Beautiful models and inconvenient facts

Macroeconomists sure like models.  When they see a new and interesting stylized fact, they instantly want to create a model to explain it.  Indeed they like their models so much that they sometimes forget to carefully check whether the stylized fact is, well, a fact.  Consider the following from a new paper by James Bullard, President of the St. Louis Fed.

They suggested that the combination of an active Taylor-type rule and a zero bound on nominal interest rates necessarily creates a new long-run outcome for the economy. This new long-run outcome can involve deflation and a very low level of nominal interest rates. Worse, there is presently an important economy that appears to be stuck in exactly this situation: Japan.

That’s a pretty bold assertion.  Is Japan really stuck in a deflationary trap?  I read on, looking for the evidence.  Unfortunately, I couldn’t find any.  There was a graph showing that Japan had experienced some very mild inflation, but I was under the impression that the Bank of Japan was an ultra-conservative bank, and liked mild deflation.  Indeed I thought that was pretty widely understood.  I guess not.

Is there any way to tell if I’m right?  Fortunately, the paper provides the answer a few pages later:

But for the nominal interest rate, most of the Japanese observations are clustered between 0 and 50 basis points. The policy rate cannot be lowered below zero, and there is no reason to increase the policy rate since–well, inflation is already “too low.””This logic seems to have kept Japan locked into the low nominal interest rate steady state. Benhabib, et al., sometimes call this the “unintended” steady state.

Hmmm.  Inflation is “too low.”  Japan is in an “unintended steady state.”  And how would we know?  The answer is simple, and is provided in the quotation.  The Bank of Japan would never raise interest rates during a period when inflation is “too low,” that would make no sense.  I agree.  The problem is that the BOJ did raise interest rates during the 2000s, indeed more than once.   So although Western economists consider Japanese inflation to be “too low,” it is quite apparent that the BOJ feels differently.

Of course there are many other reasons why we know that Japan is not stuck in any sort of deflationary gap.  They let the yen appreciate strongly during the midst of the great deflationary crisis of 2008-09.  They dramatically reduced the monetary base in 2006 to prevent inflation.  Indeed almost every time the Japanese inflation rate approaches zero, from below, the BOJ seems to do something contractionary.

But why bother with facts when we have these beautiful models?  Macroeconomists must have created a 100 models to explain how and why Japan became trapped in deflation.  It’s a pity that it never happened, as they are quite clever models.

I shouldn’t have been so sarcastic, as the Bullard paper is no worse than any other in this respect.  And in other respects it is far better than most.  Let me finish up on a positive note, by quoting Bullard’s concluding paragraphs:

When the European sovereign debt crisis rattled global financial markets during the spring of 2010, it was a negative shock to the global economy, and the private sector perception was certainly that this would delay the date of U.S. policy rate normalization. One might think that is a more inflationary policy, but TIPS-based measures of inflation expectations over five and ten years fell about 50 basis points.

Promising to remain at zero for a long time is a double-edged sword. The policy is consistent with the idea that inflation and inflation expectations should rise in response to the promise, and that this will eventually lead the economy back toward the targeted equilibrium of Figure 1. But the policy is also consistent with the idea that inflation and inflation expectations will instead fall, and that the economy will settle in the neighborhood of the unintended steady state, as Japan has in recent years.

To avoid this outcome for the U.S., policymakers can react differently to negative shocks going forward. Under current policy in the U.S., the reaction to a negative shock is perceived to be a promise to stay low for longer, which may be counterproductive because it may encourage a permanent, low nominal interest rate outcome. A better policy response to a negative shock is to expand the quantitative easing program through the purchase of
Treasury securities.

Exactly!  Promising zero rates as far as the eye can see is like promising failure for as far as the eye can see, because zero rates occur in depressed economies.   QE is much better.  And even better yet, supplement it with a price level or NGDP target.  (Good to know that monetarism is not completely dead at the St. Louis Fed.)

PS.  You might think I was quibbling about some rather small increases in nominal interest rates in Japan.  How do I explain the fact that Japan has experienced mild deflation “despite” low interest rates?  Easy, you’d expect mild deflation to cause very low interest rates, a point Milton Friedman made many years ago.  Interest rates are behaving exactly as you’d expect if they were targeting mild deflation (low, but occasional nudges upward to prevent outbreaks of inflation) and not at all as you’d expect if they were stuck in a deflationary trap (interest rates stuck at zero, and never raised.)

