Real Interest on Reserves

A central bank issues currency and wants to target the price of apples in terms of that currency. So it opens an apple window, and posts a sign promising to buy or sell unlimited quantities of apples at $1 each. Done. Arbitrage ensures that the market price of apples in the economy is always $1 each, plus or minus transportation costs to or from the central bank's apple window.

If the central bank wanted to target 2% apple price inflation, it could simply change the price on the sign daily, so it rises at 2% per year.

Suppose the central bank found it was selling too many apples, and it was scared of running out of apples reserves? It could sell something else it owned (maybe government bonds?) to reduce the stock of currency in circulation, to reduce the quantity of apples it needs to sell.

Suppose the central bank found it was buying too many apples, and was scared it was running out of room to store all its apple reserves? It could buy something else (maybe government bonds?) to increase the stock of currency in circulation, to reduce the quantity of apples it needs to buy.

(That's how the gold standard worked, except it was gold rather than apples.)

If the central bank bought or sold exactly the right amount of that something else, and did so quickly enough whenever it saw people running to the apple window carrying either currency or apples, it would never need to buy or sell any apples. The apple window could remain open, but would never have any customers.

Or it could close the apple window, and just watch the market price of apples instead of watching whether the people running to the apple window were carrying currency or apples.

(That's how inflation targeting works, except it is a basket of apples, bananas, and haircuts, and other stuff, rather than just apples.)

But maybe buying and selling something else (like government bonds?) isn't the only way the central bank could reduce the amount of buying and selling it does at the apple window. Maybe an alternative is varying the rate of interest it pays people to hold currency?

The idea is simple. If the central bank sees people running towards the apple window carrying currency, wanting to buy apples, it raises the rate of interest it pays them to hold currency until they stop wanting to buy apples from the central bank. And if the central bank sees people running towards the apple window carrying apples, wanting to buy currency, it lowers the rate of interest it pays them to hold currency until they stop wanting to buy currency from the central bank. The idea is that the central bank adjusts the rate of interest it pays on currency to vary the demand for currency, as an alternative to varying the supply of currency by buying or selling something else (like government bonds).

And paying interest on currency is a lot easier administratively if the central bank money that people hold is not in fact paper currency but electronic "currency" in chequing accounts at the central bank.

(Which is how interest on reserves is supposed to work, except that the central bank farms out the management of those chequing accounts to commercial banks, and those commercial banks are the ones who have chequing accounts at the central bank.)

Would it work? Well that depends.

If the interest on currency is paid by printing more currency, then it won't work. Because increasing the rate of interest on currency from 0% to 1% means increasing the growth rate of the supply of currency from 0% to 1%. It's like a corporation trying to increase demand for its shares to raise their price by announcing a future 101 for 100 stock split, except in this case we are talking about a corporation that lets its customers do their own stock splits any time they like, if they prefer the convenience of two $10 notes to one $20 note or vice versa. The equilibrium value of the currency is unchanged at the time that the higher interest rate is announced, but that equilibrium value will now be falling at 1% per year relative to what would have happened otherwise.

Indexing the interest rate on currency to inflation would be even worse, if that interest is financed by printing more currency. It is like a corporation that promises a 2 for 1 stock split if the market price of its shares falls by 50%. The only effect is that the market price of a share falls by 75% instead of 50%. Yes, the only equilibrium (if it exists) is if the value of the currency stays exactly the same (update: other things equal); but that equilibrium is horribly unstable. Because the further inflation rises above target the more the central bank increases the growth rate of the currency supply. And the further inflation falls below target the more the central bank reduces the growth rate of the currency supply.

[Yes, I am having flashbacks to the Neo-Fisherian "sign wars" kerfuffle.]

If the central bank wants to increase the demand for currency by raising the interest rate it pays on currency it must ensure that the interest is financed by selling something else, and not by printing more currency. This ensures that raising the interest rate has no effect on the growth rate of the supply of currency, and so will only affect the demand for currency, by increasing it.

Or it could pay interest on currency in the form of apples, which really would be a "real" interest rate. Which would mean that currency is a consul (perpetuity) paying an annual (or daily) apple dividend, which gives it a determinate real value much like an infinitely-lived apple tree. And by varying the apple dividend rate the central bank could ensure that the market value of an apple tree stays equal to a fixed number of apples (or a number of apples that declines at 2% per year).

But then it might as well just reopen the apple window, if it doesn't find it a hassle to handle apples (or baskets of apples, bananas, and haircuts, and other stuff).

[Just some thoughts on reading the recent Ricardo Reis interview (HT Scott Sumner)]

Comments

[dlr]

good post, I agree. minor note: there is an instrument known informally as a death spiral convertible bond which functions in real life like your hypothetical stock split contingent on stock performance: lower stock price means the bonds convert into more shares.

also, i think you are right about the real IOR equilibrium being unstable in most cases, but maybe a more stable equilibrium is possible under that regime? all monetary equilibria are subject to both people's perception of government incentives and that radical monetary regime changes are always possible. what if the reflexive, death spiral inflation effect of real interest on reserves was seen as such a problem for government actors that it instilled confidence in the market that the government would do much more than unusual to avoid any inflation such that it enforced a more stable equilibrium, more so than a regime that permitted inflation that was very high but didn't quite explode? it may seem hard for even a maximally motivated government to enforce such a tight inflation window but because it also has the power to cheat if needed (regime change) and people know this, might it not work to create a very stable equilibrium (without actually cheating) in the right circumstances?

[Nick Rowe]

dlr: thanks!

"minor note: there is an instrument known informally as a death spiral convertible bond which functions in real life like your hypothetical stock split contingent on stock performance: lower stock price means the bonds convert into more shares."

