The reason that New Keynesians don't like talking about the supply of money and velocity has nothing to do with problems in defining the supply of money, the instability of velocity, or the instability of the money supply multiplier. As I argued in my last two posts, similar criticisms would apply equally to the New Keynesians' actual and natural rates of interest.
The real reason is the New Keynesian view of the monetary policy transmission mechanism. They see it as working through interest rates. Given sticky prices, an exogenous decrease in the supply of money causes interest rates to rise which causes aggregate demand to fall, which causes a recession. So why not just cut out the irrelevant cr*p about money, and just say the central bank raises interest rates which causes aggregate demand to fall? Money and velocity play no role in the story. You can talk about them if you like, but they don't help you understand what's going on, and are merely a distraction.
New Keynesians have this precisely backward. Interest rates are neither necessary nor sufficient for the monetary policy transmission mechanism. Money/velocity is both necessary and sufficient.
Assume sticky prices.
Lets break this into two parts.
1. To show that interest rates are not necessary in the transmission mechanism, imagine a world with no interest rates at all. The only asset is money, which pays 0% interest. It's a backscratching economy, where backscratches must be paid for with money. Or Paul Krugman's baby-sitting coop, if you like. [Don't want to search for the link and blow my monthly ration of NYT views. Update: thanks to Phil, here's a Slate link.] Start in equilibrium. Then part of the stock of money is destroyed. At the existing PY, actual velocity exceeds desired (actual M is less than desired). So individuals buy fewer backscratches to reduce their actual velocity (to try to rebuild their actual money stocks). There's a recession, even though there are no interest rates. We can explain the recession in terms of money and/or velocity. Money/velocity is sufficient to explain the recession, while interest rates are not necessary.
Or, imagine a world in which there are non-money assets, but the interest rates on those assets (or their prices) are fixed by law, so cannot change. Start in equilibrium. Then part of the stock of money is destroyed. So individuals try to sell bonds and land etc. to rebuild their money stocks. But they fail to sell bonds and land, because everyone is trying to do the same thing, and prices of bonds and land cannot fall, so there's an excess supply. So they buy fewer backscratches instead, and there's a recession, even though interest rates never changed.
2. To show that interest rates are not sufficient in the transmission mechanism, imagine a world with barter rather than monetary exchange. The central bank controls not the supply of money but the supply of bling. Bling is the medium of account (prices are measured in terms of bling), but there is no medium of exchange. Bling does not circulate. We can't meaningfully talk about the velocity of bling. Because wearing bling is a form of conspicuous consumption of a durable good, and a way of flaunting ones real wealth, there is a demand for real bling, M/P, which depends positively on real income Y and negatively on the nominal interest rate (the opportunity cost of wearing bling). M/P = L(Y,i).
Suppose the central bank reduces the supply of bling. Interest rates rise, as people try to sell bonds and land to buy more bling. Since prices are measured in terms of bling, and prices are sticky, the real stock of bling M/P cannot rise back to equilibrium.
Does the increase in interest rates cause a recession due to deficient aggregate demand? "What do you mean we are both involuntarily unemployed because of deficient aggregate demand? This is a barter economy. Here's the deal: I scratch your back and you scratch mine. That way we are both better off, and we both get back to full employment." No, there is no recession, despite the central bank's tight bling policy causing high interest rates.
OK. Interest rates may play a role in the monetary policy transmission mechanism. For one thing, the demand for money (desired velocity) depends on interest rates. But the interest rate channel is just one way of telling the story of the transmission mechanism. It is not essential. It is neither necessary nor sufficient.
Worthwhile Canadian Initiative 
Public service announcement: Krugman’s retelling of the capitol hill baby sitting co-op parable is concisely recounted in this 1998 Slate piece: http://www.slate.com/id/1937. So no need to use up any precious NYT clicks.
Thanks Phil! Added.
For #1, I think you could go with, “There is no interest rate in Krugman’s babysitting story, and yet there is a recession. QED.” The counterargument, though, is that there actually is an interest rate; it’s just not observable. Babysitting vouchers pay a convenience yield. The yield was higher than the natural interest rate, so there was a recession.
On #2, we can’t scratch each other’s backs at the same time. If I scratch your back, how do I know you’ll scratch mine? Maybe you’ll issue a promissory note for a backscratch? That note is a financial asset. What interest rate does it pay? A low enough interest rate to prevent a recession. Even if I just trust you to return the backscratch, your credit is a financial asset, and it pays a low enough interest rate to prevent a recession.
If you want to dispense with interest rates the World’s Smallest Macromodel (not an NYT link) has the advantage that the model is clearly specified, whereas the babysitting co-op is more of a story than a model.
That model tells me that this isn’t quite right: “We can explain the recession in terms of money and/or velocity. Money/velocity is sufficient to explain the recession, while interest rates are not necessary.” Velocity is constant, given the log-linear utility function. It’s the rigid price level that does the trick, in conjunction with the impossibility of barter, not money as such. After all, you could think of M as being moonstones, desired for their appearance alone. Of course, if you want to say that “monetary economy” is a synonym for “economy in which barter is impossible” then this is a monetary economy by definition, even if the representative agent never actually parts with a single moonstone.
I’ve also done some posts that consider monetary transmission in a world without interest rates. Or a world where prices are completely flexible, and hence money supply changes don’t affect interest rates. In either case, there must be SOME mechanism that explains why increasing the money supply 10% causes prices to rise by 10%. What is that mechanism? (I think it’s the hot potato mechanism). My problem with Keynesian economics is that they don’t seem to recognize that the transmission mechanism in a interest-free economy, or in a flexible price economy, is still there when you add interest rates, or add price stickiness. And since prices are flexible in the long run, that mechanism explains why a 10% increase in M that is perceived to be permanent will cause expected future NGDP to rise by 10%. And if future expected NGDP rises, that’s all the transmission mechanism you need to affect current NGDP. It’s not necessary that interest rates change at all.
One small quibble. You CAN have a velocity of bling, it’s simply NGDP/bling. Some people erroneously believe that V measures the speed that money circulates, but that’s only the non-tautological definition of V. The more commonly used tautological definition of V has nothing to do with the speed at which money circulates, as 99% of monetary transactions are for assets unrelated to NGDP.