HT:  Benjamin Cole



29 Responses to “Beautiful models and inconvenient facts”

  1. Gravatar of dlr dlr
    29. July 2010 at 17:02

    Okay, so Japan’s deflation or near deflation is a choice and not a trap. How important a role do you think this choice has played in Japan’s two decade real malaise? Does Japan — not including the most recent recession — have an important AD problem or does it have mostly real (demographic and institutional) problems and a central bank that prefers a lower Selgin-esque inflation target? And if it does have an AD problem, is because of (very) stick wages, or has the BOJ simply never created enough money to consistently lift the demand for money out of disequilibrium during the last twenty years?

  2. Gravatar of JimP JimP
    29. July 2010 at 17:21

    Scott –

    His model may not be all that good, but he does understand the role of Fed communication and expectations. As described in the WSJ:

    begin quote
    In his paper, Bullard said that any policy regime shift intended to avoid a deflation trap needs to be “sharp and credible-policy makers have to commit to the new policy and the private sector has to believe the policy maker.”
    end quote

    Note the words “commit” and “sharp and credible” and “has to believe”.

    He is to be interviewed tomorrow at 7am on CNBC. I am curious as to what he will say.

  3. Gravatar of Lorenzo from Oz Lorenzo from Oz
    29. July 2010 at 17:37

    Scott, your comments here are somewhat reminiscent of a 1991 Daniel Goldman essay criticising macroeconomic models as not grasping the role of innovation. Indeed, your points about expectations generally seem somewhat congruent with his critique.

  4. Gravatar of Ted Ted
    29. July 2010 at 19:01

    I think you are confusing the models themselves and the interpretation of models. Modeling can be an excellent tool, but they are stylized and simplified and so it requires a bit of interpretation when looking at the real world. To me, the situation of Japan and our current situation makes perfect sense in almost any macro model with rational expectations and nominal rigidities. I think it’s easy to tell a story about Japan using even the standard New Keynesian models that seem to be favored by our Federal Reserve.

    The problem isn’t the models. It’s the interpretation of the model to the real-world many economists have trouble with. I think it’s easy to explain most of what’s gone on in the standard macro framework. I think you are confusing models versus model interpretation.

    Also, I think it’s a problem to cite Federal Reserve people since their judgment is clouded. Nobody likes admitting mistakes, so they are going to try to rationalize it away – probably unintentionally.

    I do want your opinion on something I’ve been struggling with concerning Japan. In the long-run, monetary policy is suppose to be neutral – so why would contractionary monetary policy effect Japan for so long? Was it simply because the BOJ kept inducing shocks where the bank would start being contractionary throughout the period just as recovery was on the verge? Also, why do you believe total factory productivity fell so much during the ‘Lost Decade’? Such a large reduction in TFP would certainty effect Japan’s growth – do you think this relates to monetary policy? I was toying with an idea where expectations of deflationary policy deter R&D spending and technology adoption, which reduces productivity, but I’m not sure if that thinking is correct.

  5. Gravatar of Benjamin Cole Benjamin Cole
    29. July 2010 at 19:41

    A hat tip! I am red-faced at the recognition!
    I will easily allow that the most technical aspects of Scott Sumner’s arguments and insights into the Bullard piece are over my head.
    But I contend Sumner should be happy (within this context, but hopefully in every other context as well).
    The world is talking about QE. Can Scott Sumner take credit–or was it my earnest missives to the Fed? (ha-ha). Let’s say Sumner pushed the dial a bit.
    I also offer a ray of hope. As I have posted earlier, I find most Americans do not read newspapers, could not pick Bernanke out of a police line-up, and probably confuse the Federal Reserve with Ft. Knox.

    If we go aggressive on QE, and the economy grows a bit, and small businesses and real estate types can borrow, then rates will start to edge up. Banks will sense some room to charge customers some vigorish. Most businesses know what they have to pay for credit and how their sales are doing–they do not sit around debating the finer points of monetary policy. People will become optimistic when they start to make sales, or get jobs.

