Archive for August 2013


Real interest rates are not much better

It’s widely known that nominal interest rates are not a good indicator of the stance of monetary policy.  Actually it’s probably not widely known, which is a major puzzle in and of itself.

Yes, there is a liquidity effect.  A sudden increase in the money supply will often reduce short term interest rates.  But other factors have a much more powerful impact on rates, and hence the liquidity effect can be very elusive and hard to spot.

Some economists think this can be fixed by looking at real interest rates.  But that’s false.  Soon after I criticized a recent post by John Quiggin, he added a couple more paragraphs, and shifted his argument from nominal rates to real rates:

Update As pointed out by Mark Sadowski in comments, these are nominal rates of interest. To get the real rate, which is more relevant, you need to subtract the expected rate of inflation, which fell from around 7 per cent to around 4 per cent over this period (as measured by surveys, and by the premium for inflation-adjusted Treasury bonds). So, you get a 9 percentage point reduction in the real rate from 10 per cent to 1 per cent. This doesn’t make much difference to the story. Most economists would regard policy as contractionary/expansionary if real interest rates are above/below the long-run neutral level, about 3 per cent. So, we still have a shift from strongly contractionary to moderately expansionary.

That’s sort of like saying “most economists” are ignorant of the basic principles of monetary economics.  But maybe they are.  It is a highly specialized field.

Think about why some economists might prefer using real interest rates to nominal rates.  What is the reason?  Obviously the reason is that economists understand that interest rates don’t just reflect the liquidity effect, expected inflation also influences interest rates.  Fair enough.  But why stop there? Expected income growth as well as the level of income also have powerful effects on interest rates.  Indeed if you go back to the pre-WWI period in America we had almost no expected inflation, and no Federal Reserve to raise or lower interest rates.  And guess what; rates rose and fell with the business cycle, just as they do today.  Income has an incredibly powerful impact on rates, indeed in recent years much more so that inflation.  That’s why Ben Bernanke insists that neither nominal nor real interest rates are reliable indicators of monetary policy, and you must look at NGDP growth and inflation.  But then Bernanke is not “most economists” he’s one of the leading experts in monetary economics.

Quiggin continues:

However, market monetarists want to argue that the stance of policy should be assessed relative to a policy rule (Taylor rule or NGDP) that already incorporates a prescription of cutting rates when GDP falls and unemployment rises. This doesn’t make a lot of sense to me. It’s like arguing that Obama’s stimulus was actually a contractionary policy because it wasn’t as big as (according to a standard analysis based on Okun’s Law) it should have been. It’s partly a question of semantics, but it’s associated with the claim that, if only rates had been cut even more, we wouldn’t have had the recession, or would have recovered quickly. Having been around at the time, I disagree.

No, that’s not the claim.  If rates had been cut to zero I’d guess the recession would have been far deeper.  Zero rate cuts are almost always associated with ultra-deep recessions.  With tight money.  The US in the 1930s.  Japan since 1997.  One more time:


What would have prevented a deep recession would have been a more expansionary monetary policy, not lower rates.  Interest rates tell us almost nothing about whether monetary policy is expansionary or not.  If anything, they tend to be a “reverse indicator.”  If you told me that country X had 40% interest rates, I’d guess they had a highly expansionary monetary policy.

Quiggin seems to think that MMs are sort of oddballs.  OK, but what does he make of the Volcker disinflation?

1980:3 to 1981:3:     NGDP growth = 14.0%

1981:3 to 1982:3:     NGDP growth = 3.2%

Meanwhile nominal interest rates on 3 month T-bills fell from 16.3% in May 1981 to 7.71% in October 1982.  These data are quite similar to the Australian episode considered by Quiggin.  If we use his criterion for easy money then America’s most famous tight money policy since the Great Depression was actually an expansionary monetary policy.

I don’t think so.

