Robert Hall and the Monetary Transmission Mechanism
When I discovered Nick Rowe’s blog a few days ago, I told Nick it was nice to find someone who approached monetary issues from a similar perspective. I had no idea how similar. I was all set to write a post today on the distinction between temporary and permanent monetary injections, and also Krugman’s expectations trap, when I noticed that yesterday Nick had already done so here. Because I agree with what he has to say (and he says it in a much more straightforward way than I will), I’ll try to approach the issue from a slightly offbeat perspective in the hopes that readers will benefit from both posts.
Yesterday I learned that I will soon have a short note published on the interest penalty idea in The Economists’ Voice, and in order to respond to the editor I needed to re-read some of Robert Hall’s papers on the subject. I noticed that Hall concluded a paper entitled “Controlling the Price level,” on an interesting but ambiguous note:
The reasons lie in the imperfectly understood realm of monetary nonneutrality, which may be defined broadly as sensitivity of real activity to any instrument of monetary policy whose theoretical effect is only to change the price level. As James Tobin has written:
The tail wags the dog. By gently touching a tiny tail, Alan Greenspan wags the mammoth dog, the great American economy. Isn’t that remarkable? The federal funds rate is the shortest of all interest rates, remote from the rates on assets and debts by which businesses and households finance real investment and consumption expenditures counted in GDP. Why does monetary policy [have real effects]? How? It’s a mystery, fully understood by neither central bankers nor economists.
Before continuing let me recommend Robert Hall to anyone seriously interested in understanding monetary theory. His papers can be difficult (more conceptually than technically) but if you persevere you might come to feel that you are peering into the mind of someone who understands money better than anyone else on earth. And I say that despite the fact that he never replies to my emails. Tobin is also very smart.
I think Hall is getting close to the essence of the problem when he mentions policies “whose theoretical effect is only to change the price level.” In my view we should step back from the question of how monetary policy changes current real activity, and first consider how it changes future nominal activity. If we can answer the latter question, we can make the question of real effects much easier to address, without making implausible assumptions about little fed funds tails wagging giant real GDP dogs. And as we know from any intro macro textbook, there is nothing particularly mysterious about why a 20% increase in the money supply will raise future nominal spending by 20%. We simply need to assume that there is a well-defined demand for real cash balances.
I’m not quite sure what to make of the Tobin quotation. Superficially the metaphor is similar to my Archimedean lever metaphor from the post entitled “A short course in monetary theory.” But he applies it to real output, whereas I applied it to nominal output. As far as the “mystery,” I do agree that the entire monetary transmission process seems very obscure. But I also think that if we stop focusing on the fed funds rate, it might become slightly easier to visualize.
Suppose we accept that a 20% increase in the money supply (assumed to be permanent) leads to a 20% increase in future steady-state level of nominal spending. What does that tell us about the transmission mechanism? Well there are two possibilities. One possibility is that wages and prices are completely flexible, and immediately adjust to the money supply increase. That seems very implausible. So let’s assume wages and prices are sticky, what then? In that case all sorts of real things happen to the economy today. Nominal interest rates might rise, reflecting the Fisher effect. If so, then velocity will also increase, causing nominal spending to rise sharply. If wages are sticky and nominal spending increases sharply, then output will also increase. Or, nominal interest rates might not increase, in which case the expected future increase in prices will immediately depress real interest rates. The higher expected future nominal GDP will also cause the (flexible) prices of all sorts of assets to immediately rise. These include stocks, commodities, commercial property, foreign currencies, etc., etc. And by the way, the overnight rate on interbank loans might also change a little bit. Some economists seem to think that that last minor item, the so-called liquidity effect, is the essence of monetary policy.
Suppose the economy were in a liquidity trap. Would anything in the preceding analysis change? Probably not. A permanent 20% increase in the money supply would still raise future expected nominal spending by 20%. The only exception would be if not only were current interest rates equal to zero percent, but expected future interest rates were also stuck at zero. In that case, that is if it was expected that we would be in a liquidity trap between now and the end of time, T-bills would be indistinguishable from large denomination currency notes. Then if the new money had been injected exclusively via OMOs involving T-bills, the effective money supply would not have changed at all. In that case if the Fed wanted to expand the money supply, they would have to buy some sort of non-monetary asset, some asset other than T-bills.
But even the previous example is conceding too much. Not even Paul Krugman, one of the most enthusiastic proponents of the view that monetary policy is currently out of ammunition, would argue that we will be in a liquidity trap forever. In his famous 1998 paper he assumes that eventually the economy will exit the liquidity trap and that the quantity theory will hold in the long run. But in that case a liquidity trap creates no obvious problems for the efficacy of monetary policy. A permanent 20% increase in the money supply will raise the future price level by 20%, and that will immediately trigger all the real transmission mechanisms discussed above. So what is all the fuss about? Nick correctly points out that Krugman’s argument requires the public to anticipate that the monetary injection is temporary. In that case there is no increase in the expected future price level, and thus none of the current changes that are normally triggered by higher future prices.
