John Cochrane on Monetary Policy with Interest on Reserves

Central banks are usually owned by governments, and transfer their seigniorage profits to those governments. This creates a link between monetary and fiscal policy. But the presence of that link says nothing about the direction of causation.

John Cochrane (pdf) says that central banks' paying interest on reserves allows them to conduct monetary policy independently of fiscal policy. It breaks the link. He sees it as a game-changer.

John's approach to this question is based on the Fiscal Theory of the Price level. I want to approach this same question from a totally different perspective. I will use a very standard Quantity Theory perspective and replicate his results. It's quite simple really. If the central bank increases the growth rate of the money supply, but issues new money as interest on old money, there will be the standard increase in the equilibrium inflation rate, but without the standard once-and-for-all jump in the price level due to lower equilibrium real money balances. And there will be no effect on the central bank's seigniorage profits, and so there is no link to fiscal policy. I think that's what's going on in his model.

Since increases in the money growth rate and increases in interest rates on reserves would always move together under that sort of monetary regime, it would look like increases in interest rates on reserves cause increases in inflation rates. But they don't, of course. If we increased interest on reserves, but held the money growth rate constant, the equilibrium price level would fall.

The Bank of Canada has been paying interest on reserves for 20 years. And when it wants to increase the inflation rate, because it fears inflation will fall below the 2% target if it does nothing, the Bank of Canada lowers the interest rate on reserves. And the Bank of Canada seems to have gotten the sign right, because it has hit the 2% inflation target on average, which would be an amazing fluke it if had been turning the steering wheel the wrong way for the last 20 years without going off in totally the wrong direction. Interest on reserves is old stuff in lots of other countries, outside the US. It's not a game changer for us.

Take a simple version of the Quantity Theory. The central bank sets the money supply Ms. There is a money demand function: Md is proportional to the price level P, a positive function of real income Y, and a negative function of the opportunity cost of holding money. That opportunity cost is the difference between the market nominal interest rate Rb and the nominal rate of interest paid for holding central bank money Rm. Ms=Md=P.L(Y,Rb-Rm). The price level P equilibrates the supply and demand for money. Market nominal interest rates adjust to expected inflation.

Suppose the central bank unexpectedly permanently doubles the supply of money. The price level would double. If the central bank buys assets with the new money, the central bank's real seigniorage profits will increase, at the expense of the one-time inflation tax on individuals who held money when the price level doubled.

Suppose the central bank unexpectedly permanently doubles the money supply by using a helicopter drop. The price level would double, but there would be no change to the central bank's real seigniorage profits. Lucky individuals would gain more from the helicopter than they lost in the inflation tax, and unlucky individuals would gain less.

Suppose the central bank unexpectedly permanently doubles the money supply by doubling each individual's stock of money. We can think of this as like a one-shot payment of new money as interest on old money. The price level would double, but there would be no change to the central bank's real seigniorage profits. Each individual would gain in interest exactly what he lost from the inflation tax.

Suppose the central bank unexpectedly announced that one year from today it would permanently double the money supply by doubling each individual's stock of money. (Again, there will be a one-shot payment of new money as interest on old money). The announcement would have no effect on the price level today. But the doubling of the money supply one year from today would cause the price level to double on that day. (No individual has an incentive to get rid of money the day before the doubling, because what he loses from the inflation tax he gains from interest on money, so it's a wash.) Again, there is no effect on the central bank's real seigniorage profits. It's like an announced currency reform, where two new dollars replace one old dollar.

Suppose the central bank announces that it will increase the money supply growth rate by 10% per year, and will do so by paying daily interest on money at the rate of 10% per year. The announcement would have no immediate effect on the price level, but it would have an immediate effect on the rate of inflation, which would increase by 10% per year. There would be no effect on the central bank's real seigniorage profits, or on the real stock of money. The reason is that the interest rate differential between money and other assets is unchanged, so the opportunity cost of holding money is unchanged. All nominal interest rates will rise by the same 10% from the Fisher relationship. It's like an announced ongoing currency reform.

An increase in the growth rate of the money supply will cause an equal increase in the rate of inflation. One way to issue new money is to use it to pay interest on existing money, which increases the demand for money. An increase in the growth rate of the money supply by itself will cause an increase in the inflation rate and a once-and-for-all jump in the price level and (unless we are on the wrong side of the Laffer curve) an increase in real seigniorage profits. An increase in the growth rate of the money supply plus a pari passu increase in the interest rate paid on money will cause an increase in the inflation rate but no jump in the price level, and no effect on real seigniorage profits.

What these examples show is that by issuing new money in the form of interest on existing money, and so adjusting the money growth rate and interest rate on money pari passu, the central bank can control the inflation rate without having any effect on its real seigniorage profits.

If we further assume that the central bank is owned by the government, and that the central bank's seigniorage profits are (sooner or later) transferred to the government, and so affect the government's fiscal policy, we get a further implication. If the central bank adjusts the money supply growth rate and interest paid on money pari passu, it can control inflation with zero implications for the government's fiscal policy. Monetary and fiscal policy can be independent of each other.

