Another Goldman Sachs report on NGDP targeting

Goldman Sachs always seems to have something interesting to say about monetary policy.  The following is proprietary information, so I can only quote portions:

Q: There has been increased talk recently that the Federal Reserve may soon adopt an explicit inflation target. Is this likely?

A: It’s certainly very possible. Chairman Bernanke is a long-standing advocate of inflation targeting. It was discussed at the September meeting, and Bernanke has mentioned it in recent speeches and congressional testimony. Moreover, it appears that Fed officials have the legal authority to adopt an explicit inflation target without congressional sign-off. If the FOMC decided to adopt an inflation target, it would undoubtedly be a “flexible” one””that is, it would explicitly allow for significant short-term deviations from the target for the sake of output stabilization.

Q: Would such an explicit inflation target be a good idea?

A: It probably wouldn’t make much difference compared with the current framework for monetary policy. Already, Fed officials talk about the “mandate-consistent” inflation rate of 2% or a little less. Moving from this to a flexible inflation target would only be a small step.

That said, we believe that such a step would be somewhat counterproductive at the margin under current circumstances.

That’s exactly my view.  Do something effective, or do nothing.  Don’t do things that make monetary policy look ineffective.

Q: So what should Fed officials do instead?

A: One way to address the asymmetry without resorting to a hard employment or output target””with the problems that this could entail””would be to target the level of nominal GDP extrapolated from the pre-crisis trend. This would effectively increase the weight of the employment part of the dual mandate while allowing Fed officials to hedge their bets with respect to uncertainty about the natural rate of unemployment or the level of potential output. It is a way for them to focus on a nominal variable over which they have a substantial amount of control, while leaving the split of nominal GDP between real output and inflation up to the supply side of the economy.

.   .   .

Third, Fed officials could move more incrementally, adding NGDP to a longer list of intermediate targets that might already include asset prices or bank lending conditions. Chairman Bernanke indicated some sympathy for this option in response to a question at his most recent post-FOMC press conference.

Did I miss something at the press conference?

Q: Is it likely that the FOMC will soon adopt an NGDP target?

A: We certainly do not expect a full-blown NGDP target along the lines of Exhibit 1 anytime soon. It would be quite a radical move for the Federal Reserve, an institution that typically moves in a deliberate manner.

However, we see a somewhat bigger chance of a gradual move toward an NGDP target that is “watered down” with some of the modifications described above. First, if there is one central bank in the world that could conceivably go down this route, it is probably the Federal Reserve because of its dual mandate. Most other central banks operate under mandates that focus mainly or exclusively on the delivery of low and stable inflation.

Second, the “optics” of an NGDP target are much better than those of other “unconventional unconventional” monetary policy steps such as a higher inflation target or a price level target. It is deeply counterintuitive to most people that higher inflation per se could help the economy grow faster, because they think of higher inflation as a cut in real income. This is not really correct because a higher inflation target would imply a higher target for wage as well as price inflation, i.e. it wouldn’t have direct implications for real incomes. But it is nevertheless much easier to convince people that a higher target for overall income and spending in the economy might have an expansionary impact. This means that if the Fed decided that a substantial amount of added stimulus was needed””more substantial than what’s available via “conventional unconventional” means such as further QE””an NGDP target would be a natural option.   (Emphasis added.)

Where’d they get that clever idea?  If only I could patent ideas, and charge Goldman Sachs a commission for using them.  We award patents to companies for stupid things like click to buy.  (BTW, does the word “patent” derive from “patently obvious?”)  And then we don’t allow patents for ideas that might lead to lots of jobs.  How many jobs?

However, in our model presented a month ago, which is based on empirical estimates of the linkage between spending and expectations, these expectational channels are quite powerful. Exhibit 2 shows that the combination of an NGDP target and renewed QE might lower the unemployment rate by nearly 2 percentage points by the end of 2014, compared with the “baseline” scenario, in which the Fed just follows our estimated Taylor rule. Exhibit 3 shows that the policy might boost inflation by ½-1 percentage point, although this is relative to a “baseline” prediction of inflation well below the Fed’s implicit target. Analysis of the regime shifts by the Fed and the Swedish Riksbank in the 1930s also suggests that such shifts can be powerful even if the instrument set is limited.

