Tyler Cowen’s curious curiosity

Tyler Cowen recently had the following to say about QE:

I’m unhappy with claims that “we’re not doing enough” and that therefore this is no test of the idea of monetary stimulus.  This is what QEII looks like, filtered through the American system of political checks and balances.  And if it looks small, compared to the size of our problems, well, monetary policy almost always looks small compared to its potential effects.  I’m willing to consider this a dispositive test and I am very curious to see the results.

Of course I am one of those claiming that we aren’t doing enough, but I’d like to focus on Tyler’s curiosity about the results.  I think there are more layers to this question that most people assume—many more.

Let’s start with the distinction between real and nominal GDP.  What would count as success?  Should we focus on real GDP, nominal GDP, or both?  That depends.  I think most economists would say that the whole point is to raise RGDP, and that a rise in NGDP that was not accompanied by an increase in RGDP would constitute failure.  But what happens if both RGDP and NGDP rise at rather modest rates, (which is quite likely.)  In that case, would monetary stimulus have been shown not to work?  Tyler Cowen has argued that much of the recession is real, and he has also expressed skepticism about “liquidity trap” models that say monetary policymakers cannot create inflation (or higher NGDP) at the zero rate bound.  So perhaps Tyler is interested in whether a given increase in NGDP translates into higher RGDP.

For instance, let’s assume NGDP grows at 3%, RGDP at 2% and prices at 1%.  I’d say monetary stimulus hasn’t really been tried.  On the other hand if we had 6% NGDP growth accompanied by only 2% RGDP growth I’d say monetary stimulus had been tried, and failed to produce the predicted growth in RGDP.  Does that mean it never should have been tried?  Not necessarily, it depends on the Fed’s policy goals.  I favor 5% NGDP targeting, level targeting, regardless of what is happening on the supply side of the economy.

On the other hand, as I read Tyler’s comment above, he’s setting the bar where 3% NGDP growth would constitute failure.  Even if monetary stimulus, pursued a outrance, would certainly boost inflation and NGDP, he suggests that the recent QE announcement is about all that is politically feasible in the US.  This is an argument that Krugman has made as well, and at some level I think he’s right.  But that sort of failure would not change my message, because whereas Tyler views QE2 as a partially endogenous response from real world monetary policymakers, I consider TheMoneyIllusion.com to be a sort of exogenous shock, aimed at convincing policymakers that monetary stimulus that seems highly aggressive is actually quite modest.  I’m trying to make more expansionary policies become more politically acceptable.  Yes, that’s a rather grandiose objective, and I don’t realistically think I can have more than a tiny impact on the zeitgeist, but (as I argued earlier) even a tiny impact is important when the stakes are high.

We’ve considered the distinction between real and nominal GDP, and the distinction between what’s politically feasible and what is not.  And yet we’ve haven’t really addressed the most important implications of Tyler’s curiosity.  For instance, what would be the policy implications of failure?  If NGDP rises fast, but unemployment stays high, then maybe we shouldn’t be trying to boost AD at all.  If NGDP doesn’t rise, then maybe we should have tried fiscal policy.  Or maybe not.  It’s clear to me (from market reactions) that QE2 has already raised the expected rate of NGDP growth.  I also think QE1 slightly boosted AD in 2009.  Those who claim that fiscal stimulus should have been $1.3 trillion would have to show that the Fed would not have simply offset the effect of more fiscal stimulus by doing less monetary stimulus.  Perhaps QE1 and QE2 never would have happened.  After all, the Fed’s two forays into QE both seemed motivated by a macroeconomy that was under-performing in the months immediately preceding the policy initiatives.

Most importantly, I would argue that Tyler Cowen has no reason to be curious.  We already know whether QE2 will work, we know the only effects that matter—the impact on NGDP growth expectations.  OK, that’s slightly overstating things; we know it boosted 5 year inflation expectations by about 0.5%.  Again, this shows the urgent need for a NGDP futures market.  For the moment, let’s assume an NGDP futures market did exist–my hunch is that Tyler Cowen might still have been “curious” to see the effect of QE.  Obviously I think that would be a mistake, and the only thing we will learn from observing the macroeconomy over the next 5 years is that part of NGDP growth that was not caused by QE.