Andy: “On #2, we can’t scratch each other’s backs at the same time. If I scratch your back, how do I know you’ll scratch mine?”
OK, but 3 of us could scratch each others’ backs at the same time ;-).
Kevin: “It’s the rigid price level that does the trick, in conjunction with the impossibility of barter, not money as such.”
Well, I would say that “the impossibility of barter” is the same thing as “money” (medium of exchange). But yes, if prices were perfectly flexible, the real supply of money would adjust instantly to equal demand.
Scott: yes, you have been one of the strongest advocates of the view that interest rates are a possible complicating factor of the transmission mechanism, rather than an essential ingredient. But there is maybe a difference in our views on this, nevertheless. I still can’t get it exactly right. Your comment is talking about the transmission mechanism from M to P, where the fact that money is the medium of account matters, and the fact that it is a medium of exchange does not matter. I’m talking about the transmission mechanism from M to Y, given sticky P. There the fact that money is the medium of exchange does matter.
(I don’t feel I’ve fully come to grips with your comment.)
On your quibble: Yes, there’s a “velocity” of houses too (3 or 4 years income is supposed to be normal), but that doesn’t mean that people switch houses every 3 or 4 years. My instinct is to resort to the transactions velocity. MV=PT. Y is a subset of T. When MV falls, we see all transactions falling, including sales of newly-produced goods and services.
Scott: My problem with Keynesian economics is that they don’t seem to recognize that the transmission mechanism in a interest-free economy, or in a flexible price economy, is still there when you add interest rates, or add price stickiness.
I’m not sure whose brand of Keynesian economics you are knocking here. It’s a broad church and I only know a few members of the congregation. I think you have a fair point with respect to some Keynesians, Old and New, but not all. Doesn’t your problem turn on whether the model is capable of generating a Pigou effect?
Generalising wildly: every Keynesian model worthy of the name includes, as a special case, a full-employment equilibrium. Call that special case the ‘Classical’ model. We chuck in some imperfection, almost always some kind of price stickiness and, hey presto, we have a ‘Keynesian’ model. The upshot is that pretty well every ‘Classical’ model can generate at least one kind of ‘Keynesian’ model. It might not look anything like the one Keynes sketched in the GT, but it will sort-of rhyme with it.
As an example of the sort of thing I mean, take a look at J.P. Bénassy (2007), “IS-LM and the multiplier: A dynamic general equilibrium model” (available here). It’s quite a short paper. He takes a simple monetary overlapping generations model, introduces sticky prices and ends up with something surprisingly like an old-fashioned IS-LM model. Your objection (as I understand it) doesn’t apply because an increase in M shifts the LM curve, as you’d expect, but it also shifts the IS curve via the interest rate. It’s a ‘Keynesian’ model of a sort Keynes could hardly have imagined, because it starts from a ‘Classical” model that didn’t exist in his lifetime.
Since people frequently refuse to click on links, I’d better add that Bénassy’s IS and LM curves are not derived in the same way as in your granny’s IS-LM model.
A note about the Time’s paywall. It’s incredibly porous. Just type the article title into google and you can read the article from the google link. No need for link hoarding at all 🙂
The most frustrating thing about the monetary transmission process I’ve ever had is combining them into one grand structural and self-consistent theory of 1) portfolio rebalancing and 2) hot potato process. Here is a list of the most commonly listed transmission mechanisms by group….
1. Traditional Keynesian interest rate channel (bonds)
2. Tobin’s q (equities)
3. Direct purchase of consumer goods and services
4. Direct purchase of capital goods
5. Commodities markets
6. Real estate markets
7. Exchange rate markets
8. Wealth effect for consumers
9. Wealth effect for firms
10. Anything Scott Sumner talks about, which I am unable to pigeon whole into any of these categories except for #3 and #4.
Has anyone ever written a clear and concise picture of each individual having a portfolio of such assets and how each transmission story, whether Keynesian or Monetarist, is really just looking at the same story from a different angle?
Joe
Jim: Aha! And I just heard from Niklas Blanchard that if you clear cache and cookies it resets to zero!
Joe: good comment. “Not to my knowledge” is the answer to your question. But I like the idea that, at least in many cases, it’s just looking at the same story from a different angle. Take ISLM, for example. Here’s the keynesian version. “M increases, r falls, Y increases…and, oh yes, there’s a feedback from increased Y to increased Md”. And here’s the monetarist: M increases, Y increases, … and, oh yes, there’s a feedback from increased Y to increased S and decreased r to increased Md”.
David Laidler did a good essay on the transmission mechanism, decades ago.
One thing that often gets left out is the effect on expectations. And how that depends on how people interpret the increase in M. “Why did the Bank increase M? What does it signal about the Bank’s future policy?” We really can’t answer the question of the transmission mechanism without looking at those questions. Take, for example, Scott Sumner’s favourite example, about FDR raising the price of gold. That meant something very different in those days than it would today.
The monetary transmission mechanism is a function of the institutional set-up of your model. The reason why so many have abandoned the old-school approach of counting money to predict inflation is not just the empirical problems, but the logical problems of that belief. It just doesn’t match how the institutions work. Outside money dominates inside money, and the non-financial sector can obtain more money from the financial sector whenever it wants. The financial sector is more than happy to purchase a bond and create a deposit — they will earn income from the spread. In the process, the non-financial sector always holds the deposits it wants regardless of what the central bank does. Or rather, it always holds the deposits it wants at a given rate of interest, which is set by the central bank.
So the reason why the interest rate management is the only transmission mechanism is because that is how our institutions are set up. If you change the institutions — and there is no outside money or banks in your model — then the transmission mechanism will be different.
What you are doing in your model is throwing away all the institutions that made the old-view irrelevant, in order to restore relevance to that view.
I don’t think anyone would disagree, in principle, that if the world operated differently, then there might be a monetary transmission mechanism. And you can look at less developed nations — say in which there is no banking sector due to a crisis, or the banking sector is so small that outside money is not important — and then your transmission mechanism might apply to those undeveloped or crisis economies.
But that isn’t true of modern credit-based economies and so your transmission mechanism isn’t a real transmission mechanism.
