Imagine you are a monetary economist, working in a central bank, and you wake up one morning and find yourself in an alternative universe. You don't realise it's an alternative universe at first, because everything looks the same. But when you get into work and go to the meeting of the monetary policy committee, you realise something is very different.
You are the first to give your views. You discuss all the various indicators, like inflation, unemployment, commodity prices, and give your recommendation for the overnight rate target (Feds fund rate, if you like) for the next 6 weeks, needed to keep future CPI inflation on target. The response is not what you expect. They all look at you like you're crazy. Then someone mentions that it's April 1st, and they all start laughing.
The second person to speak discusses all the same indicators you did, with two exceptions. She leaves the price of gold out of the list of commodity prices. And she adds the overnight rate of interest to the list of indicators. Then the bombshell: she gives a recommendation for the gold price target for the next 6 weeks, needed to keep future CPI inflation on target.
You then realise you are in an alternative universe (plus, she's got orange eyes).
When you regain consciousness, you head for the Bank's library. Eventually you find "The Child's Book of Monetary History" which tells you what you need to know.
The book tells you that monetary history started out the same way, with gold coins, then paper money redeemable for gold at a fixed price. Then sometime in the 1930s, the histories diverged. Faced with a Great Depression, central banks abandoned fixed gold price targets, and moved to flexible gold price targets. They adjusted the time-path of the price at which paper money would be redeemed for gold, at frequent intervals, in response to changing economic conditions. Eventually, after too much emphasis on trying to keep unemployment low during the 1960s, which caused high inflation, and the realisation that monetary policy could only target nominal variables in the long run, they adopted a CPI inflation target.
"All modern central banks target CPI inflation, using the price of gold as the control instrument" the book reads. "We remain good horizontalists (the LM curve is horizontal at a given price of gold) like our forefathers, but recognise that we must shift the LM curve up and down in response to economic conditions to keep future CPI inflation on target. All but a few "verticalists" see the price of gold as the channel through which monetary policy affects the rest of the economy. And the monetarists have never satisfactorily answered the question 'which of the many interest rates is the one that matters?' "
You are stunned. How will you keep your job, and learn to adjust to this totally new way of thinking about monetary policy?
Then you have a stroke of luck. In your pocket you have your brand new macroeconomic model, TOTEM+G.
TOTEM+G is your first revision of a standard New Keynesian macroeconomic model, which includes commodity prices in the terms of trade. You had decided to disaggregate the commodity price sector, and had just finished doing this for the price of gold when you found yourself in the alternative universe.
You plug your model into the computer, and realise it will work fine. All the same equilibrium conditions still hold true. Even the parameter values are the same. You just have to think about it in a different way. You used to solve out for the time-path of the overnight rate that would keep inflation on target, letting all other variables, including the price of gold, adjust endogenously. Now you solve out for the time-path of the price of gold that would keep inflation on target, letting the overnight rate adjust endogenously.
Actually, it is even simpler than that. While checking to see that expectations are model-consistent with your solution for the gold-price instrument rule, you realise you can get rid of all those stupid ad hoc transversality restrictions you previously needed to get a determinate price level under interest rate targeting. As long as the Bank keeps the price of gold finite, and non-negative, those conditions are automatically satisfied.
You check the robustness of your implied instrument rule, and are even more pleasantly surprised. Even if you make a permanent mistake about the natural real price of gold, all that happens is that the price level is permanently off the target by the same percentage amount, while the inflation rate would remain on target. In your original model, if you made a permanent mistake about the natural real rate of interest, the solution would never converge.
Initially you are puzzled by the non-negativity constraint on the nominal overnight rate. It is no longer a constraint on policy, because monetary policy does not operate on the overnight rate. But it is still a constraint on an endogenous variable, and the economy does behave rather differently when it hits that constraint. But then you realise that the central bank could always move away from that constraint, if it wants, by increasing the rate of increase in the price of gold. It's really no different from having a legal minimum rate of inflation on gold.
The implied instrument rule for the price of gold turned out to be very similar to the original instrument rule for the overnight rate. A sort of modified augmented Taylor Rule, with two differences.
First, the price of gold now disappeared as an indicator and was replaced with the overnight rate. Second, the Bank could not allow anticipated discrete jumps in the gold price target, but had to make it a continuous function of time. Rather than adjusting the price of gold every 6 weeks, the Bank would adjust the rate of change in the price of gold.
Of course, your Neo-Wicksellian model does not contain a stock of money, and you wonder briefly how the Bank actually controls the price of gold. You ask your colleague with the orange eyes. "Well" she says, "we just sort of announce what our target is, and it just sort of follows it. Sometimes we do need to buy or sell a little bit of gold, and we do OMOs in the bond market if our gold reserves get too big or small. The stock of money is endogenous, of course, and we never worry about that."
You start to tell her a story, about what it would be like to wake up in an alternative universe, where the Bank uses an interest rate instrument to target inflation. She asks how the Bank could control the rate of interest, and how a change in the rate of interest could affect the price of gold, and the rest of the economy. You try to explain, using her familiar ISLM diagram, which is just like yours, except it has the price of gold on the vertical axis.
"Well" she says "I'm not at all sure it would work, and I really need to think about the transmission mechanism. How does a change in the rate of interest affect the price of gold, again? It's such a strange way of thinking. We have always conducted monetary policy through the gold market, ever since the first paper money. I can't see how you could keep the price level determinate, using such a weird control instrument. But I do like a man with such a great economic imagination."
She leans towards you. Her orange eyes sparkle. Her lips open softly. And then you really wake up, and realise it was all just a dream.
Back to reality.
I am not advocating gold in particular as the good in which monetary policy should operate. In principle it could be the nominal price of any real good (or claims on a real good). I chose gold just because it made the alternative history simpler. Because gold market operations were once the way that monetary policy was implemented, the alternative universe is just one that diverged from our own history.
It is always hard to seem the same world in a different perspective. And it is the same world, because all the equilibrium conditions are the same. But we have swapped exogenous and endogenous variables, so the disequilibrium transmission mechanism becomes very different. And that story of the transmission mechanism is always just that -- a story. It is outside the equilibrium conditions of the equations.
It is perhaps easier for economic historians, or historians of thought, to think about the world in different ways. Because that's what they are forced to do as part of their work. And it is perhaps hardest for those who are closest to the current operating procedures, and who know more than anyone else about how things really work.
This is a double experiment. A thought-experiment, which I am trying out as an experiment. I don't know if it will work or fail miserably. But there's little cost in trying.
I came to write it after reading Scott Sumner's post, Jeremy Goodridge's comment on that post, and in response to many critical and useful comments by Adam P and JKH on my previous post.
Nick,
I admit to being terminally afflicted by cognitive interest rate capture. God help me.
To wit:
“Then the bombshell: she gives a recommendation for the gold price target for the next 6 weeks, needed to keep future CPI inflation on target.”
Did she also give her expectation for the “announcement effect” as it might influence the interest rate on interbank funds?
Presumably she would care about this, because she would use the interbank rate as an indicator in the next iteration of gold price targeting.
How could she not care?
Did she expect that OMO would be required to facilitate the market response to the announcement?
Would OMO be designed to affect the gold price directly, and only the gold price?
Would the central bank no longer directly affect money pricing for the banking system that it was central to?
Or would OMO be designed to affect the interbank interest rate as well?
How did this work when the US was on the gold standard? Did the Fed not affect the interbank interest rate directly at all? Did they not affect it through OMO? Or did they seek to affect it directly when they thought it was consistent with gold targeting?
As in:
“And we do OMOs in the bond market if our gold reserves get too big or small. The stock of money is endogenous, of course, and we never worry about that”.
She may not worry about it, but doesn’t this anchor the gold supply parameter to an expected effect on interest rates? Is she monitoring interest rates as an indicator for some sort of rational relationship with the gold price?
Is she monitoring interest rates for that rational relationship so that one morning, in the meeting, she doesn’t have to say:
“We’ve got a problem. I think we’re going to have to drop gold and go back to interest rates.”
Posted by: JKH | April 21, 2009 at 12:34 PM
JKH:
Yes, and unfortunately I half share your problem. Half of my brain is also captured by the interest rate approach. It makes it so hard for me to argue with you and Adam P., because I am arguing with myself at the same time! What did Keynes say about the struggle to escape?
Taking your points one-by-one, I think there is an almost perfect symmetry, until the very end. Try that old feminist "gender switch" technique, only augmented. Re-write your comment, substituting "he" for "she"; and swap "gold price" with "overnight rate" at the same time. The "He said - she said" story is then exactly how we might imagine the girl with the orange eyes responding to the protagonist in the Bank cafeteria, when he tells her his story.
But at the very end, she won't be able to understand his worries about the zero lower bound on nominal interest rates. She will say "But why can't you just raise the rate of increase in the price of gold, i.e. loosen monetary policy, so that the overnight rate increases (sic) above the lower bound, if it worries you that much?" She won't be able to imagine any circumstances in which gold-price control wouldn't work. And if it did, well, game over for monetary policy, and perhaps we should think about fiscal policy instead.
Posted by: Nick Rowe | April 21, 2009 at 01:52 PM
only just reading the post but still...
Nick: "I am not advocating gold in particular as the good in which monetary policy should operate. In principle it could be the nominal price of any real good (or claims on a real good)."
BUT MONETARY POLICY IS ALREADY WORKING ON THE NOMINAL PRICE OF A BASKET OF REAL GOODS. (Hint: the basket of goods in question is completely unrelated to the basket on which CPI is measured.) Incidently, working on a basket of real goods is far better than just one for stability reasons (I believe that was why Milton Friedman advocated that if we insisted on a precious metal banking for the currency we adopt a bi-metal standard).
The issue though is real rate of exchange between claims on the basket of goods backing the currency and all the other stuff on the goods market.
Posted by: Adam P | April 21, 2009 at 02:05 PM
Adam: you lost me. Are you referring to the "backing theory/real bills doctrine/fiscal theory of the price level"?
It's going to take me some time to re-wrap my head around that theory, but like the Bishop on heresy: "I'm agin it".
Posted by: Nick Rowe | April 21, 2009 at 02:27 PM
Well, the fiscal theory... money is backed by the basket of goods and services provided by the government.
On the post, still haven't properly read it but on the first pass I don't think the story you're telling works. You still haven't said how the bank, which is targeting the price of gold is able to control the real exchange rate between gold and everything else without using something like interest rates. Where's the transmission mechanism?
Posted by: Adam P | April 21, 2009 at 02:55 PM
I have the same sort of question as Adam, in a different way. If you go back to first principles, and assume we’re at the zero bound, what is it about the central bank’s ability to set the price of gold that would cause this to spread to a more general reflation?
Posted by: JKH | April 21, 2009 at 03:29 PM
The girl with the orange eyes replies:
"If the price of gold starts to slip below the target price, the Bank buys gold, and if it starts to rise above the target price the bank sells gold. More simply, just offer to buy or sell unlimited amounts of gold at the target price. And this is how it used to be done, in the olden days (only there might have been a small spread between buying and selling prices, called the "gold points".
Remind me again how your Bank sets an interest rate? How does the price of gold affect interest rates?
As for the transmission mechanism: even though gold production and consumption is a very small part of GDP, and the stock of gold is a very small part of total wealth, the price of gold sets the pattern for all prices of goods in the economy. If we increase the price of gold, this creates a ripple effect on the demand, supply, and prices of all goods, first the other precious metals, then base commodities, then all other goods and services. It's the cross-price elasticities of demand and supply.
(Monetarists say it's the change in the quantity of money which does the work, not the change in the price of gold. But we know they're wrong, because the stock of money is endogenous, and adjusts passively to a change in demand. And if the supply of money did ever change exogenously, so there was an excess supply of money, it would immediately return to the bank, in exchange for gold. So I really doubt if 'quantitative easing', like adding to banks' settlement balances, in a futile attempt to change the overnight rate, would make any difference at all.)
