Interest rates don’t matter as much as most people think

It’s good to see Matt Rognlie blogging again.  I’d like to take issue with the following claim:

*You may have heard from Scott Sumner, among others, that interest rates are a terrible indicator of monetary policy. This is actually true if we’re talking about current interest rates: it’s possible for monetary policy to be effectively tight because the Fed’s policy rule is contractionary, even if the current rate is very low. (Maybe it plans to raise interest rates dramatically in a few years, or whenever the economy shows signs of an expansion.) The key is to remember that monetary policy works through the entire expected path of future interest rates, not just the current rate. But interest rates are ultimately the key mechanism through which monetary policy affects the economy.

I’m told many models imply that interest rates are the key mechanism by which monetary policy affects the economy.  But I’ve never been convinced.  Start with a flexible wage/price model.  In that case a one time increase in the money supply will tend to lead to a one time increase in all nominal prices and aggregates, leaving interest rates unchanged.  Ten percent more money leads to 10% more NGDP.  A currency reform is a good example.  Obviously interest rates play no role in that process.

Now it might be argued that wages and prices are sticky, so my model isn’t at all interesting.  Not so fast.  Even Keynesians agree that wages and prices are flexible in the long run.  So it’s not obvious that interest rates play any role in explaining the long run affect of M on NGDP.  They might, but theory says we’d get to the same long run equilibrium, with or without interest rates playing a role in the transmission mechanism.  The “hot potato effect” is enough.

The more interesting question  is whether, in a sticky wage/price model, interest rates play an important role in the short run non-neutrality of money.  They might, but once again that’s not at all obvious.  For instance, suppose prices were much more flexible that wages, with commodity prices adjusting immediately to monetary shocks.  In that case you can get non-neutrality via sticky wages.  I’d guess you could also do so with monopolistic competition and sticky prices, but it’s not my area of expertise.

So theory doesn’t resolve the question, we need to look at which approach is the most useful.  And here’s where I am very underwhelmed by the Keynesian interest rate approach.  Maybe this isn’t fair, but I can’t help pointing out that I repeatedly see Keynesians make spectacularly incorrect calls by using interest rates as an indicator of monetary policy, rather than NGDP growth expectations. 

I see Keynesians argue that tight money didn’t cause the first year of the Great Depression, because rates fell sharply.  It had to be an “IS shock.”  I see Keynesians argue that tight money didn’t cause the Great Recession, because interest rates were very low, and falling, in 2008.  And I’m not just talking about a few crackpots.  These sorts of claims are made by many (perhaps most) of the most respected Keynesian macroeconomists in America.  (And don’t ask me to document this, if you disagree please produce a long list of famous Keynesians who publicly stated money was tight in 2008–and put the list in the comment section.  I think everyone honest with themselves knows I’m right.)

Now of course it’s possible that I’m wrong and the Keynesians are right.  Maybe the low interest rates really do mean that money was easy in 2008.  Then we can argue that issue.

Matt points out that the more sophisticated Keynesian models talk about the entire future path of interest rates, not just the current setting.  But I don’t believe many Keynesians understand just how knife edge those paths can be.  Take the December 2007 Fed announcement, which cut rates less than expected.  Rates obviously rose for short term maturities at 2:15 pm, but only out to a few weeks maturity.  After that rates fell, for maturities from 3 months to 30 years.  Thus a tight money policy reduced almost all interest rates at the point it was adopted.  Why?  Because it led to slower expected NGDP growth at a particularly fragile time for the economy.  Now tell me whether you think higher interest rates on maturities out to a few weeks is enough to drive important real effects in the economy, if longer term rates are going the opposite way.  Yes, using real rates makes this flaw in the Keynesian model smaller.  But most macroeconomists paid no attention to the fact that real interest rates soared in the second half of 2008.  Even worse, tight money can reduce even real interest rates in a forward-looking ratex model.  

Part 2:  While I’m commenting on Matt’s recent posts, I’d also like to discuss an excellent summary he provided of the “commitment problem.”  There are good theoretical models explaining why it might be hard for a central bank to credibly promise to inflate.  My only problem with this literature is that it’s a set of solutions in search of a problem.  No fiat money central bank has ever, ever, had the slightest difficulty in inflating.  No fiat money central bank has ever tried and failed to inflate.  And I think there is a good reason for that.  Central banks are very concerned about their reputation.  It would be humiliating to promise, say, a 5% NGDP target, level targeting, and then 4 years later when out of the recession, go “nah, nah, I was just kidding.”  Again, this isn’t to belittle the excellent models that have been developed to address the “problem,” but I’m afraid the problem doesn’t exist. 

Matt seems to have the same view as I do:

Recent events, however, suggest that elaborate commitment devices aren’t really necessary. The key constraint for the Fed isn’t sticking to its promises. The Fed cares a great deal about its credibility, and has a decent record of keeping promises whenever it has the nerve to make them. Instead, the problem is what kind of commitment to make: how can the Fed make tangible promises about future policy that don’t run the risk of creating larger problems down the road? This is the second issue in the literature, and it’s not trivial.

Paul Krugman on the other hand still doesn’t get this point:

By the way, it’s worth reading Ken Rogoff’s discussion at the end, in part for his confident assertion that raising the rate of growth of the monetary base would be enough to increase inflation. Actually, Japan tried that a few years later, without success; nor has the surge in the US monetary base since 2008 been either inflationary or indeed effective. All of which is exactly as the model predicted.

Actually, the BOJ did not want inflation, which is why they raised rates in 2000, and why they raised rates in 2006, and also reduced the monetary base by 20%.  They have the price level right where they want it.  Policy didn’t “fail.”  The problem is that they don’t want higher prices, not that they can’t get them.

And of course Bernanke repeatedly says the Fed has lots of extra tools, and is far from being out of ammunition.  They just don’t think extra stimulus is needed right now.  They paid 1% interest on reserves in 2008 to prevent that base expansion from causing inflation.  It’s clear from the market response to even trivial Fed rumors that the stock, bond, commodity and forex markets agree with Ben, Matt, and I, and disagree with Paul Krugman.  Monetary policy is highly effective at the zero bound.  It’s time to use it.