Think of the TIPS market response as not just the optimal forecast of the effect of QE, but also the optimal estimate of the effect of QE.  And not just the optimal ex ante estimate, but the optimal ex post estimate.  How can that be?  Surely we will eventually know much more about how much inflation was created than our current rather crude forecasts?  Actually no.  We still don’t know how much good Obama’s $800 billion fiscal stimulus did, because we don’t know the relevant counterfactual growth path.

Economics is all about the effect of X on Y, ceteris paribus.  The TIPS market gives us the optimal forecast of inflation, ceteris paribus.  The actual inflation rate that we observe over the next 5 years will differ from that forecast, but we have no way of knowing whether it differs because “other things weren’t equal” or because our original forecast was flawed.  Without such knowledge, we have no reason to revise our estimate of how much QE2 raised 5 year inflation expectations.  There’s no need to be curious Tyler; we know how much good QE2 did, about 0.5%/year more inflation over 5 years.  Yes, we don’t know how much that will raise RGDP growth, but only because of the bizarre anomaly that the US government has never bothered to create an NGDP future market.  Now are you guys seeing why I think this market is so important?  If we had one, the market response (TIPS vs. NGDP futures) over the past 2 months would have basically settled the dispute between Tyler and myself about whether the recession is mostly nominal or mostly real.

I seem to recall Robin Hanson complaining that we won’t spend the money needed to redo the Rand health insurance experiment—the only one that really solves the identification problem.  This is even worse, we are making enormous macroeconomic policy errors because we won’t create and subsidize trading in a simple NGDP futures market.

Update:  Here is the Hanson post.  Second update:  Oops, TGGP pointed out it should be this link:

Let’s go one year forward and consider how my reputation will fare under different outcomes:

1.  NGDP grows at only about 3-4%, and RGDP grows by only about 2%.  I think QE2 will be seen as not working, and my reputation will suffer.

2.  NGDP grows about 5%, and RGDP grows a bit over 3%.  I think I’ll do alright.  I said it was better than nothing and would slightly boost growth.

3.  NGDP grows 7% and RGDP grows at 4-5%.  I think people will see the result as strongly vindicating my policy proposals.

4.  NGDP grows 7% and RGDP grows only 3%.  I think people will think I was right about the potency of monetary policy, but wrong in assuming the recession was an AD problem.  Tyler Cowen wins twice. Krugman loses twice.  Kling wins on structural problems, but loses on the potency of QE.

But that’s not how I see things.  I hope outcome #3 occurs, but I also think the credit that I predict I’d receive would be undeserved.  I’m a “target the forecast” guy.  The markets are saying QE2 helped, but nothing too dramatic.  Outcome #3 would be more than I think the markets are forecasting, and hence would not be evidence in support of forecast targeting.  My sense is that other people don’t particularly focus on the forecast targeting part of my message, however, and that they’d see it as vindicating my constant arguments for monetary stimulus.  What do you think?

I bet you never realized there was so much complexity embedded in Tyler Cowen’s innocuous sounding comment “I am very curious to see the results.”  I’m very curious to see how Tyler interprets the results that actually occur, but I have no curiosity at all about the nominal effects of QE2.  The markets have already answered that question to my satisfaction, ceteris paribus.  But I am curious to see the NGDP/RGDP split, if NGDP rises dramatically.



18 Responses to “Tyler Cowen’s curious curiosity”

  1. Gravatar of Mike Sandifer Mike Sandifer
    11. November 2010 at 16:13

    These are the posts that make this blog so much better than any other I’ve read, and I’ve read plenty. If only I’d learned this much in my intro to macro class.

    It’s nice, even with my rather extreme ignorance on the subject, that I think I can interpret things more clearly than many media talking heads at least, thanks to you.