Remind me again how the transmission mechanism for the overnight rate works in your world. That is such a small part of the total market for loans, and hardly anyone, except a few specialists, ever trades in that market. If you change the overnight rate, why doesn't that just change interest rate spreads, and leave all the other interest rates unaffected? And what's the route from interest rates to the price of goods? A higher price of goods should leave interest rates unchanged, shouldn't it?"
Posted by: Nick Rowe | April 21, 2009 at 03:39 PM
ooops! Stick brackets around this bit from my above comment:
[And this is how it used to be done, in the olden days (only there might have been a small spread between buying and selling prices, called the "gold points".]
That was me speaking, not the girl with the orange eyes. She would say:
"We have always done it this way."
Posted by: Nick Rowe | April 21, 2009 at 03:45 PM
ooops! Stick brackets around this bit from my above comment:
[And this is how it used to be done, in the olden days (only there might have been a small spread between buying and selling prices, called the "gold points".]
That was me speaking, not the girl with the orange eyes. She would say:
"We have always done it this way."
Posted by: Nick Rowe | April 21, 2009 at 03:50 PM
"The girl with the orange eyes replies..."
Nick, are you a trekkie?
Posted by: Adam P | April 21, 2009 at 03:51 PM
And remember, the girl with orange eyes really does have much more actual historical support for her transmission mechanism than you do for yours. The price level was determined for centuries by the gold content of coins, and the exchange rate of gold for paper. Interest rate control of the price level is a VERY recent phenomenon.
That's a statement about our history, not counterfactual history.
Posted by: Nick Rowe | April 21, 2009 at 03:58 PM
"Monetarists say it's the change in the quantity of money which does the work, not the change in the price of gold. But we know they're wrong, because the stock of money is endogenous, and adjusts passively to a change in demand"
Can you expand a bit on this, Nick?
Posted by: JKH | April 21, 2009 at 04:02 PM
Unfortunately, I need my monetary history explained in terms of how central banks actually implemented their policies at an operational level. If I can't visualize it that way, it becomes counterfactual by ethereal default.
Posted by: JKH | April 21, 2009 at 04:06 PM
During the era of the gold system from about 1870 to the first world war the Bank of England stood ready to buy any and all gold tendered to it at a price of $21.85.
There were two interrelated reasons this worked. One was that no one expected the $21 price to change and this was because it was massively above any foreseeable market clearing price. Consequently, no private hedgers and/or speculators had any reason to buy gold on the possibility that the price would rise above the Bank of England support price.
But according to Robert Mundell and others, in the 1930s this $21 dollar price was too low and the supply of gold was too small to finance an expanding world economy. This was the fundamental underlying cause of the Great Depression and the depression did not end till countries went off the gold standard and FDR raised the price of gold to $35.
For a gold based monetary system to work the price of gold has to be so high that no private speculator or hedger has any reason to hold gold. Consequently, your story related above is unrealistic or will not work because it only deals with small changes at the margin. But when you get a system wide change in expectations it would be completely unable to deal with the problem.
The results would be something like when the base interest rate approaches zero in the current
system.
Posted by: spencer | April 21, 2009 at 04:25 PM
I'm not a trekkie! Why? Are there girls with orange eyes on Star Trek?
JKH @4.02 : I think the girl with the orange eyes is wrong when she says the quantity of money is irrelevant, because it's endoenous. I am just casting her in the standard "horizontalist" Neo-Wicksellian role in her universe, so that she is an exact a mirror as I can make her of our standard views on monetary policy. But she is right that the stock of paper money is endogenous in her world. If the central bank over-issues paper money, there is an excess supply of money, so people return the excess to the central bank in exchange for gold. "The Law of Reflux" it used to be called. And those who say QE is irrelevant in our world rely on that same Law of Reflux, only operating via bank reserves.
I think the way I have described the price of gold being set is about accurate historically. The bank just redeemed paper money for gold on demand, and sold paper money for gold on demand too. Or minted gold coins on demand (taking a seigniorage 'cut' as a recompense). But this mechanism pre-dates central banking, and even pre-dates banking. And it even predates much of any sort of financial system. And usury laws that would have made Islamic banking look soft would have prevented interest rates doing much of anything.
I wish a real economic historian would join in. But there are very few monetary historians. I am out of my depth in real history. (And I think I got that bit about "gold points" wrong; if I now remember correctly I think they were the cost of shipping gold across the English Channel, and so created an exchange rate band under the gold standard.)
Posted by: Nick Rowe | April 21, 2009 at 04:31 PM
Nick,
Be careful in casting your primary gold policy operator as a Neo-Wicksellian horizontalist.
Not only will you have Krugman back on your case, but a whole bunch of other more formidable plaintiffs.
:)
Posted by: JKH | April 21, 2009 at 04:44 PM
"The price level was determined for centuries by the gold content of coins, and the exchange rate of gold for paper."
Yes, but there wasn't so much monetary policy being conducted back then. Your confusing the fact that the price level was determinant with the it being determined by a monetary authority. In both cases we have a price level, the question is whether this helps do good stuff for our economy.
"And remember, the girl with orange eyes really does have much more actual historical support for her transmission mechanism than you do for yours."
No, not if the issue is monetary policy's ability to maintain full empoyment or at least avoid deep recessions/depressions.
Posted by: Adam P | April 21, 2009 at 04:53 PM
I’m interested in the role of the central bank as a gold dealer with a natural long position.
I think your link to Sumner’s comments included a comparison with the assets of today’s central banks.
Basically, a central bank must monetize the assets it chooses to hold.
A point most people don’t get is that currency issuance isn’t discretionary monetization. It depends on public demand for currency. You can’t force it except via helicopter drop, which is crazier than the dream you had.
Discretionary monetization requires forced expansion via excess reserves.
So if orange eyes wanted to reflate gold, she’s need to monetize gold via excess reserves.
Then she’d probably have to pay interest on reserves as well, especially if operating above the zero bound – or not. Either way, she’d have a decision to make on interest rates.
Would she understand this?
Posted by: JKH | April 21, 2009 at 05:03 PM
Very interesting post by Sumner. I like the idea of looking at 'futures' but still have to wrap my brain around some of it.
As for Japan? I would take a look at the price of gold in relation to the Yen. If memory serves me correctly, not only were Japanese consumers hoarding Yen they were also exchanging Yen for gold bars. I remember being fascinated by it because one would think they would hold the GBP or U.S. dollar instead of bulk gold.
In the U.S., money supply should have been tightened in the late 90s' but it wasn't. The bubble burst. In this latest case? The Fed continued with the Greenspan policy. It was a tornado in the making, just didn't know when it would hit.
Btw, I hate CPI. I remember while in Uni (seems centuries ago) and grocery shopping every day, cheaper that way. Whenever the CPI was announced the grocery store always raised one or more of the 'goods in the basket' by the announced inflation rate. Very irritating. I learnt to buy my items before the announcement. Sorry 'bout the rant but I've never seen CPI as an indicator to what was actually happening on the ground, more like permission to increase prices by a set amount while ignoring supply/demand forces.
Wasn't paper replaced because gold was too heavy to carry? The paper then evolved to legal tender, meaning the bank/government would back the value of that paper? Sellers/buyers could then in turn exchange the paper for the bulk gold and a Fee. If the bank/government wasn't issuing new paper then those holding the paper would have to pay additional fees for the storage or carry the gold. Would fees be the transmission mechanism or the inflation indicator? But if there was a rush to exchange paper for gold then the bank/government wouldn't earn any fees but would have the gold they charged and saved to live on but could charge more in fees when someone deposits gold with them.
Elizabeth I tried to rebase the currency with gold. I believe it was called Gresham's Law. Henry the VIII debased the gold coin to pay for his wars.
Posted by: Dee | April 21, 2009 at 05:24 PM
Nick,
I think Spencer above has a very important observation on gold standard cycles:
“But according to Robert Mundell and others, in the 1930s this $21 dollar price was too low and the supply of gold was too small to finance an expanding world economy. This was the fundamental underlying cause of the Great Depression and the depression did not end till countries went off the gold standard and FDR raised the price of gold to $35. For a gold based monetary system to work the price of gold has to be so high that no private speculator or hedger has any reason to hold gold. Consequently, your story related above is unrealistic or will not work because it only deals with small changes at the margin. But when you get a system wide change in expectations it would be completely unable to deal with the problem. The results would be something like when the base interest rate approaches zero in the current system.”
Sounds like a sort of “tipping point” or the gold equivalent of a “Minsky moment” when the standard reaches its inflection point of failure.
Posted by: JKH | April 21, 2009 at 05:35 PM
The girl with orange eyes replies:
"Spencer: I agree that we kept the price of gold too low in the 1930's. With an expanding world economy, the demand for money was growing. The result was deflation and unemployment. But then our history diverged from yours. We decided that it was a mistake to keep the price of gold fixed. We increased the price of gold, and kept increasing it over time to prevent deflation and unemployment.
Now some gold was held by central banks, but most gold was held by private citizens, as it always had been, for jewelry, other "industrial uses", and as a store of wealth. I don't understand what you mean by saying the $21 price was "...massively above any foreseeable market clearing price". Do you mean "below"? (Because I admit it was too low). But whatever price we set is always a market-clearing price, because we buy and sell gold so the market (including our purchases and sales) always clears. (I admit there were some episodes in which central banks mistakenly tried to use OMO to keep interest rates down, and this misguided intervention in loanable funds market did cause us to run out of gold reserves, but we soon abandoned this barbarous novelty of trying to control interest rates).
Adam P: yes, we didn't really used to do much in the way of monetary policy. We kept the price of gold fixed. But we eventually switched to adjusting the price of gold on a discretionary basis, to make sure that monetary policy served our needs, rather than fix the price of gold at some arbitrary level, and let the real economy suffer the consequences. So we were able to maintain full employment, though we did have trouble with the Phillips Curve, just as you did, until we switched to targeting inflation."
JKH: I'm going to think about your comment. Need to cook dinner.
Posted by: Nick Rowe | April 21, 2009 at 05:42 PM
I think Sumner’s idea of intervention in nominal GDP futures is fundamentally flawed.
That’s assuming the idea means that the central bank actually intervenes in the market. I don’t know that. I can’t tell because Sumner doesn’t delineate the central bank’s operational role in the proposed NGDP futures market. I don’t think there is an operational role. That means the operational requirement of some instrument like overnight interest rates remains.
The futures market for any asset, if it is traded, reflects the market’s valuation of that asset.
“Nominal GDP futures” is a bet. That’s the asset. The asset is a bet.
So the market is betting on future nominal GDP.
What does the central bank’s intervention in NGDP futures have to do with the real economy, other than it being a bet on its own policy as it affects that economy?
The central bank’s bet is not the central bank’s policy. It is a bet on its policy.
What is its policy?
It’s not intervention in the futures market.
Sumner’s idea is something the Hunt Brothers tried 3 decades ago in a related market.
Posted by: JKH | April 21, 2009 at 06:21 PM
JKH, The Hunt Brothers didn't have a printing press. The Fed would buy and sell unlimited 12 month NGDP futures at a 5% premium. Each sale by the Fed would represent a purchase by the public--a bet that next year's NGDP would rise more than 5%. This would trigger ofsetting open market sales in securities by the Fed. (ande vice versa) The process would continue until the public expected exactly 5% NGDP growth. This has been run by lots of famous economists in the past 23 years. No flaws found yet.
Nick, Are you familiar with Irving Fisher's "Compensated Dollar Plan" to adjust the price of gold up 1% each time the WPI fell 1%., and vice versa? Robert Hall likes the proposal, and I did an article about it a long time ago. I think you are right that this would eliminate liquidity traps. I still like my futures targeting idea better, as it deals with policy lags in an optimal way--but your idea will work. And I am in the "anything but interest rates" camp right now. BTW, The plan was first proposed by Rooke around 1820. So its been around for a while.
Posted by: Scott Sumner | April 21, 2009 at 08:57 PM
Scott, Nick,
Thank you for the additional information on the GDP futures idea. I don’t want to derail Nick’s post with distraction, particularly regarding an idea that apparently already has been demonstrated to be indisputably successful, although never implemented, but I’ll just make a few comments on why the idea puzzles me.
