Archive for the Category Monetary Theory


Josh Hendrickson on the problem with “moneyless” NK models

Josh Hendrickson is sort of like my mirror image; he doesn’t post very often, but almost invariably has something interesting to say.  In a new post he discusses a flaw in one common New Keynesian (NK) model, which looks at monetary policy through the lens of interest rates.  But first, a quick review of interest rates and monetary policy.  The easiest way to see the relationship is with the equation of exchange:

M*V = P*Y

When the Fed raises interest rates, pundits tend to call the policy “contractionary.” That’s misleading for all sorts of reasons, but does contain a grain of truth.  The true part comes from the fact that in the very short run, the Fed engineers interest rate increases by reducing the monetary base. Note that by itself the rate increase is expansionary, as it tends to boost the velocity of circulation, by raising the opportunity cost of holding base money.  But that expansionary effect is more than offset by the direct contractionary impact of the lower monetary base.  Here’s Josh, looking at the opposite case, an increase in the money supply:

Now consider the effects of a change in the money supply. As illustrated in the figure above, the increase in the money supply causes the interest rate to decline. This means that when the money supply increases, velocity declines. However, the interest elasticity of velocity is often estimated to be rather small. The initial effect of the increase in the money supply is that the nominal interest rate to fall and nominal spending to rise. The decline in the nominal interest rate is an effect of the change in the money supply, but note that it is not the cause of the change in nominal spending.

So far so good.  But here’s where NKs get into trouble.  Josh says that their models imply that a higher interest rate is contractionary, even if there is no change in the money supply:

The New Keynesians, however, countered that they didn’t need to use open market operations to target the interest rate. For example, Michael Woodford spends a considerable part of the introduction to his textbook on monetary economics to explaining the channel system for interest rates. If the central bank sets a discount rate for borrowing and promises to have a perfectly elastic supply at that rate and if they promise to pay a rate of interest on deposits, then by choosing a narrow enough channel, they can set their policy rate in this channel. In addition, all they need to do to adjust their policy rate is to adjust the discount rate and the interest rate paid on reserves. The policy rate will then rise or fall in conjunction with the changes in these rates. Thus, the New Keynesians argued that they didn’t need to worry about money in theory or in practice because they could set their policy rate without money and their model showed that they could get a determinate equilibrium by applying the Taylor principle.

Nonetheless, what I would like to argue is that their ignorance of money has led them astray. By ignoring money, the New Keynesians have confused cause and effect. This confusion has led them to believe that they know something about how interest rate policy should work, but they have never stopped to think about how interest rate policy works when the central bank adjusts the nominal interest rates in a channel system versus how interest rate policy works when the central bank adjust the nominal interest rate using open market operations.

Unfortunately the post is hard to excerpt; you really need to read the whole thing.

This does not mean that NK interest rate policies will not “work.”  Rather I would argue that they work for reasons that NKs don’t understand, and when they fail they will fail for reasons NKs don’t understand.  Other economists such as Peter Ireland (who is cited by Josh) have pointed out that even with interest on reserves (IOR) you can never really ignore the quantity of money, and that certain quantity theoretic claims continue to hold true.

For me, the easiest way to understand this issue is to think about (non-interest-bearing) currency. Back in 2007 the base was about $850 billion, with about $800 billion of that being currency.  Now suppose Ben Bernanke had been instructed to use interest rates to double the price level over the next 30 years.  He was not allowed to use the monetary base.  For simplicity, assume he never ran into the zero bound problem.  Even in that case, where positive interest rates allowed him to continually use a Taylor Rule-type approach, it’s easy to see that the policy objective would be impossible to achieve.  There would not be enough base money to match the demand for currency at a price level double its current value.  I think this simple truth sometimes gets overlooked in models that focus on bank reserves and IOR, and/or periods of history where there are plenty of excess reserves.

