Goldpunk, strategy space, and Michael Woodford

On the one hand: it's very good news that Michael Woodford has endorsed NGDP level path targeting (pdf). (For non-economists, Michael Woodford is the most influential living academic monetary economist; he's the one who wrote the book that defines how graduate students think about monetary policy.)

On the other hand: I'm going to rain on our victory parade. Nothing important has changed.

The fundamental problem is the strategy space. We think of monetary policy as a conditional path for a nominal interest rate. "Setting an interest rate is what central banks really really do". That way of thinking about monetary policy is what creates the problem. That strategy space fails when the nominal interest rate hits the Zero Lower Bound.  Michael Woodford's text reinforced that way of thinking about monetary policy. It helped to define that failed strategy space. His Jackson Hole paper re-endorses that failed strategy space.

Imagine an alternate monetary history where central banks had maintained direct convertibility of their monetary liabilities into gold. But not at a fixed price of gold, like under the old Gold Standard. Instead, they adjusted the price of gold at their discretion. Perhaps on a daily basis. In the alternate 1960's and 1970's there was a problem of steadily rising inflation as central banks kept increasing the price of gold in order to try to target full employment. Then an alternate Milton Friedman argued that the long run Phillips Curve was close to vertical, so central banks should not be increasing the price of gold so quickly. Eventually, in the alternate 1990's, central banks adjusted the price of gold to keep inflation on target. (Like Irving Fisher's Compensated Dollar, except with an inflation target rather than a price level target).

We could imagine an alternate Michael Woodford writing a text with the title "Gold and other prices", arguing we did not need to talk about the quantity of money, because what central banks really really did was adjust the price of gold, not the quantity of money.

We could imagine some economists in that alternate monetary history saying it would be better to adjust the price of gold to target the NGDP level path, rather than to target inflation.

But we could not imagine an alternate Michael Woodford advocating an NGDP level path target in order to escape the Zero Lower Bound on nominal interest rates.

If in that alternate history we had thought nominal interest rates were too near zero, and we wanted to loosen monetary policy, and we wanted to cause nominal interest rates to increase above zero, the central bank would just start raising the price of gold. It would be obvious to everyone. Raising the price of gold is how central banks loosen monetary policy

There's nothing special about the price of gold, of course. Except history. But the price of gold does have the right units, because it's got $ in the units. There's no $ sign in the units for an interest rate. And what central banks really really ultimately do is determine the value of that $ unit.

Update: OK, I expect you might say that Michael Woodford is trying to get himself out of the strategy space hole he dug for himself. He recognises that communications policy is important. But he doesn't seem to realise that talking about monetary policy as interest rates is just another communications strategy. It's not what central banks really really do.

Comments

[JKH]

A couple of more raindrops:

NGDP targeting is a compromise for him; his preference is a type of price level targeting that he considers too complicated for popular appeal.

And his other preference is NGDP targeting for fiscal policy rather than monetary policy. It’s very odd how that was almost a throwaway comment in his conclusion. It seems to be fundamental to how he thinks about monetary policy at the zero bound.

It doesn’t totally hang together as an endorsement, IMO.

[Andrewlainton.wordpress.com]

Dimesional analysis mistake here, as Jevons confirmed the correct unit is dollar years, and in instantaneous time that reduces to price of the numaraire, fiat nominal currency or real.

[Bill Woolsey]

If you review his section where he critiques "pure" quantitative easing, he first argues that open market purchases of zero-interest rate bonds does not immediately and directly raise spending on output. However, expectations that the level of base money will stay high in the future, after the interest rate on the bonds rises above zero, does effect current spending on output.

He then says that instead of committing to keep base money high enough in the future, it is possible instead to promise to manipulate short term interest rates in an equivalent way in the future. I think this is correct as far as it goes. However, it isn't clear that base money is being kept "too high" in the future in the same way that we can say that interest rates are being kept "too low." Further, if base money will be at the right level in the future, and so, in the future, people set prices of output and production consistent with that, it isn't clear that real interest rates need ever be "too low."

And, of course, with a base money target, it is obvious that the problem isn't that an "upper bound" on base money has been reached. It is just that people expect increases in base money to be increased when inflation rises.

[Nick Rowe]

JKH: If I thought central banks could observe potential output in real time (Simon van Norden has got the empirical evidence they are very bad at it), and if I thought I could explain "NGDP-Gap level path targeting"? to the public, then I might be tempted to follow him on that. Though he missed another argument in favour of the simpler NGDP target, which is that the danger of the ZLB is probably greater when potential GDP is growing more slowly. I don't see any big disagreement between MW and MMs on that point. I think we all see NGDP as a simple workable approximation rather than as something theoretically ideal.

Andrew: sorry, but you lost me there. When I said "$ in the units" I wanted to remain silent of what else might or might not be in the units. The price of gold (which is what I was referring to there) has the units $ per ounce (or similar). NGDP has the units $ per year. And that would be true whether you collected the data every decade, year, quarter, month, week, or in continuous time. My car's speedometer (I think) works in continuous time, but still measures kilometers per hour.

[Nick Rowe]

Bill: yep. It looked like he might be heading in the right direction when he talked about future base money. But he then backed off and said it was equivalent to an interest rate path. He wouldn't apply the same communications strategy analysis to the "instrument" that he applies to the "target".

But then, for the last dozen years or more, he's being arguing that monetary policy really really is interest rate policy.

[E. Barandiaran]

Sorry Nick but central banks don't determine the value of their currency. They determine the unit of account and that is why Milton Friedman was completely wrong when he said that central banks could add any number of zeros to the unit of account to increase the supply of currency --adding zeros is just a way to change the unit of account. Everywhere changes in the quantity of currency --assuming the unit of account has not changed-- are determined by many Central Bank transactions and we can discuss forever which transactions imply increases in currency in circulation. More important, the value of currency depends on the residual demand for holding currency for transactions (the old inventory models say more about this demand that models based on liquidity and speculation).

[Andy Harless]

I disagree. Setting interest rates is what central banks really, really do. This is especially true of contemporary central banks that pay interest on reserves, which in principle is their real instrument, with OMO's just being used to provide liquidity as necessary rather than to control macroeconomic conditions.

But let's go back to pre-IOR central banks. They basically did two things: asset swaps and communication. Usually their asset swaps consisted of swapping non-interest-bearing debt (base money) for interest-bearing debt (T-bills). (They were able to do this because they had an exclusive license to produce one of the few forms of non-interest-bearing debt for which there was a demand.) Swapping non-interest bearing debt for interest-bearing debt is equivalent to setting interest rates. Now in principle they could alternatively view it in quantitative terms -- looking at the quantity of base money rather than the price -- but I don't see that that would be particularly helpful.

In any case, such a central bank necessarily faces a problem with the zero bound, because the zero bound eliminates the distinction between interest-bearing and non-interest-bearing debt, thus taking away the bank's monopoly. (Unfortunately the central bank can't sue the treasury for violating its patent when the treasury starts issuing non-interest bearing debt.) At that point, communication becomes the central bank's only tool. (Sure, it can control the quantity of base money, but that is irrelevant: the real instrument is the communication that it intends to maintain some quantity of base money once it gets its monopoly back.)