Good Lord! "death spiral" is a great name for it.

Interesting idea about instability, sort of like a Doomsday Machine. But there would have to be some other instrument the government would use to avoid disaster.

[Market Fiscalist]

I'm imagining a situation where V never changes. The CB can hit any NGDP target just by varying M. If it sets an NGDP target of 5% then with fixed V , by increasing M by 5% , it get GDP growth of 5%. Paying 5% interest on M seems as good a way as any of increasing M and hitting the 5% target. If you chnage things so that V can vary randomly from year to year so some years you have to increase M by more and sometimes less than 5% to hit the NGDP target. In these years you have to set interest on M to above or below 5%.

On the years that you need to increase M above 5% to address the fall in V, then the increase in interest rate will probably lead to a further fall in velocity. But it seems unlikely that there won't be some increase in the the money supply that will induce NGDP to hit the target. To take an extreme example: if M increased by 1000% (by setting the interest rate on M to 1000%) it seems unlikely this would not lead to at least a 5% increase in NGDP under reasonable assumptions. I think the same thing is true in reverse for years when V is above average,

So I'm not really seeing why interest on M could not be used as the single monetary tool.

[Nick Rowe]

MF: if interest on money is financed by printing new money, and the money growth rate increases expected inflation one-for-one, then an increased interest rate on money leaves the *real* interest rate on money unchanged, and so leaves V unchanged.

[rsj]

The interest the CB pays on reserves is less than the interest it receives from the bonds. E.g. the capital of the CB is not shrinking in real life examples of IOR.

The CB can't print money and give it away. Like any other bank, it takes the income it receives from the assets it holds and spends a portion of that income on payments to its creditors and sends the rest off to shareholders -- the Treasury. Every year, the Treasury gets money from the CB, which is seignorage income. In this way, the CB is just like another bank. It still must make money.

The question is what happens in the (so far theoretical) situation that the CB wants to hike IOR above the yield of the bonds it purchased -- so that it is losing money. Here the Treasury would need to transfer bonds to the CB to make up for the capital shortfall, in which case again this can be viewed as a tax on the non-CB sector combined with a transfer to reserve holders. It would be a bailout. It stinks, from the point of view of politics and social welfare, but isn't unstable at all, because the tax would effectively remove the excess money created by the CB.

[Market Fiscalist]

If the CB could magically adjust people monetary holdings in response to changes in the demand to hold money it could hit an inflation target with no need for any other policy tools, couldn't it? If it did this but called these adjustments "interest on money" (and the rate of interest could be negative if needed) what would be different ?

[rsj]

The CB can magically adjust reserve balances, not household balances.

Reserves are accounts that commercial banks have with the central bank. The CB has no direct control over money holdings, which are demand determined.

The CB, by adjusting the interest payed on reserve accounts held by commercial banks sets an opportunity cost for the lending of reserves among commercial banks, which then propagates out to interest rates in the entire economy. It is through this interest rate channel, and only through this channel, that the non-financial sector is influenced by IOR.

[Nick Rowe]

rsj: "In this way, the CB is just like another bank. It still must make money."

I think there is a difference between a commercial bank that tries to maximise profits for its shareholders and a government-owned central bank that does not try to maximise profit for the government but instead tries to target inflation (or whatever), so that profit just becomes a residual, rather than a target. (But this difference might disappear if the central bank consistently earned negative profit, though that is unlikely.)

MF: there is a big difference between giving away new money at random, and giving away money to existing holders of money in proportion to their existing holdings of money. Because in the second case it gives people an incentive to hold more money. The second case is interest on money. Tim Hortons doesn't just give out prizes at random (in the annual Roll Up The Rim To Win thing); the more cups of coffee I buy from Tim Hortons the (proportionately) bigger my chances of winning a free coffee. It's like paying interest on Tim Hortons coffee, which gives me the incentive to buy more.

rsj: The CB can "magically" adjust money balances by buying or selling something. If it buys a bicycle for $100 (and takes no offsetting sterilisation) the money supply increases by $100 directly. Now you could say the seller of the bicycle "demanded" that $100 by willingly supplying the bicycle, but that is not a "demand for money" in the normal sense of the word. It does not mean that the bicycle seller wants to hold an additional $100 stock of money. He (probably) wanted to sell the bicycle so he could buy something else with the money, not hold it. But the money supply has nevertheless increased. *Because money is the medium of exchange, and is bought or sold whenever we sell or buy anything else, and is not like other assets*. And that is the point that people who say "the stock of money is demand-determined" never seem to get. But that is off-topic for this post, as I have blogged about it many times before. And if the bicycle seller is a bank then bank reserves rise; and if the bicycle seller is a member of the public then public holdings of money rise. And it does not matter if the bicycle is replaced by an IOU, or "bond".

[Max]

rsj, in practice the CB wouldn't need a bailout. It can recapitalize itself by simply withholding future profits. Unless the losses are larger than the franchise value of the CB.

[Market Fiscalist]

'there is a big difference between giving away new money at random, and giving away money to existing holders of money in proportion to their existing holdings of money'

I am still not fully getting this.

Lets say the CB wants 10% inflation and its only tool is printing new money. Then whether it just drops the money from the skies or if it distributes the money as 10% interest on money it should be able to hit its inflation target if V remains constant. As you say above when it pays interest it will have no effect on velocity. If people expect 10% inflation, then 10% interest is the same as 0% interest with 0% inflation.

When velocity varies it will have to vary the amount of money it prints and distributes.

