Archive for the Category Monetary Theory


Why macro stabilization policy rarely fixes problems

A big demand slump isn’t just an economic disaster; it’s also a prediction of an economic disaster. And that means it’s a prediction of policy failure.   At least that’s the implication of the Woodfordian view of macro (which I accept.)  Changes in current AD are mostly driven by changes in the future path of AD.  Changes in near-term NGDP are mostly driven by changes in expected NGDP 1, 2, 5 and 10 years out in the future.  Call it the term structure of NGDP.  And those are driven by the future expected path of monetary policy.

And of course whenever we have crashes like 1920-21, 1929-30, 1937-38, 2008-09, we also tend to have asset market crashes.  Asset markets aren’t perfect (1987) but when there’s a very big economic slump on the way they are pretty good at sniffing it out.

So here’s the problem for macro policy.  It’s good at preventing disasters, as we saw with the Great Moderation.  But when it fails, it’s really, really hard to fix the problem, because doing so requires policymakers to be more effective than the markets predict.  I won’t say that things are hopeless when markets predict disaster, but I wouldn’t put much hope on stabilization policy.  In the textbooks, the purpose of stabilization policy is to “fix problems.”  In reality it will usually fail at that.  Rather it’s good at preventing problems.  If you’ve got a problem, you’ve already failed.  Like the old joke—”if you are headed there, I wouldn’t start from here.”

I was reminded of all this while reading a Brad DeLong post that discusses a debate between Nick Rowe and Simon Wren-Lewis.  DeLong looks at the possibilities offered by monetary and fiscal stimulus when you have a “deficiency in demand.”

Let’s look at this from a different perspective.  The problem is not “demand deficiency” it’s expected demand deficiency.  Policymakers try to steer the nominal economy, they implicitly or explicitly target NGDP one or two years out in the future.  If 12 month forward expected NGDP is right on target, then no policy changes are needed.  And if 12 month forward NGDP is below target, then the markets have predicted policy will fail, and their forecast counts for far more than the views of any academic economist or government policymaker.  At that point, we really shouldn’t expect much from macro policy.  It’s likely to fail.  No wonder people are so pessimistic about monetary policy! Markets have observed the behavior of the relevant central bank (Fed, ECB, etc.) and come up with the optimal forecast of the result.  If there’s an expected demand shortfall, markets have already given a vote of no confidence to the policymaking apparatus.

From that perspective, DeLong is asking the wrong question.  It’s not, “how do we fix this problem?”  It’s, “how to we make it so that Brad DeLong and Simon Wren-Lewis never ask, ‘how do we fix this problem.’”  I see two ways, and only two ways of doing that.  Both methods involve abandoning the Keynesian policy of interest rate targeting.  Interest rate targeting doesn’t work at the very moment when good monetary policy is most essential—in a very deep demand slump. Would you buy a car that had a brake that failed just 1% of the time—only on twisty mountain roads with no guardrail? Then why do you (Keynesians) buy interest rate targeting as the appropriate policy instrument?

1.  One method would have the central bank peg the price of one year forward NGDP futures, and do OMOs until the market price is right on target.  Now you don’t have to worry about what to do if there is an expected demand deficiency, because there never is an expected demand deficiency.  At least not one expected by the market.  There may be a current demand deficiency, but if it isn’t expected to persist, then stabilization policy is right on target.

2.  Let’s say you don’t buy the “market” part of market monetarism.  You think markets are irrational.  ”Better leave this to the wise mandarins who will control policy in the optimal fashion.” What then?  It’s very simple, you do what Lars Svensson suggested, you set the monetary instrument at a position where the central bank’s internal forecast is equal to the policy target.  But which instrument?  Recall that we have abandoned interest rate targeting.  Don’t ask me, I’m the NGDP futures market guy–ask the mandarins.  Anything with no zero bound.  It might be the monetary base, it might be the trade-weighted exchange rate, it might be the nominal price of zinc. There is an infinity of possible choices.  (Now do you see why I’d rather let the markets set policy?)

In his post, DeLong cites Wren-Lewis saying he’s heard the MM arguments, but doesn’t buy them. Then he goes on to conclusively show he has not heard the MM arguments, by using the metaphor of employing both a regular brake and an emergency brake in a car careening down a hill.  This metaphor is supposed to provide justification for using fiscal stimulus “just in case” to back up monetary stimulus.

