Interest rate control as beta-anachronism

I want to make a minor point to follow up on something important that David Glasner said:

"Nevertheless, our basic mental processes for understanding how central banks can use an interest-rate instrument to control the value of money are carryovers from an earlier epoch when the value of money was determined, most of the time and in most places, by convertibility, either actual or expected, into gold or silver. The interest-rate instrument of central banks was not primarily designed as a method for controlling the value of money; it was the mechanism by which the central bank could control the amount of reserves on its balance sheet and the amount of gold or silver in its vaults. There was only an indirect connection – at least until the 1920s — between a central bank setting its interest-rate instrument to control its balance sheet and the effect on prices and inflation. The rules of monetary policy developed under a gold standard are not necessarily applicable to an economic system in which the value of money is fundamentally indeterminate."

David does not use the word "anachronism", but I think he's saying that central banks' use of an interest rate instrument (and our thinking of monetary policy in this way) is an anachronism. It made sense under the old gold exchange standard, but doesn't make sense now.

But I want to go a little bit further than what (I think) David said. An anachronism is something that's in the wrong time. There must be something equivalent to "anachronism", only in the wrong place. Like a Brit coming to Canada, and continuing to drive on the left. [Update: Brett in comments tells me the word I'm looking for is "anatopism", which is a totally new word for me.] Central banks' using interest rate control is wrong both in time and in place. [It's both an anachronism and an anatopism.] Let me try to explain why those two sorts of wrongness are very similar, in this case.

A central bank on the gold exchange standard issues monetary liabilities that it promises to convert at a fixed exchange rate for gold. It is the central bank's responsibility to conduct its operations to ensure that its promise is credible so its monetary liabilities always have a fixed value in terms of gold.

A commercial bank today issues monetary liabilities that it promises to convert at a fixed exchange rate for central bank notes. It is the commercial bank's responsibility to conduct its operations to ensure that its promise is credible so its monetary liabilities always have a fixed value in terms of central bank notes.

See the similarity between a central bank on yesterday's gold exchange standard and a modern commercial bank? They are both what I call "beta banks". Betas follow alphas. Central banks under the gold exchange standard were beta banks for whom gold was alpha. Commercial banks today are beta banks for whom central bank notes are alpha.

What ultimately determines the value of a beta bank's monetary liabilities is convertibility at a fixed exchange rate. Convertibility at a fixed exchange rate is what creates the nominal anchor. Even if convertibility is temporarily suspended, the expectation that convertibility will eventually be restored pins down the expected future value of a central bank's currency, which plays a large part in limiting movements in its current value. Any interest rate set by a beta bank plays a purely subsidiary role, in helping to ensure it maintains sufficient reserves, or access to sufficient reserves, relative to the size of its monetary liabilities, to ensure it can maintain convertibility.

But modern central banks are alpha banks. They do not promise to redeem their monetary liabilities at a fixed exchange rate to some alpha good.

But when a modern central banker looks back in time to the gold exchange standard, or across in space to commercial banks, it must look very natural to think in terms of interest rate control. Adjusting interest rates is what banks do, and I'm running a bank, so that is what I must do too.

No, you are not running a bank; you are running a gold mine. You are the alpha leader, not a beta follower. The sort of tactics that make sense for a beta follower don't work the same way for you.

When a beta bank adjusts an interest rate, on its loans or deposits, it is not trying to control the value of its monetary liabilities, it is trying to control the size and composition of its balance sheet. Convertibility is what controls the value of its monetary liabilities. When an alpha bank uses only interest rate instruments, it is one instrument short of a full set.

Comments

[Tom Brown]

Nice post Nick. A couple of things...

"...I'm running a bank, so that is what I must do too."

Wait, *you're* running a bank? There's more to you than meets the eye Nick.

"Any interest rate set by a beta bank plays a purely subsidiary role"

In your opinion then, in the US, 2008 and prior, when the "policy instrument" was a federal funds rate, can we say that the more fundamental policy instrument was the monetary base (MB)?

