Noah Smith wonders if "reality might topple a beloved economic theory". Well, if you look at Sweden, reality just confirmed that beloved economic theory. The Riksbank raised interest rates because it was scared that low interest rates would cause financial instability. Lars Svensson resigned in protest. Then inflation fell, and the Riksbank needed to cut interest rates even lower than before.
That's only one data point. But there are loads more.
If you don't know how to drive a car, and you don't even have a clue whether you turn the steering wheel clockwise or counter-clockwise if you want to turn right, one good strategy is to borrow a car, and a wide open field, and experiment. Make a random turn of the wheel, and see what happens. The recent data point in Sweden was a natural experiment like that. But Sweden is not a wide open field, and it's hard to borrow a car to experiment like that on regular roads.
An alternative strategy is to ask an experienced driver which way to turn the wheel. Preferably a driver who has managed to keep his car out of the ditch for the last 20 years. Like the Bank of Canada. And if the Bank of Canada says that it cuts interest rates when inflation is falling below target, and it wants to bring inflation back up to target, you listen. They are either right, or wrong and very very lucky.
Never ignore the advice of experienced practitioners, who have had their hands on the steering wheel for a very long time. Unless you have a very good theory about why they might be deluded.
Theory says, and the data confirm, and the advice of experienced practitioners confirms, that if it wants to raise inflation the central bank should first lower interest rates. Then, when inflation and expected inflation starts to rise, it can raise interest rates, higher than they were before. Then, and only then, does the Fisher effect kick in, and we see a positive correlation between inflation and nominal interest rates. That is the Scandinavian flick we saw recently in Sweden. [Except the Riksbank did the flick the wrong way round, so Lars "The Stig" Svensson jumped out of the Saab and watched it go into the ditch from the sidelines.]
But cars don't have to work that way, and monetary policy doesn't have to work that way either. If we wanted to, we could design a way of implementing monetary policy so that the central bank would increase inflation by raising interest rates.
For example, we could make it a rule of central banking that central banks are not allowed to issue new money except by paying interest on old money. So that if the central bank raised the interest rate it pays to people who hold central bank money from 0% to 5%, the growth rate of the money supply also rises from 0% to 5%. And the increased money growth rate would also increase the inflation rate by 5 percentage points, eventually, for absolutely standard monetarist/keynesian reasons. And that is exactly what is going on in John Cochrane's model that Noah refers to. It is really very very simple and old-fashioned, when you finally get underneath all the fancy stuff.
But central banks, right now, do not in fact work like that. They do pay interest on reserves, but they also do (very large amounts of) open market operations. ("QE" is the silly new word for the open market operations that central banks have been doing for centuries.) There is no link between the central bank raising interest rates and raising money growth rates. Under current operating procedures, the link works in the opposite direction (just look at the data). So it just won't work.
Oh God. Figuring out the intuition behind John Cochrane's paper, to see what was really going on in his model, really drained me. Do I really have to wade through that Stephanie Schmidt-Grohe and Martin Uribe paper too, and reverse-engineer their result as well? I'm too old for this. Don't any of you young whippersnappers have an economic intuition? Do you all get snowed by every fancy-mathy paper that comes along?
I expect I will have to. Pray for me.
The RBA believes that raising interest rates lowers inflation and lowering interest rates raises inflation too. And its record is even better than the BoC's at hitting its inflation target.
Of course, maybe it is just expectations: everyone expects interest rates and inflation to work like that, so ...
Still, raising interest rates increases the return on delaying spending -- but somehow can be made to be inflationary (in this amazing old/new story). And lowering interest rates lowers the return on delaying spending -- but somehow can be made to be deflationary.
Hmmm. Those expectations seem sensible, no?
Posted by: Lorenzo from Oz | November 05, 2014 at 08:43 AM
I find it hard to believe we've been adjusting interest rates in the wrong direction for the last 30-ish years and haven't noticed.