If a central bank has enough fire power, I can see how betting on the price of physical or future gold will nearly instantaneously force the price of gold to the price of the bet, although periodic policy breaks can occur. Gold is a commodity where at least some physical movement in the cash market is expected to validate the central bank’s valuation. I assume breaks in policy occur finally when the bank can no longer tolerate such physical movement as it affects its own inventory. But when the policy is in place, and holding, the cash market in the commodity validates it. And because of that, the futures market should converge to the cash market if the policy is working. To the degree that the commodity is being exchanged for actual money, it becomes a monetary policy.
I’m having difficulty intuitively seeing how a bet on future GDP will necessarily converge to actual GDP in the future, simply due to the bet alone.
Suppose the Fed announces today it will start to buy GDP futures as its monetary policy.
Say the economy yesterday was on course for a deflationary depression. The Fed starts to buy GDP futures today. The public sells futures to the Fed in exchange for securities from the Fed. Presumably the reason for such an exchange of securities is to sterilize any potential monetary base effect due to the Fed’s cash outlay, which I suppose must assume that the Fed is subject to “margin requirements” for futures purchases.
The Fed’s futures position, long 5 per cent GDP growth, is a bet.
Surely the Fed must do “something else” other than buy futures in order to win the bet.
When a central bank bets on gold, the bet is the policy. This is because the bank has the capacity to hold a cash inventory of gold of sufficient size, and at its option take futures positions if deemed necessary, to be able to enforce its chosen gold exchange rate. When the bank bets on its own gold policy, and the policy holds, it will be successful in forcing the convergence of futures to cash, because it has the physical commodity in sufficient size to be able to deliver it to or take delivery from any opposing betting counterparty on contract expiry, if necessary, in order to hold the price at its chosen rate.
When the central bank bets on future GDP, it doesn’t have the capacity to hold “actual GDP in inventory”, in order to be able to force the convergence of expiring futures to cash, and therefore enforce the “price” of a 5 per cent growth rate, if necessary. It can’t do this, because there is no such thing as “actual GDP in inventory”.
If the Fed buys futures today, and a deflationary depression occurs one year from now as the natural and predestined outcome of yesterday’s failed policy, how can such a futures market bet make any difference to this outcome, on its own? Those who are convinced that deflationary depression is the inevitable outcome in the absence of any other Fed policy will simply short futures, selling them to the Fed. With an actual deflationary depression outcome, the central bank will lose the bet and be forced to pay out the loss on its bet to the counterparties.
Therefore, I return to my original contention, which is that there must be a separate central bank policy that is used in support of such a bet. This policy hasn’t been specified.
A bet on GDP growth is a bet on an observation, not a bet on participation in that observation, unless some additional participating policy action influences the outcome of the observation. Such a bet requires some “Soros reflexivity” in order to influence the outcome. (And Soros would be just the person to bet against it otherwise.)
Conversely, a bet on cash or futures gold is simultaneously a bet on participation in that market, due to the central bank’s actual and potential participation in the cash gold market, and its ability to force convergence of futures to cash.
Where is the rest of the monetary policy that evidences the central bank’s participation in cash GDP, other than the salaries it pays its employees to design the futures bet?
Posted by: JKH | April 22, 2009 at 01:29 AM
Nick:
Sounds like your describing the scheme Fisher presented in 1920.
Just so. The classic gold-standard is premised on buying all that is offered and redeeming all notes on demand; however, I think you'd be pained to find a period of time where the CB conducted this policy with a 100% reserve. There has always been a gold cover ratio less than one. CBs conducted policy by manipulating that cover ratio. i.e., by monetizing instruments other than gold.
It cannot be so. The short-rate reflects the position of the economy relative to gold. It is or was, a short-run price-level metric.Posted by: Jon | April 22, 2009 at 03:05 AM
I must admit to not having read all the posts, but isn't there something missing from this story? Gold is internationally traded. Targeting the gold price is essentially targeting the exchange rate. This is essentially a currency board. It will eventually have crises like every managed exchange rate. Ask George Soros.
Posted by: reason | April 22, 2009 at 03:21 AM
I see spencer said something similar above. Spencer doesn't comment very often, but when he does (I know him from other sites) you better listen. Nobody knows the data better than him.
Posted by: reason | April 22, 2009 at 03:23 AM
And please - how can Canadians think only domestically. I thought that was an (US) American disease.
Posted by: reason | April 22, 2009 at 03:27 AM
"When the central bank bets on future GDP, it doesn’t have the capacity to hold “actual GDP in inventory”, in order to be able to force the convergence of expiring futures to cash, and therefore enforce the “price” of a 5 per cent growth rate, if necessary. It can’t do this, because there is no such thing as “actual GDP in inventory”.
That's my problem with the NGDP future ideas as well. It seems kind of like setting the path for election results by betting on your candidate at Intrade.
Posted by: bob | April 22, 2009 at 07:19 AM
Yes, I also agree with JKH.
Furthermore, their is nothing new in Nick's story and it doesn't resolve any of the debates that we've been having. Here's the question, why does adjusting the amount of gold that the note is worth affect the CPI? The reason is that it changes the quantity of money in the economy. Now, the issue is whether changing the quantity of money can stimulate aggregate demand directly (as Nick says) or only by changing real interest rates (as I claim). This is important because if Nick is right then we can stimulate AD today with enough money supply. If I'm right we need to credibly commit to higher future inflation.
Neither story says monetary policy can't work, but they differ in what will work. The price level determinacy issue is secondary to the question of stimulating aggregate demand. After all, in fully flexible price models monetery policy still determines the price level, we just care less what it is because money is neutral, monetary policy no real effects.
Posted by: Adam P | April 22, 2009 at 07:42 AM
Thinking about Scott Sumner’s GDP futures proposal further, it seems to me that the value of such a GDP futures contract for the purpose of monetary policy lies in its information content only. The central bank, to the degree it pays attention to a GDP futures market, should look to it as an indicator, not an instrument. In Scott’s proposal, GDP futures seem to be the only indicator. Moreover, it seems then that the idea is to “leverage” this exclusiveness by making GDP futures the designated instrument of intervention as well.
It appears the proposed policy includes conventional OMO market intervention, as a knock-on function of “required” intervention in the GDP futures markets. But if the policy includes the “kicker” of conventional OMO, there’s no reason why Fed intervention or involvement in the GDP futures market is required at all. If the market understands and has seen the announcement that the Fed has a particular GDP growth target, the futures market will vote on the expected success of the Fed in meeting this target by setting a futures price that agrees or doesn’t agree with the target. If the market price deviates from target, the Fed will interpret this as a signal that conventional OMO is required. The Fed then will intervene with conventional OMO until the futures price aligns with its target. That’s a potentially workable theory.
But on that basis, there is no need for the Fed to intervene in the GDP futures market. The required monetary policy instrument is conventional OMO, not GDP futures intervention, and not the combination of the two.
And in that case, the Fed must choose a particular OMO instrument, such as the overnight rate. But then we are back to the current system, just with a different indicator variable(s). It’s comparable to Volcker targeting broad money supply. His instrument was still OMO.
When the central bank uses conventional OMO, it is targeting interest rates, either explicitly or implicitly. It’s “target” is to have a demonstrable effect on the level of interest rates, whether a specific level or a non-specific directional change. If it sells, it expects rates to go up. If it buys, it expects rates to go down. It must make the judgement on how much OMO is the appropriate response to its indicator. And the judgement includes either a goal or an expectation as to the elasticity of the interest rate change with respect to OMO. Again, Volcker did not triple interest rates overnight. There’s always a judgement involved as to how much of an interest rate move is appropriate – using either an explicit interest rate target (Greenspan, Bernanke) or a non-specific muddle through process (Volcker).
Finally, it occurs to me that under Scott Sumner’s type of proposal, central bank intervention in the futures market for its indicator is precisely the wrong thing to do. If the central bank wants to formulate monetary policy according to an exclusive indicator, it should make extra sure that the indicator is a reflection of the true market, not one that is manipulated by the bank itself.
Posted by: JKH | April 22, 2009 at 08:32 AM
Interesting comments for me to digest. Not sure I can do them all justice.
I am going to drop the "Girl with Orange Eyes" persona. (It was probably getting irritating anyway). She is a straw woman, and a flawed monetary economist.
She had one job to do, and I think she's done it. Her job was to get people thinking about alternative monetary policy control instruments, and alternative channels of the transmission mechanism. And nobody can say monetary policy could never control that instrument, and that transmission mechanism could never work, because for centuries (millenia?) that WAS the control instrument, and that WAS the transmission mechanism (and the effect of gold discoveries, debasement, etc. on inflation proves that transmission mechanism worked).
Did it work well? Was it immune to crises? Is this a good way to conduct monetary policy? No, No, and No. But I don't need to argue that it was. Because all we need right now is some blunt control instrument, that can be controlled, and that does have some sort of transmission mechanism, to prove that QE CAN work to increase AD.
Yes, I remember reading Irving Fisher's compensated dollar plan, though my memory is a bit fuzzy. It is certainly relevant to my story, and it was at the back of my mind as I wrote it. But I wanted to keep an exact a parallel as possible between the alternative universe and our own. They have exactly the same inflation target as the Bank of Canada, and use judgment and modeling to find the right setting of the control instrument, so the protagonist's computer model of the economy will work just as well (or badly) in the alternate universe.
Now, I laid out the girl with orange eyes' theory of the transmission mechanism above. I have to say that I find her story unconvincing. The link between the price of gold and the prices of other goods in the CPI, via cross-elasticities of demand and supply, sounds like a VERY weak causal chain to me. But then I could say exactly the same thing about the interest rate theory of the transmission mechanism. The cross-elasticities of demand and supply between the overnight rate and all other interest rates also sounds like a VERY weak causal chain to me, and that's even before we get to the chain from all interest rates to the CPI. The price of gold seems loosely linked to the CPI. But then the overnight rate of interest seems loosely linked to all the other rates of interest. If she got into an argument with one of you guys, and you pointed out that the relative price of gold could fluctuate, she would immediately point to the changing spreads between the overnight rate and all the other interest rates. And empirically, she would win that argument hands down, using recent data. The price of gold relative to the CPI has been much more stable recently than the ratio of the overnight rate to the average rate of interest. Plus, as I said before, she has way more centuries of history on her side.
What's wrong with her story of the transmission mechanism? It ignores how changes in the price of gold, induced by her central bank's buying or selling gold for money, would change the stock of money, and create a generalised excess supply (or demand) of money. She is the gold-price counterpart of the Neo-Wicksellian.
But if you have a model in which the CPI, the price of gold, and the overnight rate of interest, are all linked via equilibrium conditions, then if her story of the transmission mechanism is wrong, the Neo-Wicksellian story of the transmission mechanism must also be wrong.
I haven't replied to anyone directly here, but I hope I have replied to many of you indirectly.
Some direct replies:
JKH @ 5.03. You lost me in this comment, I'm afraid. In particular, I don't understand what you mean by:
"Basically, a central bank must monetize the assets it chooses to hold."
spencer: was that a typo in your comment? Did you mean "below" rather than "above" in this: "it was massively above any foreseeable market clearing price"? If so, I basically agree with your general thrust. But I don't want to get too deeply into an argument over gold standards, and exchange rates, etc., because that's not the point of the post. It's really about the existence of ANY alternative control instrument and transmission mechanism, and gold is just a handy example, with much more history than plywood.
Posted by: Nick Rowe | April 22, 2009 at 08:37 AM
“Finally, it occurs to me that under Scott Sumner’s type of proposal, central bank intervention in the futures market for its indicator is precisely the wrong thing to do. If the central bank wants to formulate monetary policy according to an exclusive indicator, it should make extra sure that the indicator is a reflection of the true market, not one that is manipulated by the bank itself.”
This may appear to contradict what I said about gold, but it doesn’t.