Now you may argue that communications about the price of gold are more effective than communications about future interest rates, but in any case either type of communication is only effective if the central bank's communications are credible. In principle a central bank at the zero bound (if OMO's are its instrument) is unable to affect the current price of gold except via communications. Now it probably can affect the current price of gold, because its communications do have some credibility. But the current price of gold is only an intermediate target and not a particularly good one at that. NGDP futures -- or even price level futures, which already exist in many countries -- would be a better intermediate target. And the only reason such a target (be it gold or futures) is helpful is that it lets the central bank observe its own credibility concretely. If it had no credibility, it couldn't hit the target, and if it had no intermediate target, the credibility issue would be the same; it just wouldn't be observable.

[Nick Rowe]

E Baraniaran: "Sorry Nick but central banks don't determine the value of their currency. They determine the unit of account..."

If their currency is in fact the unit of account, I don't see the difference.

"More important, the value of currency depends on the residual demand for holding currency for transactions (the old inventory models say more about this demand that models based on liquidity and speculation)."

That's true if we think of the supply of money as a quantity of money. It's not true if we think of the supply of money as a price of gold.

Andy: Imagine there's a gold window at the bank, with a chalkboard sign saying "Gold bought or sold, at $P per ounce". And the "P" keeps getting erased and replaced with a different number. Sure, what it says there this very instant matters very little. It's the expected time-path of P that matters. But "doing nothing" means "not changing P". So current changes in P are "implied permanent". And if people see the bank raising P daya after day, and announcing it's going to keep on raising P until NGDP* is hit, that would be a very credible communications strategy. "My money is going to be worth less in future than it is now. Better spend it now". And the bank will be selling gold at the gold window, and will be needing to borrow gold at increasing nominal interest rates. And those higher interest rates will be a symptom of loosening monetary policy.

The price of gold is not currently infinity.

[Nick Rowe]

But yes, NGDP futures would probably work better than gold. But gold works better for me as a metaphor, since it's simpler and I can talk about alternate history. And I expect I should have talked about what Roosevelt did in 1932(?) explicitly, but it's implicit.

[Ron Ronson]

"If in that alternate history we had thought nominal interest rates were too near zero, and we wanted to loosen monetary policy, and we wanted to cause nominal interest rates to increase above zero, the central bank would just start raising the price of gold"

Quick question: In this alternative world why would increasing the price of gold (presumably by increasing the quantity of money) cause nominal interest rates to rise above zero ? Would it be because of expectations of inflation or by some other mechanism I am missing?

[Andy Harless]

Once you've acknowledged that NGDP futures would work better than gold, I have to ask, "Are NGDP futures really that important after all?" Wouldn't it work almost as well simply to target the central bank's publicly released internal forecast of NGDP? Doesn't making a market in NGDP futures (or gold) only amount to "putting the CB's money where its mouth is," a useful but not critical (and not airtight, and not costless) enhancement to credibility? And isn't that potential enhancement to credibility equally available whether or not you frame the original problem in terms of interest rates?

[Nick Rowe]

Ron: "In this alternative world why would increasing the price of gold (presumably by increasing the quantity of money)..."

In this alternate world we wouldn't think of central banks increasing the price of gold *by* increasing the quantity of money. We would think of central banks just raising the price of gold. "That's what central banks do: they offer to buy and sell gold at a price they announce." We would think of central banks changing the quantity of money *by* changing the price of gold, not vice versa.

Yep, expected inflation in the longer run, plus expected real growth in the shorter run.

[Nick Rowe]

Andy: Hmmm. Dunno. Lets set aside the question of whether the public might be better or worse forecasters than the Bank. And lets set aside the question of communications. I think I would say that it's the public expectation we want to target. Not sure though. I thought about this once, and came to a conclusion, but I have forgotten what it was.

[Nick Rowe]

Suppose the Bank had once-in-a-lifetime Top Secret information it couldn't divulge about something that would affect next year's NGDP.

[W. Peden]

Nick Rowe,

Yes, the quantity of money and interest rates would be demand-determined.

*Ducks for cover!*

[Nick Rowe]

W Peden: Bingo!

[Mike Sproul]

The trouble with central banks (or anyone else) targeting NGDP (or anything else) is that the forward price F of any financial security (including money) must be related to the spot price S by the formula F=Se^(r+u-y)T, where e=2.718, r=interest rate, u=storage/handling costs, y=convenience yield, and T=time.

Any target that deviates from that relation will put the central bank on the wrong end of someone else's arbitrage opportunity.

[Nick Rowe]

Mike: you've got the whole macroeconomic model buried in those 3 variables r, u, and y. They won't be invariant wrt monetary policy.

[Determinant]

That's the objection I have to your insistence about expectations, Nick. Sooner or later that model will work in the arbs favour and the Central Bank will lose. The government might lose in that case too.

Second, on this:

Suppose the Bank had once-in-a-lifetime Top Secret information it couldn't divulge about something that would affect next year's NGDP.

I go back to the poker-game model. Does the Central Bank never bluff? Does the Central Bank ever hold a full house while arbs on the other end hold a straight? What happens when the cards go on the table and the Central Bank loses?

[Matt Rognlie]

I agree with Andy that setting interest rates is what central banks really, really do. Arguably, one can view monetary policy through either an "interest rate" or "monetary base" lens, and you might say that it's arbitrary to elevate "the Fed decreased interest rates by 0.5%" over an old monetarist description like "the Fed increased the money supply by 0.1%". But the former is really a much more accurate description. When the Fed intervenes, it changes its positions by minuscule amounts (it's hard to even see the change in quantities when the Fed moves to a new target rate) in an incredibly obscure market -- the market for interbank reserves -- whose microstructure is understood by at most maybe 1000 people. We are affected by the Fed's policy because it's efficiently transmitted to us through interest rates and asset prices, not because we have any idea what just happened quantitatively. (This is all the more true, as Andy points out, now that we have IOR; the Fed now literally sets the short rate, without any quantitative intervention being necessary at all.)

Of course, what central banks really, really, really do is set expectations, and on that point you, Andy, Woodford, and I are all in agreement. Woodford, Andy and I like to think of the Fed as setting a contingent plan for interest rates, while you like to think of the Fed as setting a contingent (ideally non-contingent!) plan for NGDP. To some extent, there is an isomorphism between these two ways of looking at monetary policy. These are not really different "strategy spaces"; they're just transformations of the same strategy space, and the reasons for choosing one over the other are inevitably subjective.

The key difference, I think, is that you view the zero lower bound as entirely a problem of our own creation, the artificial result of framing monetary policy in a strategy space that only includes nonnegative interest rates. Woodford and I disagree with this, because we view the zero lower bound not as a mere institutional constraint, but rather as a manifestation of the lack of any difference between the non-pecuniary yields of base money and short-term liquid securities. Under the assumption that no securities except base money have any non-pecuniary "convenience yield", then, once you reach the zero lower bound there is no way that further asset purchases can do anything, except perhaps redistribute pecuniary risk slightly. The problem isn't so much that short-term rates can't go below zero, it's that for the same underlying reason that short term rates can't go below zero, all central bank operations will be completely ineffective, except insofar as they redistribute risk in incomplete markets. You could write the theory without any reference to short-term interest rates and the same conclusions would still follow.