Posted by: Michael | November 05, 2014 at 09:04 AM
Hi,
Can't believe nobody noticed this, or if they have, allow me to repeat:
http://www.theglobeandmail.com/report-on-business/economy/poloz-having-something-unpaid-on-your-cv-is-very-worth-it/article21439305/
373 comments so far, and very negative...
Posted by: sustain_ability | November 05, 2014 at 09:56 AM
Great stuff.
And thanks for doing the hard reading so others of us don't have to.
Presumably no one would deny that the institutional set-up matters, but it's remarkable that this point seems to get lost in some of these papers.
Posted by: M.R. | November 05, 2014 at 10:27 AM
Scandinavian flick? You mean "I am Curious -- Yellow"? ;)
Posted by: Min | November 05, 2014 at 11:26 AM
> So that if the central bank raised the interest rate it pays to people who hold central bank money from 0% to 5%, the growth rate of the money supply also rises from 0% to 5%. And the increased money growth rate would also increase the inflation rate by 5 percentage points, eventually, for absolutely standard monetarist/keynesian reasons.
I think that one caveat here is that we're presuming that "circulating money" is directly proportional to "base money," that being the thing that central banks pay interest on.
Reality is somewhat more complicated, since fractional reserve banking allows not-base-money to be used interchangeably with base money. That allows the money multipliers to diverge, which has indeed happened:
(Incidentally, curses to FRED's graphing tool for disabling log axes when using a formula. That's pure nonsense.)
To the extent that privately-created money is not base money, the "tail wags the dog" approach has a potential to break.
Posted by: Majromax | November 05, 2014 at 12:22 PM
There is much more to the Sweden story than that. http://ftalphaville.ft.com/2014/11/03/2024892/guest-post-the-riksbank-at-zero-lessons-for-others/
Posted by: Peter Doyle | November 05, 2014 at 12:40 PM
Lorenzo: Yes, the Reserve Bank of Australia works just as well, and provides another raft of data. (And the RBA, like the BoC, pays interest on reserves?, and raises that interest rate on reserves when it wants to tighten monetary policy?) The 1981/2 episode, in many countries, also works well.
"Of course, maybe it is just expectations: everyone expects interest rates and inflation to work like that, so ... "
Well, a lot of it is just expectations, but those expectations cannot be divorced from the "institutional set-up", that M.R. mentions in his comment. What does the Bank's current interest rate signal about its future actions? Is it raising or lowering the growth rate of base money? Has its target changed, or has its view of the shocks changed?
Michael: "I find it hard to believe we've been adjusting interest rates in the wrong direction for the last 30-ish years and haven't noticed."
Exactly. It takes an academic to fail to notice something like that. But when you look at the longer horizon data, you do see that positive correlation between inflation and nominal interest rates, just like long run theory says. The Scandinavian flick, where you turn the steering wheel first one way then the other way, complicates things. (An inverted pendulum is a simpler mechanical metaphor. You initially move the pivot point south, to make the pendulum lean north, then you move your hand north to stop it falling over.)
sustainability: I noticed it. Not worth blogging about.
M.R. And thanks to you for getting me thinking along these lines (interest on money=money growth rate) waay back.
See my response above to Lorenzo about institutional set-up.
Min: definitely not!
Majro: JC's basic model only had central bank money.
Posted by: Nick Rowe | November 05, 2014 at 12:50 PM
> Majro: JC's basic model only had central bank money.
Of course. This is just one of those devils in the details that should be specifically noticed before anything like this model's recommendations turn into public policy.
Posted by: Majromax | November 05, 2014 at 01:05 PM
Nick,
"There is no link between the central bank raising interest rates and raising money growth rates. Under current operating procedures, the link works in the opposite direction (just look at the data)."
The central bank had (prior to interest on reserves) two tools at it's disposal - open market operations and discount window lending.
With open market operations the central bank buys and sells government debt. When the central bank sells government debt, the amount of money available for holders of government debt to lend falls and when it buys government debt, the amount of money available for holders of government debt to lend rises. The standard intuition is that the central bank buys existing government debt at a premium to drive down market interest rates (drive up bond prices) and sells government debt at a discount to drive up market interest rates (drive down bond prices). The net effect of buying at a premium and selling at a discount over a significant time period is a permanent increase in the money supply. For a small central bank balance sheet, this permanent increase is negligible.