I agree with the logic of Krugman’s argument, but don’t see how it is especially relevant to liquidity traps. To see why let’s consider three examples of temporary monetary injections, that differ in ways that are often assumed to be significant (although not by Krugman.) In each of the three cases, I will assume that the Fed increases the money supply by 20% in year one, but then plans to pull the money back out of circulation in year two—exactly the sort of example envisioned by Krugman.
Case 1: Liquidity trap. Let’s assume that the Wicksellian natural rate is 0% in real terms, and the interest is stuck at 0%. When they increase the money supply by 20%, nominal rates on T-bills stay at zero, and T-bills remain indistinguishable from money. The price level cannot rise, because if it did so it would be expected to fall a year later. But that would push the expected real return from holding cash above zero, which would be greater than the expected real rate of return earned on other assets.
Case 2: Normal monetary policy. Now let’s assume that the Wicksellian real interest rate is a positive 2%. What happens to the price level when the Fed temporarily increases the money supply by 20% in year one, with the expectation it will fall back to the original level in year two? I’m not sure quite how to model this, but I can’t see how prices could rise very much, even if wages and prices are perfectly flexible. To see why, just imagine that prices did rise by 20% in year one, and fell back to their original level after the cash was removed from circulation in year two. In that case the real rate of return from holding cash between year one and year two would be 20%, far exceeding the Wicksellian natural real rate. In fact, I would guess that once the price level had risen by about 2%, the expected rate of return on cash would become high enough that, in anticipation of deflation, cash holdings would become as attractive as other assets. In that case the demand for cash would soar as the price level increase approached 2%, and this would prevent any further inflation. So unlike case 1, the price level does rise a little bit, but (unless I am mistaken) not very much. And if the money supply increase had been 200% instead of 20%, the result would have been the same.
Case 3. Fiscal plus monetary policy. I suppose one could argue that case 2 is just like case one, because nominal interest rates fell to zero after the money supply rises 20%. OK, but then let’s consider the one policy that everyone supposedly agrees will work, the helicopter drop. Only let’s make things fairer by simply having the government hand each citizen an extra amount of cash equal to 20% of the current per capita level of cash in circulation. To make it like the Krugman example, citizens will also be told that the cash infusions are temporary, and that a head tax will be imposed in year two to pay for the redemption of the currency. What happens to the price level in case three? I don’t see how anything is any different from case two. The citizens know they are no richer, so there is no wealth effect. The price level cannot rise more than two percent without triggering an enormous demand for cash as a store of wealth (in anticipation of deflation.) So prices would barely budge. In nominal terms, the policy would be virtually impotent.
Now I suppose someone could argue the following: “Of course not much happens in case three, you have assumed Ricardian equivilence. And we all know that RE is a completely unrealistic assumption.” OK, let’s say that’s true, I am still having trouble understanding exactly the assumption I made in case three, that I didn’t make in case one and two. Is it because in case one I assumed the public knew that the cash was going to be injected, and then pulled out a year later, but in case three I assumed both that the cash injection was temporary, and the fiscal stimulus was temporary? Why is the second assumption less plausible than the first? Isn’t it much easier for a central bank to make a monetary injection permanent, than for a fiscal authority to defy the intertemporal budget constraint and make a fiscal stimulus permanent? Which policy would a rational public be more likely to view as temporary?
To summarize, I can see why a temporary monetary injection might have little or no impact on the price level, I just don’t see what it has to do with the debate over monetary policy versus tax rebates. And I believe that Krugman also sees things this way, which presumably explains why he forcefully advocated more government spending rather than tax cuts. But I also notice that most Keynesians that I interact with seem to rely on simplistic arguments that tax cuts boost purchasing power while monetary policy can no longer depress interest rates, arguments that seem to lack any theoretical foundation. (BTW: I also have doubts about the expansionary impact of government spending, but I will leave that for another post.)
I would like to go back to the Hall quotation about policies that affect the price level “in theory,” by considering the impact of a permanent 20% increase in the money supply. Suppose that by year two wages and prices have fully adjusted to the permanent monetary injection, and thus nominal spending is 20% higher than before the increase. What would that do to the economy in year one, right after the currency injection? Earlier I argued that there was a complex set of changes triggered by higher expected nominal spending (real interest rates, stock prices, real wages, commodity prices, exchange rates, nominal interest rates, yield spreads, real estate prices, and so on.) It is very likely that both prices and output would rise in year one, perhaps by quite a bit.
It would be nice to know how much of the year one changes in nominal spending can be attributed to changes in the fed funds rate. Perhaps we can get some hint of the maximum impact of the fed funds rate by contrasting this example with case 2 above. Recall that in case 2 the cash injection drove nominal rates down to zero, and yet prices rose only slightly because most of the cash was hoarded. With a permanent increase in the money supply, the fed funds rate obviously cannot fall any lower than zero, so presumably any additional effects would be due to other transmission mechanisms. And I don’t see how these “additional effects” wouldn’t be pretty large. If you don’t believe me, change the example from a 20% to a 200% increase in M. Changes in the nominal fed funds rate just don’t seem very important.