No doubt the assumption that there is no non-pecuniary component to the yield of assets other than base money is exaggerated. But I'm not sure that it's so far from being right. After all, there appears to be a highly effective arbitrage between long Treasury rates and forward rates in the OIS market; and these forward rates represent the market's risk-weighted expectation of future short rates. And I'm not confident that the Fed's asset purchases can alter the distribution of risk enough to make a real difference in these rates; after all, the universe of assets subject to interest rate risk is absolutely enormous. There's $40 trillion in domestic nonfinancial debt, $15 trillion in domestic non financial equities, and so on. Maybe markets are a lot more segmented than I think (although in that case the usefulness of QE is questionable for other reasons), but I can't see how a few trillion in asset purchases is going to make much of a dent.

Now maybe this is all a distraction, because you think that the real power of monetary policy comes through commitment, not unconventional asset purchases. And I agree! But in Woodford's model, which is a pretty decent benchmark model, it is completely possible for the zero lower bound to prevent a central bank from hitting an NGDP target path. Commitment does a lot, but it isn't all-powerful. (A sketch of a limiting case where this is true: suppose prices are extremely inflexible, so that for practical purposes we can regard the inflation rate as constant in the medium term. Then an NGDP target is equivalent to an output level target. In models where (1) "divine coincidence" holds and (2) there are no shocks to potential output, the central bank's behavior under an output level target is the same as its behavior under inflation targeting or a Taylor rule. And we've already seen countless models where the liquidity trap is a problem when the central bank targets inflation or uses a Taylor rule. The central bank simply won't be able to hit its NGDP target in the short run, no matter how credible the long-term target or how hard it tries! This isn't a realistic case, of course, but it does indicate why NGDP level targeting does not magically wipe away the problems of the zero lower bound, even though it may do a great deal of good.)

This is why I think it is useful to talk about interest rates rather than NGDP as the primitive "strategy space". If it's possible that the zero lower bound will prevent us from hitting our NGDP target, then there are certain "NGDP strategies" that are not operationally possible. You could make due allowance for this and continue to talk about NGDP as the basic lever of monetary policy, but that would be difficult. It is far easier to see when an "interest rate strategy" is operationally possible or not -- you just need to ask whether the interest rates are nonnegative! This is true even if the ultimate goal of the interest rate strategy is to enforce an NGDP level target.

[Lord]

Isn't this just exchanging one bound for two, one where they own (almost) all the gold and one where they own none? While it could set a very high price, if people expect them to raise it, the remainder won't want to sell it. In that case does raising the price loosen or tighten?

[Nick Rowe]

Matt: a quick response: "The key difference, I think, is that you view the zero lower bound as entirely a problem of our own creation, the artificial result of framing monetary policy in a strategy space that only includes nonnegative interest rates."

One could see the ZLB as imposing a constraint ("Don't choose too low a level of NGDP growth for your target path, because you would only be able to hit it with equilibrium nominal interest rates below 0%"). But there's a second question about whether one could get *stuck* at the ZLB. For example: "Well we were targeting NGDP successfully, but we must have chosen too low a target growth rate, because nominal interest rates are hitting 0%. Oh well, I expect that means we will just have to raise the NGDP growth target, at least for a bit."

[Matt Rognlie]

On a much shorter and unrelated note...

And the bank will be selling gold at the gold window, and will be needing to borrow gold at increasing nominal interest rates. And those higher interest rates will be a symptom of loosening monetary policy.

If you look at historical interest rates, this does not appear to have happened when the US left gold in 1933. The series for commercial paper rates, for instance, looked like this:

10/1932: 1.94%
11/1932: 1.74%
12/1932: 1.51%
01/1933: 1.38%
02/1933: 1.38%
03/1933: 2.78%
04/1933: 2.56%
05/1933: 2.08%
06/1933: 1.69%
07/1933: 1.50%

The US left gold, and the exchange rate vs. the franc started collapsing, in mid April. Yet the trajectory of short rates following that, if anything, was downward. (By the way, the series has a few bumps along the way, but it continues downward after 07/1933; rates eventually got incredibly close to 0.) Arguably the spike in 03/1933 was unnatural and resulted from temporary factors related to the banking holiday, and short rates were just reverting back to trend after that -- so there's a decent case to be made here that the massive easing starting in April 1933 didn't lower interest rates either. But looking at the variation in this time series, you'd be hard-pressed to claim that high interest rates were a sign of monetary ease. (Even worse if you look at the tightening in late 1931 in the US, Germany, and various other countries, or the 1920-21 depression!)

Now, ultimately the general level of interest rates was so low in 1932-33 due to the Depression, and its bad monetary regime. In a sense they were a sign of tight money---the evidence certainly isn't consistent with some proto-Keynesian view that the level of interest rates entirely captures the stance of monetary policy. But I think that the best lesson to draw here is simply that the current level of the short rate is not a good indication of the stance of monetary policy, in either direction. Blanket statements like "low interest rates mean policy has been tight" or "low interest rates mean policy has been easy" are both extremely misleading in some situations. The former statement is vaguely true if you're looking at extremely low-frequency variation in short rates, or maybe long rates, while the latter statement is vaguely true if you're looking at high-frequency variation in short rates. But even then they're so loaded down with caveats that I think it's best to dispense with them altogether.

[Matt Rognlie]

One could see the ZLB as imposing a constraint ("Don't choose too low a level of NGDP growth for your target path, because you would only be able to hit it with equilibrium nominal interest rates below 0%"). But there's a second question about whether one could get *stuck* at the ZLB. For example: "Well we were targeting NGDP successfully, but we must have chosen too low a target growth rate, because nominal interest rates are hitting 0%. Oh well, I expect that means we will just have to raise the NGDP growth target, at least for a bit."

I agree that no matter what situation you're in, you can surely escape the zero lower bound if your medium-term NGDP growth target is high enough. A sufficiently high target for NGDP growth would imply a very high rate of inflation sometime in the future, which would trickle back to expectations today and increase both output and prices. Effectively, it increases the "natural" nominal interest rate. But this already involves some dynamic manipulation of the NGDP targeting rule ("oh, it's not working? let's increase the target!"), which lessens the appeal of NGDP targeting as a strictly rule-based, discretion-free approach to monetary policy.

I do think that given the NGDP target growth rates commonly suggested (say, 5%), it is likely that a sufficiently large disruption -- say, one on the magnitude of the 2008 crisis or higher -- would push the economy temporarily below the NGDP target and pin the central bank at the zero lower bound, making it powerless to help any further. Such a world would still be far better than the one we lived through -- since the central bank's commitment to doing everything in its power to hit the NGDP target would raise current prices and output far more than the wishy-washy commitments we've seen recently. But it would still not be perfect, and if we do implement NGDP targeting we should be prepared for the possibility that the central bank might not be able to stop a sizable departure from the NGDP target in the short run.

[Max]

The issue is whether the target is purely forward-looking or is history dependent.

Does thinking of monetary policy in terms of quantity (of non-interest bearing base) automatically give you a history dependent policy? Certainly not. Base money can be destroyed as easily as it's created.

Incidentally, Woodford makes the point that if QE is intended to communicate a permanent increase in (non-interest-bearing) base, then it should be vastly smaller than the QE that has actually been implemented. But I've never seen any Market Monetarists criticize the central banks for creating too much base money.

[Matt Rognlie]

Incidentally, Woodford makes the point that if QE is intended to communicate a permanent increase in (non-interest-bearing) base, then it should be vastly smaller than the QE that has actually been implemented. But I've never seen any Market Monetarists criticize the central banks for creating too much base money.