    I agree, as much as I can understand it, that the Fed should talk up the economy, and the potential for higher rates.

    Re models: A few moons back, when I was in college with my quill pen and gas-lamp, I took a few statistics classes. It turns there is a sub-discipline within statistics that can tell you have sensitive your models are to changes in data–how “rugged” the models are. If the reported data changes a little bit, what happens to the model?
    The results were eye-opening–the more complicated a model, the less rugged it was. We even used fortran cards in those days, which we thought was pretty sexy.

    All this goes to say I would not place too much faith in models. Believe you me, Wall Street had plenty of models to show that pool of residential mortgages, and the safer tranches, were triple AAA. That raises another point–models sometimes make assumptions, and you know what happens when you assume.

  6. Gravatar of Benjamin Cole Benjamin Cole
    29. July 2010 at 19:51

    And I still have two nagging questions:

    1. A global capital glut–does that not depress interest rates anyway?

    2. If the monetary base is so important, and there is roughly $500 billion of US currency floating around the globe, what happens if some of that currency comes back home? You mean to say our monetary policy is, to some extent, in the hands of drug-lords?

  7. Gravatar of Mike Sandifer Mike Sandifer
    30. July 2010 at 04:37


    I have no idea what your first two points mean. Japan’s CPI, excluding food and energy, has often fallen over the last ten years, and continues at much lower levels than in 2000.

    The story seems to be largely the same for wages. Both real and nominal wages were negative for most of the last ten years.

    How is this not a deflationary trap, unless you merely mean there’s no “trap”, because the Bank of Japan is on their desired target? And what difference does it make?

  8. Gravatar of scott sumner scott sumner
    30. July 2010 at 04:50

    dlr, The answer is clearly both. The main problem with the low inflation is that it makes monetary policy very clumsy. The BOJ does not like to use unconventional approaches, and hence is very slow to react to recessions when rates are zero. When NGDP falls (as in 2008-09), the BOJ is very slow to react.

    Another problem is downward rigidity in nominal wages.

    But I also think structural problem are the biggest factor in Japan’s slow growth.

    JimP, Yes, there were some very good things in the paper, I probably focused too much on one point. But it bothers me that all these models are being motivated by something that never happened.

    Lorenzo, I agree with his criticisms of mainstream macro.

    Ted, You said;

    “The problem isn’t the models. It’s the interpretation of the model to the real-world many economists have trouble with. I think it’s easy to explain most of what’s gone on in the standard macro framework. I think you are confusing models versus model interpretation.”

    I disagree. Macro models are ad hoc. They are built to explain certain stylized facts. The AS/AD model never would have been built if not for deflation depressions. It is a logical monstrosity, but it somehow seems to work. I’m fine with that kind of model. I am criticizing macro models developed solely to explain certain stylized facts, when it turns out that those facts are not, in fact, facts.

    See my discussion of Japan above. The Bank of Japan has acted aggressively to keep NGDP constant in the long run, which means mild deflation. But the slow RGDP growth is mostly due to structural factors.

    Benjamin, Thanks. I agree that we shouldn’t put too much faith in models.


    1. Yes, but I prefer the term “savings glut” It is possible that high savings rates in Asia have tended to depress world real interest rates. I doubt it was the only factor around 2002-06, but it might have been a factor.

    2. I planned a post on that question, and now I’ve forgotten whether I did it or not. I will check, if not I’ll do a post. The short answer is that the Fed would usually offset that change

  9. Gravatar of scott sumner scott sumner
    30. July 2010 at 04:53

    Mike, It makes a huge difference because the paper’s motivation is not to explain mild deflation (which is easy to explain), but rather why Japan is trapped, which is hard to explain. So we get all these convoluted models explaining why central banks have such a hard time escaping deflation, when they don’t have a hard time at all. They just don’t want to.

  10. Gravatar of Mike Sandifer Mike Sandifer
    30. July 2010 at 05:01

    Yes, I was about to amend my question when I realized there was a theoretical point here, apart from policy. I am too tired today to read these posts properly, so my apology for your wasted time.