Some arguments against using interest rates to adjust AD:
1. There is no relationship between central bank rates and the rate charged by credit card operators. See: http://uk.creditcards.com/credit-card-news/credit-card-interest-rates-bank-rates-1360.php
2. The Radcliffe Report on monetary policy in the U.K. published in 1960 concluded that ‘there can be no reliance on interest rate policy as a major short-term stabiliser of demand’.
3. Interest rate adjustments will not influence house purchases by much because short term interest rates have little influence on long term rates. The longest so called “fixed rate” mortgage available in the UK is around five years. As I understand it, the period is longer in the US, but this is only possibly because of taxpayer funded distortions via the Fanny and Freddie.
4. The idea that there is a close relationship between interest rates and the ACTUAL AVAILABILITY of credit has been shown to be a myth over the last two years or so: we’ve had ultra low rates with banks far more reluctant to lend than prior to the recession, when rates were higher.
5. Interest rate adjustments work primarily via entities that are significantly reliant on borrowing rather than equity finance. This makes as much sense as boosting an economy via entities whose names begin with the letters A-L and ignoring the one’s whose names begin with M-Z.
Ralph,
What you are pointing out is the fundamental asymmetry of the interest rate channel. It affects business investment, consumer borrowing, and mortgage borrowing very differently and at different time scales, with the bulk of the immediate effects concentrated on the housing market. As a result, by raising or lowering interest rates, you are also affecting the relative profitability of different industries.
I don’t think anyone would disagree with this — it goes back at least to Bernanke and Gertler.
I view the modern era of CB taylor rule policies as being a de-facto era of asset price management, specifically house price management. But I don’t know of models that are able to distinguish between these effects.
It would be interesting to come up with some mitigation policy — for example, raising downpayment requirements simultaneously to cutting interest rates, or using tax policy to give tax breaks to business investment together with interest rate management, to get a richer policy rule than the one dimensional taylor rule.
My problem with monetarist economics is that they don’t seem to recognize that the transmission mechanism in a cash-free economy, or in a flexible price economy, is still there when you add cash, or add price stickiness.
you know it has to be asked since it was brought up above. Does anyone really think that barter was impossible or very expensive in the babysitting economy? Really?
Costless barter doesn’t prevent recessions.
Adam P: I’m not sure how you envisage barter in babysitting. The couple going out for the night can give an IOU to the babysitter, avoiding the use of precious scrip. If IOUs are transferable and all issuers are creditworthy then the shortage of scrip won’t matter anymore. But I wouldn’t call that costless barter, I’d call it a perfect bond market.
Well, first of all the IOU’s may not be transferrable, it would be bilateral. But really we’re talking about the difference between inside and outside money.
The M in the quantity theory is supposed to be base money, outside money. The cashless models don’t eliminate credit, they eliminate outside money.
In general though the point I’m trying to make is that, as a theoretical point, costless barter doesn’t mean there can’t be a recession. Anyone who’s ever seen an RBC model knows this.
The fact the models don’t seem realistic is irrelevant, they can give insights that apply to the real world.
Nick, It’s not entirely about medium of account vs medium of exchange. And my mechanism doesn’t require flexible prices. You could have wages be stickier than prices, and tight money still causes unemployment. You just have to stop thinking in Keynesian terms.
Yes, you can use transaction velocity, but I have never once seen it done. Forex transactions alone are probably 100 times NGDP. So the numbers would look nothing like the published velocity numbers that we actually see. When people are talking about empirical estimates of velocity, they are NEVER referring to transactions velocity. Also note that under the tautological definition of velocity, one can have M1 velocity, M2 velocity, etc. With the transactions definition there is only one velocity.
Kevin, I don’t doubt there are some Keynesian models that go beyond interest rates, but that seems to be the transmission mechanism generally assumed in both old and NK models. At the zero bound the old Keynesians say the Fed can do no more, and the NKs say that we can still lower long rates, or create inflation to lower real rates. But it’s still all about interest rates. That’s what I was reacting to. If you define Keynesianism so broadly as to include the 9 transmission mechanisms in Mishkin’s textbook, then I suppose I am a Keynesian.
When there is a big apple harvest and the dollar price of apples falls in half, then NGDP in apple terms roughly doubles. How does that occur? I’ve never heard anyone use the interest rate transmission mechanism to explain why NGDP in apple terms doubles, they always use the excess apple balance mechanism. Now of course dollar NGDP is very different from apple NGDP, because prices are sticky in dollar terms but not in apple terms. So that makes an interest rate transmission mechanism plausible, but it doesn’t magically take away the mechanism that caused NGDP in apple terms to double.
BTW, I am pretty sure my model doesn’t require a Pigou effect. As I recall that effect tries to explain why consumption might rise via higher real cash balances. But I am assuming that only nominal cash balances change. Or have I misrepresented the Pigou effect?
Scott: But I am assuming that only nominal cash balances change.
But aren’t you also assuming sticky prices, so M/P has to change? There’s a Pigou effect, as I use the term (which is also how Bénassy uses it), if financial assets, or at least a subset of them, are regarded as wealth by households. So that’s number 8 on Joe’s list of transmission mechanisms. I think you’re right in saying that most Keynesians regard that as an unimportant transmission mechanism for practical purposes. Let’s face it, not many of us respond to news of a monetary easing by saying that household wealth is about to increase. But that’s what we see in the interest-free economy of the World’s Smallest Macromodel. It’s still there in some old-fashioned IS-LM models, with a consumption function of the form C = C(Y, M/P) where Y is real income. (But C = C(Y) was certainly more common in old textbooks.)
I think you don’t see a Pigou effect in most NK models and that is because they have immortal agents. Or maybe not. I defer to Nick and Adam where NK models are concerned.
Nick,
I will attempt to give my attempt combining your keynesian and monetarist story. Scott Sumner begins his argument saying that Keynesians are forgetting that on a poor African country with no bond markets, M increases still have real effects. This is because people just directly spend money on consumer and capital goods. Strangely, Mankiw’s textbook does not mention that, not even Mishkin’s! This is hot potato effect (HPE). But there was an inconsistency, flaw in his idea. After all, even if paradox of thrift is technically wrong, in addition to hoarding or consuming stuff, people can just save of course. So, how in the act of saving are we 1) combining keyensian and monetarist ideas of textbooks (Tobin v Friedman), 2) show portfolio rebalancing story, 3) show HPE story of quasi-monetarists.