If you still have nagging doubts about my dismissal of tax rebates, and assume it is all based in Ricardian equivalence, consider the following scenario. Imagine a country where per capita income is $40,000 and the average person carries $100 in their wallets. (Maybe they withdraw $180 from ATM machines each week, and gradually spend the money down to $20 before restocking–so average cash balances are $100.) Now consider the helicopter drop without Ricardian equivalence. That is, assume each citizen treats the free $20 bill from the governments as a gift. But they also continue to assume the money supply increase is temporary. The citizens know the $20 will be pulled out of circulation in year 2, but assume the cost will be borne by non-Americans, perhaps by a slightly eccentric Belgian billionaire. How much impact would that extra $20 of (falsely) perceived wealth have on consumption?
You might object that the effect is small because I assumed such a small fiscal stimulus. But if that same measly $20/capita represented a permanent increase in the money supply, we know it would raise expected future nominal spending by 20%, thus triggering all sorts of dramatic changes in asset prices and output in year one. The difference between an OMO and a helicopter drop is trivial, the difference between temporary and permanent monetary injections is huge. There’s reason it’s called “high powered money.”
I admit that the previous example is not really representative of the U.S. economy. Although most people hold relatively little cash for transactions, because gangsters, tax evaders, and foreigners hold lots of cash, the cash/GDP ratio is roughly 5%. The point of the example is that in a normal country where people use cash for transactions, the impact of monetary policy, even if conducted through a helicopter drop, works primarily through changes in the future expected path of nominal GDP, which then affects the current level of all sorts of relative prices and interest rates.
I don’t know why monetary economists don’t spend more time looking at the stock market, or other asset markets, rather than the fed funds rate. In my rational expectations post I discussed how although the December 2007 Fed rate cut had only a tiny impact on extremely short term rates, it had an enormous impact on both U.S. and world stock markets, with roughly 5% of total stock wealth riding on the decision over whether to cut rates by 25 or 50 basis points. And I could cite a number of similar examples. Sometimes it seems like macroeconomists treat the stock market like that crazy uncle that the family is slightly embarrassed to acknowledge. All the jokes about “predicted 11 of the past 7 recessions.” Maybe they are afraid of sounding like Arther Laffer. (Wouldn’t it be ironic if Laffer’s greatest legacy was his sophisticated understanding of monetary policy.) And does the fed funds rate predict any better than the stock market? In any case, modern macro theory suggest two different reasons why recessions shouldn’t be predictable at all. Most recessions are triggered by unanticipated nominal shocks, and if they were predictable the monetary authority would set policy at a level expected to avert them.
So to conclude, I think we need to visualize monetary policy in the following way; an effective monetary policy is policy that is expected to have a permanent effect, where the long run effect is a higher level of nominal spending (this implicitly combines Krugman and Hall’s insights.) If one accepts this perspective then monetary policy indicators are indicators of expected future nominal spending. If this is true, then monetary policy became ultra-contractionary in September and October 2008.
If not, where is my reasoning defective?
I should probably conclude with a brief comment on the difficulty of credibly promising permanent increases in the money supply. This is the hypothesis that underlies Krugman’s advocacy of fiscal expansion (although I doubt whether many of his followers know this, or would find this assumption plausible if they did.) As I have already indicated, I am not persuaded by the examples cited by Krugman. Whatever their true preferences, the BOJ has persistently behaved like a central bank that wants to maintain a deflation rate of 1% or so. So why should we be surprised that they succeeded?
It is too soon to know what is going on at the Fed. I still hold out hope that the deflation will only last a few more months, and then prices will begin rising gradually. But right now they are paying interest on reserves, which defeats the whole purpose of QE. They are withdrawing base money from circulation as it is no longer “needed,” despite their admission that expected nominal growth is far below their target. And they are refusing to set out an explicit price level path, and indeed even hint that if they fall short of 2% inflation this year, they will not try to catch up with above normal inflation next year. In other words they are doing everything wrong, if their goal is to establish inflationary expectations of 2% or so.
Because Ben Bernanke entered his job with views similar to mine, i.e. that the zero bound does not prevent monetary expansion, I much rather be in Krugman’s shoes if I was debating this issue. But I still think I am right.
Tags:
13. March 2009 at 22:20
“But they also continue to assume the money supply increase is temporary.”
Earlier you asked whether the term-structure of the Fed OMOs matters. Surely this is part of the answer: holding short-dated debt is easily reversible. Holding long-dated debt is not. In particular, if the Fed is successful the market-value of the debt will decline, therefore they must either “wait” or take the loss.
More so when the transactions are repos. Given that the Fed is lending primarily through the TAF, the intervention is very temporary because it unwinds automatically.
14. March 2009 at 03:26
Jon: Aha! (Little light bulb flashes.) Especially your last paragraph.