So we reward inventors of “click to buy,” but not persuasive arguments for policies that might create 3 million jobs at very little cost of extra inflation.  Oh well, you got to like any investment bank that takes the time to study Sweden’s 1931 experiment in level targeting of prices.

Third, an NGDP target enjoys growing support from economists on both sides of the political aisle. Several prominent economists who have recently advocated NGDP targeting, including Paul Krugman, Christina Romer, and Bradford DeLong, lean toward the Democratic side. However, many of the long-standing “market monetarist” supporters of NGDP targeting such as Scott Sumner and David Beckworth identify themselves as political conservatives. Beckworth recently wrote an article advocating NGDP targeting with political journalist Ramesh Ponnuru in the conservative National Review. Gregory Mankiw, an adviser to Republican presidential candidate Mitt Romney, has in the past also published research favorable to NGDP targeting, although he has to our knowledge not weighed in on the current debate.

That was included because I like seeing my name in print.

Weakness in actual NGDP would increase the probability that the FOMC might go down this path. Although the recent data on US economic activity have looked a bit better recently, we do expect NGDP growth to slow from the 5% pace of the third quarter to only about 2½%-3% in the first half of 2012. This is both because we see real GDP growth slowing due to greater spillovers from the European crisis and tighter US fiscal policy, and because we expect a meaningful slowdown in inflation.

Q: So what is your actual forecast for Fed policy?

A: We expect another quantitative easing program to be announced sometime in the first half of 2012. The timing, size, and asset mix will depend on how the economy performs, but it seems likely that agency MBS would be included in any new program.

We should also see the FOMC take further steps to clarify its policy framework. An explicit inflation target is possible, and a move to start publishing information about the path for the federal funds rate expected by different FOMC members is quite likely. We believe that an NGDP target could usefully complement these steps, although we recognize that it would be a large shift in the framework that is unlikely to happen overnight.

Jan Hatzius

Too bad “The Goldman Sachs” already has such a talented monetary analyst—I’d love to go work for them.  If Hatzius retires, I hope they keep me in mind.  :)



33 Responses to “Another Goldman Sachs report on NGDP targeting”

  1. Gravatar of John John
    11. November 2011 at 22:16

    I can’t stand Hatzuis’s economic analysis. He loves to talk about output gaps which are such a meaningless term. It is absolutely impossible to know what any economy’s potential output is and even if you could, that doesn’t predict inflation the way he thinks it does. Also, just because real GDP increased by X during the middle of a credit bubble doesn’t mean the economy should or could continue at that same growth rate. Typical mainstream Keynesian nonsense. Scott just likes him because he flatters his ego.

  2. Gravatar of John John
    11. November 2011 at 22:22


    What do you think about the idea of an output gap or excess capacity? I think, a lot of excess capacity was malinvested capital and that in many cases it’s best just to close down a certain factory or let a tractor rot because it has no marginal productivity. With other stuff, the price of the capital goods just has to decline. It is impossible to know how to employ all an economy’s available resources efficiently without a market context. A lot of the excess supply is completely useless.

  3. Gravatar of Dustin Dustin
    11. November 2011 at 22:54

    That was included because I like seeing my name in print.

    Yeah, I think Bob Murphy is right. Your popularity has gone to your head and Murphy is going to eat your lunch when you guys have your big debate.

  4. Gravatar of dtoh dtoh
    12. November 2011 at 02:18

    I don’t think you’d like it at Goldman. The focus there is on predicting what the Fed will do….not on hectoring them to do it. It wouldn’t be very much fun.

  5. Gravatar of marcus nunes marcus nunes
    12. November 2011 at 03:27

    Since GS referred to “Market Monetarists” the name has been officially recognized.

  6. Gravatar of Bill Woolsey Bill Woolsey
    12. November 2011 at 03:36


    Whether output gaps can be measured isn’t the same thing as whether or not they exist.

    Market monetarists generally would think that changes in the nominal expenditure lead to both changes in inflation and output gaps in the short run, but only changes inflation in the long run. Reversing the change in nominal expenditure, if sufficiently prompt, can reverse both an output gap and the immediate change in inflation, making further changes in inflation unnecesary.