The reason I bring this up is that paying interest on reserves is not the only way for the central bank to permanently increase the money supply. Central bank buys a $1 bond for $1.05 and then sells it back for $1.00. Money supply has permanently gone up by $0.05.
When the central bank raises the discount rate that it lends at, the effect is indeterminate and is really a function of the demand for credit. We normally think that the demand for credit will fall with rising real interest rates but there are several other factors at play - for instance, changes in the measure of inflation, and tax / fiscal policy changes.
Posted by: Frank Restly | November 05, 2014 at 01:19 PM
The neo-Fisherite view wouldn't apply to Sweden -- they cut their currency in circulation which is old-fashioned quantity theory. How well the quantity theory applies depends on the size of your economy and monetary base:
http://informationtransfereconomics.blogspot.com/2014/05/blowing-anti-neo-fisherite-model-out-of.html
The neo-Fisherite model applies to Switzerland, the EU, US (today), Japan, the US in the 1930s. It doesn't apply to Canada, Australia, Sweden, Russia, China (all where the quantity theory still holds fairly well).
Posted by: Jason Smith | November 05, 2014 at 01:22 PM
I love the Scandinavian flick bit.
"...issue new money except by paying interest on old money..."
Just to make sure I have this right. If a company pays stock dividends ie. dividends in the form of new stock, the price of the stock will gap down upon each dividend date. Paying interest on reserves with new reserves seems like the same idea.
Posted by: JP Koning | November 05, 2014 at 02:19 PM
Aha! Lars Svensson is The Stig's Swedish cousin! That makes Stefan Ingves Capt. Slow, and so ...
In the plums (YouTube)!
[Edited to fix link. NR]
Posted by: Patrick | November 05, 2014 at 03:27 PM
Nick, I wasn't offering "it is just expectations" as a serious hypothesis! I may think we have to be cautious about rational expectations (setting the difference between agent and model expectations to zero is an analytical choice to be justified like any other), doesn't mean I believe in irrational ones! It was half a joke and half a "expectations are based on something ..." point.
Posted by: Lorenzo from Oz | November 05, 2014 at 04:19 PM
Majro: you are right. A lot of details would need to be worked out, before one would recommend actually implementing that idea. And in a sticky price world, with commercial banks, the dynamics might look very ugly.
Frank: " The standard intuition is that the central bank buys existing government debt at a premium to drive down market interest rates (drive up bond prices) and sells government debt at a discount to drive up market interest rates (drive down bond prices). The net effect of buying at a premium and selling at a discount over a significant time period is a permanent increase in the money supply."
No. That is not the standard intuition. Stop now.
JP: "Just to make sure I have this right. If a company pays stock dividends ie. dividends in the form of new stock, the price of the stock will gap down upon each dividend date. Paying interest on reserves with new reserves seems like the same idea."
Yes. Except it pays those dividends on a daily basis, so we don't see the gaps. And the dividends are unrelated to the companies earnings. So it's more like a stock-split, in continuous time. The stock price depreciates continuously, and the faster you split it, the more quickly it depreciates. Except the stock is also the unit of account, and prices are sticky.
Patrick: Yep!
Lorenzo: Yep. I figured you meant that. But some people might not get the joke. They think MMs think that expectations come out of thin air. So I needed to spell it out a bit.
Posted by: Nick Rowe | November 05, 2014 at 06:21 PM
Peter: I had a read of your FT piece. In general I liked it. Especially that bit about needing microdata to figure out what is really going on with borrowing and lending. Because aggregate debt/income data is next to useless, unless you know who in particular has how much debt.
But I think you were much too soft on the people who wanted the Riksbank to raise interest rates to reduce the risk of financial instability. It is NOT a trade-off between targeting inflation and financial stability. That's what the Scandinavian flick is all about. If you (exogenously) raise interest rates now, so inflation falls below target, you must subsequently lower them ***even lower than they would otherwise have been***. If you want higher interest rates, then lower them.