PS.  One area I do agree with Krugman is that QE unaccompanied by explicit communication is not the most effective tool, but market reactions to rumors suggest that even that’s better than nothing.

PPS.  Another reason I don’t like models with interest rate transmission mechanism is that some of them imply the Fed can’t create 3% inflation.  It’s 2% or 4%, with nothing in between.  I suspect that’s because you need at least 4% inflation to get real interest rates low enough to generate enough AD expansion to raise actual prices by 4% (or even 3%), given a fairly flat SRAS and when in a liquidity trap.  I’d appreciate if someone would tell me whether my intuition is wrong.  I can’t read math very well.


Tags:

 
 
 

31 Responses to “Interest rates don’t matter as much as most people think”

  1. Gravatar of Andy Harless Andy Harless
    14. August 2011 at 08:09

    I repeatedly see Keynesians make spectacularly incorrect calls by using interest rates as an indicator of monetary policy

    …and I repeatedly see Monetarists make spectacularly incorrect calls by using the monetary base, or M1, or M2, or MZM, or even the growth rate of NGDP not conditioned on earlier changes, as an indicator of monetary policy. I don’t think this is so much a question of “one side is right and the other side is wrong” as that some people on either side are better than others at calibrating their measures of monetary policy. Of course you can say that the quasi-monetarists do better, on average, than the Keynesians, but that seems to me to be an unfair use of categories by using a narrower category on the monetarist side. Wicksell (arguably the real intellectual ancestor of the New Keynesians) was quite clear about the point that interest rates that seem high can actually be low and vice versa. (I don’t know if he contemplated the possibility that even zero could be a very high interest rate under some circumstances, but that’s obviously a logical conclusion that most of his followers would endorse.)

    Another point: it’s not just the future path of interest rates but the contingent future path of interest rates that matters. Woodford tends to specify monetary policy in terms of a “Taylor rule” type reaction function. That’s why I talk in terms of credible threats and retributive justice. What matters is not what the central bank is doing or even what it (unconditionally) will do but under what circumstances it will do one thing vs. another thing. This is actually true for monetarists too, because the Fed has to react to changes in velocity. You may say it’s not true for NGDP targeting, but (as I’ve argued elsewhere, but I should really credit Steve Williamson for making me aware of this) NGDP targeting is just the limiting case for a particular Taylor rule. Any of a certain class of Taylor rules, if you allow the parameters to approach infinity, will attain the same unconditional quality.

  2. Gravatar of Nick Rowe Nick Rowe
    14. August 2011 at 08:27

    Scott: “I’d appreciate if someone would tell me whether my intuition is wrong.”

    I think your intuition is right. That’s what they are saying. If the (real) natural rate is (say) negative 3.1% for output to be at “potential”, then 3% inflation would mean that nominal interest rates would need to be minus 0.1% to get output at potential.

    If Keynesians recognised that the IS curve could very well be upward-sloping (because an increase in demand causes an increase ind esired investment and so mpc+mpi>1), then they would have to talk about the relationship between monetary policy and interest rates in a very different way.

  3. Gravatar of Scott Sumner Scott Sumner
    14. August 2011 at 08:32

    Andy, You said;

    “…and I repeatedly see Monetarists make spectacularly incorrect calls by using the monetary base, or M1, or M2, or MZM, or even the growth rate of NGDP not conditioned on earlier changes, as an indicator of monetary policy. I don’t think this is so much a question of “one side is right and the other side is wrong” as that some people on either side are better than others at calibrating their measures of monetary policy.”

    I agree about the monetarists, and their use of monetary aggregates, and have said so in other posts. (Although I do think the Keynesians are even worse in this area, i.e. during 1930.) But let’s put that aside, and stipulate both are often bad.

    I deny that the quasi-monetarist use of NGDP expectations has a similar problem. I think it gives much more accurate readings of whether money is too tight or too loose.

    You can always adjust interest rate models enough to make them “work” in some sense. I could start talking about the actual DJIA relative to the Wicksellian equilibrium DJIA. Or the actual price of zinc, relative to the Wicksellian equilibrium price of zinc (i.e the zinc price consistent with macro equilibrium.) But I’m a pragmatist, and I just don’t see where that is useful.

    I do think NGDP growth expectations are very useful, and I wish we’d get on with creating an NGDP futures contract and pegging it.

    I agree that a forward-looking Taylor rule is similar to an NGDP futures targeting approach, so why didn’t Keynesians say money was tight in late 2008?

    My other problem with the Taylor Rule approach is that it sometime seems to suggest that interest rates need to be strongly negative in real terms, meaning inflation needs to be high. But I think that’s wrong, because with optimal monetary policy the Wicksellian equilibrium real rate is much higher than the equilibrium real rate appears to be with monetary policy that puts us in a depression.

  4. Gravatar of Scott Sumner Scott Sumner
    14. August 2011 at 08:40

    Nick, Thanks for explaining thaT point in terms of IS-LM–I should have cited your post. Is another way of making the point that monetary stimulus works through many channels (as discussed in Mishkin’s text) not just interest rates? I.e. it raises the relative prices of stocks, commercial real estate, commodities, etc. Which leads to more output in those industries? I do understand that there are implicit “interest rates” in those other assets. So perhaps I’m not really denying the role of interest rates here, but rather “the” interest rate on T-securities. I think it’s wrong to assume that “the” rate tells us how much inflation we need, as everyone will want to hoard T-securities if the economy is in the toilet, and equilibrium yields will be low. If the Fed is expected to quickly get things back on target, who wants to hold Treasuries?

    I also wonder what Keynesians who deny the possibility of 3% inflation expectations would make of a Fed policy that targeted CPI futures at a three percent premium. Then what?

  5. Gravatar of james in london james in london
    14. August 2011 at 09:32

    Scott you said: “I suspect that’s because you need at least 4% inflation to get real interest rates low enough to generate enough AD expansion to raise actual prices by 4% (or even 3%), given a fairly flat SRAS and when in a liquidity trap.”

    Are we really in a liquidity trap, then? I thought they didn’t exist for you?

    [Am still incredulous about your naked politicking in the endorsement of the ex-Moody’s man’s view of the S&P downgrade.]

