Scott Sumner  

Raising rates tightens money, but isn't (necessarily) tight money

A Good Krugman Post... How the Welfare State Melts Yo...

Bob Murphy has a post criticizing me for being inconsistent on the subject of interest rates and the stance of monetary policy. Sometimes I say that changes in interest rates don't tell us anything about the stance of monetary policy. Indeed more often that not, higher rates indicate easier money (as in the 1970s.) Other times I suggest that raising rates is a tightening of monetary policy, as when I said there is no justification for the Fed raising its target rate right now, as we are already likely to fall short of its policy objectives.

Perhaps it's easiest to explain this seeming contradiction by using a real world example: Fed policy during the 1970s. Since most people like to think about monetary policy in Keynesian/Austrian (interest rate) terms, let's use the best interest rate model---Wicksell's natural rate concept.

Wicksell said that an easy money policy occurs when the central bank sets the policy rate below the natural rate of interest, resulting in inflation. In contrast, if the central bank sets the policy rate above the natural rate of interest you have a contractionary policy and end up with deflation. (Substitute rising and falling NGDP, if you prefer.)

In the late 1960s and 1970s, the Fed consistently set the policy rate (fed funds target/discount rate) below the natural rate of interest. This led to ever-higher inflation rates from 1965 to 1981. And here's what's important. The higher inflation caused the Wicksellian natural rate of interest to rise steadily---via the "Fisher effect." Thus the Fed found itself in a position where it needed to set rates at higher and higher levels in order to get inflation under control.

This invisible rise in the natural interest rate is what causes confusion. It looked like the Fed was steadily tightening monetary policy during the 1970s, and yet inflation kept rising. (That's because the natural rate was rising even faster.) Hence the Fed (and many private sector economists who should have known better) concluded that monetary policy was not responsible for the high inflation. In fact, while rates were being raised, due to inflation they were falling further and further below the natural rate of interest.

OK, this shows rates are misleading, but it still doesn't address Bob's complaint. How can I say higher rates tomorrow would be monetary tightening, when I've just said the opposite? Here we need the ceteris paribus condition. During any of those meetings during the 1970s, a decision to not raise the target rate would have meant easier money than a decision to raise the target rate. Not raising the target rate would leave rates even further below the Wicksellian equilibrium, leading to even more inflation. Raising them was not enough to lead to tight money in an absolute sense, but on that particular day a rise in interest rates meant tighter money than not raising them.

Similarly, the decision by the Fed to raise rates or not raise rates in September may not influence whether money is expansionary or contractionary in an absolute sense, but a decision to raise rates in September will make money more contractionary than not raising rates in September. What would be required for a Fed move to change the absolute stance of policy? Simple, if the Fed action pushed rates from below the Wicksellian equilibrium to above the Wicksellian equilibrium, then the stance of monetary policy would flip from expansionary to contractionary. Those sorts of flips don't occur very often, and when they do they often have dramatic macro effects.

Similarly, you can't tell whether a car is going fast or slow by looking to see whether the accelerator pedal is depressed. I could be depressing the pedal and going 15 mph. Or I could have my foot off the pedal and be coasting 60mph down a steep mountain road. But I can say that depressing the pedal will make the car go faster, ceteris paribus, than not depressing the pedal.

In macro, look at the speedometer (NGDP futures prices) not the position of the accelerator pedal (interest rates.) If the Fed switched from interest rate targeting to NGDP futures targeting, the market rate of interest would always be exactly equal to the Wicksellian natural rate of interest. By analogy, if you use cruise control to drive 70 mph across Nevada, then the accelerator pedal adjusts up and down automatically, always staying at the Wicksellian equilibrium natural rate of accelerator depression.

OK, I've stretched that analogy about as far as it will go. . . .

Comments and Sharing

COMMENTS (20 to date)
Kenneth Duda writes:

Nice post Scott. I found Bob's post puzzling, because he seems to understand market monetarism, and basically resolved the "inconsistency" himself. His implication that you want to debase the dollar also made no sense, because how much dollar debasement could you possibly get if NGDP is rising only 5% per year?

D. F. Linton writes:

So without an NGDP futures market, monetary policy is technically an unobservable variable?

We can't infer it from interest rates or the size of the monetary base.

Are we at Russ Robert's position that any number of just-so stories can be told based on what we can observe in the economy?