But then we could say exactly the same thing if the central bank dropped the new money out of a helicopter, or gave it away to its favourite charity. Unless we define that as "fiscal policy". But then we could define paying interest on money as "fiscal policy" too.

John Cochrane says (page 10): "In sum, then, in this frictionless model, the government can set a nominal interest rate target; it can set that target without any adjustments to fiscal policy {E(t)S(t+j)} [expected future primary fiscal surpluses NR], and by setting nominal interest rates, the government can control expected inflation. This is the central result, which we need to digest and interpret."

This is my interpretation: If interest on old money is always paid with new money, and an increase in the rate of interest always means an equal increase in the expected growth rate of the money supply, it will cause an equal increase in the expected inflation rate (but will have no effect on the current equilibrium price level).

A little later (page 11) he says: "Even a completely fixed interest rate target does not lead to inflation instability or indeterminacy."

Yes, if a fixed interest rate target meant a fixed money growth rate target, because new money is always issued as interest on existing money, monetarists would not say this leads to explosive time paths for inflation or price level indeterminacy.

This result has nothing to do with the Fiscal Theory of the Price Level. It follows equally well from standard Quantity-Theoretic reasoning. This result has nothing to do with the presence or absence of monetary frictions that motivate monetary exchange and a demand for money when money pays less interest than other assets like government bonds. (It does depend in part on assuming perfectly flexible prices, but only because it would take longer for monetary policy to affect prices and inflation if prices were sticky.) And setting a higher nominal interest rate on money will cause a higher inflation rate only because it means the central bank has increased the growth rate of the money supply by the same amount.

John Cochrane's paper is a long one, as well as an interesting one. I may come back to it later, but that's enough for now.

Comments

[Scott Sumner]

Excellent post.

[Nick Rowe]

Thanks Scott!

[dannyb2b]

Cochrane's proposal doesnt deal with the fundamental issue that not all entities in the economy are homogenous. Some entities spending is a lower proportion of wealth than others. The wealthiest people and central bank counterparties spend little compared to their wealth therefore increases in wealth of these parties (through interest payments) should continue to be innefective. Therefore expectations shouldnt budge to the same degreee as increasing wealth of the average person through direct issuance of reserves.

[Nick Edmonds]

I'd agree with you on your general analysis of what's going on in Cochrane's model. The higher interest rate leads to higher inflation precisely because of its impact on the rate of growth of nominal quantities.

But Cochrane talks about total government debt, not just central bank money and the distinction is very relevant for his results. His equilibrium price level is derived by comparing the nominal quantity of government debt and its implied "real value". With all government debt (and money) on a short term rolling basis, when the interest rate is raised, there is an equivalent corresponding increase in the total nominal quantity of debt.

If you look instead at just reserves and allow government debt to consist of other things, specifically some longer dated debt, you get a slightly different result. Now if you increase the rate on reserves, the growth in nominal government debt will increase by less. So (in Cochrane's model) the rate of change in the equilibrium price level will increase by less than the increase in the nominal interest rate. Which means either that the real interest has to rise, or that there is an instantaneous fall in the current price level.

[Ralph Musgrave]

Another point Cochrane makes is that the recent large increase in base money as a proportion of the total money supply is a move towards full reserve banking: a system where privately created money is prohibited – i.e. the only form of money is central bank money.

Full reserve is more stable than fractional reserve because banks just aren’t allowed to have liabilities that are fixed in dollar terms on the liability side of their balance sheets (or “runnable liabilities” as Cochrane calls them). All they’re allowed is shares, thus they cannot fail. Thus as Cochrane points out, the recent increase in base money as a proportion of the total money supply makes banks more stable. However, the present situation is a messy compromise between full and fractional reserve, I think.

As to those who want somewhere to lodge a fixed number of dollars under full reserve, they put their money into entities that invest only in base money and/or short term government debt.

[Nick Rowe]

Nick E: I think there are two issues:

1. If we introduce longer term bonds into the model (as JC does later) then his (and my) results would change a bit. That's because it adds a second link between monetary and fiscal policy. A change in monetary policy, even if it held seigniorage constant, would change either the real value of each long term bond or else change future surpluses. Agreed. There is no difference between JC and me on that point, I think.

2. His simple model assumes monetary satiation, so that money and bonds pay the same rate of return. I am relaxing that assumption in my model, and explaining the intuition for his result and showing what it would mean regardless of whether money and bonds are perfect substitutes.

Ralph: yep. I was looking at the 100% reserve question in my other post. But stupidly, because my memory is failing, wasn't aware of this JC paper until Tom Church gave me the link. But the other JC paper looks even more interesting on the financial stability question.

danny: Yes, John Cochrane's model is not an MMT model.

[Avon Barksdale]

"Reserves are only non-inflationary and they are only immune from temptations, if they pay the same interest as short-term Treasuries. So, I think these concerns in the end emphasize the importance of always, and perhaps by law or rule, paying full market interest on reserves. If the Fed adds discretion to pay less than market interest on reserves, and to use the spread between reserves and treasuries as a second policy variable, as Kashyap and Stein (2012) suggest, all of Plosser’s warnings take hold."