  6. Gravatar of Bryan Willman Bryan Willman
    14. August 2011 at 09:39

    I wonder if there’s a terminology problem here – for very many people, “easy money” means “low interest rates” and vice versa. We don’t think about expanding or not expanding the monetary base, or expansionary or contractionary policy – “easy money == low interest rates” and “tight money == high interest rates”.

    Until the Great Recession, where the odd circumstance of “loans are hard to get” and “interest rates are low” appeared together for the first time in most people’s memories.

    So some number of people who think “well money was very easy because interest rates were very low” are in effect quoting what they think of an identity relationship.

  7. Gravatar of Scott Sumner Scott Sumner
    14. August 2011 at 10:22

    James, Explain to me how a country that prints its own currency can default on its debt obligations. I don’t get it.

    Regarding the liquidity trap, I was explaining the views of others, not my own views. We are not “trapped.”

    Byran, You said;

    I wonder if there’s a terminology problem here – for very many people, “easy money” means “low interest rates” and vice versa.

    Joan Robinson said the German hyperinflation couldn’t have been caused by easy money, because interest rates weren’t low. Do you agree? Was money tight during the German hyperinflation, when interest rates were high?

  8. Gravatar of Scott Sumner Scott Sumner
    14. August 2011 at 10:23

    Bryan, You are also confusing easy money and easy credit–two completely unrelated concepts.

  9. Gravatar of Jim Glass Jim Glass
    14. August 2011 at 11:55

    Explain to me how a country that prints its own currency can default on its debt obligations. I don’t get it.

    One thing that has been greatly under-reported in the entire S&P story, IMHO, is that S&P left its *short term* rating for the USA unchanged at the very top level, and lowered only the long-term rating.

    Now, being that CBO’s 75-year “long term outlook” projection for the budget ends in only 26 years in its most likely scenario, because debt held by the public then exceeds 200% of GDP and is rocketing straight up, which is deemed unsustainable (to put it kindly), it’s something of a mystery to me how anyone think this *long term* outlook deserves the *highest* credit rating in the world, and the best credit rating possible.

    The only way that could be is if one put tremendous faith in the USA’s political system to produce AAA-quality mangement of fiscal affairs, resolving that problem long before it ever came near to occurring. But what did we just go through, politically?

    Imagine a major corporation with huge fixed long-term liabilities heading it to ruin and near-certain bankruptcy in 20 years without major restructuring of its finances, which *is* possible but only with strong, able leadership. We know that its Board of Directors just held a meeting to discuss this problem, which degenerated into a Three Stooges pie fight that we all saw on Youtube. Would that corporation’s *long-term* credit rating — its 30-year bonds — deserve to be AAA? Just because it has ample cash on hand to pay its short term liabilities, ignoring the huge long-term ones? (Would it deserve AA+?)

    Remember S&P’s projection of the future sovereign ratings made back in 2006, before everything got so much worse. It’s not like there was no warning.

    As to how can a country that prints its own currency can default on its debt (the MMT line!), the Russians did it in 1998. If the options are (1) technical default getting your creditors angry, or (2) running the printing presses to inflate to create an effective default, getting the mobs in the street angry and your creditor angry too, the politicians will choose the option they see as less painful for them. Why can’t that be (1)?

    Of course with the USA keeping its tip-top short-term rating and a very high AA+ long-term rating, S&P was hardly predicting any imminent danger of default for the USA.

    But when one considers how all the long-term debt drivers on course to push annual deficts up over 20% of GDP in the forseeable future are inflation indexed or payable in real terms, in the end the USA is going to be defaulting on *some* promises it made, either these or those, sure thing. Frankly, I don’t see how the USA’s long-term credit rating deserves AA+.

  10. Gravatar of malavel malavel
    14. August 2011 at 13:35

    The Fed decreases base money by 10% each year and promises to hold interest rates at near zero for at least 10 years. Would the Keynesians still see this as easy money?

  11. Gravatar of Stew Rubenstein Stew Rubenstein
    14. August 2011 at 13:38

    I’m still trying to get my head around to what degree interest rates are a lever, a transmission mechanism, a meaningful measurement, or just an uninterpretible blinking light on the control panel.

    I get how in the flexible wage/price model, Interest rates play no role. A currency reform works because the old and new currencies are, for some period, interconvertible. I don’t see that as a useful model.

    Once you have some degree of stickiness, then it becomes interesting, but you assert that the stickiness only affects the rate of return to the one true equilibrium. While that may be true, the rate of return to equilibrium might be very slow, or an activation barrier might trap the system in a non-ideal local minimum. Can you tell that I’m a chemist by training?

    The mechanism I’m thinking of is as follows: Suppose that in a sticky wage/price model with varying degrees of stickiness for different wages and prices, we inject 10% more money. I wake up in the morning with 10% “extra” money, although I know that I’ll want to hold it as money eventually because wages and prices will adjust. In the meantime, what do I do? I buy a bond, of course, the duration and amount of which depends on what I think about the expected rates of change of other wages and prices. Since everyone else is thinking the same thing and the market for bonds is very liquid, the price shoots up very fast, assuming the demand isn’t met instantly by a corresponding increase in supply. If not, then this is what we call monetary stimulus “” we lower rates, and presumably, when equilibrium is reached with 10% more NGDP, some of it is real growth driven by actors who can use the temporarily reduced cost of capital to make productive investments.

    On the other hand, if supply of bonds is increased to prevent the price of money from dropping, this scenario reduces to deficit spending funded by the printing press. Fiscal stimulus. The increased demand for goods and services driven by this spending drives up prices and wages, and ultimately the increased supply of bonds must be managed towards the same equilibrium as before.

    The point of all this is that the varying rates of stickiness among different prices and wages are not simply minor factors which can be approximated away. They are crucial to determining how far and in what direction the economy is moved from equilibrium, how long it takes to get back to equilibrium, and who benefits from the arbitrage made possible by the shift.

  12. Gravatar of Martin Martin
    14. August 2011 at 15:05

    “Joan Robinson said the German hyperinflation couldn’t have been caused by easy money, because interest rates weren’t low. Do you agree? Was money tight during the German hyperinflation, when interest rates were high?”

