What should be on the agenda in Chicago?

The Fed plans to hold a conference in Chicago during June, with the goal of developing improved tools, targets and strategies. Obviously the major issue will be how to deal with the zero bound, which is expected to reoccur during the next recession. In my view the second goal should be to develop procedures that avoid the mistakes made during 2008 (before the zero bound was hit), which were acknowledged in Bernanke’s memoir.

One key lesson from the Great Recession is that when rates are zero it’s very difficult to be too expansionary. Almost everywhere in the world, at all times in history, central banks at the zero bound adopt policies that in retrospect look too contractionary. Contemporaneous fears of inflation prove groundless. So that’s one important lesson.

Next recession, the Fed needs to immediately stop paying IOR and be far more aggressive with QE than last time. We know that inflation isn’t the real problem at the zero bound. A recent Yahoo article suggests that bond yield pegging is another option being considered:

Fed officials would also reassess how well their policy toolkit worked in combating the deep recession that followed the financial crisis of 2008-09, and consider what additional tools might be added to prepare for the next downturn. He mentioned a crisis-time policy implemented by the Bank of Japan, which would seek to establish a temporary ceiling for Treasury debt yields at longer maturities, as a tool that might be considered.

That’s a reasonable option, but it’s not enough by itself. Indeed, doing more concrete steps at ultra-low interest rates is not enough, as ultra-low interest rates represent a sort of prediction of policy failure, a prediction that NGDP will grow too slowly to achieve the Fed’s dual mandate. The goal should be to prevent the zero bound from occurring in the first place, not just to deal with it appropriately. Once you are there, you have already (accidentally) adopted an inadequate policy.

One option for avoiding the zero bound it to raise the inflation target, but the Fed has ruled that out:

In reviewing its fundamental strategies, Clarida made clear the Fed wouldn’t change its target for inflation from the current 2 percent level. Policy makers would consider, however, whether the central bank should introduce a strategy that seeks to make up for periods of below-target inflation with periods of above-target price rises.

Level targeting is the sort of regime that makes it less likely you’ll hit the zero bound, but the devil is in the details. A promise to do whatever it takes to get back to the price level trend line in 10 years is far less effective than a promise to get back there in 4 years. Partly because the quicker rebound is more expansionary, and partly because it’s more credible that the Fed chair would still be around in 4 years.

Although I’d prefer 4% NGDPLT with no 2% inflation target, it is possible to combine NGDPLT with a flexible 2% inflation target, which puts weight on both inflation and employment. You simply set the NGDP target equal to 2% plus the Fed’s estimate of trend RGDP growth, and adjust the growth estimate periodically to reflect new estimates of trend RGDP growth. That’s not as good as a simple NGDP level targeting (inflation actually doesn’t matter), but it still gets you 95% of the benefits of NGDPLT and it’s also consistent with the Fed’s interpretation of its inflation mandate. Inflation will average 2% in the long run.

Ironically, the regime I just proposed would probably get you closer to a 2% long run trend rate of inflation than our current inflation targeting regime, which has repeatedly missed on the low side and then let “bygones be bygones”.

As noted, they also need to learn from their mistakes of 2008, and for that I recommend more reliance on market forecasts. I certainly don’t expect them to talk about NGDP futures markets, but recommending that the Treasury issue bonds indexed to NGDP growth might be a modest first step. Perhaps the NGDP bond payoffs could be based on the third GDP announcement, which occurs roughly 3 months after the quarter ends. That data is fairly complete, although of course there are occasionally later revisions as well.

The BLS might be instructed to keep two GDP series when there is a major definitional change (such as adding software to investment), to allow an internally consistent series for bond indexing. That’s a bit messy, but these major definitional changes don’t occur very often. Don’t let the perfect be the enemy of the good.



21 Responses to “What should be on the agenda in Chicago?”

  1. Gravatar of John Hall John Hall
    22. February 2019 at 12:43

    You wrote “Although I’d prefer 4% NGDPLT with no 2% inflation target, it is possible to combine NGDPLT with a flexible 2% inflation target, which puts weight on both inflation and employment. You simply set the NGDP target equal to 2% plus the Fed’s estimate of trend RGDP growth, and adjust the growth estimate periodically to reflect new estimates of trend RGDP growth.”

    You’ve discussed this before and I was actually thinking about it recently.