    Even I’m surprised that so many economists seemed to have failed to really think things like this through, especially given the importance. I’m typically a pessimist when it comes to human cognitive capabilities vis-a-vis abstract concepts.

  2. Gravatar of Bonnie Bonnie
    11. November 2010 at 17:37

    Suppose the Fed were to announce, instead of QEIII (when the time comes), a wind down of the IOR program? Might that be a more politically palatable expansion of monetary policy? It side-steps the perception of just “printing money” while perhaps accomplishing the same goal. Or can it not do that while the funds rate is .025?

  3. Gravatar of scott sumner scott sumner
    11. November 2010 at 18:19

    Thanks Mike.

    Bonnie, Yes, I think that’s exactly right. The next step should be different. Sarah Palin wouldn’t know what to make of a cut in IOR from 0.25% to 0.15%. Would she defend subsidies to big bankers?

  4. Gravatar of Mark A. Sadowski Mark A. Sadowski
    11. November 2010 at 19:00

    IOER makes it less likely that QE 2 will have much if any effect. And that means that the effect of this modest QE on NGDP will be in turn *very* modest.

    But the split? I predict it will still be statistically significant in favor of RGDP and the magnitude will be large in proportion. I’ll be astonished if it is otherwise.

    Maybe with enough of these Dr. Frankenstein type experiments with the national economy (while tens of millions remain unemployed and underemployed) the truth will sink in. But on the other hand, thanks to the VSP, the truth may never sink in.

  5. Gravatar of Philo Philo
    11. November 2010 at 19:28

    Let me second Mike Sandifer’s comment: “These are the posts that make this blog so much better than any other I’ve read . . .”; better, that is, for economic policy analysis. (Marginal Revolution is a must read for other reasons.)

  6. Gravatar of Doc Merlin Doc Merlin
    11. November 2010 at 21:56

    “Suppose the Fed were to announce, instead of QEIII (when the time comes), a wind down of the IOR program?”

    The IOR program is a lifeline to the large banks, I don’t see them ending that program.

  7. Gravatar of Contemplationist Contemplationist
    11. November 2010 at 22:11

    Perhaps the IOR vindicates the Ron Paul-Austrian view that the Fed has always been owned by the banking interests.

  8. Gravatar of TGGP TGGP
    11. November 2010 at 22:28

    I think you have the wrong Hanson post. You linked to the municipalize one before, but it doesn’t discuss RAND. Hanson argued with Austin Frakt over redoing the experiment here:
    He has a petition for redoing the experiment here:

  9. Gravatar of Liberal Roman Liberal Roman
    11. November 2010 at 22:55

    I second Mike Sandifer’s comment up top.

    Also, it is amazing as the world wonders whether QEII will have the desired effect, they are missing the point that it already has had its effect. The market is now moving on to other things, such as will Bernanke have enough guts and commitment to follow through with more measures if inflation still doesn’t pick up? And will Europe explode creating another shock and sending us all to the Dark Ages?:)

  10. Gravatar of scott sumner scott sumner
    12. November 2010 at 05:47

    Mark, I agree that it would be mostly NGDP.

    Thanks Philo.

    Doc Merlin and Contemplationist, I hope that’s not why they’re doing the program.

    Liberal Roman, That’s exactly how how look at things. But I’m going up against the entire profession. Everyone takes a “let’s wait and see how it works” point of view, I’m just about the only one who says changes on market expectations are the best test of a policy.

  11. Gravatar of ssumner ssumner
    12. November 2010 at 06:06

    Thanks TGGP, I added a link

  12. Gravatar of Luis Enrique Luis Enrique
    12. November 2010 at 07:20

    I’d like to see a lot more of this “how will my reputation fare” stuff from prominent economists, or rather “under what circumstances will I have to admit what I thought I knew about economics was wrong”.

  13. Gravatar of Bill Woolsey Bill Woolsey
    12. November 2010 at 08:32

    The important test is money expenditure (NGDP) targeting.