[JP Koning]

What about a commercial bank, say BMO, that ceases to directly convert deposits into Bank of Canada dollars but still promises to enforce a 1:1 rate between BMO deposits and BoC dollars in the secondary market? BMO would still be a beta bank following the alpha, the Bank of Canada. It would enforce the 1:1 promise by altering the interest rate it offers, sweetening that rate if BMO deposits fall to a discount and vice versa. BMO's interest rate no longer plays a subsidiary role but is now serving as the principal means by which to enforce BMO deposits' purchasing power.

The BoC is comparable to the above sort of commercial bank, not a commercial bank that enforces strict convertibility. Just as BMO uses interest rates to ensure that its deposits track the value of a reference instrument, in this case the dollar, the BoC uses rates ensure that BoC dollars track the value of a reference instrument, in this case an inflation index.

The BMO in my example never actually existed. So when a modern central banker looks "across in space space" to commercial banks, he/she won't see anything recognizable.

[Nick Rowe]

Tom: well, sometimes I *pretend* I'm running a central bank.

"In your opinion then, in the US, 2008 and prior, when the "policy instrument" was a federal funds rate, can we say that the more fundamental policy instrument was the monetary base (MB)?"

That would be one way to look at it.

JP: something tells me you are leading me into a trap of my own making.

Let me first acknowledge that there is indeed a "tension" ("contradiction" is too strong a word) between that old post and this new one. But I think there is a difference between tactics and strategy.

Suppose your imaginary BMO targeted not the level of the exchange rate, but the rate of change of the exchange rate. And suppose (plausibly) that that exchange rate is a perfectly flexible price, so there is no stickiness to help pin it down. For any equilibrium path under one path of interest rates set by the BMO, there exists another equilibrium path where BMO monetary liabilities double in quantity and halve in value. And there is nothing to prevent jumps from one equilibrium path to another.

[Tom Brown]

Thanks Nick. In your opinion then, would it be legitimate to try to determine how effective MB was as a policy instrument prior to 2008 vs after? Say in terms of % change in MB causing % change in NGDP?

[Nick Rowe]

Tom: the magnitude of %deltaNGDP/%deltaMB is not a good metric (unless paper and ink are scarce).

You know those railway things, without a motor, where you stand on them and work a handle up and down to make them move? I stood on one the other day. The crank was at the top of its travel. Pressing the handle down could have made the thing move in either direction, or not move at all. Not a good policy instrument. Neo-Fisherian railway.

[Tom Brown]

... Nick, just to be clear "the magnitude of %deltaNGDP/%deltaMB is not a good metric" goes for both pre-2008 and post-2008 (even if considered individually, and not in a comparison), correct?

[Nick Rowe]

correct. (Unless paper and ink are expensive, so "bang per buck" is important, which it isn't.)

[JKH]

“David does not use the word "anachronism", but I think he's saying that central banks' use of an interest rate instrument (and our thinking of monetary policy in this way) is an anachronism. It made sense under the old gold exchange standard, but doesn't make sense now.”

In terms of thinking about monetary policy, I’d say the opposite. David refers to interest rates only infrequently in his great series about the gold standard (although interest rates were an important and sometimes essential tool in managing reserve levels). He does refer to the price or value of gold continuously (although not so much the value of money) in describing monetary policy.

“A commercial bank today issues monetary liabilities that it promises to convert at a fixed exchange rate for central bank notes. It is the commercial bank's responsibility to conduct its operations to ensure that its promise is credible so its monetary liabilities always have a fixed value in terms of central bank notes.”

Nick, however that statement is sufficient or relevant, you can make it about commercial banks under a gold standard. There was no substantial difference under a gold standard for central bank note operations as far as I know. Notes were issued alongside gold operations.

Something seems lop-sided or at least incomplete in your analogy.

“See the similarity between a central bank on yesterday's gold exchange standard and a modern commercial bank?”