I said that NK policy might work despite its theoretical weaknesses.  That’s because asset markets would probably (correctly) interpret a rise in the IOR rate as part of a broader Fed strategy to reduce future growth in aggregate demand.  Thus markets would assume that the base would also be adjusted over time in whatever direction was necessary for hitting the central bank’s target. (Surely there must already be a Nick Rowe analogy involving steering wheels and social conventions.) As always, monetary policy is 98% expectations, and 2% concrete steps.  What are those expectations about?  They are about future concrete steps, i.e. future changes in the supply of base money, relative to changes in the demand for base money.

PS.  I have a post criticizing Charles Plosser, and 99% of other economists, over at Econlog.  And also a libertarian rant.

Other things equal, lower prices cause consumers to buy less of a good

I still see confusion about the never reason from a price change concept.  So let’s try again, looking to see whether “other things equal” helps.

Most people accept that fact that lower prices caused by less demand means something different from lower prices caused by more supply.  But what about lower prices, “other things equal?”

Screen Shot 2014-11-12 at 8.30.25 AMAs you can see the quantity demanded exceeds the quantity supplied. In that case the actual quantity bought and sold and consumed equals the quantity supplied, unless demanders put a gun to the head of suppliers.  So “other things equal” lower prices will reduce the amount that consumers purchase.  If either supply or demand shifts, then other things are not equal.  My “never reason from a price change” refers to equilibrium movements, you certainly can reason from a price or wage control set by the government, which moves prices away from equilibrium.  But as this case shows, not necessarily in the direction that you might assume.

When I said price changes have no effect, I should have said “other than to subjective states of mind.”  A price change, in and of itself, will change the amounts that people prefer to buy and sell. But a price change doesn’t affect any observable economic variable, unless it is an artificial price change that moves you away from equilibrium.

Ditto for interest rate changes and investment.  Many people stubbornly refuse to stop thinking in terms of “interest rate change equals interest rate change caused by the Fed,” and even worse, “interest rate change equals interest rate change caused by the liquidity effect of a Fed action.”  The latter claim isn’t even close to be true.  It’s not even true 20% of the time.  But it seems to be what many of my commenters and most economists assume is the case.  Here’s a typical comment from my previous post:

If interest rate rises stifle investment . . .

Stop right there; interest rate rises do not stifle investment.

I’m not sure why so many economists are confused on this point.  Perhaps they think: “The liquidity effect impacts rates in the short run.  The term ‘short run’ means roughly what’s going on right now.  And a long time series is just a long series of ‘right nows.'”  If that’s not what they are thinking, I’d love to hear alternative theories.

PS. The term “short run” does not in any way mean “what’s going on right now.”

In economics, price changes don’t have any effect, they are effects

I don’t know how many times I have to keep saying this before the rest of the profession figures it out.  Never reason from a price change.  [Update:  I mean an equilibrium price change.] It makes no sense to argue whether a higher price will increase or decrease quantity.  Here is a S&D diagram.  I dare you to show me how price changes affect quantity.

Screen Shot 2014-11-11 at 10.56.48 AM

And here is an IS-LM diagram.  Interest rates are the price of credit.  Changes in interest rates do not have any effect on quantity of output, price of output, or any other variable.  It’s not even an “other things equal” deal; price changes have no effect.

Screen Shot 2014-11-11 at 10.57.43 AM

At this point some economists will say; “I meant an interest rate change caused by a change in monetary policy.”  The problem is that higher interest rates can be produced by both easier and tighter monetary policy.  And easier and tighter monetary policy have opposite effects on prices and output.  So I’m sorry, but it’s still a meaningless debate.  It’s not that there is a right or wrong answer; there is no coherent question.  Monetary policy can shift the LM curve, the IS curve, or both.

Consider this analogy:  A debate over whether high oil prices increase or decrease global oil consumption.  The debate is meaningless.  Price has no effect.  Here’s how the issue should be discussed:

A.  An Arab oil embargo caused higher prices and lower consumption in 1974.

B.  Booming Chinese auto sales caused higher oil prices and higher consumption in 2007.

Prices are not a cause of anything; they are an effect.  And interest rates are a price.  So please stop these silly posts discussing the impact of a change in interest rates.  Talk about the effect of expansionary and contractionary monetary policies—that’s an interesting question.