Metaphysically, saving is just handing your cash to another person in return for an IOU, who then HIMSELF, consumes it. This is what paradox of thrift proponents forget. An individual saves by going into fancy smancy (American phrase) super duper complex Wall Street financial markets. They wish to purchase bonds or equities. They shout “I have more money balances than my money demand, who wants to borrow my money, at these bond and equity prices?” If no takers then the individual bids up bond prices (lowers interest rate) and equity prices to “seduce” others into taking their cash. Either the market clears with the trade of IO/equity for cash, or potential saver is unsatisfied with too high of the price and leaves the financial markets in order to continue hoarding or just consumer something. I think I have successfully included 1) combining Keynesian and monetarist ideas of textbooks (Tobin v Friedman), 2) show portfolio rebalancing story, 3) show HPE story of quasi-monetarists.
There is a flaw though. I do not know how to combine it with Scott’s idea of expectations, of which there are two main ideas (if I correctly understand).
First, monetary policy can work in three ways, 1) Increase M, 2) Increase V (lower Md) through fiscal policy (gov. spending), 3) Increase V through changing expectations by making announcements concerning current and future situation.
Second, you can know whether these are successful by the direction asset prices go in the next coming months. If they go up, then it succeded, if down, then it failed.
I do not know how to shove together 1) my story of the monetary transmission process, 2) Scott’s ideas of expectations.
Bill Woolsey described it best over on Matt Rognlie’s blog.
Suppose that money earns interest. When interest rates on bonds rise, so do interest rates on money. Changing interest rates have little influence on the demand for money. The “interest rate transmission mechanism” breaks down. However, we can still get monetary disequilibrium, an excess demand for money, and a recession.
At least, that’s what I think Bill was saying.
Kevin, No I am not assuming sticky prices. In my first example I asked what would happen if prices are not sticky. Clearly a permanent 10% increase in M causes all nominal aggregates to immediately rise by 10%. Then I asked what was the transmission mechanism if interest rates don’t changes (and of course they don’t change if money is neutral.) I speculated that the hot potato effect causes the nominal adjustment. Then I asked whether this transmission mechanism might also work in a world where prices are sticky. And it seems to me the answer is yes. Now it is also true that once you introduce price stickiness, then real money balances adjust and you get the Pigou effect, just as you indicated. And I agree with Keynesians that the Pigou effect is probably weak. But the mechanism I was thinking about was a flexible price mechanism, and hence not the Pigou effect.
The flexible price model is not merely academic, as prices are flexible in the long run. I believe the transmission mechanism for long run changes in P and NGDP is exactly the mechanism that you observe in the flex price model (the hot potato effect.) And expectations of higher long run NGDP tends to raise AD right now. Then if prices are sticky in the short run, the higher current AD leads to more RGDP, until prices and wages fully adjust.
Most people get short run effects as follows: Assume sticky prices, that makes interest rates change, and assume interest rates drive RGDP. I assume prices are flexible in the long run, M-policy drives future expected NGDP. Futures expected NGDP drives current asset prices and thus current NGDP, and sticky prices causes some of the current change in NGDP to spill over into RGDP.
Scott,
When you say, “Futures expected NGDP drives current asset prices and thus current NGDP” you refer to all asset prices – bond, equity, real estate, commodities, foreign currency, etc. You are really talking about savings, not direct consumption. But how are such changes in prices representative of the hot potato process? This has never been made precisely and metaphysically clear. You never use the word “savings” to express why these changes in prices happen.
Are you really saying the following… NGDP expectations go up -> money demand falls -> money holders decide to save -> they go into the financial markets -> they bid up prices of assets (bond, equity, real estate, commodities, foreign currency, etc) in order to get borrowers to take their cash -> the markets clear -> borrowers take the saver’s cash in exchange for given assets.
Therefore, if the Fed attempts to move MV through changing expectations, you will know that they were successful because prices of all sorts of assets will go up across the financial markets. The increase in asset prices tells you that the hot potato process was channeled through the financial markets. Is that correct?
Here’s Scott: “I am not assuming sticky prices. In my first example I asked what would happen if prices are not sticky. Clearly a permanent 10% increase in M causes all nominal aggregates to immediately rise by 10%. Then I asked what was the transmission mechanism if interest rates don’t changes (and of course they don’t change if money is neutral.) I speculated that the hot potato effect causes the nominal adjustment.”
Seriously, does nobody else notice that he’s getting his own argument wrong? (or am I the only one who bother pointing it out?)
The thing about the “hot potato” effect is that it only operates before the price rise has happened. Once prices have risen 10% then the real value of the money stock has fallen to equal the demand for real balances and money is not a hot potato.
If the price rise is truly immediate, so that interest rates never change, then money is never a hot potato.
If there is a, perhaps brief, transition period in which the money stock has risen but prices haven’t fully adjusted then yes, money is a hot potato for this time but then, for this time, interest rates will have changed.
Apparently Scott has in mind the transition period, but then the statement that interest rates never change is inconsistent with the story he’s telling.
On the other hand, Scott usually professes to believe in rational expectations (though I know he hasn’t done so here). However, the “hot potato” transimission mechanism is not consistent with ratex. Under ratex the price rise truly is immediate, basically everyone understands that the money supply increase implies a 10% rise in prices and so prices adjust immediately before any hot potatos have changed hands.
Scott can’t have it both ways, if interest rates never change then money is never a hot potato. If the hot potato transmission mechansim is at work then interest rates will change.
“Given sticky prices, an exogenous decrease in the supply of money causes interest rates to rise which causes aggregate demand to fall . . . .” This is too compressed to enable me (a non-economist) to understand the New Keynesian view. I think “aggregate demand” is just spending (or is it spending on consumer goods?), in which case a fall in the quantity of money would itself reduce AD (assuming no corresponding increase in velocity). So money supply looks intrinsically relevant to AD. But how are interest rates supposed to be relevant? If I am a debtor, continually rolling over short-term debt, a rise in interest rates will (very likely) reduce my contribution to AD; but if I am a creditor, continually receiving interest payments, won’t a rise in interest rates increase my contribution to AD, perhaps by an offsetting amount?