    I think it makes more sense to compile an NGDP level target with a price level target but with the NGDP level target putting more weight on more recent observations than the price level target. The simplest way to think of it is that the central bank could create NGDP and PGDP (or PCE price index) futures contracts. The NGDP level target might only consider the past three years and the next three years, while the price level target would consider the past ten years and the next ten years. Then, similar to your version, the NGDP target is the main thing and the trend is adjusted periodically in recognition of the price level target. So for instance, if I have a 5% NGDP target and this 20-year period (10 years before, 10 years after) has 2.5% average yearly inflation, then I could reduce the NGDP target in 25bps intervals until it is a 2% average inflation pace.

    So it is a very similar idea. The only difference is switching the emphasis from RGDP to PGDP and using futures contracts instead of Fed internal forecasts. My preference for emphasizing prices is that we want stable long-run inflation expectations. It’s at least measurable. Unlike potential output.

  2. Gravatar of Benjamin Cole Benjamin Cole
    22. February 2019 at 17:20

    Unless the Fed can be convinced to ease up on its hard-line 2% inflation target, the central blank has placed itself into a terrible quandary.

    The IMF has posited that large capital inflows, born of large and chronic current-account trade deficits, results in bloated and thus fragile asset values in the US.

    The Fed has already indicated it is uncomfortable with high asset values, but it is most uncomfortable with any inflation rate that creeps towards 2%.

    So eventually the inflation rate creeps towards 2%, or wage rates begin to show real growth.

    The Fed then tightens the monetary noose, which results in declining asset and especially property values, the most credit-dependent asset.

    Of course once property values begin to decline, banks begin to extend less credit to property. You get an avalanche in property values.

    See 2008.

    All the conditions are ripe for a replay of the Great Recession of 2008, and I suspect we were headed in that direction when finally the Fed came to its senses and stopped raising rates, and is now publicly pondering a halt to its bond sales.

    But the problem remains, and if inflation should rise above 2% the Fed may feel compelled to act and thus will bring about a replay of the 2008 situation.

    I recommend Fed adopt an inflation band target, perhaps of 2% to 3% or even a little higher.

    The best solution would be a national policy of money-financed tax cuts into the teeth of a recession, but that option appears to be off the table.

    Boy, what a world, when your best options are not even discussed .

  3. Gravatar of Dtoh Dtoh
    23. February 2019 at 00:30

    Glad to see you’ve finally accepted my argument favoring an NGDP linked note over an NGDP futures market.

  4. Gravatar of ssumner ssumner
    23. February 2019 at 07:47

    Dtoh, No such luck. I’ve acknowledged that it’s the sort of plan that appeals to people at the Fed, who you have such a high opinion of. 🙂

  5. Gravatar of Michael Sandifer Michael Sandifer
    23. February 2019 at 11:14

    I wish they would discuss setting up real time payments and daily sum of transactions reporting to the government.

  6. Gravatar of Michael Sandifer Michael Sandifer
    23. February 2019 at 11:19


    I’m curious, hypothetically, if NGDP-linked Treasury notes were issued, should the nominal interest rates on such notes be equal to the nominal rates on traditional Treasury notes of the same duration?

  7. Gravatar of Michael Sandifer Michael Sandifer
    23. February 2019 at 11:22

    Perhaps you know what I’m getting at in the question above. In a healthy economy, shouldn’t the risk-adjusted return be the same, and shouldn’t that even be reflected nominally?

  8. Gravatar of dtoh dtoh
    23. February 2019 at 14:38


    “I’ve acknowledged that it’s the sort of plan that appeals to people at the Fed”

    Don’t you think appeal to the people (i.e. drunken sailors running the Fed) who decide whether or not to implement something is an important factor in determining its relevant merit. Or is this purely an academic blog that doesn’t seek to influence actual policy.

  9. Gravatar of ssumner ssumner
    23. February 2019 at 15:27

    Michael, I’m not sure I follow. The return on NGDP bonds is some fixed rate plus actual growth in NGDP. Thus they might pay negative 1%/year, plus the actual growth in NGDP over 5 years, if it’s a five year bond. The market would determine the rate before indexing.

    dtoh, I think we are talking past each other. Of course it matters what Fed officials think if we are interested in getting something done. I’ve never denied that. I thought you were claiming I now think NGDP indexed bonds are better than NGDP futures contracts. I don’t, I just think they are easier to get important people to accept. That’s why I mentioned them here.

  10. Gravatar of ssumner ssumner
    23. February 2019 at 15:27

    dtoh, Yes, it’s mostly an academic blog—I don’t expect to influence monetary policy.