    In fact, the the article quoted, Hatzius points out that nominal GDP targeting avoids the need to extimate the output gap. Or more exactly, that was the implication of his point that this would depend on supply side factors.

  7. Gravatar of Paul Andrews Paul Andrews
    12. November 2011 at 04:19

    Just a cover for printing. I wonder how much of Goldman’s junk assets the Fed would need to buy to goose NGDP back to the pre-crisis trend? And all it takes is to find some economists who can baffle the great unwashed (and perhaps themselves), then increase their social standing so much that they can never recant.

  8. Gravatar of c8to c8to
    12. November 2011 at 04:54

    also the wikipedia debate ended in a win for the market monetarist page…we don’t need the scare quotes anymore either…

    the bank of england is already targetting nominal GDP – openly secretly

  9. Gravatar of Morgan Warstler Morgan Warstler
    12. November 2011 at 05:18

    Paul Andrews is exactly right. More importantly, that is the argument the Tea Party will use to keep it from happening.

    Just pay off the Tea Party. Make it a policy all about them.

  10. Gravatar of Tim Lee Tim Lee
    12. November 2011 at 06:48

    That was probably a rhetorical question, but I’ll answer it anyway: “patent” derives from “letters patent”—public letters from the monarch announcing the granting of a monopoly—as opposed to “letters close,” which were private letters from the sovereign.

  11. Gravatar of Becky Hargrove Becky Hargrove
    12. November 2011 at 07:11

    What dtoh said is right you’d be climbing the walls in no time. There is a reason why college professors are among the most fulfilling professions statistically, few other areas allow as much freedom of the mind. I assume the “wish” to patent your ideas was in jest, already we are struggling for technological growth as ideas there are routinely captured, speared and buried. Even so, ideas for use in the knowledge realm rely on infinity growth even more than the gadget idea. Most of us who wonder if the passing years will ever show us a “marketing plan” are happy sharing ideas for free as long as we can survive until there is a way to truly focus those ideas in the marketplace.

  12. Gravatar of John John
    12. November 2011 at 08:08


    Guys like Hatzuis and Zandi make careers out of “measuring” output gaps and “multipliers” of fiscal and private sector spending on given projects. People take these things seriously and those guys make lots of money for doing very unscientific work that is no better than an educated guess anyway.

    In the case of an output gap, if a mine opened because the price of gold spiked, then had to close and lay off its workers because the price came back down, would that be an output gap? If so, should the government print and spend until that mine becomes profitable again and can hire back the workers? Some of the more naive output gap theorists seem to think so.

  13. Gravatar of B B
    12. November 2011 at 09:25

    Scott wrote:
    “Too bad “The Goldman Sachs” already has such a talented monetary analyst””I’d love to go work for them.”

    Jan Hatzius wrote:
    “However, in our model presented a month ago, which is based on empirical estimates of the linkage between spending and expectations,…”
    “…compared with the “baseline” scenario, in which the Fed just follows our estimated Taylor rule.”

    You’re shooting yourself in the foot by not having a model. Hatzius has two. You’ve been talking about this for years. I’m shocked you haven’t teamed up with a young talented macroeconomist that knows dynamic programming. You’ve mentioned before it’s ripe for the picking. Why not have your name attached to it? Shopping a working paper would go a long way towards a job interview at Goldman Sachs.

  14. Gravatar of dtoh dtoh
    12. November 2011 at 10:00

    Beg to differ with you but the really good economists at the banks have (or at least had in my day) predicative accuracy that was a lot better than an educated guess and this made a lot of money for the firms and their clients. I don’t know much about Hatzius, but Goldman had some really smart guys like Gavyn Davies, David Morrison and Fischer Black.

  15. Gravatar of Benjamin Cole Benjamin Cole
    12. November 2011 at 10:30

    If Goldman Sachs were clever, they would hire Scott Sumner right now. The PR halo alone around Sumner, and the related use of the GS name in that context, would pay off on Day One in spades for GS.

    Imagine Scott Sumner backed up by the institutional resources of GS, and their PR machine. That would pay off in spades for the nation.

    Ben Bernanke, you are hiding in house built of straw.