It is not clear to me why this would not work if it distributes the new money by paying interest on money holdings rather than helicopter drops. Suppose V has declined. Then increasing the money supply by 10% will fail to produce 10% inflation. But if the money supply is increased further via higher interest on money - why wouldn't people think "Oh, I (and everyone else) have adjusted our money balances to meet our new requirements - now the money supply is increasing again so that means more inflation - fine, I'd better spend this new money".

I suppose its possible they might think "I'm getting more interest on my money holdings now - I'll increase my holding further". But suppose the rate on money was increased by 10% every day until the target was hit. Surely at some point before rates increased to infinity people would start spending some of the new money they were getting and inflation would get back to 10%.

[Nick Rowe]

MF: If the CB increases the rate of interest paid on money, but holds the stock of money constant (by selling something else to buy back the interest it pays on money), then V will fall, and there will be a (one-time) fall in the price level (but no subsequent change in the inflation rate).

[Deepwatrcreatur]

I'm having trouble imagining this scenario of real interest on reserves because I tend to think of the equilibrium real rate as exogenous, determined by the world market. Thus if the central bank matches that real rate somehow, we have equilibrium, but it is unstable because slight deviation from the equilibrium real rate opens up an arbitrage opportunity. This doesn't really have anything to do with a special role for money as exchange medium.

[Nick Rowe]

Deepwatr: some assets are more liquid than others, and the more liquid assets will have lower equilibrium real interest rates, because people prefer more liquid to less liquid assets. (That interest rate differential is the opportunity cost of liquidity). The demand curve for liquidity slopes down (as a function of that interest rate differential), so if there is a shift in the supply curve of liquid assets, that differential will change.

Money is the most liquid asset.

[Thaomas]

"(That's how inflation targeting works, except it is a basket of apples, bananas, and haircuts, and other stuff, rather than just apples.)"

I think this misses the key problem with US monetary policy, the confusion of "inflation targeting" (which ought to mean that downward deviations of the danger of downward deviations wold be dealt with as vigorously as upward deviations to mean = a price level trend target) with a (soft) inflation rate ceiling and one in which no one knows how much downward deviation from the target/ceiling of the price level the Fed is willing to tolerate.

I think an under-explored problem of political economy is why the Fed behaves in this way.

One issue is what objective function is an inflation target maximizing? Where does the damage of inflation come from? From uncertainty about the instantaneous rate of inflation? Or (my guess) Uncertainty about the future price level? Something else?

Another is does the Fed actually have a (soft) inflation rate ceiling policy or does it have a symmetric inflation rate policy = price level policy that is subject to constraints on its use of certain instruments of policy such as a ZLB, not making "too rapid" changes in ST interest rates, or not buying "too much" in other longer term assets or foreign currency denominated assets? And if these constraints apply, where do they come from? Are they just the policy preferences of bankers on the the Fed Board? And why might they have those preferences?

[rsj]

Max,

"in practice", the CB is not allowed to lose even a dollar, so indeed it requires a bailout even for losing small amounts when it is profitable overall.

E.g. in 2008, Treasury had to step in and explicitly pay for even minor losses at the Maiden Lane funds when the Fed was making tons of money overall that year.

This is because we do not allow the CB to do any fiscal operations, and things like giving people money or forgiving even a dollar of debt is a fiscal operation. There is this pesky thing called a constitution that reserves this to congress, and congress has not delegated this power to the CB.

For the same reason, the CB can generally only buy assets already guaranteed by the Government. It's can't just buy any asset it wants. There is a provision for giving emergency loans, but again, every dollar lost this way must be explicitly transferred to the Treasury's book and treated as a bailout, so of course the circumstances under which these emergency loans are made are very strict.

[rsj]

Nick,

"I think there is a difference between a commercial bank that tries to maximise profits for its shareholders and a government-owned central bank "

Sure, but I was not talking about profit maximization. The CB is not trying to maximize seignorage income, but it is required by law to behave in such a way as to avoid losing money. So it's more of a legal barrier against losing money. This is why the CB cannot "buy bicycles", it can only buy government bonds, or bonds guaranteed by the government. Similarly, the CB cannot just "mark up" a reserve account without either diverting interest income into that account or selling some asset. The CB cannot add a liability without also adding an asset. This means that high levels of IOR are always funded by taxes or bond sales, they are not and can never be fiscal operations.

In terms of the money supply, again it's important to be specific about what you mean. The CB controls the supply of *reserves*, which are deposit accounts that commercial banks have with the central bank. It does not control the supply of "money", which are deposit accounts as well as other zero maturity accounts overall. Indeed back in the day when (mostly non-bank economists) believed in things like "reserve multipliers" they at least had a coherent theory as to how control of reserves could lead to control of all other deposits. But that theory has long been discredited. When the CB creates $100 to buy a bond, it is *effectively* trading with the financial sector, so banks sell a bond to the CB and get some more reserves. The reason for this is that the raison d'etre of the financial sector is maturity transformation. This is why I say that, for the non-financial sector, money is demand determined. If there is a desire for more X-year maturity debt, and an availability of Y-year maturity assets, the financial sector will convert one into the other for a premium paid as a differential in interest rates. This also applies when X or Y is zero -- e.g. money. So the banking system undoes the operation as the money is refluxed back onto the banks as excess reserves.

Because of this, I think it's better to focus on interest rates and fiscal operations, since that is the only thing that the non-financial sector sees as a result of any CB operation.