Michael Byrnes writes:

Excellent post - your clearest one yet on this topic.

So NDGP targeting = cruise control? :)

Scott Sumner writes:

Thanks Ken. That reminds me that I need to do a post on why NGDP targeting is not particularly inflationary.

DF, We do now have an NGDP prediction market, but even without one I believe that actual NGDP growth is a pretty decent proxy for the stance of monetary policy. Ben Bernanke made the same argument in 2003.

And yes, you definitely cannot infer the stance of monetary policy from interest rates or the base. Those indicators make Australian policy in recent years look contractionary, whereas it's actually been much more expansionary than in most other developed countries.

Michael, Yes, cruise control.

Andrew_FL writes:
If the Fed switched from interest rate targeting to NGDP futures targeting, the market rate of interest would always be exactly equal to the Wicksellian natural rate of interest.

The rate will probably deviate away from the natural rate by smaller amounts, but it still wouldn't be exactly at the natural rate, certainly not as long as the Fed exists. But whether it tends to have small, unbiased errors around the true natural rate depends upon the target. A target of a high positive growth rate doesn't have the same implications for the market rate as compared to the natural rate as would a 0% growth rate target or a negative growth rate target.

Still it seems to me that your statement that there was no justification for raising rates based on the current rate of wage growth, was not a statement about there being no justification for making policy relatively tighter. You seemed to be saying that low wage growth was prima facie evidence that the natural rate is at least as low or lower than the present market rate, that is you seemed to be suggesting that based on the rate of wage growth, raising interest rates would be an absolutely tight policy.

But how do you know what the natural rate should be?

BC writes:

"This invisible rise in the natural interest rate is what causes confusion."

This also seems to cause confusion when people say that the Fed needs to create some "headroom" now if it wants to be able to use conventional monetary policy during the next recession. They argue that the Fed needs to raise rates before the next recession so that it can lower them to fight that recession. However, to create headroom, the Fed needs to raise the *natural* rate, not the policy rate. If the Fed raises the policy rate too early, then that will cause the natural rate to fall, decreasing headroom.

ThomasH writes:

A similar point should be made about fiscal policy. Changes in fiscal policy, changes in the extent to which taxes and expenditures depart from the NPV rule, are expansionary or contractionary ceteris paribus. If monetary policy did not depart from an NGDP, or PL rule it would automatically offset the change in fiscal policy.

Generally this is not a good assumption to make, mainly because monetary authorities follow politically constrained, quazi-inflation ceiling targets that mean they usually do not offset fiscal policy changes. This is why "Keynesianism" can work in practice even though it would not work in theory with good monetary policy.

Brian Donohue writes:


I've asked this before:

Is the Wicksellian (natural) rate a single, short-term rate, or is it a curve?

Is it nominal or real?

RPLong writes:

How do we know the natural interest rate was above the Fed's rate during the 70s? (I'm not questioning the veracity of the claim, I'm just not aware of how we measure that sort of thing.)

If the answer is, "We know because inflation kept increasing even though the Fed kept increasing the interest rate," then that feels a little like affirming the consequent. Is there some other way we know?

Andrew_FL writes:

@Brian Donohue-Wicksell abstracted away from the term structure of interest rates and the like, but logically it should in the real world be a whole set of rates.

Asking whether it's real or nominal on the other hand is a question that doesn't make sense. Like any interest rate it consists of a real component and an inflation premium. So it's both a real and a nominal rate. The real natural rate is the nominal natural rate minus the inflation premium. This is what Scott meant by mentioning the Fisher effect.

Wicksell would have said that if inflation is 0% and expected inflation is 0% (in which case, real = nominal) you'll happen to be at the natural rate but A) That'd be a coincidence and B) Wicksell was wrong. If productivity is increasing sufficiently fast consumer prices could be flat even if the "bank rate" is below the natural rate. Austrians claim this is exactly what happened in the 1920's.

@RPLong-If we had a way to measure the natural rate the Fed Funds rate should be at, it wouldn't be possible for people to disagree about whether rates are too high or too low at any point, and if we had that in real time, monetary policy would be trivial. But in fact we don't.

Benoit Essiambre writes:

Exactly! To me this language is clearer than trying to avoid speaking of interest rates at all when arguing about monetary stance.

