Is low inflation good news?

Suppose Statistics Canada reports the latest inflation data, and I am surprised to learn that inflation is lower than I had expected it to be. Or below the Bank of Canada's 2% target. Is that good news or bad news?

For most of my life I would have said that is good news. "So inflation is lower than I thought it would be. That means that the natural rate of unemployment is lower than I thought it was, and potential output is higher than I thought it was. So the Bank of Canada can and will loosen monetary policy more than I thought it could and would. And so employment and output will increase more than I thought they would, and unemployment will fall more than I thought it would, and so Canadians will be richer than I thought they would. That is good news."

And I think most macroeconomists would have said or thought something similar.

If we go back to the 1970's and 1980's, I would have added "And it's good that inflation is going to be lower in future than I thought it would be". Because I thought that inflation was too high back then.

But since the Bank of Canada started targeting 2% inflation, in the 1990's, I would not have added that extra bit. I thought that a 2% inflation target was about right, and I expected the Bank of Canada to hit that target, on average. So if inflation came in below the 2% target, I knew that the Bank of Canada had made a mistake, but I fully expected it to fix that mistake, and I expected future inflation to come back up to the 2% target. Hearing news about below target inflation did not affect my expectation about future inflation at all (except in the very near term).

Just by itself, the news about below-target inflation was (slightly) bad news, because it meant that the Bank of Canada had made a mistake. Just by itself, news about above-target inflation was also (slightly) bad news, for exactly the same reason. I didn't want the Bank of Canada to make mistakes in either direction. (These are usually "mistakes" only with the benefit of hindsight, of course, but they are still mistakes.)

But low inflation was good news because of what it told me about the rest of the economy. And high inflation was bad news because of what it told me about the rest of the economy. (Unexpectedly high inflation told me that the natural unemployment rate was higher than I thought it was, and potential output was lower than I thought it was.)

Whether some bit of news is good news or bad news depends on what you already know, and on how you expect policymakers to respond to that news.

I see no reason to view unexpectedly low or high inflation in Canada any differently today. Unexpectedly low inflation is good news, and unexpectedly high inflation is bad news, not in itself, but because of what it tells us. (Though the unexpected fall in oil prices that will probably be the proximate cause of temporarily lower inflation was bad news for Canada, as an oil exporter.)

But for some other countries, or regions, where I have less confidence in their central banks' willingness to do the right thing and fix their mistakes, low inflation may be very bad news. It tells me the danger of worsening recession has increased. Their central banks have made one mistake, and that makes it more likely they will make similar mistakes in the same direction in future.

As journalists might say, it all depends on the context. As economists would say, it all depends on the monetary policy regime.

Comments

[Tom Brown]

"I am surprised to learn that inflation is lower than I had expected it to be." Nick, Jason Smith predicted that by Dec 2015 using his information transfer model. It'll be interesting to see where it is then. http://informationtransfereconomics.blogspot.com/2014/07/worthwhile-canadian-prediction.html

[Nick Rowe]

Tom: you missed the "Suppose...." at the beginning of my sentence fragment you quoted. The full sentence reads: "Suppose Statistics Canada reports the latest inflation data, and I am surprised to learn that inflation is lower than I had expected it to be."

(And despite my efforts, I have no idea what Jason is saying. But please don't explain it here.)

[Erik Lund]

Alternative point of view: this is a freaking hair-on-fire emergency. Unexpectedly low inflation shows that our curent models are missing something pretty important about the Canadian economy. And, as we are missing it, we are not fixing it.

Spoiler: it's the low birth rate.

[Roger Farmer]

Nick
What is the natural rate? Do you have any advice on how I might ascertain its value?

[blink]

Very nice points. As usual, there are many ways one might update beliefs. You emphasize a "miss" as signaling the quality of the central bank. Another reasonable conclusion is that the economy is less predictable than I previously thought (also a negative inference).

[srin]

Nick,
I am not sure what inflation metric Canada targets. Both headline and core CPI inflation have been below the target for 2-year, 3-year, 5-year, and 10-year horizons. So, what is time period being used and what is considered being reasonably close enough to the target?

[Tom Brown]

Nick, right you are. Sorry!

[Too Much Fed]

"If we go back to the 1970's and 1980's, I would have added "And it's good that inflation is going to be lower in future than I thought it would be". Because I thought that inflation was too high back then."

And what if there was not a negative real AS shock from oil prices in the 1970's?