[rsj]

... there is however, a second order effect from QE, which is to lower the profits of the financial system when rates are zero. This is because the public demand for deposits (or MZM more generally) is roughly the total size of the Federal Debt, which is around 1 GDP. In normal times, the demand for reserves and currency is much, much lower. So in the normal course of operations, the majority of seignorage income is captured by the non-government financial system, as banks purchase safe debt and hold deposit liabilities, earning a large spread. With QE, that spread goes to zero as the Government buys up the safe assets, leaving banks with a lot less interest income (obviously non-financial deposits are unchanged by QE). By adding back IOR, that seignorage income is restored to banks.

To the degree that the seignorage income ends up as dividends to bank shareholders and bonuses to bank employees, then this is a fiscal operation, which is why it took an act of congress to authorize. That money could instead have gone to the Treasury.

But it is a very poor form of stimulus, since the recipients tend to be quite wealthy. So I would agree that IOR can function similar to deficit spending fiscal stimulus.

[JP Koning]

Good post.

In practice, I don't think we have to worry about interest on currency being paid by printing more currency. Central banks in New Zealand, Australia, and Canada have been paying IOR for decades now. We haven't seen the quantity of reserves in these nations growing at a pace that would indicate that new reserves are being printed to pay interest, nor have we seen out-of-control inflation. Instead, these central banks seem to be financing their interest payment by using flows of incoming existing reserves that they earn on their asset portfolios.

[Market Fiscalist]

'If the CB increases the rate of interest paid on money, but holds the stock of money constant (by selling something else to buy back the interest it pays on money), then V will fall, and there will be a (one-time) fall in the price level (but no subsequent change in the inflation rate).'

Yes, I get that.

But couldn't the CB also get a (one time) fall in the price level by reducing the interest on currency for a short period ? Suppose it normally pays 0%, for a year it pays -5% (and so reduces the money supply by 5% in the process) then increase rates back to 0%. It will have achieved the one-time fall. Likewise it could hit any inflation level it wanted to. Suppose it wants to get 5% inflation. It first pays 10% (or whatever) interest on currency for long enough to increase the money supply sufficiently , then reduces it back permanently to 5% and (with a bit of fine tuning) it hits its target.

My point is that there is nothing he CB can't do just by adjusting the quantity of money via paying interest (positive or negative) on it. And if it manages expectations well there is no reason why the fact that the interest rate always moves initially in a counter intuitive direction relative to the policy intent need be a concern. People would soon learn that "higher rates on money mean more higher prices" and adjust accordingly.

I see that the CB can:
1. Put downward pressure on prices by increasing the amount of interest it pays on currency
2. Put upward pressure on prices by increasing the amount of currency available (by giving it away , or buying stuff for new money).

It it increase interest on currency but makes no attempt to keep the money supply constant by selling stuff, then there will be some uipward pressure and some downward pressure on prcies at the same time frm the 2 aspcts of the policy (increased rates

[Market Fiscalist]

Please ignore the bit after "I see that the CB can:" - that was a cut and paste error.

[Too Much Fed]

"If the central bank wants to increase the demand for currency by raising the interest rate it pays on currency it must ensure that the interest is financed by selling something else, and not by printing more currency. This ensures that raising the interest rate has no effect on the growth rate of the supply of currency, and so will only affect the demand for currency, by increasing it."

How does existing currency work here? I want to go over the realistic scenario, all entities pay interest with existing currency.

[rsj]

"Instead, these central banks seem to be financing their interest payment by using flows of incoming existing reserves that they earn on their asset portfolios."

-- Which is seignorage income and belongs to the government. The CB pays its depositors FedFunds, who then turn around and pay their depositors MZM Own rate, which is a lot lower than FedFunds.

Households do not have the opportunity to earn FedFunds on their deposits while maintaining zero maturity and government guarantees. That is given solely the financial system as yet another subsidy to Wall Street.

There is no reason for the CB to pay interest on reserves in order to put a floor under lending rates, given the numerous other ways -- e.g. taxing bank assets -- that are available to force banks to lend at higher rates.

So we start with a crappy model in which banks are totally erased -- that the Central bank conducts monetary policy by buying baskets of consumption(!) goods -- and end up with a policy that subsidizes banks at the expense of the Treasury. It was ever thus. If the eternal solution to every problem of Republicans is to cut taxes on the wealthy, then the eternal solution for every issue faced by monetarists is to give more subsidies to banks.

[Frank Restly]

Nick,

"If the central bank wants to increase the demand for currency by raising the interest rate it pays on currency it must ensure that the interest is financed by selling something else, and not by printing more currency."

The interest that the U. S. central bank is paying on it's Treasury holdings is ultimately borne by the U. S. taxpayer.

One way for the central bank to increase the interest rate it pays on reserves is to give control of tax policy to the central bank. If the central bank wants to increase the demand for currency, raising both the interest rate and the tax rate will do the trick.

Another way is for the fiscal authority to reduce the number of bonds that the central bank owns. Central bank pays the same amount of net interest on fewer bonds at a higher interest rate.

[Too Much Fed]

rsj said: "In this way, the CB is just like another bank. It still must make money."

rsj, is it more accurate to say the CB is like a levered hedge fund with some different "rules" applied to it?

rsj said: "It does not control the supply of "money", which are deposit accounts as well as other zero maturity accounts overall."

rsj, do you think the CB fixes its currency and commercial bank demand deposits at a fixed exchange rate that happens to be 1 to 1?

[Matt Young]

I was confused by tine. In the Reis article he seemed to be talking about setting rates daily, based on current price leve;, but he is using the indexed bond example, except he has indexing done daily. Time, in his calculation, is assumde to drop out, it is in his inflation and interest measure, the denominator. Here is his staement:

"The central bank can say that, instead of paying 3% of nominal interest on reserve, it will pay 3% times the price level tomorrow. If it does that, note that central bank is promising a real payment tomorrow to whoever hold the reserve."