  11. Gravatar of Michael Sandifer Michael Sandifer
    23. February 2019 at 21:57


    The point is, in a healthy economy, i.e., one which is at capacity with no sovereign or foreign-denominated debt problem, shouldn’t the interest rates on government bonds equal NGDP growth expectations?

    If NGDP represents the average rate of return one can get in an economy, then wouldn’t there be an opportunity cost to holding government bonds at a lower interest rate in a healthy economy? Wouldn’t the risks of holding a government bonds and a security that pays a dividend proportional to NGDP have exactly inverse risks, assuming nominal interest rates were always allowed to reflect the natural rate? Hence, when nominal rates differ, it can be due to the differential expected rate of return on money.

    That would imply that rates on Treasuries are below NGDP expectations, because monetary policy is still too tight.

  12. Gravatar of Michael Sandifer Michael Sandifer
    23. February 2019 at 22:03

    And to expand further on the current US situation, the tight money would mean lower investment, meaning lower productivity. It also helps reduce the population growth rate.

    Then, if you buy that the unemployment rate is a function of the difference between NGDP growth and the growth of labor compensation, then there are multiple equilibrium low unemployment rates. There are those at various rates of below capacity productivity growth and that at capacity.

  13. Gravatar of Michael Sandifer Michael Sandifer
    24. February 2019 at 04:58

    Never mind the stuff on comparing Treasuries to NGDP securities. I shouldn’t post comments after drinking. Obviously, if nominal rates could fall in line with the natural rate, i.e. enough money supply expansion with sufficient credibility to offset an initial fall in NGDP, then there isn’t a problem anyway.

    And I’m not sure if the proposed NGDP-linked bonds would pay a flexible or fixed nominal rate.

    So, the question is, shouldn’t the nominal rates on Treasuries be at least as high as NGDP growth expectations, in a healthy US economy?

  14. Gravatar of dtoh dtoh
    24. February 2019 at 08:08

    “I don’t expect to influence monetary policy.”

    Haha! False modesty doesn’t suit you.

  15. Gravatar of ssumner ssumner
    24. February 2019 at 09:25

    Michael, Interest rates were zero in 2010-15 and NGDP growth expectations were about 4%. So no, they are not equal. Not even close.

    Are they positively correlated? Absolutely.

    dtoh, Neither does delusion.

  16. Gravatar of Michael Sandifer Michael Sandifer
    24. February 2019 at 09:56


    Yes, my question concerns whether the rates on Treasuries should be roughly equal to NGDP growth expectations in a healthy economy. Obviously, rates have empirically been lower than NGDP during a period of tight money. This is due to the expected rate of return on dollars versus output and investment spurring higher demand for money. Hence, the natural interest rate fell proportionally(below zero), while nominal rates were stuck just above zero.

  17. Gravatar of ssumner ssumner
    25. February 2019 at 11:05

    Michael, I don’t think tight money can explain the current low rates, but I agree that tight money explains the low rates during the Great Recession and slow recovery. The recovery is over.

  18. Gravatar of Michael Sandifer Michael Sandifer
    25. February 2019 at 13:28


    Yes, one would think sticky wages would have adjusted by now. However, money could still be tight due to the Fed’s expected reaction function, right?

    My question, poorly asked, was whether there’s an a priori reason to think that the interest rate on government bonds should equal NGDP in a healthy economy. That is, when money isn’t too loose or tight and when there isn’t a sovereign debt crisis, for example.

  19. Gravatar of Michael Sandifer Michael Sandifer
    25. February 2019 at 19:52

    In other words, in monetary equilibrium, and with no government debt crisis, shouldn’t nominal rates on risk-free debt equal NGDP growth expectations? Otherwise, isn’t there an opportunity cost?

  20. Gravatar of ssumner ssumner
    26. February 2019 at 14:24

    Michael, I don’t see why. For example, NGDP growth rates vary a lot between US states, but risk free interest rates in Ohio and Colorado are identical.

  21. Gravatar of Michael Sandifer Michael Sandifer
    26. February 2019 at 17:24


    Yes, good point, but how much should that really matter? That’s not a rhetorical question. I’m really wondering.

    The currency’s the same and it’s relatively easy to arbitrage across state lines. Plus, Treasury yields are below the NGDP growth rates of every state.

    What should we expect to see if I’m right, and the principle I suggest does apply to states, to a lesser degree? Somewhat larger holdings of Treasuries in slow growth states? I can’t find data on Treasury holdings by state.

Leave a Reply