No. I see the similarity of commercial banks in each era with respect to the operation of banknote management. Gold is separate. And therefore it makes for questionable sense (to me) to analogize - in effect - the operation of gold under a gold standard system to the operation of banknotes under a gold standard system.

[Scott Sumner]

Excellent post. During the Taylor Rule period the fed funds rate was an epiphenomenon that had just enough relationship (in the short run) to the base to allow it to be useful. Since 2008 that relationship has broken down, and monetary policy has been adrift.

[Max]

Under the gold standard, changing the fixed gold price was possible but frowned upon, and not routine. Likewise it's possible for a modern central bank to intentionally change the price level (not counting the usual drift), but it's frowned upon and not routine.

[W. Peden]

"No, you are not running a bank; you are running a gold mine. You are the alpha leader, not a beta follower. The sort of tactics that make sense for a beta follower don't work the same way for you."

I love this way of thinking about modern central banking, and especially how it links to some of your older posts on CPI inflation targets as a kind of "basket-of-goods standard".

[Tom Brown]

Nick, thanks. What confuses me though is squaring that with this multipart series on measures like industrial production in which he found strong evidence of causality (from changes in MB). What am I missing?

[Market Fiscalist]

I'm not clear why interest rate setting is optimal even for a CB under the gold standard.

A modern CB targeting NGDP or inflation could just adjust the money supply as needed to hit the target and let interest rates land where they will.

Doesn't a CB operating under a gold standard where people can convert currency into gold at a fixed exchange rate also have to adjust the money supply to maintain this rate? Why is setting interest rates any more optimal for it compared to just adjusting the money supply (via assets sales/purchases) than they are for a modern CB ?

[Nick Rowe]

JKH: "Nick, however that statement is sufficient or relevant, you can make it about commercial banks under a gold standard. There was no substantial difference under a gold standard for central bank note operations as far as I know."

If I'm understanding you correctly (as saying I could have said exactly the same thing about commercial banks under the gold standard) I strongly agree. A commercial bank operates much the same way, whether or not it is under the gold standard. In fact, if I read George Selgin correctly, you don't really need a central bank under the gold standard (and Scotland and Canada didn't really have one, and the BoE only became "central" because of peculiar legislative privileges and restrictions). Central bank paper replaced gold as alpha, but commercial banks carried on much the same way.

Thanks Scott! But I think I'm just adding a footnote to David Glasner.

Max: "Under the gold standard, changing the fixed gold price was possible but frowned upon, and not routine."

Yes. But we can imagine an alternate history where Irving Fisher's "Compensated Dollar Plan" was implemented, and central banks adjusted the price of gold more or less continuously (monthly??) to ensure the price level target (or inflation target, or NGDP target?) was hit. So the price of gold became the instrument which central banks used to hit their ultimate target, with interest rates playing a very minor role.

WP: thanks!

Tom: Dunno. To be honest, I like Mark's conclusions, because they match my prejudices, but at the same time I am very wary of VAR studies of causality, because economics is not like engineering. People have expectations, and all that.

[Nick Rowe]

MF: that's a question I didn't explore in this post. Simple crude answer: if a gold standard bank had 100% gold reserves, interest rates on loans would be irrelevant, because it wouldn't make any. Interest rates on deposits would still be relevant. Presumably it would pay negative interest, to cover its operating costs, and by adjusting interest rates on deposits it could expand or contract its quantity of deposits. With less than 100% reserves, interest rates on loans (or the spread between loan and deposit rates) would be one way to adjust its actual reserve ratio.

[JKH]

Market Fiscalist

CBs under a gold standard need interest rates as a tool in managing net gold flows in and out of their gold reserve positions - according to whatever self-imposed constraints they're under for gold reserve requirements. Interest rates affect the carrying cost of holding gold and therefore the supply and demand for gold as between monetary and non-monetary holdings.

Highly recommend David Glasner's recent series on the operation of the gold standard in the 1920's.