But implied in that, is the 3% rate, enumerated over the daily interest rate setting period. So, I do not get his thing working at all, he has no stable measure of the daily price level.

[Matt Young]

Thinking abut stable price level, it happen YoY, which is why rates are denominated yearly. The CB is setting the yearly rate if it is targeting price. If it tries to set rates monthly, it adds more pricing noise than it removes. The CB is best setting rates about every two quarters. But that answer is about, the Fed is essentially applying an adjustable, asynchronous interest charge. The price level becomes a zero mean random walk. Depositors will tend to match their personal paid prices to the aggregate. There is no arbitrage,that was the point, so all depositers equally observe and bet the month to month path of prices. On a weighted basis, everyone will adjust deposits to lower or raise consumption and realign prices. Some will be willing to suffer lower than expected interest returns,others higher; depending on the current price level and their current purchases. The inflation is going to be zero, I am pretty sure.

[louis]

"If the central bank wants to increase the demand for currency by raising the interest rate it pays on currency it must ensure that the interest is financed by selling something else, and not by printing more currency"
So the idea is that you need to commit to "mop up" the extra currency created by paying interest in the form of currency in order for the promise of that interest to raise the demand to hold money. Makes sense.
So when does paying interest on reserves make sense? You show that it can't eliminate OMOs entirely. Would you say that in an environment where ST interest rates are low (so amt of interest accruing each year is small relative to overall stock of money), and demand for money is inelastic (so it would take a very large change in central bank reserves to alter prices) that this becomes an effective tool? Pay 1% interest and promise to sell bonds equal to 1% of the money stock (incremental to the counterfactual) a year from now vs selling bonds equal to 10% of the money stock today? Is this realistic or is there some other wrinkle here?

[Nick Rowe]

louis: "So when does paying interest on reserves make sense? You show that it can't eliminate OMOs entirely."

Hmm. Good question. Your answer (it depends on the elasticity of Md) is maybe a good start.

Matt: I think we must take his "tomorrow" metaphorically. He probably means next month, or year. (Funny how we have a word "tomorrow" for next *day*, but not month or year.)

rsj: with free entry into commercial banking (not central banking) commercial bank profits (but not central bank profits) should be driven to zero. Canada has no required reserves. Required reserves, with 0% interest on reserves, are a tax on banking. But the incidence of that tax, like all taxes, depends on elasticities of supply and demand. With zero profits in banking, it will probably be borne by banks' customers (borrowers and deposit holders).

MF: OK, I think I see your point. But the CB could only change rates of change, and not do level changes, with interest on reserves financed by printing (unless interest rates were + or - infinite). That might not matter much, if it could make them very large. But I worry about how quickly expected inflation would adjust, and what happens during the adjustment.

JP: thanks! You are probably right in practice. Us theorists need to consider all possibilities though, and try to get our heads clear. And it suggests central bank communication on this question might be important.

[rsj]

Nick,

I get the argument that economic rents shouldn't exist in some models. I also get that there aren't many good models of banking used by economist. I also look around and see economic rents increasing, as well as the banking sector becoming more concentrated.

Bottom line, people are willing to 1) hold a lot of MZM. Basically a GDPs worth of MZM. I.e. the government could monetize the entire federal debt and people wouldn't notice, because that's how much money they want to hold, and they are willing to accept a rate of near zero (less than inflation) provided that they have easy access to the funds and a guarantee that there is no risk. This is in addition to currency, of course, we're talking about demand deposits or their zero maturity equivalents (money market funds and the like).

So the government is letting the banking system privatize the majority of seignorage income.

Now we can ask, what are the barriers to entry? It doesn't cost a bank more to accept an additional dollar of deposit, so there are no increasing marginal costs. In fact, the marginal costs are diminishing rather than increasing, because as the bank has more customers, the likelihood of the customers transacting among themselves increases and the need for more ATMs and branches per customer decreases in order to get the same level of convenience. E.g. If your customer base is 10% of the population randomly scattered among the population, then to make sure everyone is within 5 miles of an ATM requires covering the nation with just as many ATMs as if your customer base was 50% of the population. So the marginal costs are declining, not increasing, for purposes of handling deposits. The efficient solution, then, would be a government monopoly on banking services, just as we have a government monopoly on reserve services and clearninghouse services for commercial banks. Rather than having socialism for the banks themselves and for profit banks for non-bank depositors.

So as long as people are willing to hold a GDP's worth of money earning less than inflation, the CB should not be paying any interest on reserves provided that there is less than a GDP worth of reserves. That we do is a fiscal transfer to the financial sector which doesn't sit well with a lot of people.

[Nick Rowe]

rsj: the Canadian commercial banks used to issue paper currency ( and rather successfully too, according to George Selgin). Then the government banned them from doing so, and gave the legal monopoly to the Bank of Canada. We now have a strange demarcation of the market for money, where the commercial banks are not allowed to produce paper currency, and the Bank of Canada does not produce electronic money (except that held by commercial banks).

It is not obvious to me that this is best.

[Market Fiscalist]

' But I worry about how quickly expected inflation would adjust, and what happens during the adjustment.'

Yes, agreed. The reason I am interested in this is that I leaned from JP Konig's blog a few months ago that there are specs out there for bitcoin-like currencies that propose to maintain price stability via a mechanism very much like what is described here - I am curious if it is viable.

https://papers.ssrn.com/sol3/papers.cfm?abstract_id=2425270

[Nick Rowe]

MF: Ah! It might work for Bitcoin, since it doesn't have Open Market Operations, and it's not the unit of account, so expected inflation could adjust very quickly to interest on Bitcoin financed by printing Bitcoin.