Hate to monopolize the comment section (I'm having an unproductive day...), but I agree that this is an incredibly important point. At the current level, and certainly at any higher levels, there is no conceivable way that we could interpret the existing stock of base money as a permanent increase; if maintained, it would be extremely inflationary, far more than the Fed would possibly allow.

And once there is no longer any sense in which the marginal injection of base money is permanent, it's not clear why QE should improve expectations at all. I often hear statements about how the Fed could do some more QE to "back up" a target if necessary, but I don't understand concretely how asset purchases actually back anything up. The only explanations I can think of are

(1) institutional: if the Fed's balance sheet has extreme maturity mismatch, then it's effectively committing not to raise short rates too much for fear of going insolvent, which wouldn't matter much for the government budget as a whole but might compromise the institution's independence, and...
(2) sociological: asset purchases signal that the hard-line hawks, who see incipient inflation in every expansion of the monetary base, don't really have that much influence at the Fed.

But (1) is an extremely awkward, roundabout approach to commitment, and (2) wouldn't be an issue if the Fed had a clear targeting framework that everyone agreed upon. In this light, I don't see how asset purchases in general can really "back up" a target path for NGDP, or any other target for that matter. And once QE is off the table, you have to ask what exactly the central bank should do when the zero lower bound prevents it from hitting its target right away. (Woodford's suggestion is fiscal stimulus... I have mixed feelings about that.)

[Nick Rowe]

Matt: "Hate to monopolize the comment section..."

Keep rolling! Great comments.

[Nick Rowe]

Max: "Incidentally, Woodford makes the point that if QE is intended to communicate a permanent increase in (non-interest-bearing) base, then it should be vastly smaller than the QE that has actually been implemented. But I've never seen any Market Monetarists criticize the central banks for creating too much base money."

Whoah! I have said more than once, and I'm pretty sure Scott Sumner has said too, that a credible commitment to a decent NGDP target would result in a *reduction* in the monetary base. With proper communication, "QE" (i.e. Open Market purchases) would need to be *negative* to prevent overshooting the target. But *without* proper communication, we sure aren't going to criticise something that's better than nothing.

[David Beckworth]

Matt,

Good to see you back in the blogosphere. Since you are Woodford's proxy here, let me ask you a question about his paper. On one hand, Woodford makes the case that it is the expectation of a permanent monetary base increase that matters in terms of changing NGDP today. On the other hand, he opts for a NGDP-conditional commitment path for the federal funds rate as way to change NGDP today. Doesn't this indicate he views the NGDP-contingent path of the federal funds rate simply as means of properly managing expectations about the future path of the monetary base? If so, is he not saying monetary policy is really, really about the future path of the monetary base?

[David Beckworth]

Nick,

Yep, a number of us have been saying that the Fed's large balance sheet is in fact a sign of failure.

[Mike Sproul]

"those 3 variables r, u, and y. They won't be invariant wrt monetary policy."

I guess you mostly mean r, since u and y are unlikely to be affected by monetary policy. But if the central bank changes the real interest rate in one country relative to another, then the arbitragers will pounce and profit at the central bank's expense. You'd probably agree that with gold convertibility, the possibility of international arbitrage makes the central bank's control over the interest rate close to non-existant. The next step is to recognize that if we replace gold convertibility with bond convertibility, or CPI basket convertibility, then nothing important changes, and the central bank is still left unable to target r, or NGDP, or much of anything else.

I know of only one 'targeting' rule that works. It's the one devised by practical bankers over centuries of experience: Money should only be issued in exchange for good bills of adequate value, at not more than 60 days' date.

[K]

Matt: "I agree that no matter what situation you're in, you can surely escape the zero lower bound if your medium-term NGDP growth target is high enough."

I don't think that's obvious. What if the natural rate is 2%, inflation is -4% and the policy rate is -2%. All rates are expected to be unchanged to infinity. Then CB technology changes and nominal rates can no longer be negative. I don't see any policy path (rates or NGDP - same thing) that save us from a short term (disastrous) rise in the real rate.

[Andy Harless]

David,

Let me be a secondary proxy, since I think I know the answer. (Then we'll see if Matt likes my answer -- and if it satisfies you.)

"Doesn't this indicate he views the NGDP-contingent path of the federal funds rate simply as means of properly managing expectations about the future path of the monetary base?"

Given a model of the economy, the phrase "managing expectations about the future path of the monetary base" is equivalent in meaning to "managing expectations about the future path of the short-term interest rate" because a path for either one defines a unique path for the other. (Actually this is not quite true, because when the interest rate is at the ZLB, the concurrent size of the monetary base is irrelevant, so there would be multiple paths for the monetary base consistent with an interest rate path that includes periods at the ZLB, but the point is that these paths are macroeconomically equivalent, so the expectation of one such path will produce the same results as the expectation of another such path.)

The advantage of thinking in terms of the monetary base, as Nick would argue, is that there is this sense of "semi-permanence" whereby the current level of the monetary base arguably provides a signal about its future level -- although, as Matt points out, we can be fairly certain that the monetary base (in the US) will be smaller in the future than it is now, so it's not clear in what sense there is any "semi-permanence."

The advantage of thinking in terms of the interest rate, as Matt points out, is that it gives you more information about what macroeconomic paths are feasible. To the extent that the current monetary base is not a signal of the future monetary base, you won't always know whether changing the current monetary base has a macroeconomic effect or not. You will know that changing the current interest rate has an effect, because the case where it wouldn't have an effect is the case where you are unable to change it anyhow.

[Ryan V]

Andy,

I'm certainly no expert, but:

To the extent that the current monetary base is not a signal of the future monetary base, you won't always know whether changing the current monetary base has a macroeconomic effect or not.

Won't a permanent change in the base still have an effect? So isn't it even easier to think about monetary policy as, "A permanent change in the base always was an effect. The End"?

Of course, people need to expect it to be a permanent change. So is the difference that a change in interest rates will still have an effect even it's expected to be temporary? (At least I think that's true--maybe it's not!)

[Andy Harless]

Ryan,

A permanent change always has an effect, if it's expected to be permanent. But that condition is the same as saying that the current monetary base is a signal for the future monetary base. If that's true, it will have an effect. But in the section you quoted I was specifically talking about the case where that's not true -- where the change is "not a signal," which is to say that it is not expected to be permanent (or, more precisely, that it doesn't alter your expectations for the future monetary base: it's possible that you already expected it, and expected it to be permanent, before it happened, in which case it also has no effect, except in the sense that the anticipation may have had an effect).

The point is, if you're at the zero bound, the change in the monetary base will only have an effect if it is a signal (i.e. if it is expected, at least in part, to be permanent), whereas when you are away from the zero bound, the change can have an effect even if it is not a signal (i.e. if it is expected to be temporary). Though I guess you could argue that, even in the latter case, it has to be a signal for the immediate future (i.e. it has to be expected to be persistent even if temporary: it doesn't work if people expect it to be reversed the very next day).

[Andrewlainton.wordpress.com]

Nick we are on the same page here, GDP is a flow variable (two dimensions) and so has the same units as interest = $.years

[Pacemaker]

Andy,

"...we can be fairly certain that the monetary base (in the US) will be smaller in the future than it is now...
...To the extent that the current monetary base is not a signal of the future monetary base, you won't always know whether changing the current monetary base has a macroeconomic effect or not. You will know that changing the current interest rate has an effect, because the case where it wouldn't have an effect is the case where you are unable to change it anyhow."