Gold as a commodity has the advantage of having an actual cash/futures market relationship. This makes it a conceivable candidate for use as a central bank policy instrument. On that basis, it can potentially serve as both an indicator and an instrument.
The idea of NGDP futures includes no such cash/futures relationship. It is simply a bet. On that basis, it has no useful role as an instrument. Therefore, it can only serve as an indicator. Therefore, the central bank should not intervene in its market.
Posted by: JKH | April 22, 2009 at 08:49 AM
Scott and JKH:
Assume, just for the sake of argument, that GDP was a costlessly-storeable commodity. Would that change your arguments in any way?
I'm trying to get my head round this myself.
Posted by: Nick Rowe | April 22, 2009 at 08:50 AM
Adam P:
Suppose I were having the same argument with the orange-eyed girl as I am having with you. In other words, suppose I were trying to convince her that changing the interest rate as a control instrument could affect AD. This would be her version of your comment:
"Furthermore, their is nothing new in Nick's story and it doesn't resolve any of the debates that we've been having. Here's the question, why does adjusting the amount of [nominal interest-bearing notes] that the note is worth affect the CPI? The reason is that it changes the quantity of money in the economy. Now, the issue is whether changing the quantity of money can stimulate aggregate demand directly (as Nick says) or only by changing [the real price of gold] (as I claim). This is important because if Nick is right then we can stimulate AD today with enough money supply. If I'm right we need to credibly commit to higher future inflation."
What would make her argument against me invalid, and your argument against me valid? If it's empirical facts, she's on stronger ground, since she has more history behind her.
Posted by: Nick Rowe | April 22, 2009 at 09:02 AM
Nick,
“Basically, a central bank must monetize the assets it chooses to hold.”
This is fundamental, in my view. It’s particularly essential in understanding what’s happening to the Fed’s balance sheet right now.
Maybe it’s just my language, but here’s the thinking. I suspect you’re quite aware of it, perhaps using different language, or a different way of thinking about it.
The Fed has embarked on a range of credit programs that have expanded its balance sheet. This expansion starts with the asset side, where its so-called credit or qualitative easing is clearly apparent. The Fed has made the deliberate choice to expand its balance sheet, starting with the asset side.
The effect has been the “monetization” of these assets – the creation of new money that the private sector now holds, instead of the assets or underlying assets that it previously held.
When the Fed disburses this newly created money, the money returns to its balance sheet in the form of increased excess reserves, which are a Fed liability. This is the result of the recipients of the credit program outlays cashing their cheques with the commercial banks. That’s how the Fed’s assets end up being equal to its liabilities when the Fed is in the process of expanding its balance sheet.
“Must monetize its assets” is a description of this causality – a central bank that purchases/lends creates the money to do so, which become a liability and the source of funding to support the asset expansion. The immediate liability and funding source is in the form of new excess reserves.
This is an “other things equal” statement. The Fed has ways of liability sterilization as well, as an alternative response to monetization that is forced by its own asset expansion. E.g. it can request that the Treasury issue bills and leave the money on deposit with the Fed. This effectively converts excess reserves to Treasury deposits. But treasury deposits are still a form of money, so the monetization idea still holds at a more general level.
And I suspect we are going to hear more fairly soon about the idea of the Fed issuing its own interest bearing securities as another liability based sterilization option.
The idea is relevant here I think because a central bank’s purchase of gold must effectively monetize it in the same way, unless the new money is sterilized.
Posted by: JKH | April 22, 2009 at 09:16 AM
Nick: "Here's the question, why does adjusting the amount of [nominal interest-bearing notes] that the note is worth affect the CPI? The reason is that it changes the quantity of money in the economy.... "
NO,NO,NO. I'll repeat one of my comments from the your last post:
Since price are sticky the central bank can change real interest rates (since changing the nominal rate is a change to the real rate if prices can't move right away). Changing real rates allows the central bank to affect AD. Buy influencing AD the central bank influences inflation via a Phillips curve.
Nothing to do with the quantity of money per se. The key is being able to change real rates, by whatever means works.
Posted by: Adam P | April 22, 2009 at 09:23 AM
“Assume, just for the sake of argument, that GDP was a costlessly-storeable commodity. Would that change your arguments in any way?”
If all economic units were restricted in their expenditures to the purchase of defined GDP units (equal baskets of GDP-proportionate shares of consumer and investment products), GDP cash and futures markets would seem a bit closer to theoretically conceivable intervention instruments.
But the inclusion of consumer product components would distort the markets in the event of central bank intervention, because the central bank can't use consumer products.
And the financial system including the central bank might be a little different because the basket includes real investment products.
Another way of looking at it is that gold is fairly simple in the sense that it is an economic stock; GDP is a flow that includes the outright consumption of consumer goods and the partial consumption of investment stock.
So there would seem to be some minor impediments to the premise.
See also my 8:32 a.m. and 8:49 a.m.
Posted by: JKH | April 22, 2009 at 09:37 AM
JKH: OK. I understand you now. (I'm not sure why I didn't understand you the first time, since your language was not that unusual, but anyway.)
Sure, if the central bank uses the price of gold as its control instrument, it must stand ready to buy and sell gold for money (monetise gold) at the announced price. That's how I imagined the gold price control mechanism operating.
"So if orange eyes wanted to reflate gold, she’s need to monetize gold via excess reserves."
I don't see where the excess reserves come in. If she wants to increase the price of gold (reflate gold?) she just announces an increase in the price of gold at which her central bank will buy or sell unlimited quantities of gold.
And in her way of viewing the world (through her orange eyes) the increased price of gold would create an increase in real output in the short run, and an increase in the general price level in the long run, and either of these would increase the demand for money, and so her central bank would eventually gain gold reserves and the supply of money would expand in response to the increased demand.
But you and I might say she has the causation backwards. We might see her central bank as needing to increase the supply of money when it buys more gold to raise the price of gold. And it is the increased supply of money that causes an increase in real output in the short run, and an increase in the general price level in the long run, until the demand for money increases to match the increased supply.
We end up in the same equilibrium position as she does, but our story of how we get there differs from hers.
But then our argument with her is exactly parallel to my argument with you and Adam P, I think, when it comes to interest rates as the control instrument.
Adam P.: And here is how Orange Eyes would argue with me:
"Since price[s] are sticky the central bank can change [the] real [gold price] (since changing the nominal [price] is a change to the real [price] if prices can't move right away). Changing [the] real [gold price] allows the central bank to affect AD. Buy influencing AD the central bank influences inflation via a Phillips curve.
Nothing to do with the quantity of money per se. The key is being able to change [the] real [gold price], by whatever means works."
Posted by: Nick Rowe | April 22, 2009 at 09:54 AM
“I don't see where the excess reserves come in. If she wants to increase the price of gold (reflate gold?) she just announces an increase in the price of gold at which her central bank will buy or sell unlimited quantities of gold.”
I would have thought the re-peg of gold up in price would cause inflows of gold to the central bank and therefore monetization? Don’t central banks normally re-peg when the market has been betting against them, i.e. buying their gold and depleting their inventories? Maybe I’ve got that wrong. But you seem to suggest that in your next paragraphs.
Posted by: JKH | April 22, 2009 at 10:09 AM
Just to be clear about the rhetorical/teaching device I am using here.
I have an argument going with people on one side of me. So I have set up a persona, the Girl with Orange Eyes, on the other side of me. While I am trying to occupy the centre. By showing how their arguments against me have a parallel argument she could make against me, and trying to get them to consider what might be wrong with her arguments, I am trying to get them to consider what might be wrong with their arguments.
I like it. It helps me think more clearly about things that must be wrong with their arguments. But I don't know if it's working for anyone else.
Posted by: Nick Rowe | April 22, 2009 at 10:10 AM
"Changing [the] real [gold price] allows the central bank to affect AD"
That's the part I have a problem with.
Posted by: Adam P | April 22, 2009 at 10:12 AM
"But I don't know if it's working for anyone else."
It's sort of working, except I don't see the center as well defined. You're more a broker and adjudicator of debate, than taking a principal position, aren't you?
Posted by: JKH | April 22, 2009 at 10:18 AM
Adam P @ 10.12
So do I, if I am not allowed to talk about the excess supply or demand for money.
We know that in long run equilibrium it MUST be true that an exogenous permanent doubling of the nominal price of gold, with all other nominal variables, including the nominal stock of money, endogenous, will cause the nominal price level to permanently double. So the AD curve (in {P,y} space, must have shifted rightwards/upwards. But what is the causal chain of the transmission mechanism that shifts the AD curve?
I would talk about the excess supply of money created by doubling the price of gold, and how it spills over into an excess demand for all goods in all markets (including, inter alia, the bond market).
But Orange Eyes won't let me talk about the excess supply of money, because she says it's endogenous, and says it would immediately return to the gold window of the central bank if there were such an excess supply of money.
So here's her story: "An increase in the nominal price of gold, holding all other prices temporarily fixed, means the real price of gold is higher. The first-order condition for consumer choice sets the marginal rate of substitution between gold and other goods equal to the relative price of gold. So when that relative price increases, the demand for consumption of all other goods increases."
Maybe (if I could figure it out) she could also tell a story about how firms' investment demand increases too.
Do you buy her story? Neither, really, do I. It's not that it's logically incorrect. Her MRS=Pg/P story is the logical counterpart of your consumption-Euler equation. But it just seems too unimportant quantitatively. If we pressed her, she might give more details, about how the rise in the price of gold causes a ripple-effect and rise in the demand and price of close substitutes like silver, and other assets like land, physical capital, and stock prices, and thus into investment, and consumption (since real asset prices would rise relative to the price of current consumption). She might even mention bond prices, and hence interest rates, along the way.
But then if she pressed us on our story of the transmission mechanism, she would find it hard to believe that the quantitative effect of the overnight rate on AD could be big enough to matter, since hardly anybody borrows at that rate. And we would resort to talking about ripple-effects from the overnight rate onto other interest rates, share prices, land prices, and hence consumption and investment. And we might even mention the price and demand for gold along the way.
Again, we know that there must be some transmission mechanism from gold prices to shifting the AD curve, both from history, and from the homogeneity thought experiment of long-run equilibrium. But the actual story is less clear. I would tell it through a change in the price of gold initially causing an excess supply of money, and that excess supply of money spilling over into all markets, including the bond market, but also the stock market, land market, housing market, supermarket, and gold market.
Posted by: Nick Rowe | April 22, 2009 at 10:54 AM
JKH: OK, I (think) I understand you now. You mean that increasing the price of gold would cause the central bank's GOLD reserves to increase? Yes. Agreed.
(I don't know why I had so much difficulty understanding your comment.)
Posted by: Nick Rowe | April 22, 2009 at 10:59 AM
I will write something a little later on the ON rate transmission mechanism. I think you and/or orange eyes has been looking for something on that.
Posted by: JKH | April 22, 2009 at 11:09 AM
Target based on the number of newly employed engineering graduates (minus some defense employers). When you need to cool off the economy, imprison them and torture. When you need inflation you release them, but still torture.
Probably better to understand the real causes of economic growth and target that, before changing things now only to have to change them again later. Targetting based on copper or ammonia or novel medical/materials/communications/energy products is superior to gold. Gold would inflate in value as has been stated and has been historically noted (isn't enough gold). If I understand this correctly, gold is counter-cyclical and you are trying to target based on a cyclical GDP...cycle. If anything target inversely. For instance gold is mined as a by-product of copper so whatever massive affect on mines (5x pointless growth in mining sector my guess) you have by retarding or encouraging gold mining is going to be partially undone by the opposite price movement of copper and nickel. Gold targetting also kills India while helping Canada, South Africa, Aussie! Aussie! Aussie!, Russia, Mexico....
I like the idea of targetting a lagging indicator of GERD (only counts 50% of defence) R+D spending a decade ago.
Posted by: Phillip Huggan | April 22, 2009 at 12:42 PM
JKH: That would be worthwhile.