[rsj]

I think the whole point of bitcoin is that you can't print more. You have to mine it, and it is exponentially harder to mine each additional coin so that there are only a finite number of coins that can ever be mined. Sort of like there is a finite amount of gold in the earth. Bitcoin was created by the Ayn Rand crowd that doesn't want government messing with their money.

[Nick Rowe]

rsj: agreed. But it is also exactly those characteristics (like gold on the supply side, except useless for anything except money) that give Bitcoin a fluctuating value. But we could imagine an electronic currency like Bitcoin, except with a different supply side like MF suggests, that could have a constant value against (say) the CPI. Though how to implement it automatically with no central authority is not obvious to me.

[JKH]

“If the central bank wants to increase the demand for currency by raising the interest rate it pays on currency it must ensure that the interest is financed by selling something else, and not by printing more currency. This ensures that raising the interest rate has no effect on the growth rate of the supply of currency, and so will only affect the demand for currency, by increasing it.”

We've discussed this before, but the standard institutional set up for Treasury and the Central Bank ensures that this will happen, in the sense that an increase in IOR increases the consolidated deficit (other things equal), and that “something else” will be bonds sold by the Treasury. This covers pretty much everything that’s happened so far, including QE. The Fed in QE mode still creates new reserves by OMO/QE – not by paying interest on reserves.

JP captures this pretty concisely in his comment - although the standard revenue composition for the Fed and the Bank of Canada is interest revenue coming in mostly from the Treasury cash account (interest paid on Treasury bonds held by the CB), not from bank reserves. It's Treasury replenishing that account by issuing new bonds (or taxes) that drains bank reserves (as opposed to IOR adding reserves directly from the CB).

(The automatic nature of this money supply control gets messed up in the non-standard scenario of a CB deficit. The potential for the creation of reserves in that scenario could be obviated by Treasury recapitalizing the CB with bonds, with the CB selling those bonds for a reserve drain effect.)

We touched on this a few times. It seems that the standard fiscal/monetary institutional structure actually corresponds to what you would “like” to see happen according to your post. I assume the institutional design reinforces a certain generic approach to monetary policy discipline in that sense.

[JKH]

Conversely, you could always maintain the option of increasing bank reserves at a rate corresponding to IOR paid, simply by using OMO/QE in the corresponding amount, layered on top of existing institutional arrangements.

Or, blowing away all norms of existing institutional arrangements, by using “helicopter money”.

[JKH]

Very quick and dirty thinking about the translation of my previous comments to a real apple economy (maybe):

Existing institutional arrangements would see a CB with claims on real apples as assets (e.g. apple Treasury bonds) and the usual types of liabilities expressed in real apple terms. If the Fed were to tighten by paying an increased apple rate of interest on reserves, then its apple profits would be reduced by the increase in IOR and Treasury’s real apple cash position at the Fed (i.e. perhaps a financial claim denominated in real apples) would be lower as a result of a lower profit remittance from the Fed. Other things equal, Treasury would fund that difference by issuing apple bonds. OMO and QE could be done to increase apple reserves using real apple financial claims on both sides (i.e. apple Treasury bonds and reserve positions denominated in real apples).

Academic/theory/preference question for Nick:

To what ultimate degree can you visualize a real apple economy to include financial claims accounted for in real apples, which I’ve sort of done above – as opposed to the inclusion of "financial" positions as actual apples in some cases (i.e. in the case of cash, I think) ?

For example, the real apple "cash position" I referenced above is a financial claim denominated in real apples, with the liquidity characteristic of cash - rather than “actual” real apples. The same would hold for bank reserves held at the Fed. In that case, there is no "financial position" in that real apple economy that consists of actual real apples. Then I think the question becomes one of convertibility of real apple cash into real apples - at the central bank or otherwise.

(I don't know if any of this will make sense to you)

(emphasizing quick and dirty)

[Too Much Fed]

JKH said: "The Fed in QE mode still creates new reserves by OMO/QE – not by paying interest on reserves."

So does that mean the interest is paid using existing "currency", not new "currency"?

[Henry]

JKH,

As long as you don't start talking about red and green apples, you can do all the quick and dirty you like. :-)

[Jussi]

"IOR increases the consolidated deficit (other things equal), and that “something else” will be bonds sold by the Treasury."

Which banks may, at will, pledge against additional reserves. Could this, in theory, be enough to satiate (haircuts aside) the higher demand for reserves without any beta money (i.e. deposits) implications? Maybe there are other channels like improved banks' balance sheets, which will change things around?

[JKH]

Slightly less quick, slightly less dirty, but still quite dirty (I saw the post only 2 days ago):

First, regarding the existing institutional framework: while it is true that the CB pays interest on reserves by momentarily crediting reserve balances with additional reserve balances, the same quantity of reserve balances is also withdrawn as a result of the way in which both Treasury and the Central Bank manage their balance sheets. The effective control points for this are that the full income result of the CB is swept into the Treasury cash position, and the Treasury cash position is normally maintained at steady level. The combined flow of funds effect is that the reserve position is also kept at a steady level – unless otherwise affected by OMO/QE type operations. So to be clearer on this, IOR is paid with R, but the increase is reversed out by the full effect of institutional arrangements. In that sense, IOR does not result in a net increase in R.

My earlier point is that the existing institutional design ensures that an INCREASE in IOR and a consequent increase in the consolidated deficit is financed with something other than R – that something being debt. But that said, IOR is typically dominated by CB interest revenue, so that the consolidated CB effect is a REDUCTION in the deficit – not an increase as in the isolated case of a marginal effect of higher IOR. So while an increase in IOR is financed by “something else” (taxes or bonds), typical CB profit considered in its entirety is in itself a source of financing for the deficit – not an additional requirement.