Posted by: Nick Rowe | November 05, 2014 at 06:35 PM
Regarding the analogy with stock dividends:
The price of the stock goes down whether it’s a stock dividend or a cash dividend.
That's what happens on the ex-dividend date.
A stock dividend is equivalent to a cash dividend combined with newly issued stock at the prevailing stock price - which has already gone down in recognition of the dividend payment obligation.
Regarding the case of new money as interest on reserves:
Interest paid on reserves in the form of an increase to reserves is a debit to CB equity and a credit to reserves.
But interest earned on the CB asset side tends to more than reverse that effect, so the net equity effect is no longer negative.
So there's just less profit remitted to Treasury at the end of the day.
However, the interest that Treasury pays on the bonds held by the central bank will be greater than the CB profit that it receives back - because that bond revenue to the CB is reduced by interest on reserves.
That means that at the margin Treasury runs a deficit with the central bank.
That means that Treasury’s cash account at the CB shrinks over time. That account is a net supplier of reserves to the system, at the margin, because it pays more in bond interest to the CB than it receives back in profit.
That means that Treasury must raise revenue to avoid shrinking its CB account to zero, all at the margin.
That means Treasury will drain back reserves in order to do that – through marginal borrowing or taxation.
That means that despite the efforts of the CB to increase the money supply by paying interest on reserves in the form of new money - that alone won’t do the trick.
The CB must do OMO to increase the money supply, regardless of whatever it does by paying interest on reserves.
This makes sense intuitively because both sides of the balance sheet have to expand in order for the money supply (reserves) to increase, other things equal.
So in fact, the accumulation of interest on reserves paid as new reserves won't be sustained as a net increase in CB reserves - unless the CB on its own facilitates that increase separately through OMO.
So the increase in the money supply (reserves) is a separate decision and monetary policy action than the mere decision to pay new money as interest on reserves. The CB balance sheet and the money supply (reserves) won't necessarily expand just because of that latter decision.
(This message is sent from deep within the caves of the concrete steppes.)
Posted by: JKH | November 05, 2014 at 07:49 PM
JKH: consolidate the government's and the central bank's balance sheets. The government owns a printing press. Initially it prints nothing, and the budget is balanced. Now the government starts running the printing press, but all the money it prints it gives as interest on existing money. If it pays 5% interest per year, the money supply is growing at 5% per year. No Open market operations at all, and the budget is still balanced. But there is 5% inflation, because its just like a 1.05 for 1 stock split, every year (only on a daily basis).
Posted by: Nick Rowe | November 05, 2014 at 08:32 PM
JKH,
I understand Nick's model of interest on "old money" as the CB running a negative equity position with the interest paid not being balanced by an asset. In essence, an "off-balance sheet" deficit of the government and possible because a CB can technically not become insolvent and the treasury can drain reserves (old money) through taxation when inflation exceeds the target.
Posted by: Odie | November 05, 2014 at 08:38 PM
Odie: "...and the treasury can drain reserves (old money) through taxation when inflation exceeds the target."
No. If inflation exceeds the target, the central bank cuts the interest rate and hence cuts the money supply growth rate. No changes in taxes or government spending are needed.
Posted by: Nick Rowe | November 05, 2014 at 08:49 PM
Nick,
That's not correct.
The payment of interest on reserves is a marginal budget deficit - as much as interest on bonds.
That fact is independent of whether you account for separate Treasury and CB functions, or consolidate them.
I agree with your point that OMO is not required for a formally consolidated function. That's a natural fallout of consolidation. I kept to the deconsolidated view out of habit I suppose, and also because Cochrane makes a very interesting argument about how his view bolsters the rationale for separate institutional functions.
But the analogous point to the one I made is that a formally consolidated function also has total control over the mix of debt and reserves, independently of the choice it makes to pay interest on bank reserves with reserves at the particular time it makes those payments.