Adam P: Thinking again about Orange Eyes' story of the transmission mechanism, it would theoretically be derivable as follows:
Solve for the equilibrium nominal price of gold, as a function of the CPI, real income, etc. That's just standard microeconomics. Now invert that function, to solve for the short run AD function, by solving for real income, taking the CPI as fixed, as a function of the price of gold.
Now strictly, that AD function should have proper disequilibrium foundations, with quantity constraints properly recognised. But since we normally ignore that when we derive our AD function, we can't hold her to a higher standard.
Posted by: Nick Rowe | April 22, 2009 at 12:49 PM
Nick: "An increase in the nominal price of gold, holding all other prices temporarily fixed, means the real price of gold is higher. The first-order condition for consumer choice sets the marginal rate of substitution between gold and other goods equal to the relative price of gold. So when that relative price increases, the demand for consumption of all other goods increases."
No, this reasoning doesn't work. As a passing observation, this story with 'oil' in place of 'gold' seems at first glance to imply we should have had booming AD around the oil shock in the 1970's. If so then why all the unemployment back then?
But anyway, what's the problem with your chain of reasoning? Well, general equilibrium. We're treating gold as a consumption good here, not as money. So, we start out in equilibrium with marginal rates of substitution equalized and all that. Then we raise the real price of gold, does this raise AD? No. The demand for gold is part of aggregate demand. The rise in demand for non-gold goods is equal exactly to the fall in demand for gold.
Now maybe you might say that's ok, I'm happy to raise total demand for all non-gold goods. Is that better? No, in fact it's probably worse. First of all the amount you can increase non-gold AD is bounded by the real value of initial gold consumption. So you have your own version of a zero bound and you can't get around it with promises for future inflation. Moreover, if gold is a small part of consumption then this bound will bind quickly and limit what you can do to AD. On the other hand, if gold is a large part of the initial consumption basket then raising it's price has a large income effect that works against the substitution effect and brings AD down, that's why the effect of the oil shock was to reduce AD. Of course, if gold is an inferior good then the income effect works in your direction, still there's another problem.
Suppose somehow gold is an inferior good but still a large part of aggregate consumption. Gold is also something that requires resources to produce and arbitrarily raising it's price will cause resources to be diverted to produce more of this inferior good. That's not a way to make society wealthier.
Now, why does none of this apply to using interest rates as the intrument. Because government debt (not necessarily money) is not a consumption good, virtually costless to produce and can function as a store of value. It's also not produced by the private sector. (I'm using as my definition of money anything the government agrees to accept as tax payment.)
Posted by: Adam P | April 22, 2009 at 01:48 PM
Nick,
I don’t have a lot to say on the ON rate transmission effect, because it seems fairly intuitive to me, and I probably have little novel to say about it. First, the ON rate transmits itself through all other rates. I’d go further and say that the Fed funds rate transmits itself to every interest rate in the world, due to the central position of the US dollar.
Yield curve transmission happens in the sense that the risk free yield curve incorporates expectations for the risk free ON rate into the future. The set of risky yield curves is then built on top of the risk free yield curve with market set risk premiums.
Interest rates of all types are then transmitted to the real economy by the demand for credit. E.g. the low rates of a few years ago translated to massive “mortgage equity withdrawals” in the US, which stoked housing and other consumer markets and caused a bit of reflation at the margin. But not enough to prevent the collapse of a credit bubble that seemed foreordained by that point, due to the enormous build in outstanding debt credit over many years.
The Fed doesn’t create the money that leads to inflation. Commercial banks do. The Fed regulates the commercial bank spigot through the fed funds rate and its transmission through the term structure and the prevailing risk premium structure. The money supply that matters to inflation is a function of credit creation, which is a function of interest rates.
In my view, the monetary base is virtually irrelevant. Currency in circulation is determined by public demand, and probably follows a growth path that is reasonably close to nominal GDP. (I don’t know; haven’t checked it; don’t particularly care.) The central bank does not determine the demand for currency; not directly, and certainly not in the short run. The only part of the monetary base that really matters for monetary policy is the level of excess reserves. And that’s only a fine tuning mechanism for the short term interest rate, in that it sets the supply constraint for the competition among banks to avoid penalty borrowing at the central bank. The level of required reserves is irrelevant for this purpose. Only the differential against requirement matters in the competition for funds at a given ON rate level. The requirement can be zero, as in Canada. Countries with positive requirements are only imposing a tax on the banking system.
Economists in general have completely misinterpreted the dynamics of excess reserves at the zero bound. First, they don’t seem to fully understand that the Fed has created the extraordinary excess deliberately as a funding instrument for its unconventional asset expansion. To say that banks are hoarding reserves is absurd, in that they have no control over the creation of the excess. And to suggest that banks should “lend out” this excess is equally absurd. Banks don’t “lend” reserves. Reserves are an accounting mechanism for the balance or imbalance in the flow of funds in the daily clearing of assets and liabilities of banks. Just as the existence of reserves results from Fed creation, the transfer of reserves is almost entirely the result of non-reserve money transactions. The only exception to this is the fed funds market itself, where reserves are transferred directly between banks in order to square their clearing surpluses and deficits at the Fed in a more efficient manner. But like the rest of the market, apart from central bank transactions, fed funds transfers do not create or destroy existing reserves. And to suggest that banks should be attempting to convert the current extraordinary level of excess reserves to required reserves is beyond absurd. This completely ignores the practical reason for the enormous mathematical dimension of the excess proportion that has been created at the zero bound. Required reserves are now about $ 50 billion. The US banking system has about $ 8 trillion in deposits. That’s an average effective reserve ratio of about 63 basis points. Excess reserves now approach $ 800 billion, and will be increasing in coming months. On this basis, in order to convert $ 800 billion from excess to required reserves would itself require sufficient incremental lending to generate a $ 128 trillion expansion in the size of US bank deposits – a 16 fold increase in the size of the US banking system. O...K. Something tells me this is not within the Fed’s expectations. The mathematics of extraordinary excess reserve levels at the zero bound is useful in highlighting the absurdity of the multiplier theory, yet that seems to be how economists in general are analyzing excess reserves.
In summary, the extraordinary excess reserve position exists now, because that is how the Fed has chosen to fund its own credit easing asset expansion at the zero bound. It has nothing to do with attempting to get the banks to lend. The Fed does want the banks to lend, but it knows enough about its own system to know that banks lend on the basis of capital, not on the basis of reserves. The fact that Fed and the Treasury are putting pressure on banks to lend via TARP related moral/economic suasion, and have said nothing themselves about the banks “utilizing” their excess reserve positions, is further evidence of this.
And just as the Fed understands that the excess reserve position it has created for the banks has very little to do with what it should expect from the banks by way of new lending, it also knows that these reserves constitutes a tax on the banks unless compensated at some market rate. Right now, there might be a debate as to whether that market rate should be 25 basis points or zero. The Fed has chosen to play it “safe” (i.e. economically fair) by paying 25 basis points. Because it understands the true reason for and purpose of excess reserves, it does not share the fear of some economists who believe that paying interest on reserves somehow dissuades the banks from “lending out” these reserves. The Fed understands the dynamics of the system it operates.
I understand Scott Sumner will be preparing a response to Greg Mankiw on the subject of reserve interest. Mankiw himself has doubts about charging interest on reserves, but not for all the right reasons. One of the articles that Mankiw references in his post (Edlin and Jaffee) gets the excess reserve story typically wrong in a spectacular way, along the lines of what I’ve described above.
That’s a bit of a digression on excess reserves. But it sort of explains in part why I’m probably one of those “Neo-Wicksellian horizontalists” you speak about. Maybe I can confirm that when I figure out exactly what it means. But I do prioritize interest rates over the quantity of money in attempting to understanding how the financial system works.
Posted by: JKH | April 22, 2009 at 01:53 PM
Nick: "Solve for the equilibrium nominal price of gold, as a function of the CPI, real income, etc. That's just standard microeconomics. Now invert that function, to solve for the short run AD function, by solving for real income, taking the CPI as fixed, as a function of the price of gold."
I don't see it, see my comments on general equilibrium above. (The demand for gold is part of aggregate demand. The rise in demand for non-gold goods is equal exactly to the fall in demand for gold.)
We seem to have lost sight of the basic principle that a lack of aggregate demand is a situation where desired savings exceeds desired investment. Changing the relative prices of consumption goods causes people to substitute between goods today, it won't change desired saving or investment. Now, Nick did say that we actually target the time path of the gold price, not just today's and so if we announce future target prices and can credibly commit to hitting them there may be some scope for inducing intertemporal substitution. However, this just makes things even more distortionary as people produce more gold in anticipation of those times its real price will be high. Also, my comments just above seem to still apply.
Posted by: Adam P | April 22, 2009 at 01:59 PM
http://gregmankiw.blogspot.com/2009/04/observations-on-negative-interest-rates.html
Worth a look I think.
Posted by: Adam P | April 22, 2009 at 03:00 PM
Adam P.: Let's take your passing observation on oil prices first. I know it's not your main argument, but I find it instructive:
" As a passing observation, this story with 'oil' in place of 'gold' seems at first glance to imply we should have had booming AD around the oil shock in the 1970's. If so then why all the unemployment back then?"
The short answer is because the rise in the price of oil was not caused by the central bank's monetary policy, and decision to raise the price at which it would swap oil for money. It was caused by a decreased supply of oil. Orange Eyes would point to the current low overnight rate, and high unemployment, and ask the same question.
Just as we make a distinction between the natural rate of interest and the actual rate set by the central bank, Orange Eyes would make a distinction between the natural real price of gold and the actual real price set by the bank (assuming CPI is temporarily sticky). She would argue that if the bank wants to increase aggregate demand it needs to set the actual real price of gold above the natural real price. She would allow, that holding the nominal price of gold constant, an increase in the supply of gold (new discoveries) will reduce the natural real price below the actual real price, and increase AD. And an increased industrial demand for gold will increase the natural real price above the actual, and decrease AD. And I think history would support her in this.
It's instructive because there is a very big difference between the effect of a change in price (or rate of interest) caused by monetary policy and one that is caused by something else.
That difference matters when we apply general equilibrium theory to a change in the price of gold (or rate of interest).
Start in Walrasian General equilibrium. Just to keep it simple, suppose that gold is not produced. There's a fixed stock, that has always been here, so changes in the demand for gold are not part of AD for newly-produced goods.
A. Now raise the price of one good (gold), holding all other prices fixed. Excess supply of gold, and an equal value (by Walras' Law) of excess demand for other goods, and hence AD. As you say, the increase in AD would be very small, unless we held large values of gold.
B. Now repeat the same experiment for our transmission mechanism. Raise the price of one good (a certain type of bonds), holding all other prices fixed. Excess supply of that type of bonds, and an equal value (by Walras' Law) of excess demand for other goods, and hence AD. The increase in AD will be very small, unless we held a large amount of that particular type of bonds.
Which is larger? Total value of gold in public hands, or total value in settlement balances? I don't know. But in both cases A. and B., this is the wrong thought-experiment. In both cases, we are increasing the price (of gold or bonds) by theorist's fiat. We are not increasing the price through monetary policy. If we raise the price of gold by theorist's fiat, and then let go, the excess supply of gold would cause it to go right back down again. Same with the price of bonds.
So let me try again.
Here, massively oversimplified, is your theory of AD. Forget investment (to keep it easy for me). People can either save their income, or else demand consumer goods (AD). Savings is a function of income and the real rate of interest. Equilibrium is where S(y,r)=0. Invert that function to get y as a function of r, to get your AD function.
Here, massively oversimplified, is Orange Eye's theory of AD. Forget gold production (to keep it easy for me). People can either use their income to add to their stock of gold, or else demand consumer goods (AD). Gold demand (as a flow) is a function of income and the real price of gold r. Equilibrium is where G(y,r)=0. Invert that function to get y as a function of r, to get her AD function.