    Scott, have you considered just ending your posts in the future with a variation on the way Cato the Elder used to end his speeches (Ceterum autem censeo, Carthaginem esse delendam), by inserting at the end of each posts: “Furthermore, interest rates are an unreliable indicator of monetary policy.”. ;-)?

  13. Gravatar of Scott Sumner Scott Sumner
    14. August 2011 at 18:28

    Jim Glass, I boggles my mind that MBSs in 2007 could have been rated AAA on the assumption that house prices can’t fall, and yet US T-bonds are double AA. Something’s not right there. Sure, there’s a one in a million chance we default, but the odds of those MBSs defaulting was obviously much higher, even ex ante. So what is the criteria that they use? How can one seriously argue the US is more likely to default than some corporation. BP was almost destroyed by a single oil spill. The US government is nowhere near as fragile as a company. Microsoft’s products might be obsolete in 5 years. Who knows?

    Malavel, Yes, some of them would call it easy money. But obviously the smarter ones wouldn’t.

    Stew, I didn’t suggest that sticky wages and prices are unimportant, indeed I believe they are 100% responsible for the short run non-neutrality of money.

    Martin, Yes, I have. Sometimes I feel like I am battling the entire world. Almost everything I read in the press starts with the assumption that low rates mean easy money. Or that it “goes without saying money has been easy since 2008.” It’s an uphill battle.

  14. Gravatar of cato cato
    14. August 2011 at 19:13

    lets stop talking in terms of phlogiston.

    theres no such thing as the future path of interest rates, and even the expectation of future path is too strong. we have a vague guess about present conditions holding for some time, thats all.

    this is exactly scott’s point about the fragility of the estimates, they change based on current data so they may as well be called current interest rates, for now. does anyone really have any confidence in macroeconomic conditions 30 years from now…(or even 3 months)

  15. Gravatar of Andy Harless Andy Harless
    14. August 2011 at 19:26

    why didn’t Keynesians say money was tight in late 2008?

    I think this is mainly a problem with growth rate targeting vs. level targeting. From a forward-looking point of view, any reasonable Keynesian should have said that money was tight. But from a backward-looking point of view, not so much. The problem is that if you have a growth rate targeting regime, forward-looking policy is dangerous because there is no self-correction feature when there are large cumulative forecast errors (as in the 1960’s-1970’s and then to a lesser extent in the opposite direction in the subsequent decades). Since by most accounts the Fed did (and does) have a growth rate (specifically inflation rate) targeting regime in place in 2008, most Keynesians didn’t feel it was appropriate to describe policy in the purely forward-looking terms that would have implied it was tight.

    the Taylor Rule approach…sometime seems to suggest that interest rates need to be strongly negative in real terms, meaning inflation needs to be high. But I think that’s wrong, because with optimal monetary policy the Wicksellian equilibrium real rate is much higher than the equilibrium real rate appears to be with monetary policy that puts us in a depression.

    I’m not so sure. A Taylor rule takes the expected inflation rate as given, and if the real rate has to be more negative than the inflation rate is positive, you have to do something else (such as QE) as a substitute for having a negative nominal interest rate. But the same is true of NGDP targeting. It seems to me that a situation could easily arise in which zero nominal interest rates fail to bring the NGDP forecast up to target. Then you have to do something else as substitute for having a negative nominal interest rate. (I suppose you could try negative IOR in either case, but, because so much can leak out in the form of currency, the appropriate interest rate wouldn’t necessarily be the one implied by the Taylor rule, and there’s no guarantee that any negative IOR rate would be low enough to hit the NGDP target.)

    The critical feature of NGDP targeting (as you conceive of it), as compared to a traditional forward-looking Taylor rule, is it’s aggressiveness: most Taylor rules would let the forecast deviate from the target moderately rather than taking extreme policy actions, whereas your NGDP targeting, in its purest form, will not accept the slightest deviation from target. This means, on the one hand, that it reliably provides the stitch in time that saves nine — so you’re unlikely to wander into a depression in which there is a prolonged negative Wicksellian natural rate; but on the other hand, it means that you do have to react very aggressively in real time, which means you may need a negative interest rate (or some substitute) to implement the optimal policy that avoids that depression.

    …some of them imply the Fed can’t create 3% inflation. It’s 2% or 4%, with nothing in between. I suspect that’s because you need at least 4% inflation to get real interest rates low enough to generate enough AD expansion to raise actual prices by 4% (or even 3%), given a fairly flat SRAS and when in a liquidity trap.

    I think that’s a correct interpretation. The problem I see with this type of model is that it doesn’t account for heterogeneity in a situation where it becomes of practical importance. In real life, there is a continuum of agents with a range of expectations. For each type of agent, if (s)he were the representative agent, there might be a range of infeasible inflation rates, but since everyone’s range is different, any inflation rate the central bank might choose is almost certain to be within the feasible range for enough agents to make it actually feasible. I wonder if anyone has modeled this formally.

  16. Gravatar of FT Alphaville » Further reading FT Alphaville » Further reading
    14. August 2011 at 23:09

    […] The relative irrelevance of interest […]

  17. Gravatar of Jim Glass Jim Glass
    15. August 2011 at 00:08

    Jim Glass, I boggles my mind that MBSs in 2007 could have been rated AAA on the assumption that house prices can’t fall, and yet US T-bonds are double AA. Something’s not right there.
    Something’s not right there. Sure, there’s a one in a million chance we default…

    The mistakes of the past are irrelevant to me but for learning and rhetorical purposes. The mistakes of the present and future are what concern me.

    The USA continues to have the highest short-term sovereign rating in the world.

    In the long term, there is a good, big, fat excluded middle between AA+ and “default”.

    But when the CBO ends its budget projections in only 25 years because the 200+% of GDP and rising projected debt held by the public is unsustainable at that point, does the 30-year T-bond *really* deserve the highest rating in the world?

    Will the 30-year bond really *always* deserve the AAA rating, just because? If CBO not being able to project a future beyond 25 years because of budget unsustainability *doesn’t* justify knocking the 30-year bond all the way down to AA+, what would?