Posted by: JKH | November 05, 2014 at 08:56 PM
JKH: "The payment of interest on reserves is a marginal budget deficit - as much as interest on bonds."
If you think of money as a liability, then printing new money and giving it to people counts as running a deficit. But if you double the stock of money by helicopter, and at the same time double the price level target, and so halve the (implicit) real liability inherent in each dollar of that liability, then the total real liability of the government/central bank is unchanged, even though the nominal liability is doubled. It's just like a 2 for 1 stock split.
Posted by: Nick Rowe | November 05, 2014 at 09:26 PM
Nick,
Veering off into the accounting weeds a bit I suppose but it helps to be consistent I think.
Payment of interest on bonds or reserves is an expense for the government - consolidated or unconsolidated (assuming the CB is part of the government).
That is independent of the flow of funds arrangements around the payment of interest on reserves.
I think your translation of the basic idea would be more bullet-proof technically if you referred to it as growing the money supply (reserves) at a rate equal to the interest rate on reserves. That's a rule that supersedes any potential for any unwanted interference from all sorts of other conflicting activity that normally can affect the level of reserves. I suppose that's more to my point. The CB or the consolidated government then has to discipline itself to immunize any potential reserve impact in aggregate apart from that rule. A consolidate operation in particular has to have a rule for the bond/reserve mix. The deconsolidated operation starts with the rule that Treasury doesn't get involved in money creation other than through the central bank.
I get your point on real versus nominal liabilities.
Interesting question - what is the real deficit in your model assuming the budget is in balance apart from the payment of interest on reserves and assuming you accept my proposition (fact actually) that nominal interest on reserves constitutes a deficit at the margin? Or maybe not so interesting for you. But its not necessarily zero - because inflation depreciates the real capital value of the reserves on which interest is being paid and that's part of the real effect. And that type of depreciation doesn't get factored into the budget deficit any more than the effect of higher interest rates on the nominal value of government debt. But that really is an accounting issue.
Posted by: JKH | November 05, 2014 at 09:47 PM
JKH: "I think your translation of the basic idea would be more bullet-proof technically if you referred to it as growing the money supply (reserves) at a rate equal to the interest rate on reserves."
That is exactly what I meant. And I thought that was what I had said. Maybe not clearly enough. (Though I said "money base" rather than "reserves", because you have to grow currency too, at the same rate.)
Good question. Ignoring bonds:
real deficit = real G - real T + (M/P)(im - p) where im is nominal interest rate paid on (base) money, and p is the inflation rate, so (im-p) is the real rate of interest paid on base money (which is usually negative). And because it is usually negative, it is easier to think of (M/P)(p-im) as the government's seigniorage revenue/inflation tax.
Posted by: Nick Rowe | November 05, 2014 at 10:06 PM
Nick,
“And I thought that was what I had said…”
I’ll give you a basic example of my point.
I thought you said that the rule was that the CB pays interest at the inflation rate and pays it with new money.
But a CB can follow that rule and then do OMO to drain the reserve effect. It will still have paid the correct interest with new money according to your stated rule as I read it.
The only way that might be precluded according to your type of rule – perhaps - is if interest is paid with mathematical continuity. But in fact I don’t think that’s true either. If OMO is separate operationally from payment of interest (and there’s no reason it can’t be - and in fact it is unless specified otherwise) then the rule can still be broken under continuous interest payments. The CB accountants pay the interest and the OMO manager drains it. The CB still follows your rule since the rule is based on the flow of interest that is paid rather than on the stock of money that grows.
But if I say the rule is that the money supply must grow … then there’s no way to break that rule and you will definitely have the intended effect you want. That’s because the OMO manager in effect must follow that rule too.
The problem I’m pointing out I think is that your rule is stated in the first way but just assumes the second way as a consequence – when that doesn’t necessarily hold. You have to be specifically and sufficiently constraining in the statement of the actual rule that the CB must follow.