But that won’t work if you have monetary offset.

In any case, monetary policy is a brake that never fails, and if it does fail you don’t end up crashing, you end up with the Bank of England owning the entire world.  A level of global domination that makes Victorian-era Britain seem like a 98 pound weakling by comparison.  Global GDP is around $100 trillion.  So Piketty would say that global wealth must be around $500 trillion.  Could the Brits live on 5% of that?  I think so.  But wait until the Scots secede, those ingrates don’t deserve any of it.

As Dylan said on his greatest album:

.  .  . there’s no success like failure . . .

Lowflation? LOL. Try again!

The economics profession made a serious mistake a few decades ago when they latched onto “inflation” as a key macroeconomic indicator.  People were very upset about inflation during the Great Inflation of 1966-81, for reasons totally unrelated to the reasons macroeconomists think inflation is important.  Since everyone was talking about inflation, macroeconomists wanted to put it into their models.  Nobody was talking about nominal GDP.

[As an analogy, macroeconomists put short-term interest rates in their monetary models because central banks usually target that variable.  In the 1930s George Warren was mocked for saying that the price of gold is "the" indicator of monetary policy.  But how's that different from new Keynesians?  After all, when Warren was alive central banks did target the price of gold.]

In any case, it’s become increasing clear that inflation is not the right variable–it does not describe the nominal shocks hitting economies.  And Europeans in particular are paying a heavy price for this mistake, as when the ECB sharply tightened monetary policy in 2011 because a completely meaningless headline inflation number (including “austerity” VAT increases and imported oil) had briefly risen above their 2% target.

As inflation becomes more and more discredited, pundits try ever more clever techniques to make it seem relevant again.  Here’s a recent article from The Economist:

Against this background, it is unsurprising that inflation is stuck at just 0.5%. Although the European Central Bank (ECB) took steps to counter “lowflation” in early June, the worry is that it has still not done enough. In a survey of the euro-zone economy published on July 14th the IMF urged the ECB to adopt quantitative easing—creating money to buy assets including sovereign bonds—if inflation remains too low.

Lowflation represents a particular threat to highly indebted countries like Portugal.

It’s amazing the lengths to which pundits will go to NOT mention nominal GDP.

Let’s compare Portugal and Switzerland.  Over the past 6 years both have seen inflation fall from the low single digits, to near zero.  But look at the nominal GDP numbers (World Bank) for the two countries:

Year       Portugal   Switzerland

2007       169.3          540.8

2008       172.0          567.9

2009       168.5          554.3

2010       172.9          572.1

2011      171.1           585.1

2012      165.1          591.9

2013      165.7          603.2

Portugal’s NGDP is down over 2%, while Switzerland’s is up over 11%.  Not surprisingly, although both countries suffer from “lowflation,” Switzerland has seen RGDP grow 8% over this period, while Portugal’s RGDP has fallen by 7%.

And NGDP isn’t just the right metric for nominal shocks and the business cycle, it’s also the right variable for the debt crisis.  Nominal income is the resource that individuals, business, banks and governments have to repay nominal debt.  Why in the world would someone talking about a debt crisis mention “inflation?”  What does that have to do with debt?  What if a country has 0% inflation and 10% RGDP growth?

That’s not to say inflation is never correlated with demand shocks–most of the time it is.  But NGDP is 100% correlated with demand shocks.  So if you are worried about nominal shocks hitting the economy, why not use an accurate nominal shock measure like NGDP?  Why use a metric that is correlated with NGDP when the economy is hit by demand shocks, but not supply shocks?

Of course even NGDP isn’t going to explain all the movements in RGDP (but it will do better than inflation.)  Switzerland has better supply-side fundamentals, so even if the ECB had done enough monetary stimulus to cause Portugal’s NGDP to rise by the same 11% as in Switzerland, their RGDP performance might well have lagged Switzerland.  Nonetheless, Portugal would have done somewhat better in terms of growth, and their debt crisis would have definitely been much milder.

You want to see “lowflation?” China’s inflation rate (GDP deflator) averaged 0.75%/year between 1996 and 2003—pretty close to Japan.  Now check out the NGDP numbers for the two countries.

I’m begging the economics community.  Stop talking about inflation when you really mean nominal income.  And stop coining terms like ‘lowflation’ in a pathetic attempt to add epicycles to the obsolete inflation-oriented models of macroeconomics.