So I understand (or, at least, think I understand) the monetarist picture, but the New Keynesian model remains completely unintuitive. Though this may run against the grain, could you please do a bit more to make the NK picture plausible?
Joe, The HPE explains the long run change in prices, and the expectation of that change causes asset prices to rise in the short run. I don’t follow your comment about saving. Saving is a flow, asset prices are stocks. There is no need for more saving to make asset prices rise. Every asset transaction is both a purchase and a sale.
You said;
“The increase in asset prices tells you that the hot potato process was channeled through the financial markets. Is that correct?”
I’d prefer to say the HPE was anticipated by the markets. And I’d prefer to call them asset markets, as my mechanism works even in economies that have no financial markets. On the other hand economies that have no financial markets don’t have much of an interest rate transmission mechanism.
Adam P, OK the effect is not immediate. It takes one second for prices to adjust upward. Are you happy now? Don’t forget that in a ratex world the anticipation of the HPE makes prices rise even before people spend the money. As the cash falls out of the airplane onto Bora Bora, the natives tell the person about to buy a mango that they changed their mind, the price will now be x% higher, where x% is the expected increase in the money supply from the airplane drop. There are countries where retailers put signs in windows that everything is x% above the sticker price, after a sudden currency depreciation.
Can someone please explain to me why the fed buys bonds, usually t-bills, when it wants to ease?
Why not open a savings account at all the biggest banks, when you want to expand the money supply make a deposit, when you want to contract the money supply make a withdrawal.
The fed deposits could be lent out on fed funds market or to fund, say, mortgages. No need to hold assets in order to sell them when you want to drain reserves back out.
Why isn’t it done this way?
Scott,
My use of the word “savings” was incorrect. I apologize. I was defining “the act of savings” as “the act of purchasing assets in the financial markets.” That’s a bad use of the term. You’re simply saying that when people have excess cash balances they might spend it on anything. It can be consumption goods at BestBuy/WalMart/the mall. It could be capital goods like factories/machines/computers. Or, they call up wall street and buy bonds/equities/real-estate/commodities/currencies/derivatives/insurance/etc. You are right, better to simply get rid of confusion and simply say “asset markets” instead of “financial markets.”
When you say, “I’d prefer to say the HPE was anticipated by the markets.” Wouldn’t it be more accurate to say “Future HPE was anticipated by the markets…. and created a current HPE all on its own…. which went through all the asset markets.” Expectations go up, money demand falls, HPE happens through the asset markets, price in asset markets go up. Right?
After all, an increase in AD is just the HPE. So when you say, “expected future increase in AD increased current AD,” you’re just saying “expected future increase in HPE created current HPE.” And we can know that the current HPE is happening because we see the prices of all sorts assets, whatever they may be, going up. When we see price going up around us in the asset markets, then we know that the HPE is going through all of these asset markets.
What I just described is precisely what you explained in a brief email with me a few weeks ago (for which I am very grateful). You said “I claim that a policy that doesn’t affect the current money supply, but raises the future expected money supply, will tend to boost current velocity.” You specifically approved when I described you’re idea as “If NGDP expectations go up, then this causes the demand for money to fall, and thus the velocity of money increases, and thus nominal spending/income go up.” These two quotes are exactly what I have just described in the first three paragraphs of this post.
In the textbooks, we are told that sellers do not increase prices for any reason. They will increase it only if they get more demand (or less supply). Now, the story I presented in this comment and from our email is precisely this: Asset prices go up because those markets experience more demand from cash holders trying to get rid of their cash because their expectations just went up.
However, in response to Adam P, you seem to be saying something a little bit different. You are saying to him the following: A seller of assets says to himself “My NGDP expectations have gone up. I must immediately increase my prices, even though I have experienced no change in demand or supply…. an increase in AD will follow” But, that’s totally different. They are not increasing their prices because they experienced a change in demand or supply. They are increasing prices all on their own, without external pressure, due to a change in expectations. And somehow, that increased AD. This is strange. Page 19 of the Cowen/Tabarrok Micro textbook lists “expectations of future” as one of the things that powerfully affects the market. But that is because it causes demand to increase, not because sellers increase price all on their own before demand has changed. You seem to be presenting two different constructs/stories.
I apologize if at anytime I am rude or disrespectful. I am simply very confused, fascinated, and really really want to understand.
Adam: Deposits are unsecured. Extending credit is pretty different from repo, no?
K, I’m pretty sure that depositors are fairly senior in the capital structure so it would effectively be secured by whatever tier 1 assets the bank had, essentially those same assets that would be repo’d.
However, the point here is that the reason the Fed holds assets or lends on repo is to get reserves back out if it wants. In this case that’s not a problem, if a particular bank took the deposit and lent it out randomly to people who couldn’t pay it back the reserves would still be in the banking system somewhere (in those peoples deposit accounts) and so could still be retrieved.
In effect the Fed would put newly created reserves on deposit with a bank, the bank would keep some of them in its account with the fed anyway. Unless they somehow exit the banking system wherever they go they end up back in someones account with the fed. But ultimately, at the start of the line they belong to the fed since it has a claim on them through its deposit account. Effectively they are the asset that backs them. (And the problem of reserves leaving the system as physical cash applies equally to the case of the fed buying a bond. If the value of the bond falls due to, say, a surge in inflation then those reserves become irretrievable.)
According to Nick and Scott this would be a simpler and more effective way to conduct monetary policy, so why doesn’t any central bank do this?
Joe, Scott is being entirely inconsistent.
To be clear, what he’s saying to me is that there never is any increase in AD. The expectation of the eventual price rise (no price stickiness here remember) causes prices to immediately rise and so equilibrium is restored. When the new money shows up it is held since the real value of the newly increased money stock has already fallen to the level that agents want to hold.
Now, this is a perfectly valid story but in this story there is no HPE and no increase in AD (to get more AD you need price stickiness, the response of P is sluggish and so in the meantime people try to rid themselves of soon to be devalued money. Velocity increases.) In this story you can say nothing about what enforces the price rise if one day the ratex mechansim fails, it could equally well be the interest rate channel or the HPE.