This criticism applies to both sides of the debate.  John Cochrane and Noah Smith should not be discussing the possibility that higher rates lead to higher inflation, and Paul Krugman should not be claiming that the standard model suggests that higher rates lead to lower inflation.  Actually, both claims are true in specific situations.  But without specifying the specific situations in which each claim is true (especially the path of the money supply and IOR) the claims becomes utterly meaningless.

Truly a debate about NOTHING.

PS.  If you are still having trouble with this, consider the following.  A 1/4% rise in the fed funds rate today can be accompanied by a near infinite number of simultaneous expected changes in the future expected path of variables like the fed funds target, the monetary base, the interest rate on reserves, the reserve requirement, and all sorts of other policy levers.  Those variables in turn have a near infinite number of effects on all sorts on market variables other than short term interest rates, including TIPS spreads, commodity prices, forex prices, future expected real estate prices, stock prices, corporate bond risk spreads, etc., etc.  And all that happens immediately.

Here’s another example.  Assume Japan has run zero percent inflation for 20 years while the US has a credible 2% inflation target.  Zero inflation is expected to continue in Japan.  Suddenly they depreciate the yen 17% and set up a rigid currency board with the US.  PPP tends to hold in the very long run; so long term Japanese inflation expectations immediately rise from 0% to 2%. Interest parity holds even in the short run, so Japanese interest rates immediately rise to US levels. That’s a case where Cochrane is right, and it’s 100% consistent with IS-LM.  In contrast, in January 2001 the Fed cuts rates more than expected, and TIPS spreads rise sharply on the news.  That’s a case where Krugman was correct, and it’s 100% consistent with IS-LM.

The correct answer is “it depends.”

PPS.  Question for Nick Rowe.  Does my yen/dollar peg solve the coordination problem you discuss in this post?

HT:  TravisV.

John Cochrane on the stability of interest rate pegging

Here’s John Cochrane replying to a post by Nick Rowe:

The last 5 years have brought us a delicious opportunity for measurement. Once we hit the zero bound, interest rates can’t move any more. So the whole problem of empirically verifying long run dynamics is a lot easier.

What happened when the Fed kept interest rates at zero for 5 years? Pretty much nothing! OK, you see inflation going up and down, but look at the left hand scale — one percentage point. Given the colossal scale of other events in the economy, that’s nothing. Japan has been at it even longer, with similar results.

We seem to have in front of us a pretty clear measurement that long run dynamics are stable.

“Nothing” is astounding. This dog that did not bark has demolished a lot of macroeconomic beliefs:

  • MV = PY. Sorry, we loved you. But when reserves go from $50 billion to $3 trillion and nothing at all happens to inflation — or at most we’re arguing about percentage points — it has to go out the window.

  • Keynesian deflationary spirals. Just as much as monetarists worried about hyperinflation, Keyensians’ forecast of a deflationary spiral just didn’t happen.

I can certainly understand why Cochrane would make this argument.  It seems to make a lot of sense.  But that’s only because the standard model is infected by Keynesianism, which leaves it vulnerable to attack. Here’s one reply:

Yes, the Taylor Principle stabilizes prices during normal times, when interest rates are positive. And yes a rigid and positive interest rate peg is supposed to lead to highly unstable dynamics.  But that’s because if you target interest rates at a positive level and if there is no interest on reserves, then you lose control over the monetary base, which flies off to zero or infinity.  In contrast, at the zero bound you regain control over the base.  Thus once the Fed and BOJ lost the ability to adjust interest rates, they started using QE to prevent deflation.  (Of course with IOR you hit the “liquidity trap” that much sooner, but regain control of the base that much sooner.)

That’s the sort of reply I’d expect from Nick Rowe.  But then I’d also expect Nick to go further, and point out that this just shows how dangerous it is to describe the stance of monetary policy by using interest rates.  Here’s what actually makes the Taylor Principle work:

1.  When inflation is above target you need to reduce the growth rate of the money supply (relative to the growth rate of money demand) enough to bring inflation back down to the target.  It just so happens that the sort of policy that would reduce the money supply growth by an adequate amount is also one that will result in nominal interest rates rising in the short run (due to the liquidity effect) by roughly 150% of the rise in inflation.  But that rise in interest rates is not why inflation falls back to the target path (it’s an epiphenomenon), it falls back because you’ve reduced the growth rate of the money supply relative to demand.  Inflation falls back for monetarist reasons, not Keynesian reasons.