Second, the aspect of the institutional design and the related net non-effect of IOR on the reserve position can be separated from the aspect of how the CB manages the real interest rate – which is more to the subject of the post.

I think Nick outlines 3 general scenarios:

a) Paying IOR with the medium of exchange (e.g. dollars), and managing real effects by observing apple pricing behavior and reacting with interest rate policy

b) Paying IOR with the medium of exchange and managing real effects by intervening with apple pricing behavior – by offering a conversion window – while reacting with interest rate policy as well

c) Paying IOR with a particularized medium of exchange for that purpose – actual (real) apples - while reacting with interest rate policy as well

I think my earlier question is effectively answered in the form of option a), which as Nick says is similar to a CPI based interest rate policy system – where the CPI realistically takes the place of apples alone – and where monetary policy requires no actual (real) apple intervention

Option c) is very interesting as a conceptual exercise. It requires a CB reserve of actual (real) apples, and if IOR is the only payment that is made in actual (real) apples, it results in a systematic gross drain of apple reserves from the CB. That systematic drain must be offset by a systematic gross inflow of apples in order to stabilize the net reserve position over time. That systematic inflow could be effected by a regular conversion window, combined with interest rate policy, or by a systematic net buying of apples in the open market, combined with interest rate policy. I think Nick alludes to this sort of thing in his final paragraph.

The further intriguing thing about option c) is how it can still conform to the way in which existing institutional arrangements stabilize the quantity of aggregate reserves. In terms of required operations and corresponding accounting:

If the CB credits the bank reserve position with a quantity A of apples, then:

1. The level of CB apple reserves declines by A

2. The level of bank reserves increases by the same quantity A of actual (real) apples, BUT this component of bank reserves is no longer a liability of the CB – it is now off the CB’s balance sheet

3. The CB equity position declines by the same quantity A, which is the balance sheet offset to # 1.

So that momentary gross drain has shrunk the CB balance sheet on both sides by the quantity A

Two things then happen in the context of the institutional framework working in combination with CB apple reserve policy and CB interest rate policy:

I. If the CB makes a consolidated profit, then Treasury activity will drain an equivalent quantity of bank reserves – BUT that drain will be from that portion of bank reserves still held in the form of the (general) medium of exchange (which is the dominant portion, since only interest is paid in apples)

ii. Because there has been a gross drain of CB apple reserves, the CB will engineer a reversing inflow of apples through a combination of interest rate policy and either a conversion window or by bidding in the open market. This will cause the banks to exchange their apple reserves for medium of exchange reserves, and everything is as it was before

[JKH]

Slightly less quick, slightly less dirty, but still quite dirty (I saw the post only 2 days ago):

First, regarding the existing institutional framework: while it is true that the CB pays interest on reserves by momentarily crediting reserve balances with additional reserve balances, the same quantity of reserve balances is also withdrawn as a result of the way in which both Treasury and the Central Bank manage their balance sheets. The effective control points for this are that the full income result of the CB is swept into the Treasury cash position, and the Treasury cash position is normally maintained at steady level. The combined flow of funds effect is that the reserve position is also kept at a steady level – unless otherwise affected by OMO/QE type operations. So to be clearer on this, IOR is paid with R, but the increase is reversed out by the full effect of institutional arrangements. In that sense, IOR does not result in a net increase in R.

My earlier point is that the existing institutional design ensures that an increase in IOR is financed with something other than R – that something being debt. But that said, IOR is typically dominated by CB interest revenue, so that the consolidated CB effect is a reduction in the deficit – not an increase as in the isolated case of a marginal effect of higher IOR. So while an increase in IOR is financed by “something else” (taxes or bonds), typical CB profit considered in its entirety is in itself a source of financing for the deficit – not an additional requirement.

Second, this aspect of the institutional design and its non-effect on the reserve position can be separated from the aspect of how the CB manages the real interest rate – which is more to the subject of the post.

I think Nick outlines 3 general scenarios:

a) Paying IOR with the medium of exchange (e.g. dollars), and managing the real effect by observing apple pricing behavior and reacting with interest rate policy

b) Paying IOR with the medium of exchange and managing the real effect by intervening with apple pricing behavior – by offering a conversion window – while reacting with interest rate policy as well

c) Paying IOR with a particularized medium of exchange for that purpose – actual (real) apples - while reacting with interest rate policy as well

I think my earlier question is effectively answered in the form of option a), which as Nick says is similar to a CPI based interest rate policy system – where the CPI realistically takes the place of apples alone – and where monetary policy requires no actual (real) apple intervention

Option c) is very interesting as a conceptual exercise. It requires a CB reserve of actual (real) apples, and if IOR is the only interest payment that is made in apples, it results in a systematic gross drain of apple reserves from the CB. That systematic drain must be offset by a systematic gross inflow of apples in order to stabilize the net reserve position over time. That systematic inflow could be effected by a regular conversion window combined with interest rate policy, or by systematic net buying of apples in the open market combined with interest rate policy. I think Nick alludes to this in his final paragraph.