[Or maybe it's better to express her theory as the condition that the stock demand for Gold(y,r)=stock supply. But I could say the same about the stock of desired savings = the existing capital stock in your theory of AD. I don't know]
The two AD functions are very similar. And there's no obvious reason why the elasticities should be bigger or smaller in one case than the other.
You rightly refuse to apply Say's Law/Walras Law when thinking about savings and investment. That's because you think of monetary policy as holding the real rate of interest away from the natural equilibrium.
Orange Eyes rightly refuses to apply Says Law/Walras Law when thinking about gold demand and supply. That's because she thinks of monetary policy as holding the real gold price away from the natural equilibrium.
I'm sorry this is not as clear as it should have been. It's because I don't understand it as clearly as I want to. And my brain is old and tired.
Damn! Is anyone else having difficulty getting comments accepted, and having to sign in or sign out first?
Posted by: Nick Rowe | April 22, 2009 at 04:13 PM
My old brain is getting clearer, after a short walk. I was forgetting that Orange Eyes would invoke the multiplier.
Again, assume no gold production, to keep it simple. So the demand for gold is not part of AD. Our demand for gold is unit price elastic and unit income elastic. The central bank adjusts the supply of money endogenously so that the price of gold equals some target. Hold all other prices fixed.
The central bank raises the price of gold 1%. This causes a 1% drop in gold demand, and a tiny increase in aggregate demand. But this causes a tiny increase in income, which causes a further increase in AD, etc. (the multiplier process). We only get to equilibrium when income has increased by 1%, so that the demand for gold equals the supply, and the excess demand for money (which equals the excess supply of gold) is zero. So a 1% increase in the price of gold causes a 1% increase in AD. And it doesn't matter how little gold we own. Only the elasticities matter.
Posted by: Nick Rowe | April 22, 2009 at 04:33 PM
JKH @ 1.53:
I think yours is a perfect and clear statement of the "horizontalist" position. I wouldn't call it Neo-Wicksellian, not because it contradicts it, but because you don't say anything about how the rate of interest affects the demand for goods, and inflation. Because that's not the subject of your comment.
As an aside: I have heard talk (did I hear it again, recently, here, maybe from you?) about the Fed creating it's own T-bills? My question is this: if the Fed pays 0.25% interest on reserves, how is that different, economically, from the Fed selling T-bills offering 0.25% interest?
Second question: as far as I can see, the balance sheet of the Bank of Canada looks very different (see Stephen's recent post), in particular on the magnitude of banks' reserves at the Bank of Canada. Why is there this difference? What does it mean?
Posted by: Nick Rowe | April 22, 2009 at 05:04 PM
Nick,
Very tricky first question, but there is an important advantage to the Fed having the option of issuing its own interest paying liabilities, particularly in this environment of unconventional easing and balance sheet expansion.
The advantage lies in funding diversification and the potential difference in net monetary effect.
If the Fed monetizes through excess reserves, it is monetizing using the banking system as a forced conduit.
There are two cases of this monetization via excess reserves:
First, suppose the Fed lends to or buys an asset directly from a commercial bank. The commercial bank swaps an asset for excess reserves. Or it gets credited with excess reserves when it borrows from the Fed.
Second, suppose the Fed lends to or buys an asset from a non-bank. This is increasingly becoming the case with its credit easing program. Then, the borrower/seller deposits the funds in the banking system. This increases both M1 and excess reserves.
One of my pet peeves on the general subject is that this immediate M1 effect is very important, and it is typically overlooked by analysts. The reason it is important is that it is an automatic increase in M1 associated with this type of transaction, which doesn’t depend on any presumption of a so-called “multiplier effect”. In essence, it’s the first and essential round of the M1 effect when the Fed forces monetization though the banking system, using a non-bank counterparty.
So in either of the first two examples, we have the base case (no pun intended), where the Fed has funded its asset expansion with excess reserve creation.
To see the difference when the Fed issues interest paying securities, you really have to break it down into two steps. The first step is the banking system monetization as described above, in either the bank or non-bank counterparty case. The second step is the issuance of securities in order to “retire” excess reserves from the banking system. This is in effect a unique case of sterilization. There are two sub-cases:
The first case is where the Fed issues or sells an interest bearing security directly to a bank. The bank purchases the security and its reserve accounted is debited. In this case, the result is really the same as leaving interest earning excess reserves in place, which I think is the effect you noted. The difference between holding the security versus excess interest bearing reserves is pretty minimal, so the real effect of such sterilization is also minimal. And M1 hasn’t been affected.
The other case is where the Fed issues or sells an interest bearing security to a non-bank. The purchaser’s bank pays for the security on his behalf, debiting his bank account, and the Fed debits the purchaser’s bank’s account at the Fed. This second case is very important because it completely reverses both the M1 monetization and the excess reserve monetization involved in the second version of the base case above.
The moral of the story is that the monetary effect of the Fed’s asset expansion isn’t necessarily restricted to excess reserves. Issuing interest bearing securities and dealing with non-bank counterparties is a mechanism for “mopping up” both excess reserves and “excess” M1, as the Fed desires, particularly when it reaches the point of starting a future tightening cycle.
You won’t see anybody else talking about this M1 effect in discussing this topic generally. I don’t know why. This is not the multiplier effect, which as I’ve repeated often I believe is not the right way to think about things. I think of this as the M1 “dual” of excess reserves in the primary monetary effect.
Finally, issuing Fed securities is an alternative to requesting Treasury to issue bills and deposit proceeds at the Fed, as the latter operation has a similar monetary effect. This would strengthen the Fed's operational independence.
I think this must be very difficult to follow, and it's not as clear as I'd like, but I don't think there's an easy way to explain it. I've run out of steam for now, so I'll leave it there. I'll get back later on your Bank of Canada reserve question.
(The idea of issuing interest bearing securities has not been announced; its sort of a personal prediction of mine.)
Posted by: JKH | April 22, 2009 at 06:32 PM
FWIW for the specific example of gold I've two solutions to problems, but still think gold is a bad CPI substitute (I know gold is just being used for a thought experiment here). First you could use a mini or regional gold standard that doesn't grossly increase gold mining. Second, as I mentioned you get gold along with other base metals. I think the $$ value of a gold mine isn't too far off what you get in copper. Copper, used for industrial activity, is cyclical while gold is a hedge. So when you increase the price of gold you can simultaneously decrease somehow, the price peg of copper and nickel to undo perverse consequences of base metal price movements against whatever direction you are trying to peg gold. Neat thread for a gold mining nation.
Posted by: Phillip Huggan | April 22, 2009 at 06:55 PM
JKH writes:
Surely not the other rates pegged by CBs? The correlation between the FF-rate and CP-rates is very strong, but the correlation is fairly attenuated in other markets at other points on the yield curve. This abstraction of yields is a useful tool, but it does not always hold true. In particular, if you view a debt instrument as a product, 'i' is defined by the intersection of supply and demand. Although substitution effects (cross-elasticity) link different debt instruments together; the rhyme and reason is not captured by bare default risk+expected short-rates. Certain market participants attempt to minimize personal risk or maximize a personal preference. i.e., an insurance fund performs maturity matching. Thus the demand for 30 yr bonds depends on age-demographics. This is shown quite clearly in Treasury yields. Demand and supply of 30-year bonds are such that the yield-curve inverts at that point.
I think in a broader sense, the FF-target is not truly exogenous. e.g., Whatever correlation arises between FFs and 30yr mortgage rates arises first and foremost from the long-run inflation expectation not the other way around. In turn mortgage rates embedded an estimate of CPI, default risk, and social return on capital. If the lender has matched maturities (i.e., there is no funding risk) then the FF-target is irrelevant. If the lender has not matched maturities then he includes a premia for the yield-conversion risk. Again though this does depend on the FF-target or the time-path thereof but on the expected variance.
Posted by: Jon | April 22, 2009 at 08:01 PM
JKH:
The counterparties to NYFED OMO are typically NOT banks. Every primary dealer is either a securities trading firm not under the umbrella of a bank-holding company or is a trading affiliate of a depository institution. Depository institutions do not participate directly in normal OMO.Only recently has the situation has been different, banks are the direct counter-parties of the TAF whereas Primary Dealers have access only to the much smaller PDCF. SOMA operations have been virtually nil.
Posted by: Jon | April 22, 2009 at 09:14 PM
Jon,
I agree with most of your points.
The fed funds transmission is a small part of the explanation, but I believe it exists.
I didn't take the time to breakdown the bank/non-bank counterparty split. Your points are valid. My point was directed toward explaining the funding of the Fed's asset expansion - i.e. recently. I wasn't thinking about OMO, since that's not about balance sheet expansion. More the special programs, although I didn't break those down between bank/non-bank.
Posted by: JKH | April 22, 2009 at 09:25 PM
Nick: "The central bank adjusts the supply of money endogenously so that the price of gold equals some target."
we've come full circle.
Posted by: Adam P | April 23, 2009 at 03:37 AM
I don't understand why my post why ignored, I was serious and think I have a good point. Targetting the price of an internationally traded commodity (and let us assume that the country isn't rich enough to manipulate the entire world market) is equivalent to targetting the exchange rate. We know how this works. Why is this an alternative universe?
Posted by: reason | April 23, 2009 at 03:43 AM
reason, I for one didn't disagree with your comment. I've been arguing the transmission mechanism with Nick, or I guess with myself.
Posted by: Adam P | April 23, 2009 at 03:56 AM
And actually, (not sure if this really is good for your argement reason), but Russia has floated the idea of pegging the ruble to the price of oil.
Posted by: Adam P | April 23, 2009 at 04:40 AM
reason: I think you are right too, but it would I think just raise the same questions, of whether an exchange rate anchor works via the interest rate channel or via some other transmission mechanism. I would argue for some other transmission mechanism, the real exchange rate channel and net export demand, and real money balances.
But it doesn't help us talk about the determinacy of world aggregate demand. It gets us into arguments about "beggar they neighbour devaluations".
I didn't respond because I was working too hard trying to argue with JKH and Adam P.
Adam P. Sorry if you think I didn't engage your argument. I was certainly trying hard to. I felt you had engaged my argument, and forced me to try to think it through. It wasn't easy.
Posted by: Nick Rowe | April 23, 2009 at 06:20 AM
Sorry Nick, I certainly didn't mean to say that. But your most recent comment frustrated me. We need here to distinguish the instrument and the target. Now, in your story about the girl with orange eyes is the gold price the instrument or target?
When you say, ""The central bank adjusts the supply of money endogenously so that the price of gold equals some target", it sounds like the money supply is the instrument and price of gold the target. That would be fine but then you've gone nowhere to resolving the arguments from your last post. (And earlier in this thread I think you said that adjusting the money supply was not how the target gold price is supposed to be achieved).
If the price of gold is the instrument then you have to specify how it's controlled (I guess open market ops) and you need to address the arguments about the transmission mechanism.
Posted by: Adam P | April 23, 2009 at 07:17 AM
Nick: "Here, massively oversimplified, is your theory of AD. Forget investment (to keep it easy for me). People can either save their income, or else demand consumer goods (AD). Savings is a function of income and the real rate of interest. Equilibrium is where S(y,r)=0. Invert that function to get y as a function of r, to get your AD function.
Here, massively oversimplified, is Orange Eye's theory of AD. Forget gold production (to keep it easy for me). People can either use their income to add to their stock of gold, or else demand consumer goods (AD). Gold demand (as a flow) is a function of income and the real price of gold r. Equilibrium is where G(y,r)=0. Invert that function to get y as a function of r, to get her AD function."
Why can't people in Orange's world save?
Posted by: Adam P | April 23, 2009 at 07:29 AM
and if you're going to say that they save by holding gold then I'll ask why their's no government debt of any kind, for example money.
Then I'll point out that in this case what determines their saving is the real price of gold today relative to its expected price tomorrow which will be the real interest rate in this world (no money or risk-free bonds, just gold as a store of value).
Posted by: Adam P | April 23, 2009 at 07:48 AM
Thanks Adam!
OK, I understand your frustration now. My statement "The central bank adjusts the supply of money endogenously so that the price of gold equals some target" was really sloppy. Not totally wrong, but sloppy.