This is such an excellent point in Cochrane's paper. I like that the large balance sheet plus interest forces the treasury and the central bank into understanding the explicit trade-off between stability and discretion.

[JKH]

When invoking the dynamics of the cb balance sheet it is important to cover all accounts including most importantly equity, assets, and the treasury account.

In that context, the monetary effect of IOR is completely arbitrary as an artificial subcase of QE.

[Nick Rowe]

Avon: just how large do you want the central bank's balance sheet to be? Why draw the line at "short-term treasuries"? What exactly do we mean by "paying full market interest rates on reserves"? Why not have the central bank own all government bonds, and all commercial bonds, and all commercial stocks, and all real assets too? Here is my old post on this question. (I assumed the central bank liability was paper currency, that paid zero interest. But just change my "reduce the inflation target" to "increase interest rates on reserves" and you get the same problem.

John Cochrane avoids this problem by only having one interest rate.

With a government owned central bank, pursuing the optimum quantity of money seriously enough leads to communism! The central bank owns and monetises every asset in the economy. It's just one big mutual fund, that owns everything, and whose shares are used as media of exchange and unit of account.

JKH: you lost me a bit there, I'm afraid.

[Nick Edmonds]

Nick,

So in your case where the central bank pays a 10% interest rate on money, you are then saying that it is also taking whatever other steps are necessary to fix the money supply growth at that 10% level, right? Which would mean increasing the bill rate by a comparable amount. Is that what you meant? I can see that would have the same effect.

I know Cochrane considers long term bonds, but unfortunately only in the context of asset purchases, not in terms of how it affects his central analysis. It's a pity, because I think even a bit of non-short term debt introduces some important differences.

[Nick Rowe]

Nick E: "So in your case where the central bank pays a 10% interest rate on money, you are then saying that it is also taking whatever other steps are necessary to fix the money supply growth at that 10% level, right?"

Right.

" Which would mean increasing the bill rate by a comparable amount. Is that what you meant?"

In my mind, the bill rate (the nominal interest rate on short term bonds) would adjust endogenously, and would automatically increase by that same 10% (10 percentage points) as would all other nominal interest rates. With new money paid as interest on old money, we can get not just neutrality of money but super-neutrality too (changes in levels, and in growth rates of M, would have no real effects). But if we had long term bonds in the model, an unexpected change in the growth rate of money would not be super-neutral, agreed. There would be a real capital loss to bondholders, and future government real surpluses could be smaller.

I skimmed his later bit where he introduces long term bonds, but my brain was too tired to take it in. I had a hard enough time getting my head around this earlier bit.

[jonathan]

I'm confused. I read this post. Then I read some of the linked paper. It never mentions Canada or any other country - with passing mentions of Japan and the UK and maybe another, mostly in the context of surpluses. So I have two linked questions:

1. Is it the norm to analyze something that has already happened, that is in fact happening, for which there is a history of experience as though it hasn't? I suppose the answer to this would be yes if you're trying to explain why, but ...
2. How can that be done without reference to it happening in real life?

As I paged through the linked paper, all I had in my head was "how does this compare to what has actually happened?"

Is this a sensible reaction or should I be treating all this as an abstraction?

[Nick Rowe]

jonathan: I think that's partly a sensible reaction.

For example (and it's a closely related example) there has been a big argument in the blogosphere recently about whether the central bank increasing the rate of interest will cause inflation to rise or fall. ("Fall" is the orthodox answer.) Those arguing for "rise" seemed to totally ignore the experience of actual central bankers. Which struck me as daft.

If those on the other side had acknowledged this experience properly, and had said something like: "but that experience is an illusion, because...", or "but that only works in certain cases, because..." or "but we still don't really know why it works that way, because it won't work in this model..." I would have felt more comfortable.

Plus, we all tend to be a bit insular, and Americans a bit more than others, for the good reason that the US is a big country.

Plus, there is something genuinely(?) new(?) about the US experience, in that it pays interest on reserves PLUS the Fed has a very big balance sheet with US banks keeping very large reserves at the Fed. Canadian banks keep very small reserves, because the Bank of Canada has a (small) spread between the interest on reserves and market rates of interest, and eliminating that spread entirely might change the game.

[Frank Restly]

Nick,

"John Cochrane (pdf) says that central banks' paying interest on reserves allows them to conduct monetary policy independently of fiscal policy. It breaks the link."

It breaks the line one way - monetary policy can be conducted independently of fiscal policy. Later in the paper he discusses breaking the link in the other direction - conducting fiscal policy independently of monetary policy (fiscal Taylor rule, fiscal CPI target, variable coupon debt, Trills, government equity, etc.).

[jonathan]

Each event is unique but we look at similarities for context and to evaluate. Correct me if I'm wrong but the ECB pays ¼ on reserves plus it charges a penalty if reserves are below some amount. I think this is at the national level (?). So there are even more examples. And I see there's talk about the ECB
driving money into the market by charging instead.

I just don't understand the point of modeling without using data and without discussing existing cases. I'd call it masturbatory but it seems pernicious because it could mislead away from real data.