    Maybe the learning lesson of the past with the MBSes was that the rating agencies give ratings that are too high when everybody wants them too — both the all the politicians and all the business interests. (If one of the agencies had blown the whistle on the MBSes back when everything was looking great to almost everybody it would have been crucified by all parties — and blamed for all the damage to the housing market that followed). And that in that context it is much too easy to assume that a bad thing can’t ever happen just because it never has yet. It would really be bad to repeat these mistakes going forward, especially now after we’ve received a lesson on them.

    Really, what *would* justify a downgrade of the 30-year bond to AA+, the second-highest rating in the world, if a projection of debt exceeding 200% of GDP in 25 years doesn’t?

  18. Gravatar of PrometheeFeu PrometheeFeu
    15. August 2011 at 06:25

    “They paid 1% interest on reserves in 2008 to prevent that base expansion from causing inflation.”

    I just don’t understand why they did that. If that money is stuck in a bank account at the Federal Reserve, they might as well never have printed it. Ultimately, for monetary expansion to mean something, that new money has to leave the banking sector. I would argue that policy of paying interest on reserve effectively means they did not expand the base. (or expanded by a lot less than stated)

    Not to mention this screws up the metrics because we now have a big chunk of money which won’t leave the reserve accounts but it’s really hard to know how much this policy is affecting.

  19. Gravatar of W. Peden W. Peden
    15. August 2011 at 06:40

    PrometheeFeu,

    “I just don’t understand why they did that. If that money is stuck in a bank account at the Federal Reserve, they might as well never have printed it.”

    Well, it did add to the assets of the financial sector. IOR is paid by the US taxpayer in supplication to the banks. In return, the banks have higher reserves in relation to their lending and guaranteed returns on riskless currency. That’s a pretty good deal, at least from the banks’ point of view.

  20. Gravatar of Stew Rubenstein Stew Rubenstein
    15. August 2011 at 06:55

    PrometheeFeu, expanding the base wasn’t the point. What the Fed did with all that money was buy longer term bonds, to drive long term rates lower. Rightly or wrongly, they don’t want the inflation that would have been caused by pushing out all that money. IOR effectively sets a floor on the fed funds rate, and the huge increase in reserves soaked up 85-90% of the Fed’s balance sheet expansion.

  21. Gravatar of Scott Sumner Scott Sumner
    15. August 2011 at 10:00

    cato, I agree that interest rates are very unreliable.

    Andy. But even in early 2009 I didn’t hear Keynesians complaining money was tight. Indeed one reason I started my blog (in early 2009) was that I wasn’t hearing Keynesians complaining that money was tight. And it was obvious we needed more NGDP in 2009.

    You said;

    “It seems to me that a situation could easily arise in which zero nominal interest rates fail to bring the NGDP forecast up to target.”

    I complete agree, indeed that’s the current situation. But I don’t quite follow how this is a point of evidence against my argument. I’m not advocating interest rate targeting, I advocate NGDP forecast targeting.

    I agree with your last point about the heterogeniety of agents, but there is also the hetergeniety of assets. Even with T-bill yields equal to zero, lots of other asset prices are affected by monetary stimulus, and these affect AD.

    Jim Glass, You said;

    “The mistakes of the past are irrelevant to me but for learning and rhetorical purposes. The mistakes of the present and future are what concern me.”

    I wasn’t talking about the mistakes of the past, I was talking about the CRITERIA of the past, and of the present (corporate AAA bonds.) No one can say these corporate bonds won’t default, they could very easily default. So the criteria S&P uses cannot be risk free.

    I would like someone to explain to me why S&P even rates these bonds. Yes, I know they are required to by law, but why do we have these laws? What’s the purpose of rating government debt? I don’t get it. Can someone tell me why I should care about S&P’s opinion? What do they know that the market doesn’t know?

    PrometheeFeu, What people don’t understand is that all that money was not printed to stimulate Main Street, it was issued to bail out Wall Street banks.

  22. Gravatar of Jim Glass Jim Glass
    15. August 2011 at 14:32

    I would like someone to explain to me why S&P even rates these bonds. Yes, I know they are required to by law, but why do we have these laws? What’s the purpose of rating government debt? I don’t get it.

    We have these laws because of the goverment(s), of course. The federal govt set up and mandated use of the rating system … and created the rules under which those *being rated* (rather than investors who use the ratings) pay for ratings … and set up the cartel of SEC-determined-and-supervised rating firms (the same SEC that examined Bernie Madoff five times without catching him.) All of which agruably creates incentives all pointing in the wrong direction.

    And governments mandate that these ratings be used in a multitude of ways.

    Can someone tell me why I should care about S&P’s opinion?

    If you are considering buying a US bond you probably shouldn’t — except for drawing comfort from the fact that the risk of default on an AA+ bond is miniscule.

    For big, liquid securities the market provides the best ratings, of course, and the rating agencies typically follow the market. Though the markets were plenty wrong about the mortgage securities, trading them as risk-free, for a good long time (with the agencies again following, as typical).

    BUT governments and regulators around the world mandate use of ratings in a huge way. Basel II puts them in bank capital rules. Govts and private contracts require AAA securities for all kinds of purposes. Etc. etc.

    What happens when all these parties whip up huge demand for these securities in this kind of system? This. Pretty awful, and asking for trouble, that. It’s sort of a “let’s pretend” game with very high stakes. That puts what hit the world financial system in some perspective.

    Once the game is in play, what to do? Eric Falkenstein, who worked at Moody’s before the boom-bust, asked What if Moody’s had corrected their ratings in 2006?. Hey, that’s not a pretty picture either. No reward for the responsible there.

    So the whole system may be a big mess, but we are where we are.

    BTW, here are default rates by rating.

    And given that the system exists, I still don’t comprehend all the outrage about assigning such miniscule risk to a 30-year bond when the issuer doesn’t know how it is going to survive even 25 years — and those responsible for figuring out how are the same people who just put on the debt ceiling clown show.

    I mean as it is, in just about 15 years the USA is going to need a 50% across-the-board income tax increase (or equivalent revenue-raising VAT or some such), just to keep even with entitlements, lest the deficit be that much worse than today’s. Or entitlements have to be cut correspondingly. Just how *easy* is that going to be to manage, off of recent experience?

