym.

    I may come back to comment on content. First I have to read Glasner's latest post (I have known about it for hours, and the first sentence seems polite).

    I don't really think that macro evidence is so feeble. There is more of a problem that people wish their opinion to be treated as a null and think that means a failure of the data to reject it amounts to something, but also refuse to define their opinion precisely enough for it to be testable. I think macro data are plenty numerous enough to refute RBC and a modicum of intelletually integrity would have ended the approach by 1976 at the latest (I think it started in 1976).

    Similarly, I think that there is plenty of information about monetary policy at the ZLB and it is clear that there isn't much the FOMC can do or could have done short of sneaking subsidies for housing around the limits of the Federal Reserve act. I stress that the qualifier "the FOMC" is necessary to my claim. I think that a long lived monetary dictator could precommit further ahead than the FOMC.

    1. JOOC, I'd like to know what JOOC means.

    2. I think macro data are plenty numerous enough to refute RBC and a modicum of intelletually integrity would have ended the approach by 1976 at the latest (I think it started in 1976).

      I think the real problem with macro data is that it's just time-series. With only time-series, you have to assume stationarity, and yet all unit-root tests are low-power. Furthermore many business-cycle models depend crucially on assumptions of stationarity or non-stationarity of some finite-order AR or ARIMA-type process. Second, you have to assume ergodicity, and even if you drop the assumption and include trend breaks, the same data that lets you test for trend breaks doesn't let you estimate parameter values (or vice versa).

      On top of all this, there's the problem that no one can agree about what the "business cycle" actually IS (H-P filtered time series? First differences? Something else?), making conclusions sensitive to the way you define your trend (but nobody can test what the trend really is).

      I agree that RBC looks like toast, and I would probably be willing to be persuaded that New Keynesian (monetarist) models are toast, but is there a fully specified "old Keynesian" or ZLB-New-Keynesian model out there that gives good out-of-sample predictions and/or isn't rejected by a FIML test? I haven't heard of such a thing...

  4. Non-economist take: Too many economists think in terms of equations and infer causal relations from the equations. In the real world the relationships may be inequalities (constraints) which do not imply a causal relationship.

    If we have A = B an economist would conclude that increasing B would increase A. On the other hand if the real relationship is A < B then increasing B may or may not increase A depending on what else is influencing A.

    Monetary policy can certainly cool an economy by raising interest rates. It can stimulate an economy by relaxing the constraint but only over the range where interest rates are the operating constraint on the economy. Once interest rates are below some threshold another constraint will operate and further reductions will not stimulate the economy.

    The purchase by the Fed of mortgage backed securities is a direct intervention in credit channels and the Fed is doing it because credit channels are broken so monetary policy per se was not feeding through to borrowers.

  5. Is Waldmann's argument that the Fed can fully monetize the US debt and it not be inflationary? Then let's party!!!

    1. It would not be inflationary to monetize the short term debt (with negative real yields). Why would it - and why would it be a party?

  6. Your translation of my statement about QE# isn't exactly what I meant. It is true that we don't have clear evidence of the sign of the effect, but we have evidence on its magnitude.

    A point estimate which is very close to zero is evidence that the true effect is small (perhaps weak evidence is the stadard error islarge,but in this case it isn't). The evidence that the true effect of FOMC announcements on expected inflation is small is solid, clear, and strong.

    "nothing much happened" is not the same as "we don't know what happened". Evidence that the magnitude of the effect is close to zero is not the same as close to zero evidence about the magnitude of the effect.

    I confessed that I have in the past tried to get the null on my side. I did not mean to confess taht I was doing so in this case.

    1. There, do you approve of the update? :-)

  7. Anonymous10:41 PM

    Bottom line: We are not creating enough new jobs fast enough to return to full employment in a reasonable period. This exacerbates the GDP gap, the revenue gap and risks creating a permanent unemployable underclass who lack job experience and therefore lack job skills.

    Whether or not monetary policy is effective, it is not effective enough and needs help, a lot of help, from fiscal policy to address unemployment more directly.

    -jonny bakho

  8. So, we are really just arguing which null hypothesis is least likely to be damaging, rather than which alternative hypothesis is most likely to be right.

  9. It's exactly like debate over the ARRA (Obama's stimulus), TARP, Federal Reserve bailouts, etc. Conservatives argue that there's no proof that they worked. Liberals argue that they mitigated the downturn and helped the economy recover.