Inflation, interest rates, income inequality—the “i-words” don’t matter.  NGDP, nominal wages, consumption are what matter.  BTW, the current account deficit also doesn’t matter, as I point out in my newest Econlog post.

Andolfatto interviews Woodford

David Andolfatto is a very knowledgeable monetary policy blogger, and here he interviews Michael Woodford, perhaps the world’s leading monetary theorist.  I’ll just focus on one issue:


There is a conventional wisdom of how these tools might work. Can you explain to us the findings of your own research, how they might corroborate these findings or these beliefs? Or go against them in some manner? Is there something surprising that emerges from what you’ve discovered?


I think so. I think a lot of the discussion that you see of the point of asset purchases suggests that there should be a lot of similarity between the effects of purchasing long-term assets and the effects of cuts in the federal funds rate, the Fed’s traditional tool. People say the whole point of cutting the federal funds rate is longer-term bond yields would also go down, and if you can just buy longer-term bonds, push up their prices, that’s doing the same thing with a different mechanism. It’s a different way of doing the same thing. And if you can’t cut the federal funds rate further, then there’s an obvious reason to use the other method.

And our analysis suggests that this analogy between the two tools is not nearly as strong as you might have expected.


Why is that exactly?


Well, one reason is that the question of whether it’s clear that Fed purchases of longer-term assets can affect the prices of those assets as directly as traditional interest rate policy would. But I think the more surprising thing is that our analysis suggests that even under circumstances when the central bank finds that its purchases do affect the market price of the longer-term assets, the connection between that and spending in the economy, and then the effects on inflationary pressure, are not necessarily at all similar to those of conventional interest rate policy.


So you’re suggesting that it is possible, at least in theory, that the Fed engages in the large purchase of a certain class of assets? Injects money into the economy by purchasing a particular class of assets? And that this may, in fact, have very exact opposite sort of effects than conventional data might suggest?


Right. We clearly show that that’s at least a theoretical possibility. And obviously then deciding whether you think that’s actually happening is another thing. But I think the analysis points out that you shouldn’t assume that the mere fact that you could raise the price of the bonds answers then the question about what effect you’re having on the economy.


So can you explain the economic intuition for that effect and whether or not it has some bearing as to the conduct of Fed policy today?


I think the point is a fairly simple one, and it has to do with the question of why the central bank purchases should be able to move the market price anyway, which, again, people thought was kind of obvious. They said if you’re buying more of something, surely that will tend to make it more expensive. But when you ask whether that should actually happen with a lot of sophisticated traders out there in the market that are also trading against the central bank, what we argue is that if the other traders in the economy aren’t constrained in the financing they can mobilize to take the positions that make sense for them, they will tend to automatically have an incentive to trade against the central bank and to neutralize then the effects of the central bank’s trades.

Two things struck me.  First, Woodford has a contrarian view of the effects of QE on long-term interest rates.  Second, market monetarists have a similar counterintuitive view, but for very different reasons.

Woodford starts by pointing out that people expected QE to reduce long-term rates, just as traditional fed funds rate cuts reduce long term rates.  But MM doesn’t even accept the premise. Some of the most dramatic Fed moves toward cutting short-term rates have actually boosted long term rates, via the inflation and income effects.  We think QE also has an ambiguous impact on rates, for similar reasons.  In contrast, Woodford focuses on the risk channel (you should read the whole thing to get his explanation.)

Then Woodford suggests that the relationship between long-term rates and the economy is not as clear as with traditional tools.  We agree that it’s not at all clear (never reason from a price change), but we think that’s also true of traditional tools.  One cannot assume that lower interest rates produced by the Fed will lead to strong growth in AD.  It depends on the relative strength of the liquidity, income and Fisher effects.

In the final paragraph I quote, Woodford points out that most people think that Fed purchases “obviously” boost the price of the asset being purchased.  They misuse the S&D model.  Some commenters are outraged that the Fed is helping group X, because group X owns lots of the assets that the Fed is buying.  They see dark conspiracies.  But the purchase of bonds is also the sale of cash.  And more cash boosts inflation, which reduces bond prices.  During the 1964-81 period the Fed radically increased the amount of bonds it was buying, this led to rapid growth in the monetary base, higher inflation, and much lower bond prices.  So much for Cantillon effects. 