  16. Gravatar of TravisA TravisA
    12. November 2011 at 10:44

    @B: I don’t think a model matters that much. The GS model was a bit of a joke. It basically assumed its conclusion, which is what most models do. Anyways, I think that you overestimate the power of macroeconomic mathematical models in persuading people. Friedman didn’t really have a model, just economic logic.

    For selfish reasons, I hope Scott doesn’t get a GS job. The best blog on the Internet would disappear. What I’d like is for Scott to make a GS income from writing this blog :-)

  17. Gravatar of marcus nunes marcus nunes
    12. November 2011 at 11:08

    I think you are quite right on “models”. To me, “model” is something we carry in our minds about how (some parts) of the world works. And then you “test” them against data. I´ve just read Doug Irwin´s paper on Cassel and the Great Depression. He applied his “mental view” of the world and got it so right (before, during and after). Something he would never achieve by writing down a “microfounded model”.

  18. Gravatar of ssumner ssumner
    12. November 2011 at 11:18

    John, Output gaps are caused by sticky wages. The problem can be reduced by monetary stimulus (more NGDP.) I agree that he puts too much weight on output gaps, but still think his analysis is excellent.

    Dustin, We’ll see.

    dtoh, Maybe, but in this report Hatzius is hectoring them on what to do.

    Marcus, Yes, and Krugman did too.

    Paul, You asked:

    “I wonder how much of Goldman’s junk assets the Fed would need to buy to goose NGDP back to the pre-crisis trend?”


    C8to, That’s good.

    Morgan, The Tea Party doesn’t even have a unified view.

    Tim, Thanks for that info. It was meant to be a sort of sarcastic joke.

    Becky, Yes, I was joking about the idea patenting, as you noticed. And you are right that I probably wouldn’t be happy at GS. But I’d retire in 3 years, and that would make me very happy.

    B, I’m not really a fan of models. I believe in targeting the forecast to control NGDP (thus no model needed). And as far as the question of which nominal aggregate produces the best P and Y breakdown, we aren’t even close to having enough information about the economy’s structure to model that. For instance, most think inflation is a problem. I think the welfare costs believed to be associated with inflation are actually caused by NGDP growth (levels and volatility.)

    Thanks Ben.

    TravisA. You said;

    “What I’d like is for Scott to make a GS income from writing this blog”

    Sounds good—where do I apply?

  19. Gravatar of ssumner ssumner
    12. November 2011 at 11:19

    Marcus, Good example with Cassel.

  20. Gravatar of Paul Andrews Paul Andrews
    12. November 2011 at 16:23

    We all have models – mental models. These are inferior to formalised written models, which in turn are vastly inferior to reality.

    While no formal model comes close to reproducing reality, a formal model does have important benefits:

    a) It forces one to make his or her assumptions self-consistent.

    b) It lays out one’s assumptions clearly for all to see.

    c) It lays out one’s logic clearly for all to see.

    Through a) one can see for oneself any flaws that may not be obvious in casual thought.

    Through b) and c) others can decide whether they agree with you by first seeing whether they agree that the assumptions are reasonable, and then by following the logic of the model.

    Scott have you examined the models that Goldman have used? Do you agree with the assumptions and the logic in those models? If not, do you think it is reasonable to use their conclusions to justify your own?

    Even if you believe in “targeting the forecast to control NGDP”, you must have a mental model that leads you to believe it. In order to make a robust case you need to convert your mental model to a formal model.

    Of course, you don’t need to make a robust case to succeed politically (take Keynes for example). A vaguely plausible and politically expedient case will do fine (this seems to be the route you are taking). However if you want to be reasonably sure to have a positive effect on economies rather than a drastically negative effect, that’s a different matter.

  21. Gravatar of JimP JimP
    12. November 2011 at 18:51

    From the charmers at ZH – the complete goldman note.

  22. Gravatar of Shane Shane
    12. November 2011 at 21:37

    If the Fed is “small c” conservative, then why don’t they adopt a negative Fed Funds rate? Before we ask why there’s no NGDP target, shouldn’t we try to figure out a good reason why the Fed Funds rate isn’t -5%? Negative IOER of 6% could make that a reality. The Fed deciding to “penalize the banks” should be even easier politically than the Fed deciding to “raise incomes,” and certainly easier than the Fed deciding to “raise inflation.” The entire calculus is maddeningly abstract when it’s not even clear why the Fed isn’t continuing with its preferred paradigm.