The further intriguing thing about option c) is how it can still conform to the way in which existing institutional arrangements stabilize the quantity of aggregate reserves. In terms of required operations and corresponding accounting:

If the CB credits the bank reserve position with a quantity A of apples, then:

1. The level of CB apple reserves declines by A

2. The level of bank reserves increases by the same quantity A of actual (real) apples, BUT this component of bank reserves is no longer a liability of the CB – it is off the CB’s balance sheet now

3. The CB equity position declines by the same quantity A, which is the balance sheet offset to # 1.

Two things then happen in the context of the institutional framework working in combination with CB apple reserve policy and CB interest rate policy:

i. If the CB makes a consolidated profit, then Treasury activity will drain an equivalent quantity of bank reserves – BUT from that portion of bank reserves still held in the form of the (general) medium of exchange (which is the dominant portion, since only interest is paid in apples)

ii. Because there has been a gross drain of CB apple reserves, the CB will engineer a reversing inflow of apples through a combination of interest rate policy and either a conversion window or by bidding in the open market. This will cause the banks to exchange their apple reserves for medium of exchange reserves, and everything is as it was before

[Nick Rowe]

JKH: I'm reading your comments, thanks for them. Not sure my head is clear on them yet.

One question/point: Treasury has a chequing account at the CB. Are you including the balance in that account as part of Reserves? I would say it isn't. (Remember the old Drawdowns and Redeposits mechanism as a monetary policy instrument?)

[JKH]

hi Nick

no - I don't include the treasury account as part of reserves

I remember the old drawdown redeposit mechanism - more than you might imagine - somewhere in my brain are war stories about that - it used to be very important aspect of reserve management, announced each morning, although subject to considerable speculation how much of it included an excess reserve effect and how much of it was only a neutralizing offset to other government flows through Treasury's central bank account

I think you're the first person to remind me of that specifically in nearly 30 years (I moved to other responsibilities in 1987)

I’m slightly amazed at that :)

[Nick Rowe]

JKH: It came back to me, out of a distant haze, as I was thinking about your comment! I had to teach that damn stuff, to first year students! And I could never remember which was which.

[JKH]

Nick,

In response to your tweet, I believe the Fed had and still has a similar system of drawdown/redeposit - in effect. (I could provide some links I googled that indicate this.) Presumably the mechanism wouldn’t need to be quite so precise in a QE environment, given the massive supply of reserves to begin with. I’m guessing England and Australia have similar arrangements.

Also, you probably do/did remember this, but the language of drawdown/redeposit is consistent with marginal tightening and marginal easing respectively. E.g. if Treasury makes net disbursements from its CB account, that is a marginal easing relative to the reserve position, because it creates additional reserves. The Bank of Canada can then drawdown funds (transfer funds from the Treasury account at commercial banks to the Bank of Canada) in order to effect an offsetting marginal tightening. It can also adjust the reserve position further net of that to a desired “marginal/marginal” easing or tightening. Etc.

I read that Reis interview for the first time and I have to say my first impression (cursory) is that the argument is nonsense. There is nothing very special about the interest rate paid on reserves when reserves are plentiful, versus the interest rate paid on fed funds or short term treasury bills. Sure, there are some minor differences due to technical complications. But all of it has to do with the Fed targeting the general level of interest rates, whatever the reserve regime may necessitate in order to achieve that through a consistent IOR level. In that sense, I think the notion that IOR is special relative to the others is just incoherent at a relevant strategic level. So any discussion about the Taylor rule versus some other rule should look past the special case of IOR.

[Nick Rowe]

JKH: thanks. When I got no replies to my tweet, I assumed that probably meant that nobody knew what I was talking about, and so it must have been just a Canadian thing. But maybe they have a different name for it.

If I remember the old system correctly, the government had a chequing account at all the commercial banks, and at the Bank of Canada, but the Bank of Canada controlled those accounts, and could transfer balances from commercial to central or vice versa. My guess is that there would be squabbles between commercial banks over which one gets how much of the government deposits, but that those squabbles would be easier to manage in Canada, with the Big 5.5.

[Too Much Fed]

"Suppose the central bank found it was buying too many apples, and was scared it was running out of room to store all its apple reserves? It could buy something else (maybe government bonds?) to increase the stock of currency in circulation, to reduce the quantity of apples it needs to buy."

Gov't bonds should be considered a savings vehicle. What if the private sector is only selling them because it thinks the central bank is overpaying for them? They continue saving using currency as the savings vehicle. In other words, the private sector is just switching savings vehicles between gov't bonds and currency and not selling the gov't bonds to spend on a real good/service.

And, "Suppose the central bank found it was selling too many apples, and it was scared of running out of apples reserves? It could sell something else it owned (maybe government bonds?) to reduce the stock of currency in circulation, to reduce the quantity of apples it needs to sell.

Suppose the central bank found it was buying too many apples, and was scared it was running out of room to store all its apple reserves? It could buy something else (maybe government bonds?) to increase the stock of currency in circulation, to reduce the quantity of apples it needs to buy."

What happens if the central bank buys a bunch of apples, and the apples then rot?

[Antti Jokinen]

Nick, I've been too long away from these discussions (for my own good). Let this comment be a first sign of a comeback.

You said in one of your replies to rsj: "Now you could say the seller of the bicycle "demanded" that $100 by willingly supplying the bicycle, but that is not a "demand for money" in the normal sense of the word. It does not mean that the bicycle seller wants to hold an additional $100 stock of money."

This is connected to something that has been bothering me for long. Perhaps you can help.

For how long do people want to hold money? It could be hours, days, weeks, months or years. Anyone wanting to sell something is someone who wants to hold money instead (unless his account is overdrawn...), if only for a day or a week. But then again, nothing lasts forever. So where should we draw the line? And how exactly does it matter if the line is at one year or at one month, or one week?

(This is also partly related to anyone's motivation to rush to the apple window, instead of just buying stuff from the open market. Perhaps it would be mainly Monetarists who'd be using the window, partly for reasons others don't get?)