Similarly, wouldn't we expect the interest rate to rise if we had expansionary policy now?

[Saturos]

Nick, are you then denying that Woodford's pronouncements at Jackson Hole have brought us all much much closer to truly optimal monetary policy? So what if Woodford still doesn't really "get it", provided the Fed switches over to forward-looking NGDPLT in a few years' time (which looks about thrice as likely now)? And once that happens people will start to care less what the Fed says it's doing to the FF rate, and more what they expect NGDP to be.

[Nick Rowe]

Saturos (and Matt): Yeah, OK, I was maybe in a bit of an ornery contrarian mood when I wrote this post. "The glass is half-empty!". Especially since I was off canoeing the upper Ottawa river when the big news hit, and everyone else got to celebrate without me. Yes, there is lots of agreement between what MW is saying here and what we've been saying. It's a great paper.

[Scott Sumner]

Nick, Nice post.

Andy and Matt, Recently I argued that the real issue with the zero bound is the size of the monetary base necessary to hit the policy target. Thus the central bank has several choices:

1. Raise the NGDP target--you both seem to think that could work.

2. But let's say we don't want to raise the target. Take a country like Japan, which seems to prefer zero NGDP growth. Do OMOs "work" in that case? I argued there is always some level of OMOs that "work" i.e. keep expected NGDP growth on target (zero percent target in this case.) But I also argued that the demand for base money at a very low NGDP growth target may be so high that the central bank has to buy non-conventional assets. I don't see how even the hardest core monetarist could disagree with that. For instance, imagine a country like Australia that has a tiny national debt. Obviously they might need to buy alternative assets to hit the target. Woodford and Krugman then redefine that as "non-monetary policy." So it actually seems to just be a debate about terminology.

One other point. Bernanke has said that the Fed would have done more if it could have continued to use it's traditional signaling method--changes in the fed funds target. That supports Nick's argument that the gold price target would make a real difference. The Fed would never have to leave its comfort zone of targeting gold prices, as they'd never hit a zero bound. Hence the Fed would have been more aggressive.

[Morgan Warstler]

"We could imagine an alternate Michael Woodford writing a text with the title "Gold and other prices""

I can't imagine Woodford ever writing a text, he is super wordy.

The issue is Woodford STILL is yammering on about Fiscal which make me think he doesn't actually understand NGDPLT. I think he hasn't GRASPED it, which would fit into your point about his MO being wrong.

[Ritwik]

Nick

Once and for all, please let's clear this.

Do you think monetary policy is about creating new hot potatoes?

Or do you think it is about baking existing potatoes?

The Chuck Norris/ expectations conception is consistent with the second but not the first. Can we at least agree on the common monetarist ground between disequilibrium theories and market theories?

[Max]

"Obviously they might need to buy alternative assets to hit the target. Woodford and Krugman then redefine that as "non-monetary policy." So it actually seems to just be a debate about terminology."

I don't think it's just terminology. If you think the quantity of money matters (even when the CB has placed a floor on interest rates), then buying stuff really is monetary policy. And it makes no difference what the CB buys.

But if you don't believe that, then it's not monetary policy. It's just buying stuff, which anyone can do. The Treasury can buy stuff. Warren Buffett can buy stuff. You and I can buy stuff. Buying stuff could have powerful macro effects under some circumstances, but there's nothing _monetary_ about it.

[Matt Rognlie]

Andy, agreed completely. (I should always try to be late in following up -- someone else ends up doing it better anyway!)

For the most part, there is a one-to-one correspondence between the state-contingent path of interest rates and the state-contingent path of the monetary base. You can think of policy as setting either a complete plan for the price instrument or a complete plan for the quantity instrument, and from a formal standpoint both descriptions are completely valid. As Andy points out, the one case in which this is not true is the zero lower bound, where multiple levels of the monetary base correspond to outcomes that are both macroeconomically equivalent and lead to the same nominal interest rate. In this case, the interest rate view has an advantage, since each plan for the interest rate corresponds to a distinct macroeconomic outcome.

Woodford emphasizes this correspondence between prices and quantities, and how results that have traditionally been stated in terms of quantities might be more insightfully phrased in terms of interest rates. For instance, he says:

The demonstration by Auerbach and Obstfeld that welfare can be increased by permanently increasing the supply of base money could alternatively be used to show that welfare could be increased by committing to keep the nominal interest rate at zero until it is possible to hit a certain deterministic target path for nominal GDP, and then use monetary policy to keep nominal GDP growing at a steady rate thereafter.

The key point here is that the initial "permanent injection" in Auerbach and Obstfeld isn't really necessary to achieve the ultimate effect; it would be equivalent to leave the monetary base unchanged in the short run (where the zero lower bound renders the exact quantity irrelevant), while credibly promising to keep rates at zero later on. We can evaluate the plausibility of the latter in light of central bank operating procedures (and the prospects for changing them), whereas it's rather difficult to think about how credible a "permanent injection" is.

Another advantage of thinking about interest rates as the operational tool of the central bank is that they map more cleanly onto most macroeconomic models. If you have a simple money in the utility function specification (which embeds many richer, more specific theories of money), for instance, you get

-- one equilibrium condition that relates u_m/u_c, the marginal value of money in consumption terms, to the nominal interest rate
-- one equilibrium condition that relates the real interest rate to intertemporal substitution and investment decisions

You could, of course, rearrange these conditions and purge the nominal interest rate from the theory, using only monetary quantities instead. But this would be a needless complication of the math. And I think that this reflects a fact about the real world, not just the first-order conditions of a monetary model. As a consumer, I make both (1) a portfolio decision about my holdings of base money, based (to a limited extent) on the nominal interest rate, and (2) an intertemporal decision about how much I want to consume and save, based on the real interest rate. We could combine these decisions, and say that I decide how much I want to save based on my marginal utility of base money in both periods. But that would be a pointlessly roundabout view; base money is a fairly obscure asset, not a very large fraction of anyone's portfolio. No one, except perhaps the poorest of the poor, makes decisions about whether to consume or save by saying "well, I'll have $150 in my pocket in a year, and I have $130 in my pocket now; let's think about the marginal utility of holding that currency in both periods". They certainly don't say "well, on the interbank reserves market, there will be $20 billion in reserves in a year, whereas now there are $15 billion in reserves"! Instead, to the extent that they're doing any serious optimization at all, they consider the interest rates for borrowing and saving. The interest rate is the market device that transmits the obscurities of the interbank reserves market to the decisions of actual consumers. (In, perhaps, the same way that the oil price transmits day-to-day estimates of the probability of a war with Iran, or OPEC's changes in production quota, to consumers who know very little about these goings-on.)

[Nick Rowe]

Ritwik: "Do you think monetary policy is about creating new hot potatoes?

Or do you think it is about baking existing potatoes?"

Couldn't it be both? You make existing potatoes hotter by changing expectations about how many hot potatoes you will create (and under what conditions) in future?

[Matt Rognlie]

Scott, you say

But let's say we don't want to raise the target. Take a country like Japan, which seems to prefer zero NGDP growth. Do OMOs "work" in that case? I argued there is always some level of OMOs that "work" i.e. keep expected NGDP growth on target (zero percent target in this case.) But I also argued that the demand for base money at a very low NGDP growth target may be so high that the central bank has to buy non-conventional assets. I don't see how even the hardest core monetarist could disagree with that. For instance, imagine a country like Australia that has a tiny national debt. Obviously they might need to buy alternative assets to hit the target. Woodford and Krugman then redefine that as "non-monetary policy." So it actually seems to just be a debate about terminology.