[Nick Rowe]

MF: What JKH (and David Glasner) said.

[Too Much Fed]

"A commercial bank today issues monetary liabilities that it promises to convert at a fixed exchange rate for central bank notes. It is the commercial bank's responsibility to conduct its operations to ensure that its promise is credible so its monetary liabilities always have a fixed value in terms of central bank notes.

See the similarity between a central bank on yesterday's gold exchange standard and a modern commercial bank? They are both what I call "beta banks". Betas follow alphas. Central banks under the gold exchange standard were beta banks for whom gold was alpha. Commercial banks today are beta banks for whom central bank notes are alpha."

JKH, is that right? It seems to me with elasticity of currency and "crisis lending" if there is a bank run on ***solvent*** commercial banks that demand deposits of the commercial banks are "alphas".

I see a similarity with the gold standard in this way.

Gold starts growing 12% per year producing 7% per year price inflation above a 2% per year price inflation target.

Now replace gold with demand deposits from solvent commercial banks. Demand deposits from solvent commercial banks start growing 12% per year producing 7% per year price inflation above a 2% per year price inflation target.

If that happens, the central bank can hope interest rates rise enough, or it can raise the "fed funds rate" high enough to invert the yield curve.

It seems to me there is a demand deposit of the solvent commercial banks (instead of gold) standard with a 2% per year price inflation target.

[Market Fiscalist]

JKH,

Thanks.

On "Interest rates affect the carrying cost of holding gold and therefore the supply and demand for gold as between monetary and non-monetary holdings.". I see that a CB faced with an outflow of gold from its reserves because desire to hold its currency has fallen could raise interest rates, make holding money more attractive, and stem the outflow.

But won't it use OMO to increase interest rates, and OMO works by changing the size of the money supply ?

So if a beta CB never wanted to have an outflow of gold it could say "We will use OMO to adjust the size of the money supply to maintain the gold exchange peg, and we don't care what this does to interest rates". I still don't really see how this differs from what an alpha CB targeting NGDP or inflation could do right now. Perhaps the distinction I'm missing is that there is correlation for a beta CB between interest rates and flows of a real asset (gold) that doesn't exists under an alpha CB - but as interest rates are changed by adjusting the money supply for both alpha and beta CBs I'm not sure why this distinction is important.

[Too Much Fed]

I believe that under my scenario the commercial banks and the central bank are "alphas".

[JKH]

Market Fiscalist:

“But won't it use OMO to increase interest rates, and OMO works by changing the size of the money supply?”

Insofar as just the CB mechanics are concerned, in the pre-2008 Fed regime for example, OMO changed the size of bank reserves, but the required order of magnitude was pretty small in context. Short term interest rates were hyper-sensitive to relatively tiny changes in the system excess reserve setting controlled by the Fed. And the announcement of a Fed Funds target increase alone did much of the work in steering the market toward the new interest rate level, reducing the amount of OMO required, which was already relatively small.

In a post-2008 regime with IOR, OMO is not required to increase the Fed policy rate. The IOR and the target Fed funds rate increases are just announced jointly (or at least that’s the way it will work until QE reserves are gone, which will take a number of years.)

[JKH]

MF,

This may be of interest, if you haven't seen it:

http://worthwhile.typepad.com/worthwhile_canadian_initi/2012/04/from-gold-standard-to-cpi-standard.html

[Market Fiscalist]

JKH,

Those mechanics are interesting, and expectations set by the CB clearly are a major part of how it works (I hadn't realized quite how significant those are).

But I don't think that changes my fundamental point that (whatever the mechanics and the role of expectations) the way it works between a beta CB trying to maintain a gold-peg with no gold-loss and an alpha CB trying to hit a self-set target, the tools available to them don't seem wildly different to me.

I think that is what that (excellent) post you linked-to shows, so perhaps we agree.