    Public belief that the USA has a right to the best credit rating in the world as an entitlement, no matter how fiscally irresponsible the govt behaves, will only drive the politicians to keep behaving that way — and make the coming fiscal crunch worse and hit sooner. The public would do well to take a couple more downgrade slaps across the face, to put it in closer contact with reality, IMHO.

  23. Gravatar of Matt Rognlie Matt Rognlie
    16. August 2011 at 06:24

    Sorry that I got here so late—I’m in Korea, and I’ll be getting married a few days from now and then honeymooning, so I’m afraid my internet use will be spotty for quite some time. (Thanks for the shoutout, btw… though for the above reason, I don’t think the blogging will continue for a while.)

    Start with a flexible wage/price model. In that case a one time increase in the money supply will tend to lead to a one time increase in all nominal prices and aggregates, leaving interest rates unchanged. Ten percent more money leads to 10% more NGDP. A currency reform is a good example. Obviously interest rates play no role in that process.

    I’d contend that interest rates are still responsible at some level, although the knife-edge nature of the flexible-price case makes it easy (too easy!) to ignore them.

    My basic complaint about monetarist intuition (“10% more money leads to 10% more NGDP”) is that it leaves the mechanisms for achieving equilibrium underspecified. Your explanation is the “hot potato effect”, which is legitimate, but I think it’s just a reformulation of a story that could (and should) be told with interest rates.

    First, what is the basis for such a tight correspondence between money and NGDP? There are two possible ways of looking at the issue: (1) that money is essentially an intermediate input to NGDP, and that the ratio of input to NGDP is nearly fixed by some kind of technological constraint, or (2) although money is not really such an important intermediate input for most of GDP, money demand is roughly proportional to NGDP. I think your interpretation is more along the lines of (2), and I agree. (The dividing line between (1) and (2), of course, is not so clear, and neither is the relevance of this distinction. But I want to be clear from the outset, and I feel that trying to think about money as a factor of production is mostly unimportant and distracting.)

    So now suppose that the ratio NGDP/M is below its equilibrium level—money demand (produced by NGDP) is not high enough to match supply. How do we get to equilibrium? You talk about the “hot potato effect”: everyone has higher real money balances than they want at current prices, they spend the money in an effort to get rid of it (which they can’t do in nominal terms in the aggregate), and prices (and maybe output too) eventually rise enough that NGDP/M is at its equilibrium level.

    But why are they so intent on getting rid of the money? Even if we forget about the money/bond margin for a minute, money still has a “convenience yield”: there is a certain implicit return agents are getting from holding money, beyond the fact that it will retain its nominal value into the next period. They will only attempt to dispose of the money if this convenience yield is too low relative to the nominal interest rate that is consistent with their plans for consumption, investment, etc. over time. (The “natural rate”.) In other words, money becomes a “hot potato” if the convenience yield on money can’t match the return on physical investment, or the desired pattern of intertemporal substitution.

    If we start in equilibrium, and the only change we make is to increase current and future money by 10%, then your story about NGDP rising by 10% is certainly valid: before the money shock, the convenience yield was exactly right, and now it will be too low, which pushes up demand. Assuming the money shock does not change the natural rate, the convenience yield will continue to be too low until the previous ratio of NGDP to M has reasserted itself. The monetarist story holds, but the convenience yield is a key part of the story if you’re thinking about money in a dynamic context.

    And, of course, the kicker is that the “convenience yield” will be equal to the short-term interest rate. So interest rates really are responsible after all. You can tell the story without them, but that’s because the flexible-price case sweeps adjustment problems under the rug: after the money shock, the ratio of NGDP to M will return to its equilibrium level immediately.

    In a sticky-price model, on the other hand, NGDP doesn’t adjust right away, and we have to think about what happens in the meantime. And then, I think, we see the standard New Keynesian story: while the adjustment is still underway, interest rates are too low to match the natural rate, and we get inflation. The effect of the money shock on NGDP can be summarized entirely through the (implicitly contingent) path of interest rates—after all, that’s also the path of the convenience yield on money, which is central to the hot potato mechanism you describe.

  24. Gravatar of Matt Rognlie Matt Rognlie
    16. August 2011 at 06:44

    The more interesting question is whether, in a sticky wage/price model, interest rates play an important role in the short run non-neutrality of money. They might, but once again that’s not at all obvious. For instance, suppose prices were much more flexible that wages, with commodity prices adjusting immediately to monetary shocks. In that case you can get non-neutrality via sticky wages. I’d guess you could also do so with monopolistic competition and sticky prices, but it’s not my area of expertise.

    At some point I really need to write a post that points out the following (obvious from the math, but somehow almost no one has noticed): in a New Keynesian model, if a central bank makes an error by setting the wrong interest rate, then the damage is increasing in the level of price flexibility. The reason is simple: if prices are more flexible, then inflation changes by a greater amount in response to the central bank error, and this change compounds the original error. (e.g. more inflation from a boom pushes real interest rates even lower, making the boom even larger) It’s true that if prices are perfectly flexible, you don’t really get this, but in this case the New Keynesian model blows up and produces infinities anyway, so the proper interpretation is unclear. And even if money is neutral in the “perfectly flexible” case, it is infinitely non-neutral in the perfectly flexible limit of a sticky price model, which I’d argue is a vastly better approximation to reality than the perfectly flexible model itself. (After all, prices do not change every nanosecond; we may argue about the degree of price stickiness, but no one disputes that there is some stickiness.)

    Contrary to (almost) universally accepted wisdom, you don’t need substantial price stickiness for monetary policy to matter in a New Keynesian model. In fact, that situation is almost precisely the opposite: the stickier the prices, the less monetary policy matters (because there is less of an add-on effect from changes in inflation).

    If the central bank makes other kinds of errors, then things are different: for instance, if it has a quantity target, then price flexibility generally brings money closer to neutrality, though this is still a little ambiguous. If the central bank sets the intercept of a Taylor rule rather than the interest rate itself, then greater price flexibility also makes money more neutral (because when you obey the Taylor principle, the effect on inflation induces more than one-for-one effect on interest rates that is stablizing). But I don’t think these cases are very reasonable approximations to reality: central banks definitely don’t use quantity targets, and they also don’t set the intercept of a Taylor rule. (This is particularly true at the zero lower bound, where it is impossible to obey the Taylor principle.)