  11. Gravatar of Morgan Warstler Morgan Warstler
    30. July 2010 at 05:26

    “Exactly! Promising zero rates as far as the eye can see is like promising failure for as far as the eye can see, because zero rates occur in depressed economies. QE is much better.”

    Er, no. Raising rates and losing $4T in home value (not home equity mind you) is far preferable.

  12. Gravatar of Mike Sandifer Mike Sandifer
    30. July 2010 at 08:30


    A question occurs to me after reading Krugman’s take on this post.

    Other than the impossibility of conducting true experiments and the often erroneous uses and interpretations of statistics within macroeconomics, it occurs to me that people are looking at too many variables and situations sans the context of entire relevant economies, which ultimately should include that of the world. I suppose I see all investment options in the world, or more broadly, all econommic choices as being the weighted average of expected returns. In other words, I see one massive global portfolio, certainly with some frictions.

    There is perhaps a much larger problem with the macro perspectives I regularly read however. They apparently aren’t taking the nonlinear nature of economic interactions into account. To me, it’s not at all clear that the presumed relationships between variables should hold up as often as adherents think. I suspect there is a wild card in all linear models.

    Scott, you once compared problems with macro predictions to those in weather forecasting. Of course, one of the seminal papers on nonlinear dynamics is Deterministic Nonperiodic Flow, by Lorenz, who was a theoretical weather researcher:

    Then, there are other complexity theories that may come to bear, such as network theories.

    Do you consider these concerns to be very valid?

    I admit that the nonlinear factors in macro don’t always necessarily have to be significant, but perhaps they are sometimes overlooked when important.

  13. Gravatar of Mike Sandifer Mike Sandifer
    30. July 2010 at 08:43

    I should expand a but on my original point above, which is that economists are selective about the data they choose and the interpretations there of, even against their best intentions. I see few comprehensive approaches offered, at least in the less formal accounts that I read.

  14. Gravatar of Six of one… – Economics – Six of one... - Economics -
    30. July 2010 at 09:24

    […] SUMNER has been arguing that Japan hasn't been in a deflationary trap for twn years, from which the Bank of Japan is […]

  15. Gravatar of Andre Mittag Andre Mittag
    30. July 2010 at 09:30

    Your talks smack of the BOJ being rational actors with complete control of the situation. Yes, BOJ economists are very likely elite economists (that may be out of their depth) but you could say the same of the US Fed. Last time I checked not all Fed representatives voted in lock step, Thomas Hoenig has been voting to increase interest rates, contrary to the others on the Fed Board. It is possible that BOJ is dominated by Hoenig type economists, or very likely, BOJ economists are not fully able to comprehend the difficulty of their situation.

  16. Gravatar of Luis H Arroyo Luis H Arroyo
    30. July 2010 at 10:27

    I supose you know it, but, in any case…

  17. Gravatar of Benjamin Cole Benjamin Cole
    30. July 2010 at 11:07

    A query: Are the internal politics and decorum of the current Fed understood? Can Bullard put out a piece like he did without tacit Bernanke approval? Do Sumner, or any posters know? Is Bullard the Fed speaking, even through a hat, or is Bullard wandering off the reservation?

  18. Gravatar of Benjamin Cole Benjamin Cole
    30. July 2010 at 11:27

    Paul Krugman today goes on the offensive against Scott Sumner. Hey, it’s nice to be noticed….

  19. Gravatar of Bonnie Bonnie
    30. July 2010 at 12:49 has a story about the Fed intending to ease at its Aug 10 meeting. You might want to have a look. 😀

  20. Gravatar of JL JL
    30. July 2010 at 13:45

    Congrats with the Krugman reply. I’m anxiously awaiting your reply, and hoping Krugman will also respond to your main arguments wrt NGDP, IOR, etc.

    Anyway, it seems both of you are 100% in agreement.
    The BOJ is just inconsistent: you’re just looking at their actual policy, while Krugman is looking at their stated policy.

    And as Krugman admits, the BOJ could do more to increase inflation. So the fact that they didn’t is proof that their actual policy is to have deflation.
    Nobody is preventing them from implementing inflationary policies. Krugman himself, in the ’90s, outlined how Japan could get inflation if they wanted to.
    The fact that they have done nothing to get inflation, proves that their policy is deflationary, in effect, if not in intent.