[Thornton Hall]

Imagine someone from another planet walks in and asks, "What should I do with interest on bank reserves if I want to increase inflation?" When, exactly, does it become helpful to involve economists? Why not point to history and say, "like that"?

Similarly, if we imagine a world without Freshwater economists, what is left for the remainder to do? In such a world the history books tell us that deficit spending ended the Great Depression.

[Edward Lambert]

HI Nick, referring to response to Jonathan about Fisher effect and no one is giving reasons that acknowledge the experience of central bankers.
I wrote about this. Yes, if nominal rates are increased, inflation will react and fall in a short run time delay. Then if the nominal rate stops changing, inflation stops reacting to changes in nominal rates and inflation will drift toward the Fisher equilibrium with the natural real rate and the fixed nominal rate. The reason why we see the Fisher effect now is because the Fed rate has 1) been stable for 5 years, 2) projected to be stable in a tight range for more years, 3) low inflation potential due to low demand in relation to productive capacity (David Beckworth rejected the Fisher effect wrongly by giving the opposite situation after a war where productive capacity is low relative to effective demand) 4) well anchored inflation expectations that help inflation slide toward the Fisher equilibrium. (JC talks about inflation anchoring on p. 34)
Once the nominal rates regain the ability and forward guidance permission to move freely, then inflation will return to reacting to "changes" in the nominal rates. Then the Fisher effect will hide in the background.

[Nick Rowe]

jonathan: the Bank of Canada operates a "channel" system. Currently it pays 0.75% interest on positive balances, charges 1.25% on negative balances, and has a target for the overnight rate of 1.00%. It moves the trio up and down together, always keeping those same spreads. I think that sort of system is fairly common for modern central banks, though recently, when the ZLB became a binding constraint, we see more of a floor system, where they push the overnight rate down to the floor. But my memory keeps failing, so don't trust me on this. I learn stuff about other central banks, then forget it.

Thornton: "In such a world the history books tell us that deficit spending ended the Great Depression."

Many economic historians would disagree strongly with that statement. And even if the timing fits (it doesn't, IIRC) if many things happen at the same time, we don't know which caused which. History, unfortunately, is informed by theory.

Edward: what you say contradicts what (most) central bankers would say.

[Mike Sproul]

Not sure I understand. You mean if the central bank has issued $100 while holding $100 of 5% bonds, then the central bank would earn $5 profit (assuming zero printing/handling costs for base money). But the central bank pays 5% interest on its base money, making its profit zero. Did the central bank pay the $5 as interest to the folks who held the $100 of base money? Or did it issue another $5 of base money in exchange for another $5 of bonds?

[Nick Rowe]

Mike:

"Did the central bank pay the $5 as interest to the folks who held the $100 of base money?"

Yes.

"Or did it issue another $5 of base money in exchange for another $5 of bonds?"

No.

Damn! I thought I was writing clearly!

[Mike Sproul]

Nick:

So after the interest is paid, the central bank's T account shows $105 of bonds as assets, and $105 of currency as its liability. Since the FTPL is a special case of the backing theory, it implies no inflation.

[Squeeky Wheel]

I had a thought on how to reconcile the CB observation that raising rates reduces inflation with your IOR calculations where increasing IOR creates inflation.

In the full reserve world, there is no debt-created money (there may be debt, but the money is still fully reserved). Any IOR creates an increase in the money supply, and someone gets nominally 'richer' (the banks, their investors, etc). In this world higher rates mean more distribution of paper, and thus more inflation.

In the fractional reserve world, there is a lot of debt-created money (lots of red money floating around). The net money is small relative to the gross money. In this case higher rates, even if paid on reserves, mostly causes a reduction in the rate of money supply creation (as people borrow less, less money is created and more is repaid/destroyed). In this world higher rates mean less money creation and less inflation. To go out on a limb… an interest rate increase in this world can cause a real recession (see Volker) which reduces the real rate, meaning money is not-neutral. This also means that the real rate in the Fisher equation would not be constant, hence the equation is not a good predictor of inflation (at least in the short-medium term).

That makes sense to me, so there is probably something horribly wrong with it.

[reason]

Squeeky Wheel,
I agree with you.

[reason: I just found this in the spam filter. NR]

[Nick Rowe]

Mike: "So after the interest is paid, the central bank's T account shows $105 of bonds as assets, and $105 of currency as its liability. Since the FTPL is a special case of the backing theory, it implies no inflation."

No. Because the present value of real surpluses is unchanged, and that $105 in backing is worth in real terms what $100 in backing used to be worth.

Baseline scenario: the central bank has $100 money in liabilities that pays 0% interest, and $100 in government bonds that pay 5% interest. But since that $5 in annual interest that the government pays the central bank is immediately returned to the government, we can ignore it. Next year everything is exactly the same.

New policy: The government prints $5 worth of new bonds, gives them to the central bank to pay the $5 interest on the old bonds, and the central bank prints $5 money and uses it to pay interest on the old money. All present and future real primary surpluses are the same.