    I think that most intuition about the effect of “monetary shocks” comes from old thinking about the effects of a quantity shock. The problem is that quantity shocks are very, very, very different from interest rate shocks, or interest rate “mistakes”. Even leading New Keynesian economics haven’t really internalized this, though they are coming close. (For instance, if you look closely at equation (4.7) on page 21 of Woodford’s 2011 AEJ macro piece on the government expenditure multiplier, you’ll see that the effect of a liquidity trap clearly becomes worse as the level of price flexibility increases.)

    Admittedly, significant price (or wage) stickiness is necessary for some other nonneutralities—e.g. the problems we see in currency unions like the Eurozone. But you don’t need significant price stickiness to explain why bad monetary policy causes a recession.

    (You do need it, or something like it, to explain why we don’t get 10000% deflation during a recession.)

  25. Gravatar of Scott Sumner Scott Sumner
    16. August 2011 at 07:01

    Jim Glass, I think we agree that S&P isn’t woth paying attention to, at least for sovereign debt. I’m not really outraged by S&P, as the bond markets showed they don’t care.

    I agree the industry should be completely deregulated.

    I’m very worried about higher taxes, and somewhat less worried about higher inflation. I’m not at all worried about US default.

    Matt, I’m not really sure I understand your comment. Here’s what I am claiming:

    1. In the flexible price case the price level adjusts immediately. There are no changes in any nominal or real interest rates for maturities from epsilon to infinity. We pretty much know this from observing the effects of currency reforms. To me, that means interest rates play no role in the transmission mechanism. It’s not so much the hot potato effect, as expectations of the hot potato effect, which cause the price level rise to occur immediately.

    2. At the instant the new money is introduced the marginal convenience yield falls. But in an instant it is restored to normal as prices adjust. No time has gone by.

    In my book interest rates play no role in that adjustment process. Interest rates exist over a span of time, not a point in time. I would add that my model also applies to a world that doesn’t have any interest rates. A world that just has consumption goods and cash.

    It seems to me that your explanation stretches the definition of the interest rate transmission mechanism to such an extreme that it is almost unrecognizable. Someone could do the same for the Keynesian model, insisting that no matter how hard Keynesians like Woodford try to deny that money matters–it’s always there lurking in the background. Come to think of it, McCallum has made exactly the same criticism of Woodford.

    Most importantly I’m actually trying to make a pragmatic argument; I’m not trying to show that monetarist argument can’t somehow be reconciled with a specific interest rate path over time. My bottom line argument is that people who use interest rates as an indicator of monetary policy tend to misinterpret the nature of our macro policy, and tend to give bad policy recommendations.

    Enjoy your honeymoon in Korea! I envy you. I had a honeymoon in China.

  26. Gravatar of Matt Rognlie Matt Rognlie
    16. August 2011 at 07:05

    I see Keynesians argue that tight money didn’t cause the first year of the Great Depression, because rates fell sharply. It had to be an “IS shock.” I see Keynesians argue that tight money didn’t cause the Great Recession, because interest rates were very low, and falling, in 2008. And I’m not just talking about a few crackpots. These sorts of claims are made by many (perhaps most) of the most respected Keynesian macroeconomists in America. (And don’t ask me to document this, if you disagree please produce a long list of famous Keynesians who publicly stated money was tight in 2008-and put the list in the comment section. I think everyone honest with themselves knows I’m right.)

    Now of course it’s possible that I’m wrong and the Keynesians are right. Maybe the low interest rates really do mean that money was easy in 2008. Then we can argue that issue.

    I find that debates about the “cause” of a recession are extremely frustrating, because the definition of causation is not clear to all parties. What is causation, anyway? If party X could have prevented event Y, did party X “cause” event Y? Not necessarily: I could have prevented countless accidents across the world yesterday if I had the foresight to know when they would happen, but no one would say this makes me causatively responsible for them.

    Alternatively, suppose that I’m supervising a factory, and my job is to prevent a certain valve from clogging—which, if it’s not stopped, will ultimately destroy the factory. Some idiot leaves a wrench in a pipe somewhere, and the valve ends up clogging. Because I’m busy blogging rather than doing my job, I don’t notice, and the factory eventually blows up. Did I “cause” the disaster? You might say so. Arguably, the idiot with the wrench was the first mover, but my inaction (in a circumstance where I was supposed to be the first line of defense) was a key part of the causal chain.

    I think this is the best analogy for monetary policy in 2008. The Fed didn’t contribute the initial “shock”; that was from the financial system. But by failing to act appropriately, ease policy by a sufficient amount and aggressively set expectations, the Fed’s performance was similar to my failure to supervise the factory, and you might reasonably say that the Fed “caused” the recession.

    You can have more fun with analogies. Maybe I’m driving, and an easily avoidable deer crosses the road. I’m not paying attention, I hit the deer, my car is severely damaged and the deer dies. Was it the deer’s fault or mine? Arguably I’m the real culprit, but it’s easy to see how you could get into endless (and pointless) debates about what the “cause” really was.

    I think people sometimes react negatively to your claims about how policy was “tight” in 2008 because they’re taking a different semantic viewpoint than you, even if there are few substantive differences. Since the Fed didn’t actively “cause” the crisis by tightening its policy according to the widely observed policy variable (interest rates), and indeed brought down interest rates, someone else might say that the “Fed was easy, but not easy enough”, while you say that the Fed was tight, because it didn’t provide the accommodation that the market needed. There are certainly some deep underlying issues here (e.g. what are the most informative intermediate targets?), but much of the disagreement is really semantic.

    For what it’s worth, I think that the proper measure of “tightness” is the expected difference (including future periods) between the Fed’s rates and the natural rate. I think this is philosophically and semantically similar to your position (rather than blindly define “tightness” by how high or low some number is, I define it by where policy stands relative to where it needs to be), even though you’re not very fond of interest rates.

    By this measure, the Fed’s policy was indeed quite tight in fall of 2008.

    I don’t think your real object of criticism should be “interest rates” as a policy indicator per se; the real culprit is interest rates without context. We both see a lot of people lazily saying “oh, rates fell to near 0%, so of course money was easy!”. That’s not right, but it’s possible to think about monetary policy in a more enlightened way while putting interest rates at the center of the analysis.

