Friday, November 6, 2009

My Reply to Krugman

Paul Krugman has chimed in on my figure showing the collapse in nominal spending. He, however, is less enthusiastic about its implications:
[The figure is] certainly suggestive. But I disagree with the interpretation that this shows that the current slump is mainly about insufficiently expansionary monetary policy...


Focusing on nominal spending makes you think that low nominal spending is the problem, a problem with a monetary solution; but actually it’s the symptom, and monetary policy doesn’t matter (unless it can affect expected future inflation, but that’s another story).
Note that the part about changing expectations of future inflation in parentheses is actually an admission that monetary policy can do something about the current slump. Krugman, however, does not explore this point anywhere else in his post. That is unfortunate because it is the very argument advocates like Scott Sumner have been making in their case that monetary policy could have done more to prevent the nominal spending crash of 2008-2009. One way to think about this is to imagine what would have happened had the Fed set an explicit inflation target of say 3% in mid-2008 and promised it would do whatever was needed to keep actual inflation there. If such a policy had been adopted it is unlikely inflation expectations would have collapsed liked they did in late 2008, early 2009 as seen in the figure below (click on figure to enlarge):

And if inflationary expectations had not collapse then current nominal spending would have been far more stable. This is because if folks think that inflation will be permanently higher going forward they are more likely to spend their money today. That is, money demand will fall and velocity will pick up.

This is a point I empirically tested in a recent post. There I took the monthly expected inflation series implied by the difference between the nominal 10-year Treasury yield and the 10-year TIPs yield and put it in a vector autoregression (VAR) along with the monthly GDP series from macroeconomic advisers. Nominal GDP was turned into an annualized monthly growth rate and the data used runs from 1999:1 through 2007:9. More data would have been helpful, but TIPs only start in the late 1990s.* The two figures below show what the typical responses of expected inflation and nominal GDP to the typical sudden change or shock to expected inflation over the sample. The solid line shows the point estimate while the dashed lines show two standard deviations around the point estimate. Upon impact, the shock causes expected inflation to jump 16 basis points and occurs as the level of nominal GDP increases by 1.16 percent. In other words, a sudden positive change in expected inflation is associated with an increase in current nominal spending. Both effects persist but eventually become insignificant about 14-15 months later. (Click on figures to enlarge.)

So contrary to Krugman's claim, there is reason to believe the Fed could have prevented the great nominal spending crash of 2008-2009. The real question for me is why did the Fed allow inflationary expectations to fall so dramatically in late 2008, early 2009. My guess is they simply dropped the ball or there was too much pressure from inflationary hawks.

Update I: The Economist blog Free Exchange responds to Krugman's post by arguing that monetary policy is not out of gas.

Update II: Scott Sumner also shoots down Krugman's nominal nonsense.

*(Technical note: both series were in rates so no unit root problems, 13 lags were used to eliminate serial correlation, and corporate bond spreads were included as a control variable for the financial crisis).


  1. All the benefits you promise from targeting inflation sounds exactly like all the benefits that were promised by those who advocated targeting money supply growth in the 1970s.

    I fail to understand why inflation targeting would work any better than money growth targeting.

    I'm old enough to remember the old rule of thumb that any economic variable the monetary authorities elect to target will immediately quit working.

  2. David,

    It would be useful if you could give us some sense of what the Fed needed to do in 2008 to manage expectations.

    Take a 3% inflation target. How much QE would have been necessary in the Fall of 2008 to achieve that target? Is the reaction of expectations to QE dollar amount linear or non-linear? Specifically, can we have a situation where a moderate amount of QE produces no change in velocity, but a massive amount of QE produces a massive change in velocity/expectations? Why do you think the relationship between expectations and QE/reserve growth is either knowable, linear or constant?

    Further, market actors derive an expected value from a non-normal inflation probability distribution. If the deflation tail is quite "fat", what makes you think that a 3% inflation target (or even expectations target) would be enough to derive a negative expected value to a cash hoarding strategy? And if it wasn't sufficient, wouldn't the Fed's credibility be damaged, such that it would have to adopt an even higher target to push the expected value down?

    A post on the above would be helpful.

  3. Spencer:

    I am actually for a rule that would target nominal spending such as a nominal income targeting rule. Still, an inflation target would have done the trick since it would have change expectations of future inflation(and more folks are familiar with it). Money supply targeting, on the hand, is plagued by changing money demand and thus makes managing expectations more challenging.

  4. David Pearson:

    You ask some tough questions. I think your point is that in theory my point makes sense but how is it implemented in practice. I suspect that in times like these there are a lot non-linearities at work in the relationship between monetary policy and inflation expectations. Still, if there were a real time forecast of future inflation it seems the Fed could engage in a trial and error QE until it hits its target.

  5. David:

    What I find puzzling about this debates is the notion that zero nominal bound is binding because some interest rates are near zero. There are many interest rates that are no where near zero. I think it is very likely that it is necessary to purchase, or be seen as willing to purchase, securities with interest rates with interest rates greater than zero.

    Is the problem that too many economists minds are stuck on the one interest rate simplifying assumption? In our models "the" mominal interest rate is zero. So we are left with manipulating inflation expectations to impact "the" real interest rate.

