Raising expectations of inflation vs raising expectations of NGDP growth

Brad DeLong says: "If you believe--as I do--that the overwhelming proportion of the effects of non-standard monetary policy at the zero nominal lower bound come from reducing short-term safe real interest rates by raising expectations of inflation,..."

I believe something a bit different. Let me try to convince Brad to see it my way.

Let me take a very extreme case, a crazily extreme case, just to illustrate the difference.

Suppose the Phillips Curve, both short run and long run, were absolutely flat, forever and ever. Inflation is always and everywhere exactly 2%, regardless of monetary policy, regardless of anything. Just because. And so people expect exactly 2% inflation too, forever and ever, regardless of monetary policy, regardless of anything.

In normal times, when this economy is not at the ZLB, monetary policy would work just fine. Except now it would only affect real GDP, and have no effect whatsoever on prices and inflation.

Now suppose this economy were temporarily stuck at the ZLB, and in a recession. If Brad were right (and if we change Brad's "overwhelming proportion" to "100%") then non-standard monetary policy would have no effect at all. Because there is no way, by assumption, that anything can affect expected inflation in this crazy world.

But that conclusion seems to be wrong. Because the central bank in this crazy world could just as easily make a credible commitment to increase future real GDP (when the ZLB is no longer a constraint) as the central bank in a world at the opposite extreme with vertical short and long run Phillips curves could make a credible commitment to increase the future price level. And if the central bank's commitment caused people and firms to expect higher future real GDP, they would increase their consumption and investment demand today.

Non-standard monetary policy at the ZLB in this crazy world works not by reducing the real interest rate and moving the economy down along the current IS curve, but by shifting the current IS curve to the right. Because the position of the current IS curve depends on expected future real income and expected future real demand. Because current consumption depends on expected future real income, and current investment depends on expected future real demand. Rather than reducing the actual real rate relative to the natural rate, it raises the (short run) natural rate relative to the actual real rate.

The real world isn't like that crazy world where the short and long run Phillips curves are forever horizontal. But neither is it like the opposite extreme world, where the short run as well as long run Phillips curve is immediately vertical. It's somewhere in between. (Though recently the short run Phillips curve has been looking a lot more horizontal than I used to think it was.) And it seems to shift up and down a lot for reasons we don't understand very well, and maybe has some longish lags, and maybe some hysterisis, that we also don't understand very well.

I think it's sensible to believe that non-standard monetary policy at the ZLB works both by raising expected inflation and/or by raising expected real GDP growth, and that we don't know very much about which of the two it will be, and what the proportions are in the mixture, but that it's going to be one or the other, depending on what people and firms believe about the Phillips curve.

For example, if the central bank put a tax on holding currency, and so eliminated the ZLB as a binding constraint, we agree that looser monetary policy would then increase AD and create a recovery. We agree that recovery would be a mixture of higher inflation and higher real GDP growth. But what mixture? We don't know. My guess is that it would be mostly real GDP growth (unless the central bank really overcooked it, which I don't think the Fed will). My guess is that Brad would guess the same. My guess is that most people and firms would guess the same.

We could admit our ignorance about the actual and expected mixture and just say: 'the overwhelming proportion of the effects of non-standard monetary policy at the zero nominal lower bound come from...raising expectations of NGDP growth.'

Comments

[Alex Godofsky]

In your model, can't the central bank make GDP infinite by printing an infinite quantity of money? Or if the central bank tries to do that do we get non-monetary prices like queuing times?

[Nick Rowe]

Alex: Yep. One or the other, depending on whether real resources are scarce, or the central bank was a big blue meany that liked unemployed resources. But I did say it's a crazy world.

[Ron Ronson]

So if the CB promises to increase NGDP in the future then this will either

1) Increase inflation expectations which lower current real interest rates and get people investing which will increase NGDP in the present.
or
2) Increase expectations of future RGDP growth , which will get people to increase current spending which will increase NGDP in the present.

So it seems like if you can only make people believe that you will increase NGDP tomorrow you can increase NGDP today. But the problem is that as you may not actually need to increase NGDP tomorrow people may not take seriously your commitment to do it.