But actually a small amount of economic reasoning applied to the way monetary policy is actually conducted tells us that interest rates are the transmission mechansim.
In normal, non-crisis, non-liquidity trap times, the fed conducts monetary policy by buying a t-bill with newly created reserves. Consider the options available to the holder of the t-bill, say the bill’s initial price is 98. This agent has the portfolio choice of the bill or 98 in cash, he chose to hold the bill.
Now the fed buys the bill for 98 in cash, if the bill’s price hasn’t changed then this agent has exactly the same choice. Why doesn’t he just buy another bill? The answer is that he does, now if the bills price hasn’t changed apply the same reaoning to the agent that sold the first one the bill…
Equilibrium can’t be restored until the price of bills goes up enough that someone is willing to do something else with the cash. Only then can the cash leak out and become a hot potato. Interest rates must change.
Scott wants to say that people are rational yet completely fail to be consistent in the their portfolio choices, that’s utter nonesense.
Purchasing “savings” accounts would be neither cheaper nor more cost effective.
First, savings accounts are not demand accounts. In the U.S. you can only make 3 withdrawals per quarter from a savings account, and the bank can impose waiting fees. The CB cannot conduct real time monetary policy with savings accounts.
Second, savings accounts yields are far below FedFunds. That is seignorage to the banking sector that would otherwise be transferred to Treasury. Why should the CB spend more, and how do you select the lucky banks getting the savings accounts?
Third, banks don’t lend reserves. Banks extend loans if their cost of funds is less than the risk-adjusted lending rate. The CB manages banks cost of funds. Unless the cost of funds changes, stuffing banks with reserves isn’t going to cause a single additional dollar to be lent out, and it isn’t going to create a single additional dollar in deposits held by the non-financial sector. The cost of funds is set by interest rates, not the quantity of reserves in the system.
Fourth, why is holding deposit claims on banks easier for the CB than holding bonds? They are both financial claims. Unrolling one is not easier or harder than unrolling the other, except that withdrawing money from a savings account is more burdensome than selling a bond, due to the limit on the number of withdrawals per quarter and the waiting times imposed.
Fifth, the CB is not authorized to do this. The Treasury Act specifies what types of assets central banks are allowed to buy, and they are generally only allowed to purchase outright assets that are risk-free. A savings account is not risk free, which is why it is not an appropriate instrument with which to conduct monetary policy.
RSJ, you’ve completely missed the point.
first off you can substitute demand deposit account for savings account. It should have been obvious that’s what I meant if you actually cared to notice.
Second, it should have been equally obvious that I’m not advocating this as possible or desirable.
The point, should you care to comment intelligently, is that as far as I see Nick’s transmission mechanism implies that this would in fact be desirable. But that is a critique of Nick’s view, not an argument in favour of changing how the Fed actually operates.
LOL,
Adam, I didn’t realize your (idiotic) comments @5:05 and subsequently were ironic, and therefore “intelligent”, whereas my pointing out the flaws of said comments were not intelligent.
It is all too meta for me.
RSJ, I think the question of whether or not my comments actually were idiotic is still open. It may well be the case but you haven’t made any progress showing it.
So again, if Nick’s transmission mechansim is correct then why isn’t monetary policy done this way?
The fact that the Fed is not authorized to do it is not relevant, they could be if congress wanted them to be. The fees/waiting times garbage was just silly. I’ve had many demand deposit accounts in the US that had neither fees nor waiting times for withdrawal.
Care to say something relevant?
And of course your point 2, about deposit accounts yielding less than fed funds is also irrelevant in Nick’s paradigm. It’s supposed to be all about the money supply.
Adam,
You said
“Why not open a savings account at all the biggest banks, when you want to expand the money supply make a deposit, when you want to contract the money supply make a withdrawal.
The fed deposits could be lent out on fed funds market or to fund, say, mortgages. No need to hold assets in order to sell them when you want to drain reserves back out.”
Which suggested that savings accounts, and not demand accounts, was your chosen instrument, and that banks lent reserves. Later on, you suggested that this approach would be cheaper for the CB.
I pointed out that all three claims were wrong.
But of course, the CB could do this to manage the marginal price of reserves instead of open market operations. When banks, as a whole, are short of reserves so that they bid up the OIR, the CB can make a large deposit, more reserves are supplied, and the OIR will fall.
Except for the seignorage/risk issues, your approach would be equivalent to OMO. Central banks can manage OIR without doing any OMO per se. Apart from the practical unworkability and expense of your plan, it is perfectly equivalent to what Central Banks do today.
The only differences between the different approaches — A corridor system a la RBC, old-school OMO a la Fed, and the Adam P approach are cost/flexibility/risk.
Hence when you asked why don’t central banks switch to the Adam P approach, the response is cost/flexibility/risk.
If you want to change the approach to using demand accounts, you get the flexibility back, but at a massive increase in cost (loss of seignorage income) and you still have the risk (deposits are not guaranteed for all amounts).
With savings accounts, there is a smaller cost, but still a greater cost than using bills (as savings accounts pay less than the portfolio of bills/bonds that central banks hold) but with more inflexibility, and still the excess risk.
So that is the answer to why central banks don’t do this. When they want to avoid purchasing bills, they use a corridor system instead of your system.
But I didn’t realize that you weren’t really asking the question, you were making fun of Nick. Specifically, you were making fun of the belief that the mere existence of more deposits must somehow drive up prices. I certainly don’t agree with Nick’s transmission mechanism, for the reasons I stated previously. I agree that in a modern economy with outside money, there is not going to be a hot potato effect that is somehow distinct from an interest rate effect. The fundamental effect is an interest rate effect, but interest rates are managed by managing the marginal cost of reserves — by supplying or withdrawing reserves, and this can be done by having the CB make deposits with banks, directly increasing the quantity of reserves, or by having the CB pay interest on reserves, directly controlling the marginal value of reserves, or by having the CB buy bills or bonds or anything else. All these approaches are equivalent for purposes of managing reserves, and differ only in cost/risk/flexibility.