Interesting formula! What is neat about it from my perspective is that it follows parallel real world CB accounting. It doesn’t include the change in M/P (which is analogous to marked to market accounting) as part of the deficit! It just uses M/P as the denominator for interest payments.
(The change in M/P could be calculated in separate management accounting books – or separate economists’ valuation books – as is the case now for marked to market CB balance sheets).
Do you think I am right on that (i.e. the nature of the formula)?
Posted by: JKH | November 06, 2014 at 03:52 AM
It seems that in your setup the central bank could never handle a negative real interest rate. Inflation can't be higher than the nominal interest paid on base money. If you introduced commercial banks to the model and if a low time preference causes a negative real interest rate (forget about land for a second), then commercial banks would never find willing borrowers, since they had to offer lenders at least a nominal interest rate as high as the central bank and thus ordering borrowers to pay a positive real interest rate, which can't be done if the market rate is negative.
In short, if real rates are negative people will have an excess demand for money and all your central bank can do is tell them: "I will print money to create inflation but not higher then the interest I'm paying you." Isn't this the perfect recipe for raising money demand while raising money supply?
Posted by: libertaer | November 06, 2014 at 06:23 AM
libertaer: "It seems that in your setup the central bank could never handle a negative [natural] real interest rate." [edited by NR]
I think that's correct. But this is a very simple setup, just to illustrate my point. I wouldn't dream of putting this setup into practice, at least without some major modifications (and probably not even with major modifications).
One minor modification, to deal with the problem you raise, would be to say that the money growth rate must equal the rate of interest paid on money **plus a constant (say) 5%**. That would enable the central bank to handle negative natural rates of interest, as long as they weren't too negative. (But if the natural growth rate of the economy were positive, maybe 5% would be too low.
Posted by: Nick Rowe | November 06, 2014 at 07:29 AM
JKH: "I thought you said that the rule was that the CB pays interest at the inflation rate and pays it with new money."
Ah, no. That wouldn't work at all. That sets up an unstable feedback loop, because an increase in the inflation rate would cause an increase in the money growth rate which would cause a further increase in the inflation rate.
If the central bank were targeting 2% inflation, then if inflation drifted above 2% the central bank in my setup would reduce the interest rate paid on money, and so reduce the money growth rate, which would reduce the inflation rate back towards the 2% target.
"If OMO is separate operationally from payment of interest (and there’s no reason it can’t be - and in fact it is unless specified otherwise) then the rule can still be broken under continuous interest payments. The CB accountants pay the interest and the OMO manager drains it."
Correct. OMO would be banned (or held constant) under my setup. (Except the OMO manager would be required to buy a new bond of equal value when an old bond matures.)
Posted by: Nick Rowe | November 06, 2014 at 07:46 AM
Nick,
OK.
Thanks, my confusion on that first point.
Posted by: JKH | November 06, 2014 at 07:57 AM
JKH: yep. And my key point is, that actual central banks right now work much more like you say they do, and do not work anything like my totally imaginary setup. Is that right? So the Neo Fisherites would be right in my imaginary world, but are not right in the world as it is now. But we could, if we wanted (though it would probably be a daft thing to do), force central banks to work like they would work in my imaginary world. Right?
Posted by: Nick Rowe | November 06, 2014 at 08:15 AM
Not sure if it's off-topic here but I have long thought about asking this.
There is a continuum:
1) The banking sector is inefficient in their reserve utilization - maybe not trusting each other and only trusting the CB. Here the reserves are needed and the CB is the true alpha bank.
2) The banking sector is highly sophisticated and use collateralization to a point that they face the CB as a unified bank - they need no reserves (assuming no required reserves). The banking sector as a whole is the new alpha bank and the CB will be obsolete.
Now reality is surely in between but how stable is the need for the reserves and thus does this pose a problem to the price level targeting through the base money?
Posted by: Jussi | November 06, 2014 at 08:27 AM
Nick,
Yes - I agree that what you suggest is feasible from a technical implementation standpoint.