    My view as a non-economist is that if the economy improves in 2013H2 and 2014, then it will be partly due to Bernanke. My prediction is that if does improve, then the QE skeptics will argue that the economy improved on it's own without the help of the Fed. Conservative will argue that the economy improved despite Obama's policies. If the economy doesn't finally improve then yes the QEs were a dud.

    "Policies Are Working

    Today, we are seeing evidence that our accommodative policies are working. Financial markets recognize that low returns on the safest fixed-income assets are an economic reality that will continue for a long time. Not being satisfied with these low returns, investors are repositioning into other asset classes that will more directly bolster spending. The stock market has rebounded significantly and is back near its pre-recessionary level. In an encouraging development, commercial and industrial lending by banks is picking up while their credit terms, which had been quite restrictive during the recession, are easing. Indeed, from a variety of sources, markets are increasingly channeling money to promising projects that had previously lacked funding.

    Accommodative monetary policies also are supporting some more-familiar cyclical recovery mechanisms that lead to a virtuous cycle of growth in economic activity. Although bottlenecks certainly remain for many households (particularly those whose home mortgages are under water or those that have imperfect credit scores), low mortgage rates have facilitated a pickup in refinancing and a slow but steady rise in new residential construction. Some further increase in demand will spill over from the improvement in housing markets, coming in the form of higher demand for construction materials and appliances. Furthermore, for years now, consumers and businesses have made do with old household durable goods and capital equipment, and there likely is much pent-up demand for replacing such items with new ones. Low interest rates are helping to support some pickup in this demand to replace old items; for example, I’ve heard from several carmakers how low-interest financing has contributed to an improvement in auto sales."

    Evans has three more paragraphs on how Fed policy is helping.

  10. I don't think Macro Data is the problem. Ideology is the problem. Just because people lie, we don't say words are at fault. Words can be used to deceive, or illuminate, data is no different. Words can fail us even when we are trying to illuminate, and indeed we don't yet know everything in order to be able to fully illuminate it in any case, words are not the problem.

    Economics has many distracting conversations going on in the language of choice among Economists, math. With care and time it becomes clear who is seeking to deceive (or more generously persuade) and who is seeking to illuminate. Clever constructs aimed at persuasion are a distraction.

    Math is a few words short in its capacity to articulate what is really going on in the economy as well. But in the case of looking at monetary policy there are two problems, the first has been pounded home by Bernanke, Fed policy can not overcome current bad fiscal policy on it's own, the second is a corollary, few things are true absolutely, most things are true conditionally, monetary policy effectiveness depends. It depends on what else is going on; are nominal rates high or low, where are real rates, what is going on with investment, exports, debt levels, fiscal policy. No man is an island, and neither is an economic truth, nor is it iid, or stationary, etc.

  11. Rather than a dichotomous decision to reject H0/H1, why not conducting the LRT (assuming MLR?) to calculate which side has more favor?

  12. Pedro2:11 PM

    Very interesting way of summing up the discussion, and one that I would agree with, for the most part: Macro data (and analysis) is problematic at best and if you can state that you have the null, so to speak, then you certainly have the advantage.

    But I'm very curious, Noah, about one thing: are you a fan of Takeshi Kaga by any chance?

  13. This comment has been removed by the author.

  14. I'm reading Robert's arguments differently.

    I think the main thrust of his argument is contained in a sentence by him lifted from the comments above, "I don't really think that macro evidence is so feeble. There is more of a problem that people wish their opinion to be treated as a null and think that means a failure of the data to reject it amounts to something, but also refuse to define their opinion precisely enough for it to be testable."

    That is exactly how I understand it. The point is not that macro data sucks. I really don't see why you think it sucks. Because it's low frequency? Then you should have low-frequency null hypotheses to go along with the type of data available, no?

    As Robert says, there's plenty of macro data available to debunk silly theories. The problem is when people come up with theories (and especially when they keep "fixing" their theories as evidence accumulates) that aren't designed to be tested: the "falsifiability" Popper type argument.

    Just as Freudian and Marxian theories can't be falsified, some economic "theories" are designed to body-swerve falsifiable nulls: if you think you have a point, then frame it in a way that if A happens your null holds and if non-A happens your null is refuted.

    Or in other words, take your theory to the test instead of blathering.