2.  Now suppose you are at the zero bound.  The Keynesian model requires something radically different, whereas the monetarist model just keeps plugging along with the same old method—adjusting the money supply relative to changes in money demand.  No ad hoc additions are needed for the monetarist model, unlike the Keynesian model.

But what about Cochrane’s criticism of the M*V=P*Y model?  What about his observation that reserves have soared while prices have hardly budged?  Unfortunately, there is no such model.  No one has ever proposed an M*V=P*Y model.  No one has ever claimed that inflation is proportional to reserves.  The most distinguished monetarist of all time was Milton Friedman, who famously argued that reserves are not a good indicator, and that money was tight during the 1930-33 deflation despite the huge rise in bank reserves due to Herbert Hoover’s QE.  There’s nothing new under the sun.  Nothing to be explained here; just move right along.

Sophisticated quantity theorists have always understood that the demand for bank reserves can be very large at zero nominal rates, especially when interest is paid on reserves (but even if it is not.) At the zero bound expectations (which are always about 98% of monetary policy) become 99.5% of monetary policy.  So I guess I was exaggerating a bit; the zero bound really is slightly different, but only quantitatively, not qualitatively.

An interest rate peg can lead to hyperinflation or hyper-deflation precisely because it can lead to upward or downward spirals in the monetary base.  That spiral does not generally occur at the zero bound, and hence there is no “stability” mystery to explain.  Unless you are a Keynesian.

I don’t want to sound too negative here–for some reason I had never really thought about that flaw in the standard Keynesian model until I had to think about how to respond to Cochrane’s post. That’s what makes the blogosphere so great.


Does monetary policy “cause” changes in NGDP?

Suppose that there are 6 ways the government could have prevented the 2008 financial crisis.  One might be abolishing moral hazard (FDIC, TBTF, the GSEs, etc.)  Another might have been to ban subprime mortgages made with taxpayer-insured funds.  Another might have been higher capital requirements.  Another might have been to require bankers to post personal bonds in case their bank failed (or needed to be bailed out), like on the old days.  Another might have been NGDPLT. Etc., etc.  In that case one might argue that failure to do one of those things “caused” the crisis.  Of course there are many ways of thinking about causation, but I find policy counterfactuals to be one of the most useful ways of describing causation.

Before considering whether money causes NGDP, I like to consider some related questions.

1.  I would argue that between 1879 and 1968 monetary policy “caused” the price of gold.  For most of the period the price was fixed by the monetary policymakers.  But even in 1933-34, when the price rose sharply, I’d argue it was “caused” by monetary policy, in the sense that the Fed was targeting the price of gold.  In contrast, the Fed was not targeting the price of gold in recent years, so the increase was not caused by the Fed in any useful sense of the term ’cause.’

2.  So if the Fed targeted gold prices from 1879-1968, does that mean other variables like the price level and NGDP were endogenous, and hence not caused by the Fed?  It’s debatable whether they were completely endogenous, but let’s say they were.  I’d still argue that the sharp decline in the price level and NGDP during 1929-33 was caused by the Fed.  And that’s because I’d argue that gold price targeting was not a wise policy, and that a desirable counterfactual policy would have been to stabilize either the price level of NGDP during 1929-33.  The Fed’s failure to do so caused the Depression.