Scott: And I had just been thinking to myself: “Funny how Scott hasn’t said much about the temporary/permanent distinction. Oh well, I had better do it myself”.
Case 2: The standard answer is that the current price level rise will be inversely related to the interest-elasticity of the demand for money (assuming all interest rates adjust to a full Fisher effect). If the interest-elasticity is 0, a 10% increase in M will cause a 10% increase in P. If the interest-elasticity is infinite, it will cause a 0% increase in P.
Implicit assumption that makes the difference between cases 2 and 3: could be many, but the most plausible is that in 2 the helicopter drops money on people who already hold 0% bonds, and will be taxed back from the same people. In 3 the helicopter drops at least some of the money on people who do not hold 0% bonds, and who are borrowing-constrained and/or liquidity-constrained, and face an (unobserved) >0% implicit interest rate on borrowing.
I need to read and think through the later parts of your post more carefully before responding.
By the way, the next question is: “How could we tell, from watching nominal interest rates, whether quantitative easing was or was not working? Would a rise, or a fall indicate that QE was, or was not, working?” I think that’s another way of looking at what you are talking about in part of your post. (It’s just a variant of the old monetarist question: “do high nominal interest rates signal tight or loose monetary policy?”)
14. March 2009 at 04:28
How can such a small tail wag such a large dog?
Consider an analogy: suppose Canada has a fixed exchange rate with the US. But instead of the Bank of Canada committing to peg the C$ to the US$, it’s the other way round; the Fed promises to covert the US$ into C$ at par. The Bank of Canada promises nothing. Then the Canadian tail would wag the US dog.
(I got this insight from a working paper from somewhere in New Zealand, about 8 years ago. Can’t remember the author. Maybe RBNZ.)
Why does the Bank of Canada control Canadian monetary policy? Why isn’t it the Bank of Montreal (or one of the other commercial banks), which is bigger than the BoC? Because the BMO makes its liabilities redeemable into BoC liabilities, not vice versa. BoC makes no promises to redeem its liabilities for anything. It can (and does) do what it wants to do.
14. March 2009 at 04:37
A question from a non-economist not directly related to this post:
I don’t understand your suggestion to impose a penalty rate on reserves. I assume there something I’m missing from how the banking system works. Here’s how I thought it worked.
Base money / monetary base / central bank money = cash + bank reserves at the Fed. Total reserves only change if the Fed buys or sells some asset or if a bank borrows of pays back a loan from the Fed.
Deposit money is then created through fractional reserve banking when banks ‘pay’ money to non-banks.
If this is correct, how would a penalty rate on reserves increase deposit money? The only way for a bank to decrease its reserves is to push it to some other bank, pay back a loan from the Fed or buy some asset from the Fed.
So what did I get wrong? Could someone please enlighten this poor confused brain of mine?
14. March 2009 at 04:41
Jon, My argument here is that in the Krugman model what really matters is whether the injection is temporary or permanent, and that what you buy doesn’t matter very much. The Fed could buy a $100 billion in gold bars, but if they were expected to sell them back a few weeks later, it wouldn’t do much good. On the other hand if the currency increase had all been through buying zero rate T-bills, but was expected to be permanent, we’d already be off to the races. Even repurchase agreements can be done again and again, if the Fed intends to preserve a higher level of base money. And I think Krugman is right, within the framework of his model.
But there are other equally good models where it does matter.
Since buying something other than zero yield T-bills is more likely to trigger an excess cash balance effect right now, it might also be a better signal of the Fed’s intention to pursue QE, and thus might be more effective than buying T-bills or repurchase agreements. So I think your intuition is right, but mostly because of the signals it sends about Fed intentions.
Nick, I guess I was implicitly thinking that the interest elasticity of demand for cash is finite until prices rose 2% (at which point the expected deflation rate equals the expected real return on riskier assets.) Then the demand elasticity becomes infinite, and caps the rise in prices at 2%.
I probably wasn’t clear enough in case two. That should have been an ordinary OMO, not a helicopter drop. I wanted the three cases to be:
1. OMOs starting from 0% interest rate environment.
2. OMOs starting from a 2% interest rate environment
3. Helicopter drop.
Since all three price increases are pretty similar, it seems that whether you start from zero interest rates isn’t that important, nor is it all that important whether there is a fiscal component, was really matters is whether it is temporary or permanent. So I think that Krugman is right about one thing here, the distinction between temporary and permanent is very important, and not just important in a liquidity trap, but always. And that tells me that the only indicator of an EFFECTIVE monetary policy action, is one that changes the long run growth path for nominal GDP.
If you go back to my “Do you believe in rational expectations” post, you will find that I strongly believe that any effective action by the Fed to get expected NGDP growing much faster would raise long term interest rates (and probably medium term as well), and raise them sharply. That’s why when people say the Fed should buy long bonds to depress long rates they are only half right. They should buy long bonds (perhaps) but with the goal of raising rates.
14. March 2009 at 05:24
Scott,
I believe that your analysis is hampered by a focus on trying to cause an increase in the price level from an initial equilibrium.