PS.  Real interest rates on 5 year Treasury debt (ex ante, risk free) rose from a low of 0.57% in July 2008 to a peak of 4.2% at the beginning of December 2008. Throughout this period the US was NOT at the zero bound.  Maybe you guys can help me.  Find examples from Quiggin or any other Keynesian blogger in late 2008 complaining about the Fed’s ultra-tight money policy.

Can’t find any examples?  Keep trying; they must be out there somewhere.  After all the smarter Keynesians like Quiggin insist that while nominal rates are not reliable, surely real interest rates are.  So find me some examples.  To paraphrase Bob Dole; “where was the outrage?”

PPS.  Quiggin’s fiscal policy analogy makes no sense, as the change in the money supply would be the closest analogy to deficit spending.  Both interest rates and NGDP are market variables that are affected by changes in the money supply.  Of course there are other differences.  Fiscal stimulus is costly whereas monetary stimulus is not.  Hence monetary policy is best viewed as setting the steering on a ship, where no one direction is more costly than another, just different.  And NGDP is the best way of measuring direction; interest rates are almost meaningless.

The Fed gives itself a triple mandate

Every once and a while you need to stop and take stock of just how insane current monetary policy really is:

The tepid demand dampened inflation pressures last month. A price index for consumer spending edged up 0.1 percent, slowing from a 0.4 percent increase in June. Over the past 12 months, prices rose 1.4 percent compared with 1.3 percent in June.

It was the biggest increase since February.

Excluding food and energy, the price index for consumer spending nudged up 0.1 percent after advancing 0.2 percent in June. Core prices were up 1.2 percent from a year ago, rising by the same margin for a fourth consecutive month.

Both inflation measures continue to trend below the Fed’s 2 percent target. That, combined with the lacklustre consumer spending, would argue against the U.S. central bank trimming the $85 billion in bond purchases it is making each month to keep interest rates low.

Yes, so it would seem.  Indeed they should sharply boost the pace of QE.  But they will probably taper anyway.

Now I suppose you could argue that the Fed doesn’t have a single mandate, they are supposed to also focus on employment, so while easier money is needed to hit the inflation target, the employment target requires . . . umm, easier money?

Reuters continues:

Many economists, however, believe the Fed will make an announcement on the tapering at its September 17-18 policy meeting, starting off with a small cut to the bond-buying program, known as quantitative easing.

“This does nothing to alter our view of tapering. Fear of unquantifiable financial risks within a QE regime that offers diminishing returns is driving the policy agenda, not strong growth and inflation,” said Eric Green, global head for rates, foreign exchange and commodity research at TD Securities in New York.

Yes, I forget about the third (and top secret) mandate that Congress gave the Fed; “unquantifiable financial risks.”  Here’s a general observation.  When big cumbersome institutions are given the mandate to target “unquantifiable” anything, don’t hold your breath for good outcomes.

This is really absurd on so many levels.  The asset markets are much more stable when you have steady NGDP growth, than when you have wild swings in NGDP.  Anyone recall asset prices in 2008-09?  The asset markets are suggesting that tapering will create instability in emerging markets, and to a lesser extent the US stock market.

Ben Bernanke is a student of history, and knows full well that this reasoning is EXACTLY WHAT THE FED DID WRONG in the 1930s.  And I mean exactly.  You could find dozens of similar articles in the NYT from the 1930s.  I hope he’s quietly pushing back against this view from within the Fed.

PS.  The current frontrunner for the job of Fed chair is the most famous proponent of focusing Fed policy on reducing unquantifiable financial risks.

PPS.  The ECB raised rates in 2011.  How’d that reduce “unquantifiable financial risks?”  How about the Fed’s tight money policy of 1937?

PPPS.  Right now I want all readers to cross their fingers.  That’s it.  That’s the Fed’s secret plan to bring inflation up to their 2% target.

PPPPS.  Marcus Nunes has a good post on the declining inflation rate in America.