    Or is there an assumption that open market operations must use T-bills, and that the interest rates on all of them are low, and so low that they could easily be driven to zero by a more aggressive policy? I am willing to grant that additional open market operations in zero yield T-bills will have little effect, though they won't do any harm. And when they are all purchased, other assets must be increased if the quantity of base money needs a futher increase.

    Sumner has convinced me that a committment to do this will raise all interest rates, including those on short and safe assets. But if the Fed must actually purchase long term bonds and risky bonds.. so be it.

  6. Pearson:

    Your argument isn't that quantitative easing can't raise inflation expectations (or nominal expenditure, which is the point really.) It is instead that errors are more likely. Overshooting or undershooting are more likely with unsettled conditions.

    I think the "let's not try because we might overshoot" is a mistake. I wonder if it is the policy of targeting inflation expecations, and some fear that inflation expectations will get out of control, that leads to this error.

  7. David:

    Congradulations on the original post and the attention received.

    I still think total final sales is the best figure.

    When you aggregate, then total final sales should still work, but you could also use total purchases.

  8. Bill:

    I definitely was not expecting the nominal spending figure to attract so much attention. In fact, I had posted it some time before and was just using it to motivate the second figure of the OECD nominal spending. Yes, I would now go with final sales of domestic product but the figure was originally done with final sales to domestic purchasers and for convenience that is what I went with in the post.

  9. David,

    I think the strategy vs. tactics construction glosses over too much.

    The question is, "can the Fed successfully manage inflation expectations in a debt crisis." It is not, "can QE raise inflation expectations."

    I am not sure if you believe the answer to the first question is "yes". I agree with your observation: it comes down to whether "trial and error" is a feasible strategy (not tactic) for expectations management. And that, of course, has to do with the character of the "overshooting" risk.

    I would say if overshooting lends itself to corrective iterations, the risk is small. In other words, if its like sailing a boat to a fixed point, then any deviation from that point can be corrected by a slight move of the tiller.

    If, on the other hand, overshooting creates self-feeding feedback loops, then "course corrections" are either not effective or self-defeating. Here's a concrete example:

    Let's say the Fed starts out with a 3% target and increases reserves by a $1tr, and that short term inflation expectations do not budge. The Fed injects another $500b. Longer term (10yr TIPS) inflation expectations climb to 3%, but near-term expectations again remain near zero. The Fed injects another $1tr as "overkill" to ensure it has an impact on expectations. Expectations, however, jump to 5% inflation. The Fed must now decide whether to remove reserves or raise its inflation target. The market knows the Fed will not remove reserves as long as unemployment is 10%, so velocity increases as actors hedge against a loss of purchasing power, and inflation expectations rise further.

    When you say, "the Fed can engage in real time trial and error", what you are really saying is, "the market will be okay with the Fed tacking (strongly) in both directions -- contraction and stimulus." The reality is, I can construct many scenarios where the Fed can only tack in the direction of stimulus -- it will always be seen by market actors as an asymmetric set of choices as long as unemployment is high. This is exactly what is happening know with popular market perceptions of the "carry trade" as "guaranteed as long as unemployment stays over 8%".

  10. One thing that seems missing from a lot of the discussions on inflationary expectations is the fact that the Fed started paying interest on reserves last October, with the explicit reason being to keep inflationary pressures down (at least from public statements they've made). When they started, this interest rate was 0.75% when the Fed Funds rate target was already 0-0.25%. They cut this rate down to 0.25% earlier this year but even that is at the upper end of the target range. Since then, other public statements have made it clear that they intend to use this as a tool to keep inflation low in the future.

    Not being a macro person I am not familiar with any existing literature on the effectiveness of interest on reserves as a policy tool. All I can think of is that if I'm a bank & I can get a risk-free 0.25% by leaving money in reserves then an effective Fed Funds rate lower than that makes the alternative awfully unattractive. Notice the sudden crash in the M1 multiplier as soon as the policy came into effect.

    Of course I may be missing something glaringly obvious, which you may need to explain to this micro hack. But otherwise this is one more piece of evidence supporting you & Sumner -- that monetary policy is still too tight.

    How all this plays into Roubini's very scary assertions on the carry trade, I do not know...

  11. Wafa:

    I agree that the Fed's payment on interest on excess reserves is problematic. At best it does nothing to spur nominal spending and at worst it is contractionary. To some extent banks probably would be sitting on higher levels of excess reserves anyway given the frail economy. The question is how much more did the Fed cause excess reserves holdings to increase because of this policy. It will be intersting to see how history will be written on this one.

  12. The liquidity freeze in markets during the crisis in late 2008 distorted the yields of nominal and real rates... and TIPS are very iliquid, more than T-Bills or T-Notes. So this was a technical subject, not an economical. We should not try to read more than that, inflation expectations did not collapse, i think is better to see the inflation expectations from the University of Michigan.

  13. Yes, TIPs has liquidity premium which distorts the interpretations but even a look at the survery of forcasters 1 year inflation forecast as seen here shows a sudden drop in inflation expectations.