The issue at hand is preventing a decrease in the price level. Basically, the actual, unchaning price level, is above a lower, equilibrium price level. Presumably, at some point, the actual price level would fall to that point. But that never happens, instead, the expansive monetary policy raises the equilibrium price level so that it doesn’t fall.
I will grant that the current price level is already below the “target” growth path. And so, getting it back up to where it “shoud” have been, is an icnrease in the price level.
And so, I suppose that is where we are right now. But, can monetary policy keep nominal income on target in the face of decrease in velocity caused by financial difficulties?
Anyway, I find the constant focus on these simple cases Of “suppose the Fed causes a 20% inflation.” to be unnerving.
If you are Krugman, and are looking for excuses to change the U.S. to Sweden, then treating monetary policy as if it’s goal is to cause 70s level inflation (as just a simplifying rhetorical technique, might have some advantages.) Krugman is adovacated using government spending to fill the gap in demand, returning nominal spending to trend, real output to trend, employment to trend, all with a shift in the allocation of resources to more worthy areas–other than private investment and consumption.
My goal is to get nominal spending back to trend, real production and employment back to trend, with any reallocations being between different sorts of consumption and investment.
Explaining how all of this works by discussing 20% inflations over a year or two, or massive ociliations in the price level might be helpful in some ways, but just because Krugman thinks that way, doesn’t mean that you should.
14. March 2009 at 06:50
Scott: you were clear enough, and I understood your 1,2,3. I wasn’t clear enough.
An OMO is equivalent to a helicopter increase in the money supply, plus a vacuum cleaner reduction in the supply of bonds, EXCEPT that the helicopter/vacuum cleaner only flies over people who already own bonds. 3 has the helicopter fly over everybody.
14. March 2009 at 07:48
Scott: I think Nick got my point more than you did. My point was that the method/target of the OMO affects its perception as being “temporary” or “permanent”. So yes, maybe it is the permanent nature that is critical. That’s consistent with my qualitative observation that 1) policy is currently ineffective and 2) the term structure has shifted short.
Nick: I believe the Fed actually refers to repos and purchases of Bills as TOMOs while calling purchases of notes and bonds POMOs.
Nick notes, “EXCEPT that the helicopter/vacuum cleaner only flies over people who already own bonds.”
By this analogy, the TAF only flies over depository institutions. The “AIG” money flies over European Banks. The currency swaps fly over depository institutions through the world… Where as historically the NYF OMO flew over the bond holding public.
14. March 2009 at 09:58
Deflationary expectations –
This is what they feel like:
http://online.wsj.com/article/SB123689292159011723.html?mod=googlenews_wsj
Noonan says there is no pill for this kind of depression. But there is of course – and Bernanke can provide it – if he chooses to do so.
Right now people are hording cash and gold, looking for a small house to buy with no mortgage – out in the country so they can grow their own food. And buying guns. Literally.
If Obama and Bernanke don’t do something about this they are completely crazy.
China says they want a guarantee for a strong dollar. Good luck with that.
15. March 2009 at 05:34
Bill, A few thoughts on your comments:
I do think monetary policy should be easily able to overcome any drop in base velocity caused by the financial crisis, but equally importantly I think it should be able to sharply increase base velocity by putting a penalty rate on excess reserves. I know people don’t like the Zimbabwe example, or even the 1933 FDR example, as they represent high inflation, and I don’t want high inflation. But I do think they tell us a lot about how monetary policy works to boost nominal aggregates even when much of the real economy is malfunctioning. It’s like a scientific experiment where you expose something to extreme conditions to better understand its properties. (BTW, the 20% number is roughly the rise in the WPI after FDR devalued the dollar.) If you don’t use a large nominal example, some real factor (like the difference between a 0% Wicksellian real interest rate and a 2% real rate can seem very important, but when you run the experiment with a big monetary shock, you realize that none of these real factors play a significant role in limiting fiat money regimes ability to target NGDP growth at 3% or 5% or whatever they choose.
Regarding Krugman, I am forcefully advocating a monetary solution, precisely so that we don’t end up with Krugman’s preferred big government solution (and this relates to my newest post.) BTW, although I don’t support high inflation, I think high inflation tends to make voters move to the right, whereas deflation makes them move to the left.
Nick, I do now see your point about the helicopter drop plus removing the bonds. Yes, you are correct. I also re-read your other comment about Canada being the tail wagging the dog if we pegged the U.S. dollar to the Can$, and think that is a nice way of expressing the idea. I had never thought of it in that context. I wonder if the UK/US relationship circa 1913 comes close to that example.
Jon, Yes, I agree with what you say here. I did try to mention that the method could matter, but I should have highlighted that point better, because the first part of my response suggested you were wrong, and I really wanted to indicate you were right, but primarily because of how methods of monetary injection are correlated with the temporary/permanent distinction. I did not know about the Fed terminology you mention, but that makes your point even stronger.