PPPPPS.  Here’s Ryan Avent:

There are limits to what reporting can uncover about a decision that is the president’s to make. Mr Obama may in fact be enthralled by Mr Summers’ private monetary views. For all I know the president is a die-hard Scott Sumner fan and Mr Summers has whispered to him that market monetarism is now his macroeconomic lodestar. But there are few signs that anything like that is going on.

Here are some signs that that is not going on:

1.  I favor strict rules, Summers favors discretion.

2.  I believe monetary policy is highly effective at the zero bound, Summers believes it’s ineffective.

3.  I oppose fiscal stimulus, Summers is in favor.

4.  I believe the Fed should ignore asset “bubbles,” Summer disagrees.

5.  I believe wage flexibility is stabilizing, Summers thinks it’s destabilizing.



[As Krugman would say, wonks only.]

So the recent post I did defending Tobin led to a comment section discussion with Nick Rowe, George Selgin, Saturos, etc., which eventually ended up on this knotty problem of what is “money.”  Is it all about the medium of account, or the medium of exchange?  They are almost always the same, so it’s hard to find good real world examples.  And it’s even hard to construct thought experiments, as if you separate them, or drop one entirely, it’s hard to know what auxiliary assumptions to make.  I’ll try to show this with four examples, three of which support my view, and two of which support Nick’s (one is consistent with both.)  I’ll let the comment section sort it out:

1.  I’ll start with a sticky wage model (which I prefer), but don’t worry the other examples will all be sticky prices.  We have a tropical country called “Barteria”, where the workers produce fruits and coconuts on large plantations.  They are paid in what they produce, and barter these goods for other fruits, so they can have a diverse diet.  However their wages are denominated in coconuts, and are “sticky” in coconut terms.  But they aren’t paid in coconuts!  Thus suppose they were all paid a wage equivalent to one coconut per hour, and their productivity is 1.2 fruit per hour.  To make the math easy, assume the market price of all types fruit is initially equal to one fruit per coconut.  Thus initially the equilibrium allows the owners to earn 0.2 fruit/worker/hour in profit, and everyone has a job.  Their contract specifies they be paid 8 coconuts worth of fruit per 8 hour day, which at initial prices means 8 fruit for an 8 hour day, and they trade fruit with each other in flexible price markets.

Then a coconut blight kills many of the coconuts, so coconut worker productivity falls in half.  The price of coconuts in terms of other fruits soars, and yet workers continue to insist on being paid (the equivalent of) one coconut per hour.  So if the price of bananas fell to 1/4 coconut, the workers would insist on being paid 4 bananas, so that they would be earning the one coconut that their contract specified.  Unemployment results, as they are just not that productive.  (Although liberals would accuse me of blaming the victim.)

Think my example is far-fetched?  Think the workers would be happy getting their usual one fruit per hour?  Think again.  In 1930 the workers of the major industrial countries were paid in gold.  Not gold itself, rather their wages were denominated in terms of so much gold per hour, paid in some other medium.  Then in 1930 global gold hoarding caused the value of gold to soar.  The price of the commodities that workers buy fell in terms of gold.  You might think workers would say to their boss; “We understand that gold has become more valuable, and thus we don’t need as much to buy our usual purchases, so you can pay us an amount of MOE that buys less gold, as long as we can buy our usual goods.”  But the workers did not say that.  Why not? What’s the title of this blog? They said “We insist on being paid in gold, even though there isn’t enough gold in the world for full employment.  We don’t care that our gold wages will now buy more goods and services, we demand payment in gold.”  And keep in mind that many workers never owned a gold coin in their life.  They were poor.  Yet it remained a token with mystical powers to the workers, a sort of barbarous relic.  The workers crucified the owners on a cross of gold.  (Wow, I’m inventing some great metaphors phrases today!)

BTW,  America’s labor leaders opposed FDR’s devaluation.

In this coconut example there is no medium of exchange.  It’s a barter economy.  But there is a medium of account.  And changes in the value of the MOA cause business cycles.