The price of gold is the (short-term) instrument. The central bank announces a (temporary) target for the price of gold. It hits this target price by buying and selling gold at that price. It buys gold with newly-printed money, so the money supply expands if it buys gold. And when it sells gold, it gets money in exchange, so the money supply contracts.
I think that OMO would only be needed if the bank's reserves of gold got too large (it owns all the gold, and so cannot buy any more to keep the price from falling) or too small (it runs out of reserves, so cannot keep the price of gold from rising by selling more gold).
[Possibly off-topic, but I am uncertain what fiscal regime I have at the back of my mind. I think I want the central bank to be dominant, and be able to do what it wants, which means the fiscal authorities have to adjust taxes or government expenditure to take any bank profits/seigniorage into account and satisfy the LR budget constraint. That's a "Ricardian regime", IIRC. Or is it the other way round?]
Now: "Why can't people in Orange's world save?" They can, of course, but that's not really answering your question. My gut intuition tells me that if I understood that issue fully and clearly, and could answer it fully and clearly, I could get to the root of the whole thing. But I don't.
Let me try, and fail, anyway.
The answer has to do with Say's Law, and how new money enters the economy, and how the natural rate can differ from the actual rate (and with Austrian Business Cycle Theory, though I will try to stay away from that).
Suppose we are initially in equilibrium, so that aggregate demand equals output (desired savings equals investment, and the supply of loanable funds equals the demand for loanable funds). Then the central bank decides to lower the rate of interest. At the initial level of output, there is now an excess demand for output (desired investment exceeds desired savings). What is happening in the market for loanable funds? The private sector has an excess demand for loanable funds (since desired investment exceeds desired savings), but that private excess demand is being met by new money being printed and supplied as loans, in order to push down the actual rate of interest.
What's happening in the market for gold? We don't really care. If we assume the price of gold is sticky, there is probably an excess demand for gold. If we assume it is perfectly flexible (while other prices are sticky) the price of gold will rise, getting equilibrium in the gold market. But I don't think it matters much which we assume. Output has to increase until desired savings equals desired investment at the new lower rate of interest, so that the private demand and supplies for loanable funds are equal, and no new money needs to be lent out to augment the supply of money.
Now let's run the same thought experiment in Orange Eye's world. Start in equilibrium, then the central bank decides to raise the price of gold. At the higher price of gold, the demand for other goods now exceeds output, and there is an excess supply of gold from the private sector, which is being met by a demand for gold from the central bank, and the money supply is increasing.
What is happening in the market for loanable funds? Do we care? We could assume the interest rate is sticky. In which case there could be either an excess demand or an excess supply of loanable funds. I can't see any particular reason to suppose one rather than the other. If there's excess demand, borrowers are rationed in how much they borrow. If there is excess supply, lenders are rationed in how much they lend. In either case, the actual quantity borrowed will equal the actual quantity lent. Or we could assume the interest rate is perfectly flexible, in which case the loanable fund market clears, and doesn't seem to affect Orange Eye's story.
Output has to increase until the demand for gold rises to equal the supply at the new higher price, so AD equals output, and new money stops being issued to buy gold.
I am not satisfied with that answer. But what is happening to the market for loanable funds, savings and investment, doesn't seem to matter for Orange Eye's story of the transmission mechanism any more than what is happening in the market for gold seems to matter for our story.
Posted by: Nick Rowe | April 23, 2009 at 09:06 AM
Nick,
I've been reading up on some Stiglitz and just came across this line. I'm not sure exactly how this fits into the discussion, but I think you might find it relevant:
"We have presented a model of credit rationing in which observationally identical borrowers some receive loans and others do not. Potential borrowers who are denied loans would not be able to borrow even if they indicated a willingness to pay more than the market interest rate, or put up more collateral than is demanded of recipients of loans. Increasing interest rates or increasing collateral requirements could increase the riskiness of the bank's loan portfolio, either by discouraging safer investors or by inducing borrowers to invest in riskier projects, and therefore could decrease the bank's profits. Hence neither instrument will necessarily be used to equate supply of loanable funds with the demand for loanable funds. Under those circumstances credit restrictions will take the form of limiting the number of loans the bank will make, rather than limiting the size of each loan or making the interest rate charged an increasing function of the magnitude of the loan, as in most previous discussions of credit rationing.
Note that in a rationing equilibrium, to the extent that monetary policy succeeds in shifting the supply of funds, it will affect the level of investment, not through the interest rate mechanism, but rather through the availability of credit. Although this is a "monetarist" result, it should be apparent that the mechanism is different from that usually put forth in the literature."
Posted by: bob | April 23, 2009 at 11:22 AM
Scott,
Sorry I’m late getting back on your Bank of Canada question. I’ve been banging my head against a wall at Sumner’s blog, which is normal. Unlike yourself, people there seem less willing to engage different viewpoints than their own, and debate them and question them in an interesting, vigorous, and constructive way.
Ironically, I tend to follow the Fed in much greater detail than the Bank of Canada. In fact, I spend little time following the BoC; I should be more balanced in that regard.
I thought Stephen’s post was quite good, with some useful comments.
I can venture only a guess, but maybe a reasonable one, as to why the Bank of Canada balance sheet looks different.
You noted the main difference, which is the liability composition. The bank has funded its balance sheet expansion with government deposits. I imagine the Bank of Canada has the option of funding with excess reserve creation if it so desires. Perhaps they don’t want to do this because it would create confusion and raise all sorts of particularly thorny questions about what they’re doing, given that Canada has a zero reserve requirement. Maybe the government funding route is a simpler operational and political decision in that sense. And it may be more convenient than it is for the Fed, given that the Bank of Canada’s balance sheet expansion has been proportionately smaller. But if they pay interest on reserves, either route effectively sterilizes the potential theoretical alternative of zero interest bearing reserves. Finally, the Fed has used government funding for some of its requirements, but is trying to wind it down. There’s a political/independence aspect involved as well I guess, and who can gauge the difference there between Washington and Ottawa? As noted earlier, depending on how much more balance sheet expansion the BoC undertakes, perhaps it will consider issuing its own interest bearing securities, but I doubt things will get that far in the case of Canada.
Posted by: JKH | April 23, 2009 at 12:31 PM
Nick,
I meant Nick, not Scott.
:)
Posted by: JKH | April 23, 2009 at 12:45 PM
bob: I think that's called the "credit channel", as opposed to the interest rate channel. But it is very similar really, since it tells us that monetary policy impacts the rest of the economy via its effect on the market for loanable funds. If we think of those who are on the margin of being credit-rationed having a shadow price of loanable funds, then monetary policy would work by lowering that shadow price, which acts much the same as a fall in the rate of interest. Stiglitz formalised it, and created a proper theory, but I think the insight goes back at least as far as the Radcliffe Report (?) on monetary policy in the 1960s in the UK, which was the heyday of old-Keynesianism.
JKH: And it has clearly damaged your head enough that you have got my name mixed with Scott Sumner's ;-) . You meant to start your post with "Nick," !
Part of the problem is that the rest of us are just so much less familiar with the details, and thinking in terms of balance sheets. Anyway.
What you say about the Bank of Canada vs the Fed makes sense to me. But I think one more difference might be that the Bank of Canada can directly choose where to allocate the Federal government's balances, whereas it cannot directly choose chartered bank reserves? Also, since the old "drawdowns and redeposits" mechanism was how it used to control monetary policy, it is just going back to a familiar method, which the old guys will remember how to do.
Posted by: Nick Rowe | April 23, 2009 at 01:01 PM
Seriously, am I the only one who sees a problem with this? Here's Nick:
"Now let's run the same thought experiment in Orange Eye's world. Start in equilibrium, then the central bank decides to raise the price of gold. At the higher price of gold, the demand for other goods now exceeds output, and there is an excess supply of gold from the private sector, which is being met by a demand for gold from the central bank, and the money supply is increasing.
...skip this paragraph...
Output has to increase until the demand for gold rises to equal the supply at the new higher price, so AD equals output, and new money stops being issued to buy gold."
First we have "there is an excess supply of gold from the private sector, which is being met by a demand for gold from the central bank" so the higher price of gold causes people to demand less but there's already less supply by just the amount required to equate supply and demand (why? because that's how much the price would have adjusted by).
Then we have "Output has to increase until the demand for gold rises to equal the supply at the new higher price". Demand and supply were already equal at the new higher price, the point at which demand and supply are equal is how the market chose the new price! If the bank has put the gold it bought back on the market then the price comes back down. No change in AD.
Now, we also had "the money supply is increasing" in that paragraph but that's not supposed to be the story here.
Posted by: Adam P | April 23, 2009 at 01:32 PM
Remembrance of things past: drawdown and redeposit: a sign of grizzled veterans.
Not sure about your allocation point, Nick. Both are forms of funding. The asset allocation is a separate decision, I think.
There is a liquidity risk with government deposits, in that the government might want to use its money in the economy, which would essentially convert government deposits to excess reserves. Perhaps that’s what you’re thinking. The government in that sense has a callable deposit. But this would only be technical to the degree that the government had agreed to leave the deposit with the Bank of Canada in the first place in cooperation with and as an instrument of monetary policy.
Excess reserves are callable by the Bank of Canada, in the sense it can drain them from the system by selling an asset. Thinking about it this way, the government has a call option in the case of government deposit funding; the bank has a call option in the case of excess reserves. A potentially interesting but not particularly useful observation, perhaps.
Posted by: JKH | April 23, 2009 at 01:36 PM
I'm thinking about doing a follow-up post, by putting gold into Paul Krugman's Liquidity Trap model. That model stacks the deck against monetary policy, I think, because the underlying problem by construction is a negative real natural rate. So you have to increase expected inflation to escape the trap. But a gold price control instrument could work despite this.
Assuming there's a fixed stock of gold, with no production. People wear gold as jewelry, which gives them utility. You can rent gold.
There are two additional equilibrium conditions:
1. The real rental rate on gold will equal the MRS between gold and consumption, which depends on the ratio g/c.
2. The nominal rental rate on gold, divided by the nominal price of gold, plus the rate of increase in the nominal price of gold, equals the nominal interest rate on bonds.
The central bank, stuck in a liquidity trap, where OMOs don't work, switches to the gold price instrument. It buys and sells gold. It can also announce a time-path for the price at which it will buy and sell gold.
Assume the price of gold is perfectly flexible, but the price of consumption is temporarily sticky.
What path for the price of gold should the central bank commit to, to escape the liquidity trap, in that model?
Posted by: Nick Rowe | April 23, 2009 at 01:42 PM
“Output has to increase until the demand for gold rises to equal the supply at the new higher price". Demand and supply were already equal at the new higher price, the point at which demand and supply are equal is how the market chose the new price!.”
I’m late to this discussion, but that revision on its own sounds right to me. I'm not sure how you match an instantanous gold reset to a discrete time AD increase; but matched discrete time adjustments would work.
Posted by: JKH | April 23, 2009 at 01:46 PM
"The central bank, stuck in a liquidity trap, where OMOs don't work, switches to the gold price instrument."
Nick, I think an interesting model in the very abstract would be a central bank with both government bonds and gold as assets. Then specify which asset would be used in intervention at which times, very generally.
Posted by: JKH | April 23, 2009 at 01:57 PM
Nick, I just realized it. You're talking about fiscal stimulus, it doesn't matter if the bank buys gold or cars. The source of our mis-understanding was that I thought you were talking about monetary policy. Why didn't you just say so?
You want the central bank to create government debt and buy real goods with it whenever AD is too low (money is gov't debt remember). It sells the goods back and retires the debt when AD is too high, a price target serves as indicator of when AD is too high or too low.
Well, Keynes would like the idea. On the other hand it's not what central banks are supposed to do.