JimP, Noonen is a very graceful writer, but she doesn’t really understand monetary economics (I’m not blaming her, as even many economists outside macro don’t really understand how powerful a fiat money regime is.) Nevertheless, she has a good ear for detecting changes in the national mood. I don’t know if anyone noticed this, but in some of my early posts one of the most important points I was trying to make is that crises created by excessively contractionary monetary policies almost never look like they were caused by a lack of money. To the average person, and even to many economists, the problem seems to be caused by what are actually symptoms of this monetary policy—financial panic, fear, consumer pessimism, etc.
I think my only chance to ever develop a reputation in economics for an idea I developed, is if at some time in the future economists begin to think it was a monetary policy failure in September and early October that triggered the sharp NGDP contraction and worsening financial crisis. That’s a real long shot however, because it goes against common sense much more than even the more plausible view that monetary policy can do more today. So I am facing a real uphill battle on getting people to accept my conjecture.
15. March 2009 at 08:56
Scott:
I favor negative interest rates on excess reserves. (And other low risk, short term assets too. The lower bound is the cost of storing currency.)
If banks are capital constrained, the negative interest rates on excess reserves should get banks to hold government bonds of various sorts, that also have a zero risk weight in calculating capital requirements. This should expand checkable deposits and required reserves. To the degree that people who were holding these goverment bonds just increase their holdings of checkable deposits, the demand for money rises to match the supply. But, of course, some of those who were holding govenrment bonds might spend the money on something else, which is the point.
Also, some years ago, there were estimates that sweep accounts resulted in 1/2 of trasactions accounts to be reported as savings accounts. There is no reserve requirement on them. I would guess that if there were a penalty on excess reserves, then the motivation would reverse, and there would be a shift to reported checking and away from reported savings accounts. (This is all to the good, as far as I am concerned, but this shouldn’t impact spending.)
Rather than solely focusing on the money muliplier (or base money velocity) there should be an impact on the interest rates paid on checkable deposits–lowering them. Which should help by reducing the demand for money.
On the other subject:
Your position appears to be that the failure of Lehman Brothers increased the demand or reduced the supply of money. The Fed failed to clearly signal that it would increase the quantity of money enough to maintain nominal income. This caused a larger increase in the demand for money/worsened financial crisis. And so nominal income fell by more.
I think rather than focus on the word “cause,” would would stick to, “failed to properly respond.”
I find your faith in rational expectations a bit peculiar. Why should we assume that market participants agree with us rather than with the vast majority of “experts’ who don’t seem to know the difference between money and credit, who claim that when the Federal Funds rate hits zero, monetary policy is impotent, etc? While I think it would help if the political leadership, including the Fed, expressed confidence in quantitative easing, I think it is important to understand how this all works out when market participants don’t know what to think. Perhaps, it is even worthwhile to think about how it works when they believe in a liquidity trap. Or they believe that total spending is the sum of individual spending and “monetary policy” works only through providing more loans for people to fund their spending.
Everyday I see people, especially people closely involved in financial markets, who dismiss my understanding of monetary economics.
Last fall, there were polls showing that a majority thought that a return of the Great Depression was likely or highly likely. Do you really think it was because they saw that the Fed wasn’t committed to quantitative easing?
16. March 2009 at 05:42
ok, “penalty rate on excess reserves to increase velocity”. 1 – have we even gotten to a point where the Fed has recognized VELOCITY to be the problem. I don’t think they’re there yet. they’re throwing everythign they got at it and i don’t think they get that velocity is the problem. maybe the fed gets it, but politicians sure don’t. 2 – what about bank loss reserves? aren’t you playing with fire for penalizing loss reserves? 3 – why not create a 3rd party such as the TALF, RFC, etc… to do the lending that would be done by creating a penalty rate on excess reserves? who cares if the lending is done by C, BAC, or USA
16. March 2009 at 11:43
“I find your faith in rational expectations a bit peculiar. Why should we assume that market participants agree with us rather than with the vast majority of “experts’ who don’t seem to know the difference between money and credit, who claim that when the Federal Funds rate hits zero, monetary policy is impotent, etc?”
The Austrians used to distinguish between social-science knowledge and knowledge of time and place–by which they meant the notion that agents are generally ignorant of global information and modeling but know how much the smith, the machinist, and the smelter cost.
Just as in 1776, when Adam Smith published the wealth of nations… no man really knew what was happening but he knew his local situation, and his local situation was precisely the distributed knowledge that makes the ‘free-market’ effective when ‘central-planning’ is not.