Although the overarching condition I think is that interest be paid on reserves (as emphasized in the title and substance of Cochrane's paper)
I got confused back there because I haven't revisited this in a few weeks
Spent quite a bit of time on Cochrane's paper though
Was hoping to do a post on it sometime, with reference to your simpler approach
Don't know if I can manage it though - not yet anyway
Posted by: JKH | November 06, 2014 at 08:35 AM
Jussi: The Canadian banking system is very very close to your 2, in the sense they hold almost 0 reserves. (Ask JKH). But, IMO, they are still beta banks because they promise to redeem their money at par for alpha BoC money. Even if they held no reserves at all, they would still be beta.
But a bit off-topic for this post.
Posted by: Nick Rowe | November 06, 2014 at 08:37 AM
@Nick:
> Ah, no. That wouldn't work at all. That sets up an unstable feedback loop, because an increase in the inflation rate would cause an increase in the money growth rate which would cause a further increase in the inflation rate.
> If the central bank were targeting 2% inflation, then if inflation drifted above 2% the central bank in my setup would reduce the interest rate paid on money, and so reduce the money growth rate, which would reduce the inflation rate back towards the 2% target.
But I thought the point of this was that this wasn't really a feedback loop at all?
The CB interest-on-money acts as a hard constraint on growth of the money supply. That means that subject to short-term departures, the long-term inflation rate would have to equal the CB's target. From quantity theory, short-term fluctuations in inflation/price level would have to be interpreted as short-term fluctuations in money velocity, but that's constrained by real economic factors.
If off-target inflation is bound to be a short-term phenomenon, the CB should turn the wheel the opposite your suggestion: decrease interest rates but announce that its long-term money supply target is unchanged. This would (by your argument in interest-on-money) cause a drop in the price level back to target but maintain 2% inflation as an equilibrium path.
More "likely", we'd see the opposite happen, since your simplified model deliberately assumes steady output. With growing output, inflation would be below-trend over time, requiring the CB to make periodic but temporary interest-rate adjustments on the upside. This is introduces new money to the system to account for the new, real-terms output.
> The Canadian banking system is very very close to your 2, in the sense they hold almost 0 reserves.
I wonder, given the difference in monetary systems, how "large" the open market operations have to be in the US and Canada to effect a change in interest rates.
Through its IOR spread, the Bank of Canada could potentially change interest rates without actually buying or selling any government of Canada bonds, whereas until recently that was not an option for the US Fed.
Posted by: Majromax | November 06, 2014 at 09:29 AM
Nick: "No. If inflation exceeds the target, the central bank cuts the interest rate and hence cuts the money supply growth rate. No changes in taxes or government spending are needed."
If inflation is above target (assume a steep drop in money demand) and interest rates are at zero the CB cannot reduce M any further as it lacks assets to sell. Negative interest rates are the same as a tax on the banking sector just being paid to the CB directly instead of first to the treasury and then forwarded to the CB.
Question: Why would deflation not have the same (real) effect as paying interest on old money?
Posted by: Odie | November 06, 2014 at 09:45 AM
> If inflation is above target (assume a steep drop in money demand) and interest rates are at zero the CB cannot reduce M any further as it lacks assets to sell.
That's no problem. Remember that we're assuming a world where we can effortlessly pay interest on all base money, currency included.
In such a world, a negative interest rate is just as easy as a positive interest rate.
Since this interest is the only (or at least the primary) way the CB affects the money supply, we don't have to worry about the CB "having assets to sell."
Posted by: Majromax | November 06, 2014 at 10:07 AM
Majro: in a normal world, with no interest on money, an increase in the money growth rate causes a one-time upward jump in the price level, as well as higher inflation. But in this world, where the money growth rate equals interest on money, we don't get that one-time upward jump in the price level. The opportunity cost of holding money stays the same, because the higher interest on money exactly cancels the higher inflation.
But a change in the growth rate of the economy, or a change in the growth rate of velocity, would also affect the inflation rate. So the central bank would need to adjust the growth rate of money/interest on money to keep inflation on target.