[JKH]

Nick,

I'm saying that if you want a rule that increases interest earning bank reserves at the same rate as IOR then you just do it.

What that looks like in terms of balance sheet depends on going in accounting specifications.

Conventional cb accounting requires asset acquisition in order to neutralize the negative margin effect of growing IOR on growing reserve balances. This is to avoid the cb from developing it's own negative equity position.

Unconventional accounting consolidates at the transaction level so that both bonds and reserves are issued from a consolidated net liability position, which amounts to a consolidated negative equity position. This is the implied accounting choice of helicopter drops but you economists avoid being strait forward about this sort of thing.

I think you understand this and your post looks coherent to me for what it's worth.

But I think you guys underestimate the power and value of starting from cb accounting - real world or hypothetical.

[Frank Restly]

Nick,

"and the central bank prints $5 money and uses it to pay interest on the old money"

Typically when we think of interest we consider it the time cost of money - emphasis on the word cost. This presumes that the borrower paying the interest has a fixed amount of time (lifetime) and thus income to make good on his / her debt. With governments / central banks that have infinite lifetimes (infinite income) or lifetimes significantly longer than any individual, the time cost of money can either be:

1. Zero - government can print the bonds, central bank can print the money to buy the bonds - both are infinite life agents and so government can defer payment of interest to central bank and central bank can wait an infinite amount of time to be repaid. No need for government to "print $5 worth of bonds to pay $5 interest on the old bonds". Government sells 5% gazillion year accrual bonds - "interest" and principle are repaid to the central bank after a gazillion years.

2. Tax revenue made available to pay interest / total debt outstanding

[Nick Rowe]

Squeeky: I'm not sure. I think it comes down to: what are we holding constant, and what are we changing, when we suppose the central bank increases interest rates? In particular, are we assuming that everyone expects the central bank will increase the growth rate of the money supply by the same amount (effectively raising the central bank's target rate of inflation)? Or are we assuming the central bank will use open market sales to prevent this happening? But my head is not yet fully clear to think of the clearest way of saying this.

JKH: I think I am following you now (not sure). But what is key is that when it pays a higher interest rate on reserves, and makes the stock of reserves grow at that same rate, the central bank also cuts its transfers to the government. It earns less profit so it pays less profit to the government, and this is what prevents the central bank having a negative equity position.

Frank: "Typically when we think of interest we consider it the time cost of money - emphasis on the word cost. This presumes that the borrower paying the interest has a fixed amount of time (lifetime) and thus income to make good on his / her debt."

No we do not. Only confused accountants/finance types think of interest that way, and you are even more confused than them. http://worthwhile.typepad.com/worthwhile_canadian_initi/2013/09/thoughts-on-teaching-the-time-value-of-money.html

We (economists) think of (one plus) the rate of interest as the relative price of present goods in terms of future goods. If the goods are real goods (apples) we are talking about a real (apple) rate of interest. If the good is money we are talking about the nominal (money) rate of interest.

But do not continue the argument on this point here. It is off-topic.

[Squeeky Wheel]

Let me see if I can be more clear about how I was thinking about it. All I'm attempting is to explain the sign of change in rates vs change in inflation - why are they opposite sign empirically, but your example of full reserves has the same sign.

I was imagining that the independent variable is the CB target interest rate. The CB selects this target and lets everything else adjust accordingly.

So I start with a simplified version of Canada: total reserves ~ 0. Let target rate in period 0 = r0 with a channel system for balances. Inflation in period 0 is i0. Required reserves ~0% i.e. the extreme version of 'fractional' reserves. Let r1 = target rate in period one. From history we expect that if r1 > r0 then i1 < i0 and if r1 < r0 then i1 > i0. The story I usually hear from interest rate CBs (Fed, etc) is that lower rates mean more *borrowing* by the real economy. Since borrowing creates money, the monetary supply increases and hence inflation increases. The story is about debt creation of money. I assumed ~ no reserves and no open market sales/purchases to change that.

Then we look at the full reserve system (however we get there): total reserves ~ 100%. Target rate = R0 implemented by IOR. Inflation is I0. If we assume a constant real rate then from theory we expect that if R1 > R0 then I1 > I0. Why would the sign flip? Well 1) loans don't create more money, so low rates don't mean higher money growth, 2) repayments don't destroy money, so high rates don't mean less money growth, 3) higher IOR means more money growth. It seems to me the story in this case is not about borrowing but about the interest 'created' by the CB. It looks like I'm assuming no open market operations here either - I suppose that could affect things, including the sign.

[Frank Restly]

"We (economists) think of (one plus) the rate of interest as the relative price of present goods in terms of future goods."

The relative price for who? If an agent with an infinite lifetime is deciding between present goods and future goods, the real interest rate on apples will always be zero (assuming that apples are fungible and not supply constrained). Same with money and the nominal interest rate.

Economists make the assumption that all agents are identical.

[Nick Rowe]

Frank: "Economists make the assumption that all agents are identical."

You are now making even more idiotic comments, in an attempt to troll, and ignoring my very clear: "But do not continue the argument on this point here. It is off-topic."