  27. Gravatar of Scott Sumner Scott Sumner
    16. August 2011 at 07:10

    Matt, That’s another reason I don’t like the NK model. Summers and DeLong made a similar argument in 1986, and used it to support FDRs NIRA program (aimed at making prices less flexible) But the NIRA was a disaster, sharply slowing the recovery.

    In my view price flexibility is a good thing, as if NGDP is going to fall 10%, I’d rather it be mostly lower prices. the NK response is that price flexibility makes a given demand shock even worse, causing a faster fall in NGDP. But I think their mistake is to assume inflation is the relevant nominal aggregate, whereas I think expected NGDP growth is. Of course my argument also applies to all sorts of real world monetary regimes with nominal anchors (inflation targeting, NGDP targeting, gold standard, etc.) In all those cases increased price flexibility is stabilizing. I think it’s also stabilizing with an interest rate peg, but that’s harder to prove. (It relates to my criticism of inflation as “the” nominal indicator, rather than NGDP growth.)

  28. Gravatar of Scott Sumner Scott Sumner
    16. August 2011 at 07:11

    Matt, I do agree that you don’t need mush inflexiblity to get the “Keynesian” result. That’s where I disagree with people like Tyler Cowen.

  29. Gravatar of Matt Rognlie Matt Rognlie
    16. August 2011 at 07:34

    I’ve seen DeLong and Summers, but I think the result is even clearer in the current New Keynesian model. For what it’s worth, I also think NIRA was pretty clearly a bad idea: even if it had some positive effects from creating inflation expectations, there were direct and negative real effects from cartelization, and the effect on prices could have been replicated in a much better way using monetary policy.

    BTW, I don’t think this really works as an argument against the New Keynesian model. In the basic NK model, the negative real effects of monetary contraction are achieved completely through markups… so the basic NK model says pretty clearly that NIRA is insane policy, since it’s using an increase in markups to offset a monetary contraction that’s harmful due the high markups it causes!

    (Now, I think that this particular aspect of the basic New Keynesian model deserves some criticism—although markups are hard to measure (what’s the marginal unit labor cost, anway?), it doesn’t seem intuitively sound that the increase in (implicit) markups is the sole channel through which monetary contraction affects the economy. This can be alleviated by introducing either nominal or real wage rigidities into the model, though.)

    In all those cases increased price flexibility is stabilizing. I think it’s also stabilizing with an interest rate peg, but that’s harder to prove.

    I’m pretty certain that it’s not stabilizing with an interest rate peg. The “pure interest rate” peg case can’t really be analyzed because in theory it leads to nominal indeterminacy. But if you have an interest rate peg that will last N periods, followed by some rule that achieves some kind of nominal anchor, you have nominal determinacy and can analyze what happens. It’s pretty straightforward in this case that price flexibility is destabilizing: for the duration of the interest rate peg, price flexibility accentuates any “mistake” by moving the real interest rate even further.

    This is a particularly important case because of the zero lower bound problem—being stuck at the zero lower bound is in effect an interest rate peg.

  30. Gravatar of Scott Sumner Scott Sumner
    16. August 2011 at 08:01

    Matt, You said;

    “I think people sometimes react negatively to your claims about how policy was “tight” in 2008 because they’re taking a different semantic viewpoint than you, even if there are few substantive differences.”

    I don’t think so. I got into blogging precisely because almost no one was blaming bad Fed policy for falling NGDP. That seems like a very substative policy point to me, unless I am missing something. (Of course now lots of Keynesians are.)

    Regarding the philosophical question of “causation” I think I am well within the mainstream. If Russians hoard currency equal to 10% of the base in 1992, and if the Fed doesn’t accommodate that increase in base demand, and if real interest rates soar much higher, I think most economists blame the Fed for tight money. (Of course the Fed did accommodate base demand in that case.) In late 2008 base demand soared for different reasons, the Fed did not fully accommodate, and the real interest rate soared. To me, that seems like a very common sense definition of causation.

    You said;

    “I don’t think your real object of criticism should be “interest rates” as a policy indicator per se; the real culprit is interest rates without context. We both see a lot of people lazily saying “oh, rates fell to near 0%, so of course money was easy!”. That’s not right, but it’s possible to think about monetary policy in a more enlightened way while putting interest rates at the center of the analysis.”

    I agree that it’s possible, but I don’t see any value added. The most useful indicator of easy and tight money is the NGDP growth rate (ideally a futures contract price.) Once you have that, interest rates add nothing useful. Without it, you try to infer the market forecast of NGDP growth. You might use 5 year bond yields in doing so, but of course they are positively correlated with NGDP growth, meaning lower rates usually mean tighter money. (This is one point I was trying to make in my next post, BTW, the one on the Wicksellian zinc price.)

    You said;

    “It’s pretty straightforward in this case that price flexibility is destabilizing: for the duration of the interest rate peg, price flexibility accentuates any “mistake” by moving the real interest rate even further.”

    Yes, but the “real interest rate” is exactly the aspect of the NK model that I am criticizing. I think what matters is not i minus inflation, but i minus expected NGDP growth. I agree with most of the rest of what you said about price flexiblity. But perhaps now you begin to see where a pragmatist like me gets his views. You and I agree about the NIRA. But DeLong and Summers are pretty smart guys, so I assumed they were correctly utilizing the Keynesian implications about price flexibility in defending the NIRA. I’m glad to hear they weren’t, but again, I see example after example of smart people claiming to use Keynesian ideas, and reaching really bad conclusions. And BTW, I also have lots of problems with old-style monetarists, Austrians, MMTer, etc. I try to develop my own (admittedly crude) modeling ideas in a way that seems useful. And it all starts with NGDP futures, and idea that seems extremely useful.

  31. Gravatar of Scott Sumner Scott Sumner
    16. August 2011 at 08:05

    Matt, One other point, it’s not clear whether the “inflation rate” in the Fisher equation refers to the actual measured rate, or the equilibrium inflation rate. The arguments that the real interest rate matter implicitly assume the inflation rate measures actual transaction prices, free of queuing cost. They assume the number is meaningful in some sense.

Leave a Reply