2.  Now for a sticky price version.  We will assume that prices are sticky in terms of coconuts, but coconuts are not the MOE.  Goods are bartered.  Once again, we assume a coconut market shock that makes coconuts worth more.  What happens?  I seem to recall Nick arguing that nothing happens, people continue to barter as before.  There may be a problem in the coconut market, but no generalized problem of deficient AD.  Yes, that’s one possibility. Score one for Nick.

3.  But it depends on what you mean by “sticky prices.”  In case 2 all transactions occur at market clearing prices.  So in a sense the problem of stickiness is being assumed away.  So I like to imagine a world with a MOE and MOA that are different.  No more barter.  And also assume flexible prices between the MOA and MOE.  Prices are sticky in terms of the MOA, but a varying amount of MOE is needed for transactions.  Real world analogies might be with US dollars as the MOA and Zimbabwe dollars as the MOE, or gold as the MOA, and silver coins as the MOE.  Let’s do the latter.

Assume once again that the demand for gold rises, and gold become more valuable in terms of both silver and all other goods.  Assume the equilibrium value of silver in terms of all other goods is unchanged.  Recall that prices are denominated in terms of gold, the MOA.  Thus something that used to cost 1 silver coin might now cost two.  Because the quantity of silver is unchanged, and thus it’s relative value in terms of goods is unchanged, this is a negative demand shock.  But there are two ways to visualize this case:

a.  The MOA got more valuable while nothing happened in the MOE market; hence the MOA is the key variable.  (My view)

b.  The stock of MOE measured in terms of gold has fallen in half, thus it’s a MOE shock that causes the recession.  (Nick’s view.)

I think in terms of the quantity of MOE, whereas Nick thinks in terms of the value (in MOA terms) of the MOE.

So I’ve considered four cases.  Case 1 and 3a and 3b are all consistent with the MOA driving the cycle.

Cases 2 and 3b are consistent with Nick’s view; the MOE is the key variable.

So it’s three to two, but somehow I don’t expect to maintain the lead when he replies.

PS.  I vaguely recall Keynes saying around 1930 that the basic macro problem was that British wages were too high relative to the amount of gold in the world.  Does anyone recall?  It was when Keynes served on a commission looking at monetary problems.

Update:  Commenter Arthur pointed to a good example of a split MOA/MOE in Brazil, from Marcus Nunes.

Reply to George Selgin

In a recent post George Selgin made the following comment:

Scott Sumner, like Milton Friedman, forthrightly denies that there’s such a thing as booms, or at least of booms caused by easy money, to the point of taking exception to a recent statement by President Obama to the effect that, among its other responsibilities, the Fed should guard against “bubbles.” But here, and unlike Friedman, Sumner basis his position, not merely on the claim that prices are more flexible upwards than downwards, but on a dichotomy erected in the literature on asset price movements, according to which upward movements are either sustainable consequences of improvements in economic “fundamentals,” or are “bubbles” in the strict sense of the term, inflated by what Alan Greenspan called speculators’ “irrational exuberance,” and therefore capable of bursting at any time. Since monetary policy isn’t the source of either improvements in economic fundamentals or outbreaks of irrational exuberance, the fundamentals-vs-bubbles dichotomy implies that monetary policy is never to blame for changes in real asset prices, whether those changes are sustainable or not. If the dichotomy is valid, Sumner, Friedman, and the rest of the “monetary policymakers shouldn’t be concerned about booms” crowd are right, and the Austrians, Schwartz, Taylor, and others, including Obama and his advisors, who would hold the Fed responsible for avoiding booms, are full of baloney.