Stop commenting now. Go and waste someone else's time.

[Nick Rowe]

Squeeky: "I was imagining that the independent variable is the CB target interest rate. The CB selects this target and lets everything else adjust accordingly."

Normally we think of central banks as having one instrument. For example, the central bank can set the quantity of central bank money, or the rate of interest, or the exchange rate,...but only one of that list. Everything else has to adjust accordingly. One degree of freedom.

But if the central bank also pays interest on central bank money, that gives it two instruments. For example, the central bank could set both the quantity of central bank money *and* the rate of interest on central bank money, and let everything else adjust accordingly. Two degrees of freedom.

" It looks like I'm assuming no open market operations here either - I suppose that could affect things, including the sign."

I think that is what's at the root of it. Back to one degree of freedom, given that assumption.

[Vaidas Urba]

Central banks have two instruments - IOR and quantity of money. The question is - what are two targets?
The first one is the nominal anchor. Nick, you often write as if the second target should be the market share of central bank in the liquid liabilities market. My view is different - the second target should not be concerned with static prevention of financial crises. It should be a dynamic tool instead - central bank should increase the real quantity of money during the financial panic, subject to profitability constraint.

[Thornton Hall]

Nick,

Thank you very much for responding to the comments of the uninitiated like myself. Nonetheless, I think you demonstrated the failure at the heart of your profession in doing so.

My point is that if you strip away all the discourse about theory, policy makers would be forced to make their choices based on the correlations of history. Would such people know the truth about cause and effect? Who cares? Certainly not policy makers, nor their subjects. The question is rather, would their decisions have the desired effect?

Imagine a medical experiment where everyone starts out with a a virus induced cold they caught on the same day. The control group makes no change, but the experimental group, relying on correct theory about germs an the immune system, takes an anti-biotic. Both groups recover at the exact same rate. Over time, however, this misuse of antibiotics leads to a global disaster as antibiotic resistant bacteria kill 30% of the population.

You can look at the experiment and say that without theory we can't understand the cause and effect involved with disease. I look at the experiment and say, theory didn't help anyone get over a cold any faster and it ultimately lead to global catastrophe. Keep your theory in a tower, please. The rest of us will muddle along just fine with correlation and induction.

[Nick Rowe]

Thornton: economists aren't stupid. We have thought about these questions.

If we could do experiments, it would be easy. Give me 100 countries, I will toss a coin to divide them into two groups, and will raise interest rates in one and cut them in the other, then wait a couple of decades. If my theory is right, the first group will suffer a deflationary recession, and the second group will suffer hyperinflation. We can test my theory easily that way, and learn a helluva lot about economics.

But it would never get past the Research Ethics Board.

Countries don't like having their monetary policy determined by the toss of a coin. In *any* sensible central bank, interest rates are only changed to offset what they see as a shock, so we can never disentangle the effect of the interest rate change from the effect of the shock.

Some economists do search through the history, or write history books (Milton Friedman's Monetary History of the US for example) trying to hunt down "natural experiments" where policy changed for what looks like some "random" reason.

There is a whole branch of economics, called "econometrics", which tries to help us "muddle along...with correlation and induction". But the speed of my car is correlated with the position of the speedometer needle. If I grab the needle and move it, will that cause my car to speed up? Those econometricians have even invented "causality tests" which try to disentangle cause and effect. But they know their tests aren't perfect, because their tests say that people carrying umbrellas causes rain.

[Tom Brown]

Nick, what do you make of this way of looking at the issue of causality:

http://informationtransfereconomics.blogspot.com/2014/05/causality-in-information-transfer.html

Am I right to think that this is a unique view of economics? Or does this remind of you somebody else's research from the past?

The idea of an "entropic force" (really no force at all):

http://en.wikipedia.org/wiki/Entropic_force

possibly being applicable to economics is something I haven't seen anyone else discuss, but then my exposure to economic literature is very limited and maybe I'm just ignorant of well known literature along this line. In any case, the implications for cause and effect are thought provoking. I thought his footnote was interesting too:

"[1] This represents a huge break with traditional economics. It also makes traditional economics seem kind of silly if you try to use the language of utility in thermodynamics ... The pressure of an ideal gas falls because an atom has diminishing marginal utility of extra volume. Ha! This break is also interesting because it means that all the properties of supply and demand are emergent. Atoms don't have feelings about pressures and volumes and the idea gas law isn't even true for an individual atom. Bringing this insight back to economics: diminishing marginal utility is not a property of an individual economic agent, but rather an ensemble of agents. Supply and demand is not about incentives, but rather a property of ensembles of people performing market transactions."

I'll be interested to see if Jason's theory has any success. AFAIK, he's the first to attempt to apply information transfer models to economics.

[Tom Brown]

Nick, you write:

"If we could do experiments, it would be easy. Give me 100 countries, I will toss a coin to divide them into two groups, and will raise interest rates in one and cut them in the other, then wait a couple of decades."