Yes, there are cases where large asset purchases are associated with low inflation (such as recently); my point is that there is no consistent relationship between Fed asset purchases and the price of that asset.

HT:  Tom Brown, TravisV.

David Glasner on me on IS-LM

Here is David Glasner:

Scott is totally right, of course, to point out that the fall in interest rates and the increase in the real quantity of money do not contradict the “money hypothesis.” However, he is also being selective and unfair in making that criticism, because, in two slides following almost immediately after the one to which Scott takes such offense, Foote actually explains that the simple IS-LM analysis presented in the previous slide requires modification to take into account expected deflation, because the demand for money depends on the nominal rate of interest while the amount of investment spending depends on the real rate of interest, and shows how to do the modification. Here are the slides:

.  .  .

Thus, expected deflation raises the real rate of interest thereby shifting the IS curve to the left while leaving the LM curve where it was. Expected deflation therefore explains a fall in both nominal and real income as well as in the nominal rate of interest; it also explains an increase in the real rate of interest. Scott seems to be emotionally committed to the notion that the IS-LM model must lead to a misunderstanding of the effects of monetary policy, holding Foote up as an example of this confusion on the basis of the first of the slides, but Foote actually shows that IS-LM can be tweaked to accommodate a correct understanding of the dominant role of monetary policy in the Great Depression.

Was I really so unfair?  Did I actually accuse Foote of not understanding that tight money can reduce nominal rates.  You be the judge:

Note the very last comment on the slide, about the significance of deflation.  The rest of the PP slides develop this idea further, and correctly show that while tight money might raise real interest rates, it could lower nominal rates through the Fisher effect. Thus it could shift the IS curve.  That helps, but it seems to suggest that the IS-LM model can be rescued by switching the argument from nominal to real interest rates. Alas, that won’t work.  The Fisher effect is only one of the ways that monetary shocks impact interest rates.  Tight money also reduces expected future real GDP, and this also shifts the IS curve.  So it isn’t just nominal interest rates that fall, real rates also fell during the 1930s, as expected future real GDP plunged.

It sure looks like I’m suggesting that Foote “correctly” understood the distinction between the impact of monetary policy on real rates and nominal rates.  I’m not sure where David got the idea I was being critical of Foote.  Then David simply ignored my observation that switching from nominal to real interest rates in no way rescues the IS-LM model.  Tight money can also reduce real interest rates.  Ex ante real rates did fall during the 1930s.  So IS-LM cannot be “tweaked to accommodate a correct understanding” of the role of money Depression.  It’s rotten to the core.

David continues:

The Great Depression was triggered by a deflationary scramble for gold associated with the uncoordinated restoration of the gold standard by the major European countries in the late 1920s, especially France and its insane central bank. On top of this, the Federal Reserve, succumbing to political pressure to stop “excessive” stock-market speculation, raised its discount rate to a near record 6.5% in early 1929, greatly amplifying the pressure on gold reserves, thereby driving up the value of gold, and causing expectations of the future price level to start dropping. It was thus a rise (both actual and expected) in the value of gold, not a reduction in the money supply, which was the source of the monetary shock that produced the Great Depression. The shock was administered without a reduction in the money supply, so there was no shift in the LM curve. IS-LM is not necessarily the best model with which to describe this monetary shock, but the basic story can be expressed in terms of the IS-LM model.

I agree that the best way to visualize the Great Contraction is through a large increase in the demand for monetary gold.  But I’m confused by David’s claim that this would not shift the LM curve.  Gold was the medium of account.  Unless I’m mistaken, an increase in the demand for gold would certainly be expected to shift the LM curve to the left, wouldn’t it?  (But then IS-LM is not my forte, so please tell me if I am wrong.)

PS.  In case you think I cherry-picked a quote, here’s the opening to my post, where I describe the quality of Foote’s PP slides:

Commenter Joseph sent me an excellent set of PP slides by a professor at Harvard named Chris Foote.  He has a very clear derivation of the AD curve from the IS-LM model.

If I’m going to be accused of being unfair to someone, at least give me the satisfaction of trashing their work!  My post had no criticism of Foote at all.  Just some criticism of ideas he listed on one slide as things other people have claimed might have caused the Depression.  I never assumed that was his view, and indeed he himself criticized some of those arguments in later slides.