If money dominated all other assets in rate of return (so we hit an equivalent to the ZLB where everyone prefers holding money to any other asset), cutting interest on money won't help. Because the inflation rate falls by the same amount, leaving the real rate of return on holding money unchanged.
Odie: "Question: Why would deflation not have the same (real) effect as paying interest on old money?"
It will.
Posted by: Nick Rowe | November 06, 2014 at 11:14 AM
Nick, It wasn't just Sweden, the ECB made the same mistake in 2011. With even worse results. And thanks for deciphering the Cochrane paper; it would have taken me much longer than you.
Posted by: Scott Sumner | November 06, 2014 at 12:06 PM
Good question.
A response on "Grumpy Economist."
http://johnhcochrane.blogspot.com/2014/11/the-neo-fisherian-question.html
John Cochrane
Posted by: John Cochrane | November 06, 2014 at 12:56 PM
@Nick:
> But in this world, where the money growth rate equals interest on money, we don't get that one-time upward jump in the price level.
I think I understand the difference now. The "interest on reserves" model also doesn't include the idea of one-time central bank actions, like skipping or doubling a single interest payment.
I think the crux of the matter is that the Fisher view is clearly correct over the long term, but central banks don't irrevocably deal in long-term policy. I suspect that Canada's central bank would look a great deal less skillful if it had tried to set the 30-year rate rather than the overnight rate.
Posted by: Majromax | November 06, 2014 at 01:12 PM
"Do you all get snowed by every fancy-mathy paper that comes along?"
Every right-wing fancy-mathy paper.
At this point the evidence for malice outweighs the evidence for stupidity.
Posted by: Barry | November 06, 2014 at 01:14 PM
Barry: "Every right-wing fancy-mathy paper."
No. I am far more right-wing than any of those mathy guys. Most of them aren't into politics at all. And Neo-Fisherite views are not right-wing (nor left-wing).
Only daft lefties think that everything is a right-wing conspiracy. (OK, maybe some daft righties think that too.)
Posted by: Nick Rowe | November 06, 2014 at 03:07 PM
You can use seignorage to expand spending or shrink the debt.
Posted by: Miami Vice | November 06, 2014 at 04:25 PM
In real terms
Posted by: Miami Vice | November 06, 2014 at 04:26 PM
Though, I think, a monetarist would say, TRY to use seignorage to expand government spending in real terms.
Posted by: Miami Vice | November 06, 2014 at 05:18 PM
I think Irving is rolling in his grave watching the prefix "neo" being attached to Fisher!
Posted by: marcus nunes | November 06, 2014 at 09:10 PM
I just posted this over on Cochrane's blog, if anyone here can help I'm all ears.
Forgive me if I'm being daft (and some commenter point me in the direction of some reading if I am please), but I just google the Fisher equation. The Fisher equation DEFINES the real interest rate as the nominal rate less expected inflation. It's shorthand for what your interest is going to be worth when you get it. And that's ALL it is. If nominal interest rates move up, so do real interest rates. Full stop. It is nothing like a gravitational force pulling inflation towards it.
There is also a concept called the natural rate of interest. This is the (theoretical) interest rate at which desired savings = desired investment. This interest rate would be rigid (in the short term) to a change in the nominal interest rate. But it doesn't have to follow any sort of relationship to nominal interest rates and inflation. And it's not the same thing as the real interest rate in Fisher's equation.
I guess what I'm saying is, this Neo-Fisherian stuff seems to me to be nothing more than confusion over definitions of things.
Posted by: Andy Wall | November 06, 2014 at 09:20 PM
Andy: unfortunately, you can't learn money/macro with 10 minutes Googling. Especially not this argument. The Fisher equation is not just a definition. It's also an equilibrium relationship between nominal interest rates and inflation rates. And the question is whether causality in that relationship is reversible, and whether the economy will get to that equilibrium.
"I guess what I'm saying is, this Neo-Fisherian stuff seems to me to be nothing more than confusion over definitions of things."
No, it isn't.
Posted by: Nick Rowe | November 06, 2014 at 11:42 PM