[Too Much Fed]

Tom, have you seen JKH lately?

[Nick Rowe]

Hi Roger: I don't have anything new to say about the "natural" rate.

It's a theoretical construct, that is only defined within a particular class of models. Your models, for example, I would call (for want of a better word) "natural *range*" models. You have a (roughly?) vertical LRAS curve or LRPC, but it's a *thick* curve, right? So they aren't strictly natural rate models, which have a thin vertical curve, but they aren't unnatural rate models, which have a sloping curve.

And like all theoretical constructs, we can't directly observe it, but can only infer its location (more or less accurately) from the data and the model.

Erik: looking at Canadian data over the last 20 years of inflation targeting, I see inflation come in below target roughly half the time, and above target the other half the time. The Bank of Canada always misses something, because it doesn't have a crystal ball.

blink: thanks! " Another reasonable conclusion is that the economy is less predictable than I previously thought (also a negative inference)." And that is an interesting thought too. Not exactly sure how to quantify it, because I always expect the BoC to miss (but don't know which direction it will miss), so it has to be a bigger miss than I expected, for me to change my beliefs about the BoC's ability.

srin: "Both headline and core CPI inflation have been below the target for 2-year, 3-year, 5-year, and 10-year horizons."

No they haven't. Here's the data. Unless I misunderstand you.

[srin]

Yes, it is the3 same data. I said 3, 5, and 10 year horizons, not the latest year. Also, I recognize the band is 1-3 but I expect over 10-year horizon to be close to the middle of the band, not 30 basis points below it. That is nontrivial. From what I can see, the 2% target appears be like a soft ceiling, not the middle of the band.

[Tom Brown]

Too Much Fed: I think he's commended fairly recently on one of Nick's posts.

[Majromax]

@Srin:

> Also, I recognize the band is 1-3 but I expect over 10-year horizon to be close to the middle of the band, not 30 basis points below it.

Not necessarily. The BoC's inflation target is a forward-looking target and not a price level target, so we wouldn't expect the bank to correct for previous errors. That means that we should expect to see the price level as a random walk imposed upon a 2% trend, if the BoC is doing its job.

As a random walk, the error in price level compared to 2% growth should become unbounded with time, following sqrt(#periods). That means that the error in observed inflation rate should trend to 2% only slowly, proportionally to 1/sqrt(#periods).

The only thing we can really say is that it's clearly observed that the BoC's errors over the past 10 years have not been symmetrical, but that should not come as a surprise in the context of the entire financial crisis thing. Canada did have significantly above-target total inflation (1y period) around 2002, but that's outside your 10-year horizon.

If anything, looking at the BoC's data suggests to me that total inflation and "core" inflation are less correlated than the BoC media would tend to portray, in that the latter doesn't look a great deal like a time-average of the former.

[srin]

Majromax,

I was under the impression that central bank's are omnipotent in hitting nominal targets!! Your explanation is nothing but poor rationalization. How exactly are we supposed to judge whether a central bank is hitting a target--what is the error bound and time frame? Nobody can see forward looking performance. All we can see is backward looking performance.

Whoever said that it is a price level target? Dont obfuscate. If you dont hit a target over any reasonable time horizon, then it is either not the target or you are failing. I am not talking about making up for being below inflation. I am simply judging performance and the lofty claims made about the BoC's supposed ability to unfailingly hit its target. In my book, if you have not hot the target in the past 3, 5, and 10 years, you are simply failing. The rest is rationalization.

[Nick Rowe]

srin: the BoC targets total, not core, to 2%. Except in the shorter term, where it uses core as an "operational guide" (which means it ignores fluctuations in total, provided core is smooth. So if total averages 2%, but core averages below 2%, that is how it should be. Here is an old post with graphs of total and core, compared to a 2% trend, since 1995. I don't see any systematic undershooting there. Except of course in core, which is supposed to undershoot 2%, if total is trending higher than core.

[Phil Koop]

Further to Nick's observation, the average of the last 20 annual inflation rates is 1.82%; the standard deviation is 0.75. Obviously, we cannot reject the hypothesis that the "true" (i.e. long-run) average inflation rate is 2% based on this evidence.

Have the years since the crash been special? Well I think they have been. But you could not prove this with historical inflation figures. Dividing the most recent 20 years into 4 5-year periods, the 5-year averages are 1.3%, 2.4%, 2.1%., 1.48%. None of these figures is significantly different from 2% and the most recent is actually higher than the first.