[JP Koning]

"For any equilibrium path under one path of interest rates set by the BMO, there exists another equilibrium path where BMO monetary liabilities double in quantity and halve in value. And there is nothing to prevent jumps from one equilibrium path to another."

Even though BMO deposits are not convertible on demand, they may have a senior fixed claim on BMO's assets in the event that the bank is wound up. Would that pin down their current value and prevent jumps?

[Felipe]

Hmm, but isn't the central bank beta to the CPI basket? Or is your argument that because it targets a rate and not a level that it becomes alpha?

[Frank Restly]

Nick,

"But modern central banks are alpha banks. They do not promise to redeem their monetary liabilities at a fixed exchange rate to some alpha good."

Can government bonds be considered an alpha good? True the exchange rate between a central bank's monetary liabilities and a government's bonds can float. However, I can't imagine a central bank denying a request by a beta bank if that beta banks asks that the central bank buy some of it's government debt holdings.

For instance, if I (beta bank) buy a 30 year bond from the government at 5% and several years later, 30 year government bonds are being sold with an 8% interest rate, what price does the central bank pay if it buys my 5% bond from me? Likewise, If I (beta bank) buy a 30 year bond from the government at 8% and several years later, 30 year government bonds are being sold with a 5% interest rate, what price does the central bank pay if it buys my 8% bond from me?

I would suspect that the central bank is market price indifferent and always buys bonds from the market and sells them into the market at par. The reason I suspect this is that the central bank (as a lender of last resort) has an interest rate tool that it can use independently of open market operations. The central bank can simultaneously buy bonds at par (increase in liquidity) and raise interest rates (discount window rate). Likewise it can simultaneously sell bonds at par (decrease in liquidity) and lower interest rates (discount window rate).

Is the central bank a beta bank to the federal government?

[JKH]

MF,

I personally find the alpha-beta model a bit awkward in the interpretation, but we probably agree.

[Nick Rowe]

JP: "Even though BMO deposits are not convertible on demand, they may have a senior fixed claim on BMO's assets in the event that the bank is wound up. Would that pin down their current value and prevent jumps?"

I don't think so. Again, double the quantity, and halve the value of each, (just like a 2-for-1 stock split), and the senior fixed claim of the total BMO deposits stays the same, in real terms.

Felipe: you could think of the central bank's notes as beta to the CPI basket's alpha, if the central bank promised to directly redeem its notes for CPI baskets, on demand. It would then be exactly like under the gold standard, except there would be CPI basket reserves in the central bank's basement, not gold.

Frank: "Can government bonds be considered an alpha good?"

No. Bonds are beta.

[JKH]

MF,

Another stray thought about a related technical issue while I’m on the topic.

If you go back in history covering about 30 or 40 years leading up to the 2008 Fed regime change, you will find absolutely no correlation between the level of excess reserve balances in the banking system and the level of interest rates. Zero correlation. I guarantee it.

Yet the level of excess reserve balances is what the Fed controlled in order to maintain the Fed funds rate at its chosen level – and/or to change it. And the level of excess reserves balances is what is most directly impacted by OMO.

(I’m not saying anything there about either the level of currency in circulation or the level of commercial bank deposits.)

[Brett]

The word for a situation out of place is anatopism.

[Nick Rowe]

Brett! Just the person who would know the answer! I knew there had to be a word, but I had never heard of "anatopism". English is an amazing language. Post updated. Thanks!

[JP Koning]

"Again, double the quantity, and halve the value of each, (just like a 2-for-1 stock split), and the senior fixed claim of the total BMO deposits stays the same, in real terms."

Ok, but a doubling in quantity is usually carried out by BMO getting an equivalent number of loans and mortgages in return. So a doubling in quantity is not like a stock split. Wouldn't that pin down the value of non-convertible BMO deposits?

[Nick Rowe]

JP: double the nominal value of loans, measured in BMO dollars, and halve the nominal value of the BMO dollar, and it's the same number of bricks and mortar.

[e abrams]

anaholic ?
from greek holos, whole or complete