Interest rates tie the concepts with micro foundations. As I mentioned before, people and businesses don't budget for debt payments in fractions of GDP, nor do they calculate profits in that unit. That is, at least until central banks start following an NGDP target and people use that unit whether they know it or not.

Travis Allison writes:

Scott, I haven't heard of the Wiksellian rate changing with inflation. Are you saying something different than Nick Rowe? Or has Nick left out some details?

Andrew_FL writes:

@Travis Allison-That's just a plain old ordinary inflation premium. Why would you think the natural rate would not account for expected inflation?

Philip George writes:

You might also add that not raising rates does not mean monetary policy is not contractionary.

As the graph on shows, the YoY growth rate has been falling since January 2014.

It explains the weird comments by some market analysts that the Fed's easing policies have led to a contraction in liquidity.

Jose Romeu Robazzi writes:

Prof. Sumner
You said "but a decision to raise rates in September will make money more contractionary than not raising rates in September"

Does that mean your model has lags as well (meaning first raise rates and then observe that monetary stance is tighter after the fact)

Or does it mean that fulfilling the market's expectation with a rate hike it will create the expectation of further tightening (the leads)?

By the way, don't know if anybody has said this before, I think your idea of whatching NGDP and not rates can be integrated into the Autrian Business Cycle Theory, in my opinion, with very lottle adjustments ...

RPLong writes:
If we had a way to measure the natural rate the Fed Funds rate should be at, it wouldn't be possible for people to disagree about whether rates are too high or too low at any point, and if we had that in real time, monetary policy would be trivial. But in fact we don't.

What I'm trying to figure out is whether it is common knowledge that the natural rate of interest increased rapidly in the 1970s, and if so, how/why we know it to be true.

If we don't know it to be true, then I might be correct that Prof. Sumner's argument affirms the consequent, and this might be one of the problems Dr. Murphy is pointing to in his post.

Brian Donohue writes:

Thanks Andrew_FL.

Here's why I'm asking:

The Wicksellian rate is unknown, but presumably the policy rate is known.

The policy rate is reflected in the current term structure of interest rates.

If we're talking about a nominal rate, this is the Treasury curve; if it's a real rate, this is the STRIPs curve.

Now, if we assume away noise related to "unexpected inflation", I presume it's fair to think of an (unknown) "Wicksellian real curve" as compared with the Treasury STRIPS curve.

I presume such a "Real Wicksellian curve" would be fairly stable over time, as it represents aggregate time preference factoring out inflation.

Over the past few years, the (observed) STRIPs curve has bounced around and changed its steepness quite a bit.

If the stance of monetary policy can be inferred from the difference between the (known, but volatile) policy rate and the "unknown but presumably stable" Wicksellian rate, it appears that monetray policy has swung quite a bit the past few years.

Of course, the Wicksellian rate is not so stable, it is harder to infer the stance of monetary policy or the direction of change (tightening vs loosening), but anyway...

/ramble over

Andrew_FL writes:

@Jose Romeu Robazzi-Yeah, if you read Prices and Production, Hayek emphasizes that the key variable that needs to be stable for business cycles of that kind not to occur, is the "effective money stream" or "amount of money payments. This is (sort of) like NGDP, but it's probably closer to Fisher's PT.

@RPLong-I think it arguably does, but I don't think that's the problem Bob has with it.

Though, if you're willing to accept a bit of a priori reasoning, it seems highly unlikely that people could have "anti time preference" so a negative real fed funds rate is pretty strong evidence you're below the natural rate. Of course, that would also be true now...

@Brian Donohue-I think it's a little risky to assume the natural rate in real terms is stable over time. It probably doesn't fluctuate wildly, but if it's unknown, there's no way to verify our assumptions about it.

Travis Allison writes:

@Andrew_FL if you read Nick Rowe's piece, he talks about the Wiksell rate as a real rate. That is what I assumed it was. I suppose Scott is using the Wiksell rate as real + expected inflation, since his comment makes the most sense in that context.

Still, I wonder if the Wiksell real rate changes with inflation. Perhaps if inflation/deflation changes labor force participation, the Wiksell real rate will change.

OK, I've stretched that analogy about as far as it will go. . . .

That is the best explanation of NGDP targeting I have read. Keep stretching it! It won't brake! (a pun!)

Seriously though it is very intuitive and helpful for the lay people (and Bob Murphys) out there.

Comments for this entry have been closed
Return to top