The last sentence in the above should read “All these approaches are equivalent for purposes of managing the marginal price of reserves …”
This is getting silly but I guess I was unclear if you take the comment out of the context of the post.
The point was not to make fun of Nick, or to advocate that this would be a good idea. The point was whether it would really work, to say it’s equivalent to what they already do is a reasonable answer but not an obviously correct one.
The real issue is when you say that policy can be conducted “by supplying or withdrawing reserves, and this can be done by having the CB make deposits with banks, directly increasing the quantity of reserves, or by having the CB pay interest on reserves, directly controlling the marginal value of reserves, or by having the CB buy bills or bonds or anything else”.
I’m suggesting that perhaps it matters what the central bank buys and that it buys something at all, that things would be different if it just put money into the system by putting it on deposit. Now, Nick’s post clearly implies that it doesn’t matter and you’ve agreed with that. I’m questioning whether that’s true or not.
I tend to think the portfolio effects of actually buying the bill, as I described above in the context of disputing what Scott Sumner said, matter at least a bit.
Anyway, since it’s really a theoretical point I was getting at whether or not it’s more expensive to do things this way is also not really the point.
Imagine it was exactly the same cost, which would be better policy?
Finally though, I don’t think it’s as obvious as you think that trading in the bond market is cheaper. The NY fed has a team of traders that are decently paid and would not be needed if policy was implemented by simply making deposits and withdrawals.
Adam,
I don’t see anything special with having the government directly buy bills, or having the government ensure that the financial sector has a marginal cost of reserves. In the case of the latter, they will buy the bills up until no arbitrage is possible.
Remember that the “interest rate” that central banks set is the overnight interest rate charged by one bank to another. E.g. for Canada, “The Bank of Canada’s target for the overnight rate is the rate on collateralized, market-based overnight transactions”. Central Banks can control this rate by agreeing to lend an unlimited amount at the policy rate to other banks, and by paying interest on reserves at the policy rate. That creates a type of corridor, in Canada, a 50bp corridor, and the actual overnight interest rate is typically in the mid-point of this corridor.
None of this requires bond purchases or sales. The OIR can be managed quite effectively without it. I think the Bank of Canada keeps OMO as a backup option, but it certainly does not control interest rates via OMO — it controls them by implementing the corridor.
The important point is that what matter is the marginal cost of reserves, not the quantity of reserves. This is basically the dispute here. Nick focuses on quantity, and needs some God-given assumption about the relationship between quantity and marginal cost. He should be focusing on the marginal cost of money rather than the quantity of money.
In Canada, the quantity of reserves are targeted at near zero — I think they want a system wide reserve position of 25 million — which is near zero as a fraction of deposits. Nevertheless they control the marginal cost of reserves very well, and that propagates to other rates.
Now real-time adjustments of quantity will succeed in adjusting the marginal cost of reserves, obviously, but it is less accurate because you are always driving the system to be in a deficit or surplus reserve position. Banks as whole either have “too much” or “too little” reserves, in which case the reserve rates are going down or going up. Adjusting quantity is akin to controlling the sign of the time derivate of the OIR, in order to keep the actual OIR bouncing around the target. It is better to directly control OIR by adjusting the marginal cost of reserves directly.
In all of these cases, you are controlling lending rates in the interbank market, which propagate out to rates on bills and such. In a modern economy, a currency issuer does not need to intervene in the bond markets in order to control lending rates for its currency.
And IIRC, the ECB also does not perform a lot of OMO, but primarily manages rates by lending to banks at the policy rates — but that system is a bit hard for me to decipher, so others can clarify that.
The expense that I was referring to was not transactional costs, but the lost interest income, which in normal times is turned over to the treasury as seignorage income. There is a huge spread between what depositors get and what banks pay, and we want the CB to be on the right side of that spread.
And the main point of all the above — getting to criticizing Nick’s transmission mechanism (which I am sorry you were not mocking) — is that there is no a priori relationship between quantity of reserves and their marginal cost that is independent of the institutional framework.
Therefore you cannot talk about a transmission mechanism from the quantity of reserves to prices that is independent of the regulatory framework.
Canada decided to have $25 million in reserves, and it can have any marginal cost that the BoC wants. If it quadrupled the system level of reserves to $100 million, then Scott S. would probably argue that the price level would also quadruple.
But it wouldn’t, as long as the marginal cost of reserves remained the same.
The non-financial sector, which sets prices, doesn’t know or care about how many reserves are in the banking system. All it cares about are the lending/borrowing rates offered to the non-financial sector. The financial sector only cares about its marginal cost of funds when making a loan or buying up bills.
Similarly, if the marginal cost of reserves is the same (e.g in a zero bound), but the CB floods the system with reserves, then still there isn’t going to be an effect on prices.
In any case, only the effect of interest rates on prices can be said to be independent of the institutional framework governing the financial sector, because this is what the non-financial sector sees. The quantity of reserves is an input into the blackbox, and the box spits out a borrowing rate for the non-financial sector. The relationship between quantity of reserves and the output rate is implementation dependent.
Therefore if you are going to be an old school quantity person, then you had better become an expert on the institutional framework, and you need to recognize that your conclusions will change as soon as the regulations change.
You can’t use quantity of high powered money arguments in a backscratch economy and then assume you are saying something relevant about our economy. But you can use interest rate arguments in a backscratch economy and say something relevant about our economy.
Adam: “And of course your point 2, about deposit accounts yielding less than fed funds is also irrelevant in Nick’s paradigm. It’s supposed to be all about the money supply.”
Right. In Canada (and elsewhere too) overnight deposit rates have been stuck at token amounts (25 bps or so) since the early nineties. Even non-guaranteed deposits (over $100K) have earned basically nothing (40 bps?) for about 10 years. There’s a total market failure in deposit rates – they don’t respond to policy rate changes at all. So, what if the BOC tried to stimulate the economy via direct deposit? Worst case, the banks would just take the free money and buy T-bills, resulting in the same drop in rates as if the BOC had spent the same amount of money on T-bills. So same thing, except the banks earn the seignorage instead of the BOC. But the banks could, if they really wanted, just leave the deposits on balance sheet and do nothing with them. It doesn’t cost them anything. So the money supply only affects the economy through the rates channel.