    There is no target for the growth path of NGDP.

    There is a promise to incease base money by 600 billion over the next year. There is a continued promise to keep the federal funds rate very low for an extended period of time. There is a stated hope that long term interest rates will fall.

    There is a call for inflation to return to 2%.

    But no target for a growth path of money expenditures.

  14. Gravatar of Tom Dougherty Tom Dougherty
    12. November 2010 at 12:32


    I am wondering why you are curious about the NGDP/RGDP split. Aren’t TIPS markets telling us that the NGDP/RGDP split due to QE2 to be 0.5% per year higher? So, hypothetically, next year if NGDP is 3.75% and RGDP is 2.0%, shouldn’t QE2 be responsible for boosting NGDP from 3.25% to 3.75% with QE2? I thought your point is that the portion of NGDP growth due to QE2 is already known to the extent that we will ever know. Is this right, wrong, completely and hopelessly wrong?

  15. Gravatar of Doc Merlin Doc Merlin
    12. November 2010 at 12:59

    ‘ I thought your point is that the portion of NGDP growth due to QE2 is already known to the extent that we will ever know. Is this right, wrong, completely and hopelessly wrong?’

    No, rational expectations doesn’t say people can’t be wrong, just that free markets aren’t systematically wrong. Scott tends to lend more credit to the TIPS spread than I do, however, as I think that fed intervention keeps it from being a free market.

  16. Gravatar of Tom Dougherty Tom Dougherty
    12. November 2010 at 18:09


    I didn’t mean to ask if people can’t be wrong. I meant am I wrong. My question is regarding how much NGDP rises and RGDP rises and the difference between the two. What difference does it make what the difference between NGDP and RGDP is, if you assume that whatever the actual difference might actually be, QE2 is only responsible for 0.5% of the difference. So what if NGDP grows by 5% and RGDP grows by 3.25%, once the assumption is made that QE2 is responsible for increasing NGDP by 0.5%? What difference does it make how much NGDP grows by if QE2 is responsible for 0.5% or the increase?

    Maybe I am looking at this completely wrong? One could say that if NGDP grows by 7% next year, then QE2 did a great job. But how do we know it wouldn’t have grown very fast, say 6.5% without QE2? Or if NGDP grows by 3.0% then, well, QE2 wasn’t so great. Yeah, but without QE2 it would have grown 2.5%? So, in both cases QE2 contributed the same amount 0.5%, but when it NGDP grew 7% it was successful, however, when NGDP only grew 3% it was a failure.

    So, isn’t Scott Sumner saying, we already know how successful it is by the TIPS market, all that is left is how it will be perceived latter on? This seems to be a sort of shocking statement to know how successful the QE2 right away by looking at the TIPS market before it has had its full effect. But I like his point about because so many other variable will be changing over time there will be no way to extract what part of the increase in NGDP is due to QE2 and how much is due to everything else that has changed.

    But I guess there are other things that can be learned by looking at RGDP and how it reacts to QE2, but how to tell how much of RGDP is reacting to QE2 and how much to everything else?

  17. Gravatar of scott sumner scott sumner
    13. November 2010 at 09:53

    Luis, I agree, as long as we aren’t asked to forecast market outcomes.

    Bill, I agree that the lack of an NGDP target (and futures market) is a big problem. We have to guess expected NGDP growth by looking at other markets, which isn’t easy.

    Tom, I didn’t state my point clearly. I meant that the portion of expected price level growth due to QE2 is already known, and the portion of NGDP growth due to QE2 would be known if we had a NGDP futures market–as we should, but do not.

    Because we don’t know how much QE2 is expected to boost NGDP, we also don’t know how much it is expected to boost RGDP.

  18. Gravatar of A Question for the Money Printers A Question for the Money Printers
    15. November 2010 at 07:37

    […] help me out. We’ve got lots and lots of economists (like Paul Krugman but also Scott Sumner and even Tyler Cowen) who think QE2 might help a little bit, but that it’s not enough to get […]

Leave a Reply