3.  Now suppose the Fed is targeting inflation.  In that case any change in RGDP growth will lead to an equal change in NGDP growth.  Changes in NGDP will have nothing to do with monetary policy. Or at least it seems that way.  But not necessarily.  Suppose inflation is a bad target, because labor market instability (due to sticky nominal wages) and debt market instability (due to nominal debts) is much more closely related to NGDP fluctuations than to inflation fluctuations.  In that case a central bank that is targeting prices could be said to have “caused” a deep recession by letting NGDP fall, even if it was hitting its inflation target perfectly, and RGDP was falling for other reasons.  Of course it’s debatable as to whether its useful to consider the central bank to be to blame for the deeper recession, but it’s certainly not implausible, and indeed is the prediction of the standard AS/AD model used in econ textbooks.  In the graph below when you are targeting P and there is a negative supply shock, then NGDP will fall.  In that case you go to point E6, a deep recession.  If you were targeting NGDP when this supply shock hit, you’d have a mild recession (E7)Screen Shot 2014-11-04 at 8.28.43 AM.

It’s certain fair game to blame the Fed for depressing AD in response to a negative supply shock, even if 100% of the fall in NGDP is “caused” by a fall in RGDP.

Nominal GDP growth is the sum of a real and a nominal variable; real GDP growth and inflation. And yet (paradoxically) NGDP growth is a 100% nominal variable, completely under the control of monetary policymakers.  Whether it is useful to think in terms of NGDP fluctuations “causing” recessions depends on how you think the business cycle would be affected by a counterfactual policy of NGDPLT.

Similarly, changes in the nominal price of a gold (a 100% nominal variable) are the sum of a nominal and a real variable (inflation, and changes in the relative price of gold.)  In 1933-34 it was useful to think of the rising price of gold at an indicator of easy money.  More recently (in the early 2010s) it was not useful to think in terms of the rising price of gold as indicating an easy money policy.  In 1933-34 the price of gold was used as a signaling device for the future path of monetary policy.  That was not true in 2010-12.  I’d like to use some of these ideas to analyze part of a Tyler Cowen post from a couple years back:

My worry is that some Market Monetarists speak of ngdp as if it is some block of stuff, handed down from on high (of course in the past our central banks have not been targeting ngdp).  It’s as if ngdp determines the size of the room, and a carpenter is then asked to build a house within that room.  If the room is too small, a large house cannot be built.  Or, if you are not given enough clay, you cannot build a very large sculpture.  Along these lines, if the growth path of ngdp is not robust enough, the economy cannot do well.

I get nervous at how ngdp lumps together real and nominal in one variable, and I get nervous at how the passive voice is applied to ngdp.

My framing is different.  My framing is that the private sector can manufacture its own ngdp.  It can do so by trade and it can do so by credit and of course velocity is endogenous to the available gains from trade.  Most of the major central banks are, today, not obsessed with snuffing out recovery and increases in real output.

The value of money can be defined in terms gold, other currencies, all goods and services, and share of NGDP that can be bought with a dollar.  These all involve an inverse relationship; 1/Pgold, or 1/price of foreign currency, or 1/price level, or 1/NGDP.  Thus NGDP can be viewed as a single thing (one way of describing the value of money), or of course it can also be viewed as a composite (P and Y, or M and V).  If someone is used to viewing policy in terms of money supply targeting, or price level targeting, then an NGDP discussion can seem odd—adding together two very different things.  As I said, if you target P (or M) then NGDP will seem to fluctuate due to “non-monetary factors” like supply shocks (or V shocks.)

But that tells us nothing about whether it is useful to think in terms of NGDP being causal, i.e. whether NGDPLT is a useful policy counterfactual.  So while Tyler’s argument is defensible, the average reader would probably assume that Tyler has spotted a logical error in the NGDP fanatics, which is actually not there.  (In fairness, elsewhere he explicitly denies doing so.)

Go back to the opening analogy in this post.  I favor eliminating moral hazard.  But it wouldn’t be fair of me to accuse someone who favored higher capital requirements for banks as having ignored the real cause of banking crisis.  At best I could argue my solution was more useful. Perhaps Tyler should have discussed which alternative monetary policy targets are the most useful.  The commenter who sent me this post thought (probably wrongly) that Tyler was criticizing NGDPLT.  I don’t see that.  He was claiming it’s not the cure-all some of its proponents seem to think it is.