Adam: “if a particular bank took the deposit and lent it out randomly to people who couldn’t pay it back the reserves would still be in the banking system somewhere (in those peoples deposit accounts) and so could still be retrieved.”
But that money would belong to other people now (not the deadbeats who borrowed and spent it). The CB loses an asset if a bank at which it has made a deposit goes broke. And the associated liability can’t be made to go away. End of story. (Unless… you are saying that the Fed doesn’t need assets and that M0, since it doesn’t mature, should be thought of as Fed equity rather than debt. But then, if the wants to tighten, the government will have to tax it back. This would be pure helicopter money that totally discards any vestiges of the real-bills doctrine. Maybe not a bad idea.)
There’s a big difference between secured and unsecured lending. If a bank goes bust, the CB gets to keep the security and then claims any loss (hopefully small) on that security as a bankruptcy claim pari passu with other senior unsecured creditors. Unless the collateral is worthless, recovery will be higher.
Historically, at the discount window, the Fed only took extremely high quality, liquid collateral (typically highly rated sovereigns or government guaranteed). Post crisis, almost anything investment grade apparently became acceptable and the discount window spread declined to 25 bps. So Bagehot’s principle of lending against good collateral at penalty rates has gone out the window lately. But the principle, at least, is not to take bank credit risk. Imagine if a bank at which the Fed has made significant deposits, suddenly has problems. Is the Fed supposed to withdraw its deposits in response (and kill the bank) or ECB/Greece style double up because Trichet would rather take horrific losses later (on somebody else’s watch) rather than lesser losses now (on his watch)? Lending based on collateral quality is fair to all participating institutions and far less fraught with moral hazard.
If the CB lends money that is not repaid, the loss is fiscal policy — it is paid for by Treasury, not the CB.
Treasury supplies capital to the CB and receives the seignorage income from the CB’s operations. So depending on how the CB decides to recognize the loss, that will correspond to either less seignorage income for Treasury as the CB rebuilds its capital or for a one-time replenishment of capital supplied by Treasury.
A good example is the Maiden Lane assets, in which case Treasury has already agreed to absorb the losses. Of course those losses are small in comparison to the seignorage income that the Fed is now providing.
Adam P, Your argument that you need sticky prices to in order for money to affect AD makes no sense. Imagine prices rise immediately in proportion to M. In that case nominal expenditure also rises in proportion to M. That means the AD curve must shift to the right in the AS/AD diagram. That’s the only way you can get a higher NGDP. But that means AD has risen. So the entire premise of your argument is flat out wrong.
You discussion of interest rates and monetary policy combines two unrelated issues. One is the fact that when prices are sticky, even a helicopter drop of cash will affect interest rates. Because prices are slow to adjust, interest rates become the shock absorber in the short run. But if prices are flexible, a helicopter drop doesn’t produce any “liquidity effect.” There is no change in interest rates, all nominal prices immediately rise upward in proportion to M.
Now assume we don’t have a helicopter drop, but instead do an OMO. In that case monetary policy reduces the interest-bearing part of the debt, which has a (very tiny) affect in interest rates. But this isn’t the effect most people associated with monetary policy, which reflects price stickiness. After all, most OMOs are tiny. Instead it reflects the fact that debt held by the public falls. It’s fiscal policy–an inflation tax. And that’s true whether the new base money is used to buy fighter jets, or T-bills. If they buy fighter jets, then the government doesn’t have to borrow as much, and there are fewer T-bills held by the public. I generally ignore this effect because it is so small.
It can’t be true that interest rates cause the long run adjustment in the price level, because interest rates don’t change in the long run. Only prices change. So why do prices reach a new equilibrium rising in proportion to M in the long run? Do you seriously believe that low interest rates caused the German hyperinflation? It boggles the mind. Even the 100% inflation rates observed in Latin America a few decades back can only be explained with money growth, not interest rates.
Sorry to intrude from the peanut gallery, but I’m confused by the first paragraph in Scott’s last post: Isn’t AD a function of M/P?
Patrick, there’s a host of confusions here. Scott and Adam seem to be talking past each other. My impression is that Scott thinks of AD in purely nominal terms. So of course money matters even in a flexible-price, Panglossian setup. A k% increase in M means an instantaneous k% increase in NGDP. But there is no impact at all on RGDP in such a model; and Adam is thinking in terms of RGDP. To me it makes no sense to speak of a transmission mechanism in such a model. Everything happens instantaneously.
Kevin, I think you’re exactly right. I’d also submit that Scott is the only economist in history who really believes that AD measured in cents is actually higher than AD measured in dollars.
As for the thing about monetary policy by deposit, I think what I’m really wondering is why Nick hasn’t written any posts criticizing the Fed for buying long-term bonds in QE. In Nick’s view even with T-bill yields less than the IOR rate, increasing the money supply by buying bills should work just as well as increasing the money supply by buying long-term bonds. And with long-term bonds the fed takes more risk.
Patrick. M/P fell by 99% during the German hyperinflation. Would you and Adam argue that monetary policy was depressing AD during the German hyperinflation? M/P is a useless indicator of AD, right up there with nominal interest rates.
Kevin, Whether you want to call it a transmission mechanism is irrelevant, but there must be a REASON why a monetary increase causes prices to rise in a flexible price model. Surely it doesn’t happen by magic. I claim that the HPE is the most plausible REASON. Call it a transmission mechanism, or something else, I don’t care. It happens quickly, because when rational people see the HPE is going to work, prices rise in anticipation (in a flexible price model.)
Adam, I’m assuming the standard AS/AD model used in every macro textbook. You are the one confusing AD with quantity demanded. In a flexible price model the AS curve is vertical, and when AD increases only prices rise. That’s what we teach all our undergrads. I’m talking about the mainstream view of AD, you have in mind some alternative definition, which seems to me to be quantity demanded, not AD.
I think your mistake is in assuming that because output generally rises in response to more AD, that the extra purchases ARE the increase in AD. But that’s crazy. It’s be like saying that the fall in computer prices has caused more demand for computers. No, the extra supply of computers causes more quantity demanded. And again, flexible prices are not an uninteresting assumption, as prices are flexible in the long run.
BTW, I thought the Fed should have just bought more T-bills.