  23. Gravatar of Steve Roth Steve Roth
    12. November 2011 at 23:02

    “So we reward inventors of “click to buy,” but not persuasive arguments for policies that might create 3 million jobs at very little cost of extra inflation. ”

    So much for efficient allocation of resources by the market…

  24. Gravatar of ssumner ssumner
    13. November 2011 at 10:16

    Paul, I’ve certainly written down simple models in the past, but I don’t think others find them to be of much interest. I certainly don’t rely on GS to justify my policy preferences, I simply note that other experts, using other models, reached similar conclusions.

    My basic model is that hours worked depends negatively on nominal wages divided by NGDP per capita. And nominal hourly wages are sticky. In that case monetary policies that affect NGDP, have a short run effect on hours worked, but no long run effect. Of course you can make the model more complex with hysteresis, etc. And you can model NGDP with the supply of money, op. cost of holding money, etc.

    Thanks JimP.

    Shane, Good question.

    Steve, Yes, patents are government-created monopolies, not the free market.

  25. Gravatar of Paul Andrews Paul Andrews
    13. November 2011 at 16:42


    If monetary policies can increase hours worked in the short run, but have no long run effect, must there not be a dip in hours worked after the short run completes (compared to what hours worked would have been), so that the total long run remains unaffected?

    Don’t you ever wonder whether the things left out of your models (mental or otherwise) may have more of an effect than the things included in your models?

  26. Gravatar of ssumner ssumner
    13. November 2011 at 19:24

    Paul, You asked,

    If monetary policies can increase hours worked in the short run, but have no long run effect, must there not be a dip in hours worked after the short run completes (compared to what hours worked would have been), so that the total long run remains unaffected?

    It depends where you start from, and other aspects of the model. If you start from a depressed condition, and just push the economy to “normal,” I’m not sure there has to be a relapse.

    You asked;

    “Don’t you ever wonder whether the things left out of your models (mental or otherwise) may have more of an effect than the things included in your models?”

    Every single day.

  27. Gravatar of dtoh dtoh
    14. November 2011 at 00:38

    You say…”My basic model is that hours worked depends negatively on nominal wages divided by NGDP per capita.”

    Isn’t it NGDP per capita divided by nominal wages and what do you mean by “depends negatively on”

    Sorry if I’m being dense.

  28. Gravatar of ssumner ssumner
    14. November 2011 at 18:30

    dtoh, If wages/NGDP rise, then each worker is grabbing a bigger share of NGDP. That means fewer jobs to go around.

  29. Gravatar of Dtoh Dtoh
    15. November 2011 at 04:11

    I get that. I just think you had the denominator and numerator reversed when you described it in your post.

    Also have you given any thought to how the model is impacted by:

    A) increasing trend to performance based comp, and
    B) shift from higher rate overtime hours to lower rate base hours when output contracts

  30. Gravatar of ssumner ssumner
    15. November 2011 at 13:19

    Dtoh, Maybe I reversed it.

    I don’t have views on the issues you raise, although I presume they’d have some effect on the proposal. My hunch is that the effect would be small.

  31. Gravatar of Doc Merlin Doc Merlin
    17. November 2011 at 04:45

    Inflation targeting means that the central bank responds to adverse supply shocks by tightening money.

    It also means that the central bank responds to positive supply shocks with monetary loosening.

    Both of these are /bad!/
    The central bank should NOT respond to supply shocks, but only to monetary problems. NGDP targeting takes care of that.

  32. Gravatar of Andrew C. Andrew C.
    17. November 2011 at 18:29

    “That’s exactly my view. Do something effective, or do nothing. Don’t do things that make monetary policy look ineffective.”

    In other words, do or do not, there is no try. Yoda for Fed chair.

    Sorry, I know you’ve made that reference before (I looked it up before commenting), but I couldn’t help saying it again anyway..

  33. Gravatar of Scott Sumner Scott Sumner
    17. November 2011 at 19:17

    Doc, I agree.

    Andrew. That’s right.

Leave a Reply