Sure, to some extent it's just a debate about terminology. But I think there is also a real difference of opinion between Woodford and you on the effectiveness of unconventional asset purchases. Woodford suggests that the "portfolio balance effects" cited by Bernanke (and implicitly invoked by you as well) are small to nonexistent, to the point where their effects are possibly so miniscule in comparison to forward guidance that they're not an important component of monetary policy.

Simply buying long-maturity assets, for instance, will not necessarily do much to bring down long rates (independent of its effect on expectations). There is a very effective arbitrage between forward short rates and the long rate, and those forward rates are risk-weighted expectations of future policy. Even if the exact Ricardian neutrality result in Woodford's model doesn't hold, it's very hard to see how QE, or even QE*5, is going to redistribute risk enough in the vast, $50+ trillion market for financial assets with interest rate risk to change these rates substantially.

Now, you could maybe make some serious headway with QE*10 or QE*20, or maybe a program with magnitude closer to the current QE that invests in riskier and more obscure asset classes, where the central bank's power to absorb risk will not be offset so much by an unwinding of private positions. But the former starts to become politically implausible: is the Fed really going to get away with making its balance sheet larger than all commercial banks combined, or the entire market capitalization of the NYSE? (And I suspect that we'd start to see other effects long before we got "portfolio balance effects" of necessary size; for instance, if the Fed had a balance sheet 10 times its current size, then any higher-than-expected increase in short rates would instantly tip it into insolvency. The "Fed cuts off its own arm" signaling effect from such massive asset purchases is probably more significant than the portfolio balance effect that's supposed to be doing the work!)

And I think it is not just semantic quibbling to question whether the latter is really "monetary policy". If the Fed gains traction by investing in less liquid markets fraught with informational frictions, it's starting to act less like a central bank and more like an activist fiscal authority. A lot of the advantages we normally attribute to monetary stabilization -- fewer political economy problems, less waste from half-baked stimulus projects -- become less clear.

[K]

The point about the irrelevance of the path of the quantity a the ZLB has been made many times before on this blog but in particular I said it here and in the comments that followed in many different ways. I think it's worthwhile rereading that previous debate. 

[JKH]

Matt Rognlie:

“You can think of policy as setting either a complete plan for the price instrument or a complete plan for the quantity instrument, and from a formal standpoint both descriptions are completely valid. As Andy points out, the one case in which this is not true is the zero lower bound, where multiple levels of the monetary base correspond to outcomes that are both macroeconomically equivalent and lead to the same nominal interest rate. In this case, the interest rate view has an advantage, since each plan for the interest rate corresponds to a distinct macroeconomic outcome.”

This is what I don’t understand.

Why are there not other cases where it’s true?

Specifically, why does it matter what the excess reserve component of the monetary base is after “lift-off”, when the Fed must pay interest on excess reserves in order to support the new target funds rate, and when in doing so it makes excess reserves essentially indistinguishable from short term treasury bills, just as they are at the zero bound?

And therefore why is this idea of a “permanent” increase in the monetary base so important?

[JKH]

sorry, meant:

"Why are there not other cases where it’s not true?"

[Nick Rowe]

Matt: Let me just put an Old Keynesian twist on this debate: Old Keynesian models have a multiplier. (I'm not talking about the *fiscal* multiplier; I'm talking about *the* multiplier -- the thing that multiplies *any* exogenous increase in desired expenditure. The thing that's bigger than one.)

How big is the multiplier at the ZLB? Specifically, how big is the multiplier when the central bank promises in advance it will do nothing to choke off successive rounds of the multiplier process until the NGDP target is hit? When you include positive feedback from: 1. the standard effect of income and expected future income on consumption; 2. the additional effect from greater investment when expected future demand is higher; 3. the effect of higher expected inflation on reducing real interest rates.

With a big enough multiplier a tiny increase is all you need. With an upward-sloping IS curve, the multiplier isn't finite.

MW seemed to be talking about something like this multiplier process (he used the word "exponential") early on in the paper, but he didn't really flesh it out.

[Scott Sumner]

Matt, That's all fine, but then my response is that there has never, ever, been a single monetarist in all of world history, at least as Woodford defines monetarism. No one believes that temporary swaps of cash for T-securities of any maturity have significant effects. I'll do a post to more fully explain.

I've been reading Woodford's paper. It's great, but he attacks a straw man in the Japan section. The debate is not at all as he claims.

[K]

JKH,

"when the Fed must pay interest on excess reserves"

In that case this whole conversation is moot. None of monetarist logic works at all unless the non-pecuniary convenience yield on money is offset by the pecuniary cost (FF-IOR) of holding it. How can money be a hot potato if it's like T-bills *and* it provides liquidity?

[JKH]

K,

Thanks.

I’m not sure I understand this at all.

When you say “this whole conversation is moot”, do you mean the case where the Fed must pay interest on excess reserves? Because that case is a certainty - if there are excess reserves of any magnitude more than the historic level (less than $ 10 billion) after what Woodford describes as “lift-off”. There’s just no question that that is what has to happen, if excess reserves are meaningful and we have “lift-off”. And the FF –IRR is going to be zero in that environment, or close enough to zero to be irrelevant to anything else that matters. There’s also no question about that. And that means that the idea of a “permanent” increase in the excess reserve component of the base means a permanent commitment to the zero bound, without lift off. In which case, the idea of a “permanent” increase in excess reserves is moot itself, if not meaningless.

Is that what you’re saying?

[wh10]

JKH, remember that older thread at NEP where Fullwiler and Sumner were debating about this (I believe you also chimed in)? If I recall correctly, Sumner refused to accept that "there’s just no question that that is what has to happen."

[JKH]

wh10

I don't recall that; but I know which side Fullwiler would take

I'm just trying to understand the motivation behind this idea of a permanent increase in the monetary base and whether it actually depends on the opposite view; maybe that's what K was saying but I'm not sure; it seems to be very ingrained in the literature from various accounts

It's also one area of Woodford's paper where I know I'm not sure what he's saying also - because in the rest of it he's quite remarkably close to a post Keynesian view of monetary operations - not all the way there, but fairly so

[K]

Scott: "No one believes that temporary swaps of cash for T-securities of any maturity have significant effects."

"Temporary" meaning "ending before the short rate rise above zero," right? I understand that you don't like to talk about rates, but in this case that seems like a concise and accurate way to quantify what you mean?

I think it's worthwhile considering the monetary mechanics of the moment the short rate leaves the zero bound. Lets assume the Market Monetarists are appointed to run the FOMC and you make some commitment to maintain the balance sheet at some level (presumably significantly reduced from the current level). Perhaps the level is where ever it was at the end of 2008 plus 5% per year until NGDP is on target (much like Milton Friedman would have done, *but* with the NGDP "target" in place). For the first while there are excess reserves. Since IOR is zero (otherwise this whole conversation is moot), that means that the FF rate must also be zero, or there is no way you can have excess reserves. Eventually, lets assume you reach your NGDP target. At this point the short rate will have to rise, or NGDP will rise above target. Since the short rate cannot rise with excess reserves in the system, you will presumably have guided total reserves down to the level of mandatory reserves at this point.

