Don't even try to "Normalise" interest rates

If you think that the rest of the economy is normalised, so it is time to normalise interest rates too, you are wrong. If the rest of the economy is normalised, then interest rates must already be normalised.

Wicksell said there was some underlying "natural" rate of interest. If the central bank sets a rate of interest below the natural rate this would cause inflation; and if the central bank set an interest rate above the natural rate this would cause deflation. So any central bank that wanted to stabilise the price level should aim to set the actual rate of interest equal to that underlying natural rate of interest. And the actual rate of interest would need to be (roughly) equal to the natural rate on average over any extended period of time if we wanted price stability.

We haven't advanced much since Wicksell. The only important thing a Neo-Wicksellian would add is that it's important to distinguish between nominal and real rates of interest (real = nominal minus inflation), so if we have a 2% inflation target we add 2% to the natural rate to get the "neutral" nominal rate. And the actual nominal rate would need to be (roughly) equal to that neutral nominal rate on average over any extended period of time if we wanted to keep inflation stable at 2%.

You can see how this matches the idea of "normalising" interest rates. We've got "normal" both in the sense of an average over time, and in the sense of a desirable norm that central banks should aim for.

But when I hear the words "The Bank of Canada (or some other central bank) needs to normalise interest rates soon" I reach for my shovel. Let me try to tell you why.

If you want to argue against some view, as I am doing here, it is usually a good strategy to try to state what that view is. This is my best guess:

'The economy was humming along normally, with the Bank of Canada setting the actual rate of interest roughly equal to the natural rate of interest, so the output gap was roughly zero, and inflation was roughly at the 2% target. Then the financial crisis happened, so the output gap turned negative (a recession), and inflation started to fall below target. So the Bank of Canada needed to cut the actual interest rate below the natural rate temporarily to offset the recession. But now the economy has recovered, with the output gap back to roughly zero, and inflation roughly on target, it is time for the Bank of Canada to normalise interest rates, by raising them back up to the natural rate again.'

I don't want to quibble with the details of that view -- whether the output gap really is zero, or whether inflation really is back to target -- because the only difference that makes is whether we should normalise now or wait a few more months. I want to attack the idea of "normalisation" itself.

1. The natural rate of interest is a moving target. What was normal in the past won't be normal today or in the future. Almost anything that changes desired saving or investment, both in Canada or globally, will change it.

2. The natural rate of interest is a target we can't see, except looking backwards; we can see (roughly) where it was in the past, by looking at past output and inflation, but we can't see where it is now, and where it's going to be in the future.

3. The natural rate of interest is a target that moves when we miss it; even worse, it tends to move even further away from where we thought it was.

4. There is not just one natural rate of interest, because there is not just one actual rate of interest. There are many rates of interest, depending on the liquidity, risk, and term, of the financial asset in question. Each of those many rates of interest has its own natural rate, and the spreads between them will change over time.

5. But it's the bit I bolded above that shows what is really wrong with the "normalisation" view: "So the Bank of Canada needed to cut the actual interest rate below the natural rate temporarily to offset the recession."

Here's an alternative way of viewing the world: something (the financial crisis) happened that caused the natural rate to fall suddenly. The recession happened because the natural rate fell more quickly than the Bank of Canada cut the actual rate, so the actual rate was above the natural rate. Now that the economy has recovered from the recession, with the output gap roughly zero, and inflation roughly at target, this means that the actual rate of interest is roughly equal to the natural rate of interest. Normalisation of interest rates has already (roughly) happened. The new normal is not the same as the old normal. And what is normal next year, and the year after, will probably be different again.

If the natural rate of interest really were a constant that never changed over time, or always changed slowly and predictably, it would make sense to talk about "normalisation" of interest rates. But then recessions would only happen if central banks suddenly decided to raise interest rates above the unchanged natural rate.

If we had some sort of crystal ball, however imperfect, that could tell us what the natural rate was, it might make sense to talk about "normalisation" of interest rates. But we don't. We can only look at output and inflation to try to surmise where the natural rate is. If the output gap is zero, and if inflation is on target and looks like staying there, we can only surmise that the natural rate must be whatever the actual rate currently is.

If you think that the rest of the economy is normalised, so it is time to normalise interest rates too, you are wrong. If the rest of the economy is normalised, then interest rates must already be normalised.

Comments

[Iskander]

Great post Nick. Do you ever find the wicksellian framework useful? Simplicity seems to be one of its advantages and I think the focus on interest rates is a residual from the days of the gold standard, once we left our golden fetters central bankers have continued with it ever since (aside from a very brief departure in the 1980's).

[louis]

Important, and very clearly stated.
I am amazed at how many otherwise intelligent and informed finance people don't get this principle. They talk of rates (including LT rates) as being "artificially low" due to central bank policy, as opposed to being a consequence of real forces in the economy to which the central bank is reacting.

