IT vs PLPT vs NGDPLPT for the Bank of Canada

The Bank of Canada's 2% inflation target comes up for renewal every five years, and it's up for renewal now.

To my mind, there are only three serious candidates:

IT. Inflation Target. Keep the existing 2% Inflation Target (with maybe minor amendments).

PLPT. Price Level Path Target. Like inflation targeting, and with the same 2% inflation, but if inflation drops below 2% one year you aim for higher than 2% inflation in subsequent years until the price level returns to the original path.

NGDPLPT. Nominal GDP Level Path Target. Like PLPT, and choosing a target growth rate for NGDP that gives roughly the same 2% inflation on average, but with NGDP replacing the price level as the target variable.

Before saying anything more, let's have a look at the historical data, which Stephen has kindly presented for me in two graphs. Thanks again Stephen!:

The first graph is CPI. Stephen chose the 2% trend line because 2% is the existing target for inflation. If the Bank of Canada had been targeting the price level path instead of inflation, it would have been trying to keep the economy on the trend line:

Here is NGDP (the years covered by the two graphs are slightly different). Stephen told me he plucked the 5% number out of thin air, but it seemed to work. If the Bank of Canada had been targeting the NGDP level path instead of inflation, it would have been trying to keep the economy on the trend line:

These two graphs meet the interocular trauma test.

Look at the first graph, for CPI. What do you see? Or rather, what don't you see? Unless you have your glasses on, you don't see the recent recession. An economist who looked carefully at that graph, and nothing else, would surmise there had been a short sharp boom in early 2008, followed by a short sharp correction, and a return to normal in 2009.

You don't need glasses to see that something big happened in 2008 on the second graph. An economist who looked at that graph would see a mild boom from mid 2005 to mid 2008, followed by a sharp recession and slow recovery that is still continuing in 2011.

Which of those two pictures gives you a more accurate and useful summary statistic for the Canadian macroeconomy?

I'm tempted to finish by saying "NGDPLPT is better. The second picture shows monetary policy was a little too loose before mid-2008, and too tight after mid-2008. Which it was. I rest my case". But I had better not.

Let me first say that I was surprised to see Stephen's pictures. I really didn't expect them to tell such different stories.

I have been supporting NGDPLPT for the US. But Canada is not the US. The US is starting from scratch; Canada has 20 years of successful IT behind us. "It ain't broke, don't fix it" is an argument to be taken very seriously, especially in macroeconomics, where we don't really know much with any degree of certainty. Plus, we have all gotten used to IT; the Bank of Canada has learned how to do it, and everyone else has learned how to live with it. A lot of people have made their plans for the future based on a 2% IT. We don't mess with that quasi-constitutional rule without good reason. There's a lot to be said for small-c conservatism, especially when it comes to monetary policy, which we don't understand well enough, and which should create a stable rule that people can rely on.

So I've been mostly sitting on the fence when it comes to IT vs PLPT vs NGDPLPT in Canada. But the recent experience of the Bank of England did give me some reason to prefer NGDPLPT over IT (see my old post).

The Lucas Critique is real, and important. We can draw any line we like on a graph, but that doesn't mean the Bank of Canada can actually hit that line. And even if it tries to hit that line, or at least come closer, we don't know what else would have changed if the monetary policy regime had been different than it in fact was. Theory tries to tell us what else would have been different, but theory doesn't always speak with an honest unambiguous voice.

So we can't say for certain that Canada's macroeconomic experience would have been better over the last few years if the Bank of Canada had tried to follow an NGDPLPT rather than IT. But those pictures are awfully suggestive that it might have been.

And theory does add one important thing: for any given nominal interest rate or monetary base set by the Bank of Canada today, current NGDP is positively related to expected future NGDP. So if the public thinks that the Bank of Canada will eventually get future NGDP back to trend, that helps to keep current NGDP on trend. The mild boom before 2008 would have been even milder if people had expected a reversion to trend. The sharp dip after 2008 would have been less sharp if people had expected a reversion to trend.

Comments

[brian]

Shouldn't CPI be compounding. Wouldn't a 2% compound growth line end at around 35%

[Scott sumner]

Great post. It will help me when I testify in Canada

[K]

Nick: The market monetarists think central banks are omnipotent. But lots of the rest of us a far less optimistic. And *that alone* ought to give you MMs pause in your faith. And lack of credibility is exactly why this is a very bad time for the Fed to be switching target (it's better than *not* switching target, but nevertheless crappy timing). They have to get us to believe that they'll buy lots of stocks or that when things get moving they'll permit an inefficient quantity of inflation. And frankly, we don't believe it. If they wanted us to believe that, they should have proven their mettle by creating some inflation well before they they found themselves at the ZLB, and they could easily have delivered. The BoC is not at the ZLB (though perhaps already too close), and should take this opportunity to get even further away by raising the rate of inflation and then later, the policy rate. NGDP targeting would be a good policy for achieving that, but the time to implement it is now while they still have a credible policy instrument at their disposal.

Brian: I'd assume those are log changes.

Scott: Fantastic! Best of luck. Let's hope it gets the ball rolling at the BoC.

[Nick Rowe]

brian: I think it's in logs, so a constant growth rate means a straight line on the graph. But I'm the token math-challenged economist on this blog.

Scott: great! By video, or are you coming up here? I think I'm testifying too.