So what have you done? You have kept the short rate at zero until you reached your NGDP target. After you reach your target, you can no longer control the quantity of reserves since total reserves=mandatory reserves. So afterwards *all* you've got is short rate policy.

Before you exit the ZLB, the path of excess reserves was irrelevant. After exiting the ZLB, the path of excess reserves was flat at zero. Mandatory reserves are always irrelevant since they cannot be used for liquidity. So no relevance of reserves before or after exiting the ZLB.

JKH,

I wrote the above before reading your comment. I assumed that excess reserves would have to be taken to zero at the moment of liftoff. I think that can be done quite easily. As Woodford points out, the BoJ took off massive quantities of QE in a matter of months, so its all highly reversible.

We can run your scenario in which IOR=FF and excess reserves can stay on post liftoff. But then where's the hot potato if money yields T-bills + liquidity convenience??? So no present hot potato and no future hot potato. Again no relevance for the post liftoff path of quantity.

[JKH]

K,

We may be saying the same thing; I don't know.

You can't have excess reserves of any material amount after "lift-off" unless you pay interest on excess reserves. And that for all purposes has to be the target rate for fed funds.

I don't question the ability to drain excess reserves prior to lift-off, although the Fed certainly isn't going to do that. They're going to pay interest and work the reserves off gradually as they increase the funds rate. They've signaled that ages ago. The whole idea of "exiting" the excess reserve position is a non-issue as far as I'm concerned. They can lift the funds rate as high as they need to with or without excess reserves.

So I still don't understand why this idea of a permanent increase in "the base" is relevant. If its permanent after lift-off, you must pay interest on excess reserves. If its permanent without lift-off, who cares, because that means ZIRP is permanent.

Nick, perhaps you understand the question I'm asking?

[K]

JKH,

"We may be saying the same thing; I don't know."

We are.

"You can't have excess reserves of any material amount after "lift-off" unless you pay interest on excess reserves."

Correct.

"So I still don't understand why this idea of a permanent increase in "the base" is relevant."

It's *not* relevant. As I said, there's no relevance for the path of the quantity of reserves under all possible scenarios. The quantity of excess reserves either doesn't matter *or* it's zero and it's the short rate that matters.

We are in 100% complete agreement.

[K]

JKH,

Maybe my comments were confusing if you read them with the assumption that IOR=FF. I assumed IOR=0. Otherwise, as I said, the whole thing is moot, which is your point. I don't see how you can have *any* kind of monetarism if IOR=FF. Scott has claimed to me that it doesn't matter. I don't think I've ever heard Nick elaborate a clear position.

[JKH]

K,

Excellent. Thanks.

[Ritwik]

Nick

But changing expectations about future hot potatoes is supposed to make existing potatoes hotter. The actual future potatoes do not come into picture, expect in a passive *demand-determined* way - which is then consistent with the actual operations of the monetary system. You can talk about unexpected actual base changes- where the actual quantity is important - or talk about changes in expectations, where the quantity is not important. Unless you have a model/ theory that links the amplitude of the signal with the actual quantity of money, you can't talk about both at the same time. But your theory is explicitly that the amplitude of the signal is not related to the size of the base.

Nothing is a hot potato if it is anticipated.

[Ritwik]

Nick

I am being so nitpicky about this because Scott and you always seem to like what the other has to say (he acknowledges hot potatoes and your Chuck Norris is very ratex-y) yet the *revealed preferences* that you display when actually making your formal arguments is very different. Scott always talks about asset markets,and you always talk about operations of the payment systems. The asset markets view of the world is compatible with the mainstream view of the world, though the explanation is different (expected inflation and income vs wealth effects/ Tobin's q). The hot potato operations view of the payment systems is not compatible with the mainstream, and not compatible with Chuck Norris/ Scott Sumner/ ratex. The *only* way I can reconcile them in my mind is through expected inflation/income, or making existing potatoes hotter. Do you agree with this?

[Ritwik]

K, JKH

It should not all be surprising that Woodford seems to sound almost post-Keynesian, though I would say that someone like Mike Sproul would find more common ground with him than a post-Keynesian.

As Perry Mehrling so wonderfully illustrates, Woodford's Interest & Prices, like all the great monetary theory texts of their generations, was written in response to the challenge presented by contemporary developments in the macro-economy/ monetary system. In Woodford's case, the challenge was not simply interest rate rules, but Fischer Black and the challenge of finance. People tend to think that Woodford is simply Taylor++ because he has Keynesian models where fiscal policy works and is somewhat a darling of the saltwater crowd, but he has also written extensively on the fiscal theory of the price level, where Cochrane finds common ground with him. His is a *banking school* view of the world - he believes in reflux (suitably modified in today's world as Wallace neutrality), he models systems where bank deposits, reserves andindeed all forms of money pay interest. But while the Post-Keynesian views are closest to someone like Alvin Hansen (banking school ergo monetary authority should be passive and fiscal authority dominant) Woodford's views are probably closest to someone like Edward Shaw(who was a fiscal conservative), with a bit of Keynes thrown in.

Here's the source(s) of my views :

http://economics.barnard.edu/sites/default/files/inline/the_money_muddle.pdf

http://economics.barnard.edu/sites/default/files/inline/mr_woodford_and_the_challenge_of_finance--revised.pdf

[JKH]

thanks, Ritwik

he seems to have gargantuan status in macro, from the reactions I see to his paper

what would you attribute that to?

his capacity to span freshwater, saltwater, etc.?

other?

[JKH]

Ritwik,

I took a quick scan of the second link paper and it looks interesting, but I couldn't find anything at the first link.

On the second, I read the section where Mehrling describes the nature of the channel system as the CB making an unlimited two way market, which is pretty familiar to me. That reminds me of a question I have about the Woodford paper for you or anybody here.

As I understand it, Woodford is pretty sceptical of the “portfolio balance” theory of interest rate determination, as it relates to LSAP effects on bond yields.

Suppose the CB makes an unlimited two way market in the 10 year bond, just the way it would in a channel system for the overnight rate. Then there’s no question that it determines the yield on the 10 year, just as it does for the overnight rate in a channel system. That would seem to be the limiting case of the portfolio balance model, the way I see it, with respect to LSAP intervention at 10 years.

My question is – where’s the inflection point between not believing in and believing in the portfolio balance model, with respect to the size of an LSAP intervention? And if that sort of function is continuous (which it seems to me it logically should be), then doesn’t that support the portfolio balance model in effect, qualified by the size of the intervention?

[Ritwik]

JKH

Pg 298 in the first paper gives the typology of monetary thought, one which I have found very useful.

Will think about your question later.

[Scott Sumner]

K, I've never claimed the quantity of reserves is important, what matters is the monetary base. After you exit the zero bound, base money becames a hot potato. The Fed controls the base in such a way as to hit their NGDP target. The don't do that by directly targeting the base, but rather endogenously adjusting it to keep expectations of NGDP growing at about 5% per year.

I do agree that ERs are roughly zero once short rates rise above zero, if there is no IOR.

[K]

Scott: "I've never claimed the quantity of reserves is important, what matters is the monetary base...The Fed controls the base in such a way as to hit their NGDP target."