[Benoit Essiambre]

I like to point out that low or negative real returns on stores of value have historically been the norm. Before financial systems existed, almost all investment had negative returns if you didn’t put work and energy into them. To store value, you had to accumulate stuff, buildings or land. Most options either had high maintenance costs, were subject to risk of damage from natural causes and theft, were very volatile or required hard labor to get production out of.

Even more recently, it has often been difficult to get low risk, hassle free, liquid, positive real returns. From a basic science perspective this seems to reflect the laws of thermodynamics that tell us that everything tends to decay without a constant supply of work and energy. In general, most things require maintenance to keep their worth.

The 20th century may have been an outlier. Because of unprecedented demographic and technological growth, positive risk free real returns on liquid assets were easy to find. It is possible that under favorable conditions, wealth can have positive returns and even compound into very good long run returns but it is not a guarantee and there is nothing natural about it. It may not continue forever, particularly amidst an aging and retiring population in a world no longer as rich in easy to exploit natural resources.

While people are used to get negative returns on very short term purchases, you buy fresh vegetables at the supermarket even if they degrade over time, they can’t seem to accept the normalcy of negative returns on longer term assets despite the tendencies inherent in the laws of physics. In nature, squirrels’ nut caches have a certain percentage of losses from theft and spoilage. Returns tending towards the negative are natural even if they can seem unusual for people just out of the 20th century.

I'm almost entirely talking supply side here. As you mentioned, demand for investment can also push down rates in some situations such as when a wave of people are about to retire and are doing last minute build up of their nest egg.

[Nick Rowe]

Iskander: Thanks!

If we think in terms of central banks setting an interest rate, then some sort of Wicksellian approach seems not just useful but unavoidable. And that's how the Bank of Canada talks, and how people think, and in this post i wanted to talk to them in their own language. But ultimately, no, I would prefer we thought about monetary policy some other way.

louis: thanks! Yep, they think in terms of rates being "artificially" low, which seems to imply that the natural rate is some sort of physical constant that never changes. It's some sort of bowdlerisation of Austrian economics, I think, but I can't get further than that.

Benoit: "Before financial systems existed, almost all investment had negative returns if you didn’t put work and energy into them."

I like that point. Wealth doesn't just manage itself.

[Ralph Musgrave]

So Wicksell, according to Nick, said the central bank should set the rate of interest at the “underlying natural” rate. Isn't that a contradiction in terms? If the central bank, or rather the state as a whole (i.e. government and central bank) simply desists from interfering in any way with interest rates, than those rates will be at their “natural” or “free market” rate won’t they?

But that in turn begs the question as to how much governments should borrow because the more a government borrows, the higher the rate of interest. Strikes me Warren Mosler and Milton Friedman got the answer to that one right: governments should borrow nothing. I.e. the only government or state liability (if you can call it that) should be zero interest yielding base money.

As to what the base rate (at least as defined by the Bank of England) should be, that should be Walter Bagehot’s “penalty rate”: if a commercial bank messes up and is short of base money, it should be charged a penalty for messing up.

[Nick Rowe]

Ralph: we can imagine a different world, where central banks set something else (e.g. the stock of base money, or an exchange rate), and it might be a better world, though their actions would still affect interest rates. But I'm talking about the current world here.

"Strikes me Warren Mosler and Milton Friedman got the answer to that one right: governments should borrow nothing. I.e. the only government or state liability (if you can call it that) should be zero interest yielding base money."

You keep saying that, but you don't seem to understand what it means. As George Selgin explained to you on Twitter, Friedman in one paper suggested the central bank make inflation sufficiently negative so that nominal interest rates on government bonds would be 0%, even though real interest rates (which are what matters for the government's liability) are positive. He proposed that policy to make the opportunity cost of holding currency (as opposed to bonds) zero, not to make it cheap (in real terms) for the government to borrow. It would actually *raise* the government's borrowing cost, because it could no longer borrow at negative real rates by issuing currency.

No more on that topic in this post. It's off-topic.

[JKH]

I really haven't thought this through, but I'll pose the question anyway:

Consider a term structure of natural rates (for a one chosen natural rate type):

Assume the central bank sets the actual overnight rate.

Two scenarios:

a) The actual overnight rate equals the natural overnight rate
b) The actual overnight rate is less than the natural overnight rate

Question:

Do you think the term structure of natural rates should be different for those two scenarios?

If so, I assume the more general implication is that monetary policy (almost) always affects the natural rate of interest (considered comprehensively across terms structure). Maybe that’s supposed to be obvious. But it’s not clear to me.

[Nick Rowe]

JKH: Good question. I think the answer is yes. Take a simple case, with just a "short" and a "long" interest rate. And suppose desired saving and/or investment depend on both interest rates. Starting in equilibrium, if the central bank lowered the short rate, there would need to be (counterfactually, because it wouldn't actually happen) a rise in the long rate to bring desired saving and investment back in line at "full employment".

I think this is a particular sort of application of my point 3 above.

[Matthew Young]

The central bank knows the uncertainty bounds on the natural rate.