I’m happy to reassure George that I do not believe the things he claims I believe.  I believe the Fed often creates booms, and that these booms often lead to recessions. So in that sense my views are quite Austrian. I am particularly surprised by his claim that I don’t believe that monetary policy affects real asset prices, as he recently commented on a post that was devoted to exactly that proposition:

Now here’s where I part company with Keynesians who might have been with me so far.  Although short term interest rates are one of those “asset prices” that cause the money market to achieve near instantaneous equilibrium, even as the goods and labor markets are in disequilibrium, they actually have very little role in moving NGDP and prices to the level necessary to restore long run macro equilibrium (and to move interest rates back to their original level.)  In my view 60% of the heavy lifting is done by what Keynes called “confidence” and I call “expectations of NGDP growth” and Ford Motors economic forecasters call “expected nominal incomes in 2014 available to buy Ford cars.”  Another 35% of the transmission is done by asset markets like stocks, forex, commodities, real estate prices, junk bond yield spreads, etc.  And maybe 5% by risk-free short term rates.  At most.

So I just claimed that 35% of the transmission effect of monetary policy works through changes in real asset values, and have been saying similar things all along. George is a smart guy, so clearly something I said was misleading, or created a false impression. Perhaps it’s my denial of “bubbles.” I believe in the EMH (i.e. no bubbles), but only for asset markets. Because goods and labor markets have sticky wages and prices, they are not efficient, and monetary stimulus creates booms and busts in terms of output.  In some cases, such as the 1970 recession, the blame is almost 100% the preceding boom. Indeed the preceding boom also played a big role in the next few recessions. Where I differ from some Austrians is that I believe the preceding booms in 1929 and 2007 were not major factors in the subsequent slump. In those two cases I think tight money is mostly to blame, perhaps 90% or more.  It’s hard to be more precise as the trend line is a judgment call (in the absence of NGDPLT.)

I see booms and bubbles as unrelated phenomenon.  If boom means “excessive nominal spending” then I don’t see a strong correlation between booms and bubbles.  The biggest excesses in AD tended to be associated with almost no bubbles.

In the standard AS/AD model, which I accept, business cycles can be created by either increases or decreases in AD.  It’s symmetrical.  The only asymmetry is that the SRAS curve probably becomes somewhat steeper to the right of the LRAS, and hence overshoots do less damage.  That’s one reason booms feel less bad that recessions.  The other reason booms feel good is that they overcome government distortions that normally hold employment below the optimal level.  If only they could do so forever!  But alas, there is a price to pay, as Selgin has ably pointed out.

I do agree that a few market monetarists may underestimate the importance of symmetry, the importance of avoiding excessive expansion in NGDP.  It’s a big group of people now, and many came on board with this crisis, which obviously affects their overall worldview.

Here’s George again:

With so many old-school monetarists switching sides, the challenge of denying that monetary policy ever causes unsustainable booms, and of claiming, with regard to the most recent cycle, that the Fed was doing a fine job until house prices started falling, has instead been taken up by Scott Sumner and some of his fellow Market Monetarists.

Housing prices started falling in 2006, but I never criticized the Fed until it was clear to me that NGDP was falling in late 2008.  I wrote posts pointing out the broader economy did well during the first 27 months of the housing bust.  If my memory is correct, some Austrians were still skeptical that money was too tight in November 2008 (at the SEA meetings), just as Hayek was skeptical that money was too tight in the early 1930s, even when his NGDP norm would have called for easier money.  I will be glad to stand corrected if wrong.

I want central bankers to ignore “bubbles” and focus like a laser of stabilizing NGDP.

Here’s George:

In the meantime, it seems to me that there is a good reason for not buying into Friedman’s view that there is no such thing as a business cycle

This strikes me as odd.  Friedman clearly believed there are business cycles as I understand the term (ups and downs in GDP.)  Maybe Friedman criticized the term “cycle” as implying regularity.  But he clearly believed monetary stimulus was unhealthy in the 1960s, and his natural rate model of 1968 implies that the easy money in the 1960s was to blame for the 1970 recession (which could only have been avoided by steadily higher inflation leading to hyperinflation.)

I do disagree with George on a few points (the EMH, the importance of Cantillon effects, etc) but not on nearly as many points as he assumes.