16. March 2009 at 17:27
Bill, The causation question is difficult. If you consider a stable money supply (but which Money?) to be a stable monetary policy, then you could argue that the error was an error of omission. Indeed, “Errors of Omission” is the first three words of the title of the paper I wrote on the recent crisis. I would point out, however, that if you make this argument, you might have to discard the hypothesis of Friedman and Schwartz that the Fed “caused” the Great Depresison, as the monetary base increased sharply between 1929 and 1933. BTW, Bernanke agrees with Friedman and Schwartz, so I think by his own criterion, the Fed is to blame for not increasing the base enough. There is also an action the Fed took, interest on reserves, that made the crisis much much worse, and that wasn’t just an error of omission. There is also the issue of whether the collapse started with Lehman. I am inclined to think that policy was already a bit too tight, and some of the causation (I’m not sure how much) went from the economy to Lehman. In my view AD started falling slightly in August, and the decline picked up speed in September, even before Lehman could have had much impact on AD.
Regarding Ratex, the information you provide actually makes me more likely to believe in Ratex, not less. In my work I use Ratex in one context, and one context only–markets (or any aggregation of people) tend to be smarter than the experts. You can meet very smart people, like Krugman, who think monetary policy is impotent right now, so how can we expect mere average people to know what’s going on? Well individually they don’t, but collectively they do (as Jon mentions.) When the Fed was more expansionary than expected in December 2008, the stock market rallied strongly, despite the fact that “experts” like Krugman said the action should have had no impact on equity prices. In my study of the Depression I was frequently surprised by how far-sighted markets could be. Of course there are times when they are wrong, nobody can predict the future perfectly, but if you ask “Are market expectations consistent with my model of monetary theory?” I would answer “Yes, markets seem to respond to policy shocks as if they get it.” Or take your example of polls that show a Great Depression is likely. If it were true the public really believed that, I might begin to question Ratex. But they don’t. How do we know? Because markets are not signaling another GD, they are signaling roughly 1% expected inflation, not the negative 5-10% per year we had in the early 1930s. When people have to put money where there mouth is, they suddenly become very “rational.”
The financial markets provided the best estimate of the problems we faced last October, they turned out to be right, and they provide the best estimate today. Ex post, they may be right or wrong, but on average they are more likely to be correct than any single expert. I learned to believe in Ratex through painful experience; if the markets don’t believe your model, don’t blame the markets for being stupid, re-examine your model. In the 1980s and 1990s the markets didn’t believe Friedman’s predictions that high inflation was just around the corner (because of rapid money growth.) And although Friedman was a very smart guy that I greatly respect, the markets were right, and Friedman was wrong. That taught me a lesson.
There are many other arguments for Ratex, for instance they clearly fit the way markets forecast inflation better than adaptive expectations, but I’ll leave those examples for another day.
Alex, I really don’t know what they are thinking, but at some level they have to understand that velocity has fallen sharply. It’s really just an accounting issue–if money goes up rapidly and NGDP is falling sharply, velocity has to have fallen. They know that. Perhaps they see something I don’t see (most probably a policy lag concern.)
The loss reserves are no problem. As I indicated, the Fed should decide how much reserves are prudent, how much they’d like banks to hold, and only charge interest above that level. If they do so, reserves become capped at that prudent level, and further injections of MB become effective at boosting AD. Again, there are no good arguments against an interest penalty on excess reserves, that cannot be addressed by slightly modifying the plan
Jon, I think we agree. If everyone invested in stocks solely on the basis of their local business, then the stock market would signal aggregate slowdowns a month earlier than the experts would see it in the data. Manufacturing data for early February only just came out.
Individually, every neuron in your brain is more stupid than a worm, but collectively those neurons are pretty smart.
22. March 2009 at 18:45
While I appreciate the theoretical grounds of the monetary arguments presented here, I’m going to go off the reservation to pose an alternative issue. As I’m new to the blog, I have some worries that it’s not kosher theory, but I’ll throw it out there none the less.
On a macro scale, I would argue that a major source of irrational behavior (via limited rationality) arises in perceptions of risk: “irrational exuberance” and the offsetting irrational risk aversion. I’ve seen some data sets that appear to have the potential, either as a complement to monetary policy or treated as a “factor of production” to be highly explanatory.
Put simply, imagine that the effective, short term interest rate (risk adjusted) is a composite of two factors:
cost of money
+ cost of risk
= total interest rate
Central banks (like the Fed) control the cost of money. Individual lenders control the cost of risk based on their perception of risk. Assume, for the sake of argument, that the individual lenders are overly pessimistic and consequently perceive too much risk and calculate in too high a risk premium.
cost of money
+ cost of risk
+ cost of pessimism
= excessive interest rate
Unfortunately, the central banks cannot fix the problem directly since they have no control over individual perceptions of risk. Consequently, they lower the cost of money:
cost of money
– pumping money into the system
+ cost of risk
+ cost of pessimism
= interest rate that is closer to optimal
If the risk perception is too great, even a zero interest rate cannot fully normalize the effective interest rate. If this can occur, it would cause a contraction in lending (cash at 0% displaces risky, but theoretically efficient investment) and subsequently nominal demand.
Since falling demand is deflationary:
– the nominal, risk-free interest rates actually rise
– the effective interest rate rises
– we get into a negative feedback loop, a “liquidity trap”
On theoretical grounds, I think Sumner’s argument *should* still hold. If you convinced the public that the deflation would eventually unwind with massive inflation, businesses should refuse to sell inventory at a discount… for cash that has, buried inside it, a huge negative interest rate. Unfortunately, under standard operating procedures the Fed is committed to *preventing* the subsequent and massive inflation and would actually undermine this check and balance.