The central bank can look at seasonally adjusted price variation across its domain and that price uncertainty is a distribution of price. The shape of the distribution should mirror the shape of borrowers balanced against depositors.

So, the central bank can look at the imbalance between loans and deposits, discover the imbalance relative to any price variation and set interest charges such that the distributions match within the known error bounds.

[rsj]

"If you think that the rest of the economy is normalised, so it is time to normalise interest rates too, you are wrong. If the rest of the economy is normalised, then interest rates must already be normalised."

You could have stopped right here, it's so obvious.

[Nick Rowe]

rsj: funny thing is, I had a really hard time writing this post, and only wrote that bit when I had finished writing everything else. And then I realised it belonged at the beginning, as well as at the end!

[Colin Percival]

If "the economy is normalized" means "inflation is at 2%", then you're absolutely right. If anyone is making the argument "inflation is currently at 2% therefore we need higher interest rates", they're crazy.

If "the economy is normalized" means "economic conditions are such that, over the next few years, holding the overnight rate at 1% will result in inflation increasing beyond the 1-3% target", then you're absolutely wrong.

I don't know what people mean when they say "the economy is normalized"...

[Nick Rowe]

Colin: if someone said "inflation is 2% now, and output is at potential now, ***but output is growing faster than potential, and so inflation will rise above 2%, unless we start raising interest rates now***", I agree that would be a good reason to start raising rates (assuming I agreed with their conditional forecasts).

But I don't remember hearing that additional bit in ***...***.

I was hoping someone who was in favour of the "normalisation" argument might come on and spell out the case better than I can.

[Thomas Hutcheson]

One factor that should affect R* is whether or not the market expects BOC/Fed to actually follow their inflation target. In 2008- inflation was allowed to fall below target. Would having held inflation at 2%, with catch-up if necessary have prevented the recessions? We do know that the failure to hold to the target was associated with a failure to prevent the development of an output gap. And since the price level has yet to return to target, it is not clear whether the Fed is committed to an inflation target in the future. Does thin not make future recessions more likely and therefore suppress the present desired level of investment and therefor R*?

[Nick Rowe]

Thomas: "Would having held inflation at 2%, with catch-up if necessary have prevented the recessions?"

I used to think the answer was (at least roughly) "yes" (though it would have taken a very clever or lucky central bank to be able to keep inflation exactly on target despite a big shock). But after seeing the data from the last recession, I have changed my mind on that one. The BoC kept core inflation roughly on trend (though not precisely), but we still had a recession.

" Does thi[s] not make future recessions more likely and therefore suppress the present desired level of investment and therefor R*? "

Central banks' failure to prevent the last recession would presumably make future recessions more likely, and this could indeed reduce desired investment currently and so reduce r*. I think that is very plausible. But what I don't know is how big this effect would likely be.

[Frank Restly]

Nick,

I think Wicksell limited his definition of "natural rate" to the interest rate that generates price stability without regard to GDP or output gap measurements.
I realize that Taylor rule type guidelines recommend interest rate adjustments that use both output gap estimates as well as inflation rates.

"But when I hear the words The Bank of Canada (or some other central bank) needs to normalise interest rates soon I reach for my shovel. Let me try to tell you why."

I understand your rationale, but I also understand that independent monetary policy serves as a check on expansive fiscal policy.

Is it possible that a central bank can forgo independence to actively support fiscal measures?

How do you tell the difference between a central bank that is setting an interest rate based on a government's fiscal position and a central bank setting a rate based on some measure of the natural rate?

[Frank Restly]

Nick,

My post above is a political-economy type question that goes beyond inflation and GDP measurements and so if you want to pass, I understand.

[Nick Rowe]

Frank: I think you are asking 2 separate questions:

1. The output gap can't strictly be observed, because we don't observe "potential" output Y*. Y* (and the natural rate of unemployment u*) is a bit like r* in that regard. Which opens up a whole can of worms about what we mean by saying the rest of the economy (the output gap, or unemployment) is "normalised". Which I wanted to avoid for this post. So you can read me as saying that *even if* you could convince me that Y=Y* and u=u*, it does not mean we should raise r to where it was before the recession.

2. Since we don't observe r*, the only way in practice we can prevent a central bank becoming subservient to the government's fiscal position, and keep it accountable, is to give the central bank a target that we can observe, like 2% inflation.

[Frank Restly]

Nick,

"Since we don't observe r*, the only way in practice we can prevent a central bank becoming subservient to the government's fiscal position, and keep it accountable, is to give the central bank a target that we can observe, like 2% inflation."

That presumes that government expansion leads to higher inflation rates. In socialist regimes where government takes over private enterprise, higher inflation rates are not a given.
Instead, consumer choice may become limited or other undesirable outcomes may follow.

It seems that a central bank that is striving to stay independent of fiscal policy should be looking at the government's demand for credit in addition to the prevailing inflation rate / inflation expectations.