Interest rates versus the base

Here’s Paul Krugman:

But in this more complex world, where even the definition of the money supply becomes highly dubious, why even talk about an LM curve? Well, before 2008 most macroeconomists didn’t! They talked instead about interest rate targets, Taylor rules, and all that. Mike Woodford, who is probably our leading macroeconomist’s macroeconomist, has even made one of his signature modeling tricks the building of models in which there is (almost) no outside money. Sensible macroeconomists have known for a long time that quantity-theory type models, if they were ever useful, aren’t much use in the modern economy.

In normal times central bank monetary policy is conducted in terms of, and best thought of in terms of, the target interest rate

This is certainly the conventional view, but I think it’s wrong.  Let’s start with the fact that just as there are no atheists in a foxhole there are no non-monetarists during a hyperinflation.  When prices rise 8700% (almost) no one tries to explain the path of prices by referring to the path of interest rates. Even Wicksell and Keynes became quasi-monetarists during the early 1920s hyperinflations.  The reason is simple.  Interest rates tell us nothing about the level of prices and NGDP, whereas the base does.  Thus huge changes in the price level and NGDP can only be explained by looking at changes in the base.

[Matt Yglesias denies that money causes hyperinflation.  But all he’s really saying is that the monetary deluge that causes hyperinflation has a REASON.  I.e. Latin American countries would choose to spend more than they received in taxes, and printed money to cover the deficit.  Countries with identical deficits, but good access to credit markets, would not print money and would not have hyperinflation.  It’s not the deficit, it’s the money printing.  As an analogy, Matt’s claim would be like asserting that fiscal stimulus did not boost employment in 1942, WWII did.]

And if money explains hyperinflation, it also determines the path of prices and NGDP at lower growth rates, it’s just that the effect is disguised by the relatively greater importance of money demand (or velocity) fluctuations.  Conventional economists would claim that because velocity is volatile, interest rates are “more useful” way to think about monetary policy.  But they are not.

In the standard model, fluctuations in NGDP are caused by movements in interest rates.  And yet rates tend to be high during booms and low during recessions.  So how is the interest rate approach “more useful?”  Here’s where the NKs get clever; it’s not the interest rate that matters, it’s the market rate relative to the Wicksellian equilibrium rate.  Does this sound familiar?  Sort of like the monetary base relative to velocity.  Except that MV = PY is a tautology, whereas they merely have a theory.  At least unless you define the unobserved equilibrium rate as the one that produces steady growth in aggregate demand (PY).  In that case it’s also a tautology.  But how is it “more useful?”

Another problem with the nominal interest rate is that the policy lever locks up at zero rates, and hence central banks have trouble communicating at the zero bound.  In contrast, the base has no zero bound, and hence there is no point at which central banks are unable to communicate by adjusting a policy instrument.  For example, suppose the Fed indicated that they would keep increasing the size of their monthly QE purchases by 20% each month until expected NGDP growth got back on target.  That’s a clear strategy.  It would work very quickly (or else they’d own the entire universe quite quickly).  But a promise of low interest rates until some objective is met is basically consistent with Japan’s performance over the past 15 years.  You might do better, but you might not.

Another argument used in favor of the interest rate approach is that, out in the real world, people and policymakers think in terms of interest rates, not the base.  But that’s exactly the problem.  People think money has been easy since 2008 (even though according to Ben Bernanke’s criterion it’s been the tightest since Herbert Hoover was President), precisely because they’ve been taught that monetary policy is “best thought of” in terms of interest rates. They’ve been taught to look at not just a poor indicator, but one that is actually negatively correlated with the actual stance of monetary policy.  A poorly informed public will make bad public policy decisions.  And not just the public, even economists are confused.

At the same time the supply of base money is also an unreliable indicator (although less so than interest rates.)  Thus we should also not view changes in base money as a good indicator of the stance of monetary policy.  Rather changes in the base supply relative to base demand are what is important. As Ben Bernanke said, NGDP growth is the best indicator of whether money is easy or tight.

PS. OK, Bernanke said NGDP growth and inflation.  But he had to say inflation, NGDP was his choice.