Particularly if you accept a zero bound, I’m concerned that there is no standard monetary tool available to reverse this trend over the short term. I suppose a sustenance level of expenditure could eventually put a floor on demand, but this seems like a poor solution. Indeed, it only feeds back into irrational risk perceptions over the short term.
Further, even an interest rate gap of ½% would probably exceed the ability of fiscal stimulus to temporarily close the gap. Rebates would be equally ineffective since they should face the same “cash vs. risky investments” issue. In fact, fiscal stimulus lacks both the leverage of fractional reserve banking and, unless massively successful over the short term, the ability to influence the underlying risk perceptions.
===
Obviously, by changing the assumptions, I change the rules… but if we’re willing to violate an assumption of efficient risk perceptions, I think the overall boom and bust cycle, the limits of monetary policy, and the impotence of fiscal policy becomes very strait-forward.
23. March 2009 at 06:06
Clayton, Your risk theory may be right, but I disagree with the standard view of the zero lower bound. Monetary policy is fundamentally about changing the monetary base. The base should be increased fast enough to get inflation expectations (or NGDP growth exp.) equal to the target. There is almost no limit to money creation, so inflation expectations can even be increased at the zero bound. This effectively lowers real interest rates below zero (BTW, it also cuts aggregate risk, so it addresses the problem in two different ways.)
23. March 2009 at 15:23
Again, I don’t disagree that it should be theoretically possible to force real interest rates on cash into negative grounds by generating inflation. That said, how do you convince the public to act on that assumption? Particularly recently, the Fed has made an effort to develop credibility as an inflation targetter.
Perhaps more importantly, if the Fed loses credibility as far as inflation targets go, wouldn’t that generate inflation-related risk in loan portfolios? Now returns aren’t just variable as a result of default… they’re variable in terms of inflation rates… plus Fed-led inflation would be positively correlated with default (assuming that arises mostly in poor economic times).
Wouldn’t it be more practical to shift inflation expectations from 2% to 5% — albeit transparently and slowly for practical reasons? If you added the same 3% to the target fed funds rate, real interest rates and returns throughout the market (with some exceptions like tax distortion) shouldn’t change.
As a side effect, however, you would gain the ability to drive real interest rates 3% lower without deviating from inflationary targets.
24. March 2009 at 04:53
Clayton, I’m not sure what you mean by “shift inflation expectations from 2% to 5%” Inflation expectations are not at 2%. If they had been over the past year, we would never had had the recession. It occurred because inflation expectations fell far below 2%, and still are far below. So let’s get to 2% before we think about 5%. (Which I oppose in any case.)
You are wrong about the Fed having an explicit inflation target. They have never announced one, What they need is actually an explicit target path (called level targeting) which means if inflation comes in 1% below target one year, they shoot for one percent above the next. They definitely are not doing this. They are also taking highly contractionary actions like paying interest on reserves, so I can’t take seriously that they are aggressively trying to boost inflation. If they are, they are very incompetent.
Don’t worry about whether the public “believes” the Fed. If the Fed starts doing the right things, the public will believe them.
24. March 2009 at 13:35
Inflation expectations are not at 2%. If they had been over the past year, we would never had had the recession.”
I think I’m missing an assumption in your argument as I could easily see a rational expectations perspective where the exact opposite was true:
– If expectations are *lower* than inflation, prices are too low and the actual monetary impulse (larger than anticipated to cause higher inflation) should have a temporary, positive real effect (mitigating recession).
– Vice versa, if expectations *exceed* inflation, then excessive pricing is met with inadequate money supply, contributing to contraction in real terms.
In this view, 0% inflation with 2% expectations will create or intensify a recession (closer to what I meant)… while 0% inflation with 0% expectations would be RE neutral (closer to how I interpret your statement).
What am I missing? You could easily inflate by 2% an economy that has contracted (in real terms and due to non-monetary factors like risk) by 10%… which would translate to a reduction of ~8% or so in money supply (all other things being equal).
Hoping I don’t wake an even more ruthless dragon… under NGDP pathing, I think you could at least theoretically generate a 2% NGDP growth by inflating at 12% an economy that shrinks by 10% if again (1) the real contractions are not monetary in nature and (2) expectations do not exceed 12%.
Given the high level of transparency, the latter seems extremely likely. Perhaps you should point me to a paper of yours to clarify if I’m wrong about the former?
26. March 2009 at 05:17
Clayton, I answered this in another post you commented on.
26. November 2010 at 14:36
[…] Geld soll welche Folgen haben, genau? Dazu gibt es zwei sehr lehrreiche Einträge von Nick Rowe und Scott Sumner, und Krugman sieht es in seiner Forschung genauso: temporäre Erhöhungen der Geldmenge sind […]