Posted by: Adam P | April 23, 2009 at 02:54 PM
There is a problem though, why doesn't fiscal stimulus happen by buying gold? Why does the gov't buy bridges and roads instead. Well, Nick introduced the problem when he assumed that there was no gold production. If you don't employ people making gold then you won't improve employment in the economy. Thus no increase in AD. The fiscal stimulus boost works first of all by buying something that wasn't already there and employing people to build it. The fiscal spending is part of AD and so it increases it directly. You can also get a multiplier from the increased incomes of the newly employed workers. However, fiscal stimulus does not work by the sort of spillover that Nick was describing. It works by directly being part of AD.
Posted by: Adam P | April 23, 2009 at 03:03 PM
JKH: That would be an interesting model, where the Bank could switch between two different instruments. And it is sort of at the back of people's minds when they recommend QE in times like now, but would use the standard overnight rate instrument in normal times. But it's hard enough arguing and thinking through how alternative instruments work.
Adam P : I find it hard to draw a clear theoretical dividing line between monetary and fiscal policy. In practice, we think of monetary policy as whatever the central bank does, and fiscal policy as whatever the government does. But even that doesn't really work in the long run, since the long-run budget constraints of government and central bank are linked (assuming the government owns the central bank, so eventually gets the seigniorage revenues or profits).
Helicopter money, for example, can be seen as a money-financed temporary tax cut (transfer increase), and so is partly fiscal, by that count.
An open market operation, where the bank buys bonds, increases the interest earnings of the central bank, which it transfers as profits to the government, which means a cut in expected future taxes (or increase in expected future government expenditure), via the long-run government budget constraint.
I assumed that gold was not produced, partly for simplicity, as a said, but also to make central bank operations in the gold market less "fiscal". If gold production were large, and the supply were elastic, then the central bank buying gold and increasing the price of gold would have a direct effect on AD, because gold production would be part of GDP. This would be even more clear if we assumed a sticky price of gold. Buying gold would employ unemployed gold miners. Just like if the Bank hired lots of unemployed economists. Fiscal policy by the central bank.
Assuming gold was not produced made my job harder, in explaining the transmission mechanism. But it seemed like "cheating" for me to assume gold is produced. Gold production is a small part of GDP. I don't see how it can be an important part of the transmission mechanism.
Posted by: Nick Rowe | April 24, 2009 at 08:53 AM
I would think the distinction is whether or not new government debt is created. Helicopter money is always fiscal stimulus, QE is monetary because you buy gov't debt for cash so the total value of debt outstanding is unchanged.
You want to create new government debt when the gold is bought, this is entirely fiscal.
Posted by: Adam P | April 24, 2009 at 09:18 AM
Nick: "An open market operation, where the bank buys bonds, increases the interest earnings of the central bank, which it transfers as profits to the government, which means a cut in expected future taxes (or increase in expected future government expenditure), via the long-run government budget constraint."
you sure?
Posted by: Adam P | April 24, 2009 at 09:55 AM
Nick,
I’ve just posted another string of comments on Sumner’s latest piece. You’ve might be interested, at least because I’ve drawn on a couple of quotes from the New York Fed itself that support my general position. At the least, it provides some useful factual information on operational matters. Also, the general debate on the technical interpretation and motive for the Fed’s excess reserve easing (a debate which seems mostly to consist of my own monologue with myself, otherwise unheard) may be relevant to the question of similar potential easing at the Bank of Canada. Carney I gather is approaching this one with caution.
If interested:
http://blogsandwikis.bentley.edu/themoneyillusion/?p=1032#comments
Posted by: JKH | April 24, 2009 at 10:10 AM
I would agree helicopter is fiscal, QE is monetary, and gold is monetary if bought from existing stock.
Posted by: JKH | April 24, 2009 at 10:25 AM
Adam P: if you define money to be part of government debt (which it is in one way, but isn't in others, since it pays no interest, and there is no commitment to redeem it for anything), and you define "fiscal" as anything that changes debt (though that means a balanced budget fiscal stimulus is not fiscal), then gold purchases with money are by definition fiscal.
(I would not define debt to include money, because when you try to construct the long-run government budget constraint using money finance, with no interest on money, and a positive growth rate of nominal GDP, it does not converge. Money is like a stable Ponzi scheme, since the growth rate of nominal GDP is greater than the zero nominal interest rate on money. Bond-finance, if the nominal rate of interest on bonds exceeds the growth rate of nominal GDP, means you can't have a stable Ponzi scheme, and the long run government budget constraint bites.)
'Nick: "An open market operation, where the bank buys bonds, increases the interest earnings of the central bank, which it transfers as profits to the government, which means a cut in expected future taxes (or increase in expected future government expenditure), via the long-run government budget constraint."
you sure?'
Um, yes, I think I'm sure, if you'll excuse the oxymoron. I am assuming it's a permanent increase in M, and doesn't trigger some future offsetting action by the Bank. I am assuming the Bank is dominant, in other words. (Is that a Ricardian, or non-Ricardian fiscal regime? I always get the two muddled.)
JKH: I would say helicopter is monetary + fiscal, but otherwise agree.
I will check over at Scott Sumner's blog. More people read than comment, by the way. (At least, I hope that's true!)
Posted by: Nick Rowe | April 24, 2009 at 02:13 PM
well, "if you define money to be part of government debt (which it is in one way, but isn't in others, since it pays no interest, and there is no commitment to redeem it for anything"
The only thing you redeem a government bond for is money.
The next point is what I had in mind with my "you sure?" question.
"I would not define debt to include money, because when you try to construct the long-run government budget constraint using money finance, with no interest on money, and a positive growth rate of nominal GDP, it does not converge...."
It converges if you determine the price level endogenously. That's the point of the fiscal theory of the price level. Your comment on the central bank buying back a bond and using the interest to reduce tax burdens seems to imply that we'd be best off if the bank bought back all outstanding government bonds. But this can't be the case, it can't really matter. Budget constraints are REAL constraints, they are not nominal constraints. Our real tax payments must cover the real expenditures of the government (even if the tax is collected by seigniorage).
Posted by: Adam P | April 24, 2009 at 03:21 PM
It's all starting to come together now, I think.
Adam, at the back of your mind, do you have the Fiscal Theory of the Price Level? If so, that the root of our differences, I believe. It is not what is at the back of my mind.
I could have talked about those long run government budget constraints in real terms. Just subtract inflation. So money is a stable Ponzi because the real interest on money is less than real GDP growth. Bonds are not a stable Ponzi because real interest on bonds is greater than real GDP growth. I toyed with doing so, but did then in nominal for simplicity.
I am trying to remember the author of a paper criticising the Fiscal Theory of the Price Level. It was either Buiter, or someone else whose name I associate with Tobin. The point of the paper was that budget constraints must hold both in and out of equilibrium. So the price level cannot be determined by adjusting to satisfy the LR government budget constraint. He gave a counter-example, by putting forward the theory that the price level must adjust so that Mrs Smith does not go bankrupt.
The reason this may be at the root of our discussion of the transmission mechanism is that when we talk about the causal chain from a control instrument to the price level we are outside of equilibrium, in what Patinkin called a "stability experiment".
Starting in equilibrium, and holding all prices fixed by theorist's fiat it WOULD be true that real wealth would be higher if we did an OMO purchase of bonds. That's what we mean when we say that "money is net wealth". We would be richer. And that's one of the reasons we would demand more goods, so AD increases. And that's how we can prove that there would be upward pressure on the price level, so that our holding the price level fixed by theorist's fiat was unsustainable. That's a stability experiment.
Posted by: Nick Rowe | April 24, 2009 at 03:48 PM
I'm familliar with the Mrs.Smith example but it's quite clearly wrong. The difference is that Mrs.Smith can't print money, the government can. Mrs.Smith does have a nominal budget constraint, the government has a real constraint but the government is nominally unconstrained. That's why the government's budget constraint determines the price level but Mrs.Smith's doesn't.
Posted by: Adam P | April 24, 2009 at 04:04 PM
Nick, yes, of course, helicopter is fiscal + monetary; obtuse error on my part
Posted by: JKH | April 24, 2009 at 04:06 PM
Here's Chris Sims explaining why the gov't BC determines the price level buy Mrs.Smith's doesn't.
http://sims.princeton.edu/yftp/Cancun/DebtEquity.pdf
Posted by: Adam P | April 24, 2009 at 04:13 PM
Adam: you are very quick on the theoretical draw! I feel I'm doing well, keeping up with you as much as I am!
I have skimmed the Sims paper. Not read it.
Here's the main difference, I think. He assumes the government can print money. I assume the central bank can print money, but the government can't. What it amounts to is that I (implicitly) assume an independent central bank. So if the central bank does something (like an OMO) that affects its stream of profits it sends to the government, the government just has to adjust taxes or spending to compensate. If instead you assume the government does not adjust taxes or spending endogenously (so the government is dominant, and the central bank is not independent), then if the central bank does an open market purchase of bonds now, people expect it will eventually do an open market sale of bonds in the future, with the same present value. That will clearly have very different effects from an OMO that is not expected to be reversed in the future. An OMO increase in the money supply that is expected to be reversed in future will generally have two offsetting effects: the current increase in M increases AD; the future decrease in M decreases expected inflation which decreases current AD. Sargent and Wallace Unpleasant Arithmetic, I think.
But that means that if I say there is a permanent increase in the money supply, via OMO, I must implicitly be assuming an independent central bank.
In terms of the Fiscal Theory of the Price Level, only the government can do quantitative easing (change the total quantity of government debt = bonds+money). The central bank can only do qualitative easing (change the mix of debt between bonds and money).
That explains a lot of our arguments. I think in very different terms. It's not just my use of words, but different things I'm implicitly holding constant.
I'm wondering how gold fits in to the budget constraints. Unlike bonds, gold reserves of the central bank do not pay interest. But they may appreciate (in nominal, and maybe real terms).
Posted by: Nick Rowe | April 24, 2009 at 05:18 PM
Well, I think the fiscal theory only requires that the government be able to print money. The key is that although the government has real constraints it is nominally unconstrained, this implies directly that in the long run the price level must adjust to satisfy the government's real budget constraint. This also resolves the problem about the fact that budget constraints are still constraints out of equilibrium. The gov't has a real budget constraint that it satisfies in or out of equilibrium. The gov't has no nominal budget constraint, in or out of equilibrium.
But I take your point about the independant central bank. Seems as though the Eurozone is the canonical example. Are the Euro gov'ts nominally constrained? I don't know, for one thing they may control the physical printing presses and can then physically start printing Euro notes.
To me though this is an important point, my definition of money is (was) anything that the gov't accepts for tax payments. This depends on the fiscal theory to make sense. This definition also resolves why changing the gold/dollar exchange rate was an effective tool under the gold standard but might not be today. Under the gold standard gold was money, prices could be expressed in terms of gold and thus changing the amount of gold each dollar was officially worth changed prices of all things directly (in a sense).
Posted by: Adam P | April 25, 2009 at 06:21 AM
Adam P: that makes a lot of sense of our argument.
My definition of money is anything "players" accepts as a medium of exchange, where the government is just one of the players, and tax receipts are just one of the things the government plays with. But it's perhaps not so much that our definitions differ. The difference in definitions is perhaps a reflection, rather than a cause, of different ways of thinking about money.
Yep, I think the independent central bank thing is crucial. And thinking back to my post a couple of months back on why the Eurozone could get into big trouble fits in with your observation about the ECB. Because I was there assuming that the ECB would not bail out the Eurozone fiscal authorities in a crunch (like a "run" on their bonds, or failure to sell bonds to finance a deficit).
Funnily enough (and wandering off topic), many other economists say that the Eurozone needs a central fiscal authority to support the ECB. I see it the other way round. The ECB could support a central fiscal authority.
Feel free to ignore this question. I'm just curious. My guess is that you are a fairly recent (<10 years) PhD graduate from a top US or Canadian school. Am I totally wrong?
Posted by: Nick Rowe | April 25, 2009 at 06:57 AM
Yes, that's right. I must say I've been enjoying the blog (seems like most of are).
Keep up the good work.
Posted by: Adam P | April 25, 2009 at 09:13 AM