    Q for Krugman: Does the Fed want high or low unemployment in the US?
    Because all of their actions indicate that they want higher unemployment.

    And does Zimbabwe want high or low inflation? I assure you, Mugabe would love his paper to hold its value. But his actual policy is to decrease its value.
    So it is not unreasonable to say that Mugabe has a highly inflationary policy.
    Just as Japan, and now the US, has a deflationary policy.

  21. Gravatar of 美しいモデルと不都合な事実 by Scott Sumner – 道草 美しいモデルと不都合な事実 by Scott Sumner – 道草
    30. July 2010 at 15:22

    […] Illustion “Beautiful Models and Inconvenient Models” by Scott Sumner on July 29th […]

  22. Gravatar of JeffreyY JeffreyY
    30. July 2010 at 17:37

    Shorter Krugman: “The BOJ doesn’t want deflation, they’re just incompetent.”

  23. Gravatar of Luis H Arroyo Luis H Arroyo
    31. July 2010 at 00:28

    Reading the Krugmang´s comment on Scott, I don´t see where is the great diference that he imagines.
    But, I would ask Krugman why BOJ has the chance to be much more agresive against deflation, But not so the FED. In Japan, yes, ok to monetary policy; in US, only fiscal policy would be effective. When the CBO is proyecting Debt/GDP ratios above to 100%, and there is a real risk of deflation, I think it is much better an “inflationary” policy.
    What’s more, I would say that the expected debt has much to do with the $ trillion hoarded by Banks & Corporate. A fiscal consolidation announced by the government would be a definite aid to erase incertitude.

  24. Gravatar of scott sumner scott sumner
    31. July 2010 at 06:06

    Mike, No need to apologize.

    Morgan. I disagree.

    Mike, Complexity may be important in some areas of economics, but I think less than many people assume. I don’t see it as being in factor here. BTW, I responded to Krugman.

    Mike#3, It is impossible to be comprehensive, you always have to focus on what you think is important.

    Andre, I agree. But what are the policy implications of your view. Is it that there is nothing they can do? Or that they should act differently? I’d say the latter.

    Luis, Yes, I saw that.

    Benjamin, That’s a good question.

    Bonnie, Unfortunately, I will be on vacation, and won’t be able to comment. I’ll try to leave an anticipatory comment before I leave.

    JL, Jeffrey and Luis, Check out my response. Ryan Avent of The Economist says Krugman argument actually supports my point. I take that as a win for me.

  25. Gravatar of Morgan Warstler Morgan Warstler
    1. August 2010 at 10:37

    Scott, yes but I wish you could tell me #3:

    1. I know you want to socialize the losses (target NGDP).
    2. I even know and agree that no action will be the long way home.
    3. What I don’t know is why you think that intentional liquidation of the insolvent banks, and mark to market in the home market is bad.

  26. Gravatar of scott sumner scott sumner
    2. August 2010 at 05:03

    Morgan, I don’t object to liquidating banks. Laws should be enforced if they are bankrupt

  27. Gravatar of If It’s Good for the Goose… «  Modeled Behavior If It’s Good for the Goose… «  Modeled Behavior
    2. August 2010 at 13:46

    […] ~ August 2nd, 2010 in Economics | by Niklas Blanchard Earlier last week, Paul Krugman took Scott Sumner to task for the assertion that the Bank of Japan didn’t actually want inflation, and instead, […]

  28. Gravatar of Contemplationist Contemplationist
    11. July 2011 at 10:27

    As far as I can tell the fault in the US is solely the Fed’s and especially Bernanke’s. Long before Sarah Palin heard the name “Mises,” Bernanke was pimping for fiscal stimulus in Congress and the bank bailouts, instead of doing his job targeting inflation (isn’t that what the current regime is?) or failing that, price level or NGDP. Now that right-wingers have caught the Austrian flu, he’s under the scope, which would have never been the case had he done his job

  29. Gravatar of ssumner ssumner
    12. July 2011 at 18:28

    Contemplationist. That’s right, but it’s not just Bernanke, the entire Fed was clueless in late 2008.

Leave a Reply