I've often had the thought of trying to utilize on online multi-player game to perform economic experiments. Have you ever heard of such an approach? I write a little more about the idea in this comment, but I admit my thoughts on the matter are pretty vague:

http://informationtransfereconomics.blogspot.com/2014/06/money-unit-of-information-and-medium-of.html?showComment=1402007970047#c2052506831700679325

[Tom Brown]

Nick, in response to myself two comments up, Jason has a link to this paper in his latest post:

http://www.minneapolisfed.org/research/sr/sr218.pdf

Maybe that's close to the "well established literature" I was wondering about. Are you familiar with that Kockerlakota paper?

[Nick Rowe]

Tom: some economists do run experiments, like psychologists do. Volunteers play games for real money (sometimes quite large amounts). We were behind the curve on this, but are catching up.

[Jacques René Giguère]

Or pigeons or rats
Quinine pellets are a Giffen good for rats
http://www.ncbi.nlm.nih.gov/pmc/articles/PMC1323011/

No wonder given that tonic water has earlier been shown to be a Giffen vs root beer.

http://theamateurecon.blogspot.ca/2011/03/beer-rats-and-giffen-goods.html

[Jacques René Giguère]

Though now lawyers are sometimes used as they are more numerous, cheaper to buy, nobody cries when they suffer or die and there are things that rats won't do. (Apologies to lawyers present...)

[Thornton Hall]

Doctors are very smart, too. But it turns out that eating cholesterol has nothing at all to do with blood cholesterol. A generation and a half was taught that eggs are bad for you and skim milk is good for you. The exact opposite is true.

[Thornton Hall]

But more importantly, there's another point that I am making. Why is the question posed as if "doing economics" is either useless or beneficial. The third possibility is, logically, that the world is worse off as a result of the existence of economics. Krugman obviously is aware of a scary third possibility, which is why he is at such pains to demonstrate the difference between what he calls "textbook economics" and... er... things that were demonstrably wrong. SWL has joined this project. Yet, to the lay outsider, it looks like there is no principled distinction that policy makers could have seen before the crisis.

I think the kerfuffle over Larry Summers and the Fed is instructive. Us dumb non-economists could plainly see that his support for deregulation was part of the cause if the crisis. Interestingly, no one (with the possible exception of DeLong, whose comments on the subject were totally unintelligible) defended him on the basis that he was simply applying the standard textbook economic wisdom of the time. But as near as I can tell, that is exactly right. The smartest guy in the room applying the state of the art wisdom in the field took deliberate steps to enact policies that directly led to the immiseration of millions of people.

I think the comparison to the physics of weather is an apt one. The thing is, back before super computers and complexity theory, meteorologists weren't walking the halls of power advising that the best way to protect rural Americans from severe weather was to collect them into concentrated locations and give them cheap portable shelter (ie, trailer parks).

[Nick Rowe]

Thornton: "Us dumb non-economists could plainly see that his support for deregulation was part of the cause if the crisis."

A very narrowly American perspective. Canadian banks have been much more stable than US banks for a very long time. But it is not at all clear to me that Canadian banks have been more tightly regulated than US banks. If anything, they have been less tightly regulated. No restrictions on multi-branch banking, for example. No legal reserve requirements for the last 20 years. Differently regulated maybe, but not more regulated.

I am old enough to remember when many educated people thought that centrally-planned economies would obviously do better then every man for himself. They sneered at economists' talk about invisible hands.

Governments will make economic policy one way or another. Someone will always be giving them advice. Economics won't go away. Keynes said something apt about so-called practical men.

But this is now way off-topic. So let's let it drop.

[JKH]

Nick,

I've read this about 6 times and it gets better every time.

Just a couple of points on the economic argument that uses some sort of accounting foundation.

As I understand it the economic argument uses a money supply growth target that becomes the inflation rate. The cb then pays that same nominal rate as the rate of interest on nominal money balances. That decreases the demand for money in such a way that there is no initial jump in the price level.

I hope that's roughly correct.

Then you talk about a delta seigniorage function where the real value doesn't change because the inflation effect is offset by the increased nominal money supply.

And the fact that the cb may buy government bonds doesn't change this because the interest paid by the government to the cb gets paid right back to it by remittances.

In making this last connection you sort of back into real world balance sheet accounting from hypothetical helicopter drop accounting where the cb balance sheet shows no assets - just as the typical treasury balance sheet profile is a net liability position.

The interesting thing is the seigniorage from that zero asset negative equity starting cb concept is actually a negative number - considered as anti-derivative of the delta - but that your delta reasoning holds in any case.

[Nick Rowe]

JKH: "I've read this about 6 times and it gets better every time."

Thanks! I thought you were an old guy. My old brain gets too tired to do that!

"I hope that's roughly correct."

That is exactly correct. Different to how I would explain it, but it works fine.

"The interesting thing is the seigniorage from that zero asset negative equity starting cb concept is actually a negative number - considered as anti-derivative of the delta - but that your delta reasoning holds in any case."

That's where you lose me. But you are probably right.