OK. But in order for the hot potato effect to work, there needs to be some point in time (now or in the future), in which the Fed controls the quantity of some *non-interest bearing* money. If it's interest-bearing it can't be a hot potato. But they have no control over the quantity of currency, which is entirely at the discretion of the representative household. What they do control is the total base, i.e. the sum of reserves and currency.

If we have IOR away from the ZLB, then currency will just be converted to reserves as interest rates rise. Why would I spend it, if I can just convert it to reserves and back to currency if I want to spend it later?

If IOR=0, but FF is above zero, there will be no excess reserves. The CB is forced to buy back any excess reserves in order to maintain FF above IOR. So now if people feel they have too much currency they will convert to reserves and the CB will convert that to T-bills. So instead of an equilibrium between currency and reserves, there's an equilibrium between currency and T-bills. Makes no difference. The quantity of non-interest bearing money is chosen exclusively by the representative household as a function of its preference for interest vs relative liquidity of currency over deposits/t-bills. There is simply no state of the world in which the Fed controls the quantity of anything that can conceivably be described as "high-powered money."

[Ritwik]

K

The absolutely only way I can make sense of the hot potato story is through increase in inflation (or increase in income), thereby making existing (or even the path of) monetary potatoes hotter than is previously anticipated. It's about reducing money demand, not increasing base money. I'm never quite sure what Nick and Scott are on about when they talk about expectations and the hot potato, if they don't mean this.

[Bill Woolsey]

For anyone for whom it isnt' obvious, the gold price mechanism is that the central bank buys at sells gold at a target price, with that price being adjusted regularly--perhaps daily as Nick said, but every 6 weeks would be natural.

I would assume that such a central bank would do ordinay open market operations in bonds so that its actual gold reserves are near zero. So, if it raises the price of gold and people actually sell gold to the central bank, the central bank would undertake open market operates until people buy that gold back from the central bank. And vice versa. If people buy gold from the central bank, it would then sell government bonds until people sell that gold back to the central bank.

Now, in a liquidity trap, interest rates on some kind of government bonds might be driven to zero, but the central bank could still make periodic increases in the price of gold. However, I think the "problem" would show up as the central bank accumulating gold reserves and not being able to get rid of them. It would be buying government bonds and expanding base money. The interest rate on those bonds would stay at zero.

Now, assume that the target was unchanged--say a nominal GDP growth path. Nominal spending is below target right now, but everyone beleives that we will eventually get back to target. And that belief has impacts now. It is one reason why nominal GDP is not quite as far below target as it otherwise be--say if we had inflation targeting.

Further, everyone believes that the price of gold in the future will be set to keep nominal GDP on target.

The issue, then, is whether having the price of gold increase now will do any good, when nominal GDP is below target, despite the expectation it will eventually return to target. While people may not be sure as to whether any current increase in the price of gold will be eventually reversed, they know it will be reversed if it would result in nominal GDP going above target.

Do temporary increaces in the price of gold at the central banks window cause spending on output to rise, when the T-bill yield has already been driven to zero?

I think there is an income effect for those in the gold mining industry. And don't those folks who would have purchased gold, but who are deterred by the high price purchase other goods instead?

If it is true that the central bank cannot get rid of the gold reserves, does the risk of capital loss on those gold reserves translate into less fiscal transfers to the treasury and so higher taxes and so a desire to accumulate more money to pay the taxes?

Anyway, it is an intersting thought experiment.

[Nick Rowe]

Bill: "I think there is an income effect for those in the gold mining industry. And don't those folks who would have purchased gold, but who are deterred by the high price purchase other goods instead?"

Yes, exactly. If we were in my "goldpunk" alternative history, people would not be talking about the transmission mechanism from the overnight rate rippling out through other interest rates and asset prices to consumption and investment demand. Instead they would be talking about a rise in the price of gold encouraging gold mining, and rippling out through silver and platinum and other metals, and all the other goods etc. etc. And in that alternative transmission mechanism people would argue about whether or not the monetary hot potato is or is not an essential part of the story, just as they do with interest rates, and argue about the role of expectations etc.

[Nick Rowe]

Ritwik: "The hot potato operations view of the payment systems is not compatible with the mainstream, and not compatible with Chuck Norris/ Scott Sumner/ ratex. The *only* way I can reconcile them in my mind is through expected inflation/income, or making existing potatoes hotter. Do you agree with this?"

Good question. I think they are compatible. I think they are different aspects of the same thing. But it is currently beyond my abilities to provide some over-arching perspective that could capture both at the same time. It's too damned hard to think about this stuff.

[David Beckworth]

Thanks Andy and Matt. It seems to me that one of the practical difficulties of tying the NGDP target to an interest rate interest instrument is how clearly the Fed's intentions can be signaled. Would the public and markets really respond forcefully from the Fed saying its policy rate will be kept low until some NGDP target is hit? Although different, the experience from the Fed's long-term ffr forecasts don't give me hope. The problem with this approach is that is not clear how to interpret the stance of monetary policy without knowing the natural rate. Maybe if the Fed published the natural rate too it would help, but then it gets even harder for the public to understand. They have to know the conditional path of the ffr, the expected path of the natural rate, and the NGDP target.

If the Fed did conditional LSAPs every week tied to a NGDP target that would be far easier for the public to understand. The experience from the past QEs and expected future ones (e.g. Draghi's comments about doing whatever is necessary) make me think this approach would pack more of a punch. It would provide the slap to the market's face needed to catalyze robust nominal spending.

[Alex Godofsky]

@K

If IOR=0, but FF is above zero, there will be no excess reserves. The CB is forced to buy back any excess reserves in order to maintain FF above IOR. So now if people feel they have too much currency they will convert to reserves and the CB will convert that to T-bills. So instead of an equilibrium between currency and reserves, there's an equilibrium between currency and T-bills. Makes no difference. The quantity of non-interest bearing money is chosen exclusively by the representative household as a function of its preference for interest vs relative liquidity of currency over deposits/t-bills. There is simply no state of the world in which the Fed controls the quantity of anything that can conceivably be described as "high-powered money."

I think the hidden piece of this story is the "excess" in excess reserves. What if, concomitant with the rise in the interest rate, the quantity of required reserves increases to absorb the entire excess (i.e. bank lending increases)? What if when the central bank even implements a "permanent" monetary injection by promising not to raise rates until this has happened?

If you had a central bank that didn't target interest rates at all, that completely ignored them, that's precisely what would happen if the central bank promised a permanent expansion of the base while rates would zero - they would stay zero for a while, and eventually rise above zero after all the excess reserves had become required reserves.

[Ritwik]

Nick

Off the topic, but have you read Tyler's post on the rise of barter in Spain? Well it's not truly barter, more like parallel currencies but an important bit was about some payments being done in 'hours' rather than euros.

You must have noticed how I often talk about the atemporal exchange vs. inter-temporal coordination theories of recessions. I think this provides a neat natural experiment.

If the atemporal exchange theory is more likely, we should see these localised parts of the Spanish economy overcoming the recession. If the inter-temporal coordination theory is more likely, this may not happen and the 'hours' might be hoarded.

We might never know because post-facto data will might be sparse andmixed with many other signals, but do you agree that this is indeed a good test of Wicksellian vs nominal theories of the recession?

[Nick Rowe]

Ritwik: No I hadn't. (I've been otherwise occupied recently). But it sounds important. Thanks for the tip. Need to get my brain going before even thinking about the answer to your question.