To say “ngdp is low,” or “ngdp is on a low growth path,” or “ngdp is below trend,” and so on — be very careful!  Those claims do not necessarily have causal force.  Arguably they are simply repeating, in a new and somewhat different language, the point that the private sector has not seen fit to engage in more trade, credit creation, velocity acceleration, and so on.  Formally speaking, the claims are not wrong, but I don’t find them useful as an explanation for why economic growth or recovery, at some point in time, is slow.  It is one way of repeating or re-expressing the slowness of economic growth, albeit with some transforms applied to the vocabulary of variables.

This paragraph touches on both the “causality” and the “usefulness” perspectives I discussed earlier, but in what seems to me to be a somewhat unsatisfactory fashion.  Start with the final two sentences.  He’s saying that by 2012 enough time had gone by so that wages and prices should have adjusted.  That’s an eminently plausible argument.  The final sentence can be thought of as an alternative hypothesis.  Say we are targeting P or M, and output is slowing for Great Stagnation reasons.  In that case NGDP will also slow (or might slow in the M targeting case) and MMs like me might wrongly attribute the slowdown in RGDP to a slowdown in NGDP.  And that’s certainly possible.

When Tyler wrote the post the most recent reported unemployment rate was 8.1%, for August 2012.  Now it’s 5.9%.  I believe the most useful explanation for that sharp fall is that NGDP has been rising faster than nominal wages in the US.  Tyler says central banks were not trying to snuff out the recovery.  But we do know that in 2011 the Fed was trying to speed up the recovery while the ECB was trying to reduce inflation by raising their target interest rate.  And we know that NGDP in 2011-13 grew much more slowly in the eurozone than the US.  And we know that unemployment rose sharply in the eurozone, while it fell sharply in the US.  So while Tyler is right that movements in NGDP need not have any causal effect on RGDP, I believe that it just so happens that we live in a universe where it does have a causal effect, or at least that it is useful to talk in terms of causal effects from NGDP.

This matters when we consider sticky nominal wages.  Sometimes it is suggested that the “inside workers” have frozen up or taken up so much ngdp with their sticky wage demands that the outsiders cannot find the ngdp to fuel their activities.  It’s as if there is not enough ngdp to go around, just as there was not enough clay to make a sufficiently large sculpture.

Yes, workers and firms can behave in a way that overcomes any shortages of NGDP. Indeed Tyler and I agree that in the long run they WILL behave in a way that overcomes any shortage of NGDP. But these metaphors are expressed in a slightly misleading way.  The average reader would have a great deal of trouble figuring out whether Tyler is expressing a new classical argument that nominal shocks don’t matter, and that real GDP is determined by real factors, or a NK/monetarist argument that nominal shocks matter in the short run but not the long run, and that September 2012 is now the long run, or the hybrid theory that nominal shocks might matter in the short run, but not if the short run NGDP changes are caused by real factors such as less aggregate supply.  A close reading of his other posts shows he believes the nominal shocks matter in the short run, but many of the (skeptical) NGDP metaphors employed here are also applicable to the short run.  Do they apply to a case where AS and AD are entangled?

So for instance, suppose a central bank is targeting inflation.  Then suppose an adverse supply shock reduces RGDP by two percent in the short run, even in the best of circumstances (i.e. stable NGDP).  But also suppose that NGDP falls with the supply shock (as Tyler correctly noted might happen.)  My claim is that in that case RGDP would fall by more than 2%, perhaps 4%. What does Tyler believe?  The metaphors he employs seem to suggest that he is skeptical of this claim, but elsewhere he argues that nominal shocks do matter in the short run, so falling NGDP should make the recession worse.

PS.  Two years ago I did a post in response to this same Tyler Cowen post.  I did this post without looking at my earlier post, and ended up with something very different.  Maybe in two more years I’ll do a third.

PPS.  I do agree with Tyler that framing effects are important.  In microeconomics there is one dominant framing method, and hence you don’t see as many cases of microeconomists who are Nobel Prize winners call each other idiots as you do in macro, where different framing effects lead to almost a complete failure of communication.

PPPS.  Election today?  All I care about are the referenda.  Both major parties have sharply declined in quality over the past 20 years.  Both deserve to lose.