[Jacques René Giguère]

Nick: my trolling quotient being rather low, permit me one slightly O/T on bank stability:
" No restrictions on multi-branch banking, for example. " Meaning that a bad mortgage in a faraway province stay on your book instead of being unloaded. May induce slightly induce more prudence.
I'll try to find the link but last year, the Atlanta Fed did a study on why Canadian and Australian banks withstood the crisis better, Fascinating for surprising reasons (among others, a wish to diminish lower and middle classes influence on the power structure of a colony). IIRC, Mark Thoma linked to it in mid-april 2013 and there was a mention in The Economist.

[Nick Rowe]

Jacques Rene; you're allowed!

A month ago, David Price from the Richmond Fed emailed me this interesting article comparing US and Canadian banking stability (pdf). I was in the UK, and my brain was fried, so I never got around to doing a post on it.

The money quote:
"One might suspect that it’s because Canadian financial
institutions tend to be more tightly regulated; they have
higher capital requirements, greater leverage restrictions,
and fewer off-balance sheet activities. But Canada’s financial system was largely unsupervised until the late 1980s. In a period in which both Canada and the United States had virtually no official supervision or regulation of bank risk-taking — from the 1830s to the advent of the Fed in 1913 — America experienced no fewer than eight systemic banking crises, while Canada had only two short-lived episodes in the 1830s relating to problems here. That suggests regulation alone can’t explain Canada’s stability."

[Thornton Hall]

I think this is the money quote:
"Nowhere did Canada’s structural and regulatory
differences manifest themselves more clearly than in mortgage
finance. Canadian banks tend to hold on to mortgages
rather than selling them to investors. Fewer than a third of
Canadian mortgages were securitized before the financial
crisis, compared to almost two-thirds of mortgages in the
United States. Some have argued that this, combined with
tight regulatory standards, gives Canadian banks stronger
incentive to make those mortgages safe. Fewer than 3 percent
of Canadian mortgages were classified as subprime
before the crisis, compared with 15 percent here. In Canada,
banks can’t offer loans with less than 5 percent down, and
the mortgage must be insured if the borrower puts less than
20 percent down. Mortgage insurance is available, moreover,
only if the household’s total debt service is less than 40 percent
of gross household income. Not only did Canada have a
much smaller housing boom than us, but its mortgage delinquencies
barely rose above the historical average of less than
1 percent. At the peak, 11 percent of American mortgages
were more than 30 days overdue."

Which is another way of saying: housing finance is better regulated in Canada.

But more importantly, I'm not just talking about the repeal of Glass-Steagal. I'm neither an insider, nor an economist, so my understanding may be wrong. But it seems to me that the most telling event of the crisis was the bailout of AIG. Why did an insurance company need a bailout? Because otherwise General Electric would fail. Why did GE have anything to do with anything? What sort of insurance was this? Isn't the answer to those questions: textbook economics says that the more "complete" a market is, the more stable it will be? Credit default swaps and all the rest of the totally unregulated explosion of derivatives were praised as "innovation" by Alan Greenspan. At the time I thought this was Ayn Rand influenced nonsense. But now I think it was just run-of-the-mill economics.

I guess what I am driving at is this:
What guidance did the field of economics provide to the well intentioned policy maker in 1995 who wanted to know if derivatives should be allowed to proliferate, linking risk across banks, non-banks, and non-financial entities? Isn't it the case that economics had a more or less clear answer: it's a good thing that makes the system more stable? Is so, is there a principle that could be applied in retrospect that would have told us to ignore that advice, a principle that could guide us in the future?

Because if there isn't, it seems to me that the field should simply refrain from comment on policy. Plenty of academic fields are like that.

[Tom Brown]

Nick, re: games, have you seen this?

http://boardgamegeek.com/boardgame/113328/youre-the-banker

Check out the publisher. Those folks appear to be having a ton of fun don't they? I wonder if they have any copies left... sitting in their vault somewhere.

[Tom Brown]

Nick, I just noticed that Noah Smith recently touched on the online game proposal too:

"Will huge multiplayer online video games give us a laboratory to study recessions?"

Here:

http://www.bloombergview.com/articles/2014-06-05/what-happens-when-the-economy-baffles-economists

Since I've been blathering on about that for about a year now, I'm going to assume he got the idea from me. :D

[Too Much Fed]

“Baseline scenario: the central bank has $100 money in liabilities that pays 0% interest, and $100 in government bonds that pay 5% interest. But since that $5 in annual interest that the government pays the central bank is immediately returned to the government, we can ignore it. Next year everything is exactly the same.”

“But if the central bank also pays interest on central bank money, that gives it two instruments. For example, the central bank could set both the quantity of central bank money *and* the rate of interest on central bank money, and let everything else adjust accordingly. Two degrees of freedom.”

“But what is key is that when it pays a higher interest rate on reserves, and makes the stock of reserves grow at that same rate, the central bank also cuts its transfers to the government. It earns less profit so it pays less profit to the government, and this is what prevents the central bank having a negative equity position.”

Let’s make the scenario this:

Baseline scenario: the central bank has $100 money in liabilities (one day) that pay .25% interest, and $100 in 10-year government bonds that pay 2% interest.

Something happens so that the whole yield curve goes by 300 basis points 4 years later.

What does the central bank do?