This seems a weird way of looking at things and is based presumably on the premise that either you can't actually raise NGDP today without these tricks or if you could (and did) it would lead to just inflation and no RGDP growth. Its seems pretty unlikely that the CB couldn't raise NGDP (otherwise it could buy all the assets in the world without nominal spending increasing). And if the CB can increase NGDP, but this resulted only in inflation, wouldn't that allow real rates to be below 0% and lead to an increase in investment, which would increase RGDP ?

I guess I'm just missing the reasons behind the "committing to be irresponsible" thing that Krugman often mentioned and may be part of Delong's thinking too.

[Nick Rowe]

Ron: forget about monetary policy for a minute, and think about ordinary life. We make promises, and usually we keep those promises, and usually people expect us to keep those promises, even when we wouldn't do what we had promised to do, if we didn't feel an obligation to keep our promise. In fact, if we we going to do it anyway, we wouldn't need to have made a promise.

Is the practice of promising puzzling, at a philosophical level? Yes, deeply. But at an ordinary commonsense level, it isn't puzzling at all.

[Ron Ronson]

If someone makes a promise to do something in the future in order to get out of doing something in the present, and the thing they promise to do is against their nature (A bookworm says "I promise to go out dancing tomorrow as long as you let me stay home and read tonight") then they may have trouble being believed.

There doesn't seem any reason why central banks have to take that chance. They have the power to boost NGDP in the present. If monetarists are wrong and this doesn't boost RGDP directly by causing a movement up the AS curve then it has to reduce real interest rates by increasing inflation , which will increase investments and get things moving the Keynesian way.

[Nick Rowe]

Ron: we make promises to change other people's expectations about our own future behaviour, because we want to change their current actions. If we thought that they thought that we would want to do the promised thing anyway, we wouldn't need to promise. We only promise because they wouldn't otherwise believe we would do the thing we promised to do.

"They have the power to boost NGDP in the present."

Suppose people expected that central banks would reduce near future NGDP back down again to its current level. Would they still have the power to boost NGDP in the present? In other words, how much does current NGDP depend upon expected future NGDP? I would say "a lot, especially at the ZLB".

[Ron Ronson]

Ah, now I see the dilemma. Present and future policies cannot be as easily separated as I was assuming. But couldn't this be overcome by using monetary policy for fiscal ends ? For example: Increasing the money supply via funding tax breaks on investment or (if all fails) funding direct government projects. These would boost RGDP irrespective of people's expectations about future monetary policy

[Brad DeLong]

Yes. If I could go back in time and persuade Larry Summers in 1999 to create a Treasury security that was a nominal GDP future rather than one that had a CPI future embedded in it, I would point to the fall in expected nominal GDP in 2018 and call Bernanke's policies a failure for that--superior--reason.

But I can't. So I have to point to the TIP-Treasury spread.

And if I had a time machine, also, I would request both real and nominal consols. I really want to see both real and nominal consols right now...

[Nick Rowe]

Thanks Brad. I think we are on the same page.

Ron: "Present and future policies cannot be as easily separated as I was assuming."

Yep. And that's true for fiscal policy too. Yes, fiscal policy can work, but whether or not it works depends also on expected future fiscal policy.

[Frank Restly]

Brad,

"If I could go back in time and persuade Larry Summers in 1999 to create a Treasury security that was a nominal GDP future rather than one that had a CPI future embedded in it, I would point to the fall in expected nominal GDP in 2018 and call Bernanke's policies a failure for that--superior--reason."

What you want instead is a government liability that has a high return on investment during slow growth times and a low return on investment during higher growth times. You want an output gap future, not a GDP future.

[Frank Restly]

Did the whole concept of countercyclical fiscal policy die when I wasn't looking?

[Nick Rowe]

Frank: see my recent post on countercyclical fiscal policy

[Frank Restly]

Nick,

Yes I read it.

"Suppose we want a countercyclical fiscal policy to help smooth out those fluctuations in the natural rate, to help prevent us hitting the ZLB. What would countercyclical fiscal policy look like, in the same picture?"

Are nominal GDP futures pro-cyclical or counter-cyclical in a New Keynesian framework? I would think they would be pro-cyclical in that the return on investment would be positively correlated with the present growth rate of the economy.

[rsj]

Levying taxes is not monetary policy.

But I agree that if we increased taxes on capital income or even wealth, then this would increase aggregate demand.

[rsj]

... assuming that this was offset by tax cuts on labor income or consumption