PPS.  And how does David know I am “emotionally committed” to the view that IS-LM is worthless? Perhaps I have rational reasons for holding that view.  I’ve claimed that monetary policy can shift the IS curve, or else one has to assume the IS curve is upward sloping (Nick Rowe’s view.) I haven’t seen anyone rebut that view, emotionally or unemotionally.

IS/LM and AD

Commenter Joseph sent me an excellent set of PP slides by a professor at Harvard named Chris Foote.  He has a very clear derivation of the AD curve from the IS-LM model.  This slide caught my attention.

Screen Shot 2014-07-16 at 3.32.54 PM


The “spending hypothesis” obviously cannot explain the Great Depression.  In 1987 we saw an equally big stock market crash, and GDP kept booming.  (And don’t bother trying to concoct excuses about how things were different in 1987—I’ve swatted them all down 100 times.)

An investment backlash? Does that mean people switch from investment goods to consumer goods? If so, why did consumption also fall?  

Hoover actually favored public works projects, so fiscal policy explains nothing until at least 1932, when he raised taxes.   Was there a Great Depression in 2013?  I don’t think so.  And yet we had “savage” austerity in 2013.

OK, so it was tight money.  We’ve known that for 50 years, ever since Friedman and Schwartz. What interests me is the suggestion that the “money hypothesis” is contradicted by various stylized facts. Interest rates fell.  The real quantity of money rose.  In fact, these two stylized facts are exactly what you’d expect from tight money.  The fact that they seem to contradict the tight money hypothesis does not reflect poorly on the tight money hypothesis, but rather the IS-LM model that says tight money leads to a smaller level of real cash balances and a higher level of interest rates.

To see the absurdity of IS-LM, just consider a monetary policy shock that no one could question—hyperinflation.  Wheelbarrows full of billion mark currency notes. Can we all agree that that would be “easy money?”  Good.  We also know that hyperinflation leads to extremely high interest rates and extremely low real cash balances, just the opposite of the prediction of the IS-LM model.  In contrast, Milton Friedman would tell you that really tight money leads to low interest rates and large real cash balances, exactly what we do see.

Note the very last comment on the slide, about the significance of deflation.  The rest of the PP slides develop this idea further, and correctly show that while tight money might raise real interest rates, it could lower nominal rates through the Fisher effect.  Thus it could shift the IS curve.  That helps, but it seems to suggest that the IS-LM model can be rescued by switching the argument from nominal to real interest rates. Alas, that won’t work.  The Fisher effect is only one of the ways that monetary shocks impact interest rates.  Tight money also reduces expected future real GDP, and this also shifts the IS curve.  So it isn’t just nominal interest rates that fall, real rates also fell during the 1930s, as expected future real GDP plunged.  The IS-LM model is useless, and should be discarded.  The fact that it is used to derive the AD curve probably explains why I can’t seem to understand how Keynesians use the concept of AD; they are basing it on ideas that make no sense to me.

Here’s a simpler model.  AD is a hyperbola (a given level of NGDP).  This model does not assume NGDP targeting, just as the current AD model does not assume money supply targeting.  Changes in NGDP are caused by monetary policy.  The P/Y split for changes in NGDP is determined by the slopes of the SRAS and LRAS curves.  The LRAS curve is vertical.  Interest rates?  Yeah, they fluctuate a lot.

BTW, the PP slides end up with an excellent Greg Mankiw post from December 2008.  Mankiw recommends the Fed commit to a price level 30% higher in 10 years time.  Had they done so the recession would certainly have been much milder.

On the other hand if Mankiw had been head of the Fed in December 2008 they would have done almost exactly what they did under Bernanke, not what Mankiw recommended in 2008.  Mankiw’s policy views are pretty similar to Bernanke’s views.  Unfortunately there is also the formidable “FedBorg.” Had Mankiw been in charge, it would have been Bernanke who wrote the blog post suggesting the Fed should commit to a 30% higher price level in 10 years.  We are all just puppets, filling out the roles determined for us by blind fate.

Off topic:  Ramesh Ponnuru is right, I see lots of interest in NGDPLT among conservatives.

I have a post on AD over at Econlog, for those not burned out on the issue.

Just to show you that MMs don’t agree on everything, David Beckworth has a excellent post arguing against the Great Stagnation hypothesis.  Marcus Nunes is also skeptical.  In contrast, I accept the Great Stagnation hypothesis.  Fortunately it doesn’t matter for our monetary policy views, as we all favor targeting NGDP and ignoring RGDP.