K: yep. I recognise that the fact that many people don't think monetary policy can do anything makes it harder for monetary policy to actually do something. But for the BoC this is much less of a worry. Expected inflation doesn't vary much from the 2% target, so the Bank has credibility. (Over a shorter horizon it may vary from 2%, but then the Bank has a 2 year horizon, so that's not evidence of any lack of credibility.) And the BoC did not cut to 0.25% until April 2009, and raised in June 2010. And it did not make full use of unconventional policy. So it could have done more, if it had wanted to. And the expectation that it would do more would also have been effective. But looking at the top graph, you can't really see any massive policy mistakes, except far too easy and then far too tight for 2008 alone.

[Nick Rowe]

K: "NGDP targeting would be a good policy for achieving that, but the time to implement it is now while they still have a credible policy instrument at their disposal."

Yep. That's important. Get people used to it, and believing it, before it's really needed. Indeed, if you look at the pictures before 2008, it's not obvious that there would be much difference between any of the 3 policies. It's only when the crunch comes it really matters.

[Stephen Gordon]

Brian - as others have guessed, the graphs are for the logs of CPI and NGDP.

[Luis H Arroyo]

Very great post. For once, a cuasi(market?)monetarist recognises that

"The mild boom before 2008 would have been even milder if people had expected a reversion to trend. The sharp dip after 2008 would have been less sharp if people had expected a reversion to trend."

If NGDP is useful in recession, it should work also in excess of growth!

[K]

Nick: "Expected inflation doesn't vary much from the 2% target, so the Bank has credibility."

The BoC's adherence to policy is truly impressive. But the market may confuse credibility in enforcing policy with credibility in enforcing hard money. Personally though, I would be inclined to take them at their word *if* they announced NGDPLPT in the current environment. But *not* after NGDP shrinks 5% in 12 months. Also, it's an open question whether moving inflation around over a reasonable time scale is anywhere near as easy as maintaining a long term fixed inflation policy (no mean feat itself).

[Ritwik]

Umm.. so the nominal problem is real? :)

Not trying to be flippant - just wondering what your conclusion would have been if NGDP would have been on trend and CPI above the trend line?

[Nick Rowe]

Luis: thanks! Yep. We always tend to tell the story one way, about mitigating the depths of a recession; but it works the same the other way too.

K: the BoC tries hard to tell people the target is symmetric. 1% inflation is as bad a miss as 3% inflation. It's difficult to say whether most people believe in that symmetry, though I do. If it weren't symmetric, it would be a total fluke that inflation has in fact averaged almost exactly 2%. (Though, actually, that still looks a bit of a fluke, even if the BoC is symmetric.)

Ritwik: and real GDP behaving roughly as it did? You have roughly described the UK economy. I would conclude that NGDP is better than CPI, because NGDP is giving no signal and CPI is giving a false signal, but that something else is problematic (in the UK it's probably VAT, exchange rate, and oil prices).

[jesse]

I tend to agree with Krugman: NGDP targeting is a cynical but politically-viable way of increasing inflation beyond the 1-3% band.

[jesse]

"Which of those two pictures gives you a more accurate and useful summary statistic for the Canadian macroeconomy?"

But of course you won't be able to determine much from the graph because you're measuring the output of a system under feedback control. The Bank has been successful at controlling inflation; you need to plot "control effort" -- the amount of monetary stimulus meted out through interest rates as well as other forms of stimuli -- to see how severe the recession was. The small blip around '08 is perfectly explainable under control theory too: the tools the BoC and government have at their disposal have a lagging response so there will be "tracking error" trying to react to acute and severe shocks like GFC.

All the first graph shows me is that whatever the Bank of Canada is doing to control inflation, it's working. The tracking error overall is low as we would expect in a system under proper control.

[Andy Harless]

Another point in favor of NGDPLPT is that it would tend to give more legitimacy to that approach internationally and therefore make it more likely that the Fed would seriously consider adopting it.

I also like K's point that the best time to change regime is when you already have enough credibility to make the new regime effective. I recall Greg Mankiw, at some point in the mid-90's, arguing for replacing discretionary monetary policy with a mechanical rule, saying, "Alan Greenspan has done such a good job that it's time to take the job away from him." His point IIRC was more about the politics, that it's easier to undertake a regime change (especially the one he was advocating) when things are going well enough that there isn't a lot of pressure. That point might apply here too.

[jesse]

Hi Nick, another point here is that the Bank of Canada is in a way doing GDP targeting, it's indirectly there in their mandate actually, around financial stability. They have taken on the impetus to concern themselves with real GDP growth as it pertains to cause inflation to be less controllable in the medium/long run. Carney makes statements about the economy in general, never mind that the Bank of Canada is capable of controlling inflation under a wide variety of economic environments. That is, if the economy is underperforming or overperforming, the Bank can control inflation in both scenarios, but it's not "complacent" when GDP growth is above or below potential, partly because adverse shocks on either end make their standard tools less effective. In that sense there is some targeting going on, mostly by advising the government on what it can do to regulate and facilitate economic output.

[brian]

How much does NGDP have to do with population growth and should the BoC target fertility rate and immigration to help? OK I'm not serious but I wonder how a NGDP target would work with a declining working age population.

[Ian Lippert]

Hey Nick, I have another question about NGDP targetting and real interest rates. If the real interest rate is based off the expected inflation rate would an large enough increase in NGDP push the real interest rate down to zero? Doesnt this mean there is a lower bound to NGDP targetting in a similar way to what we are facing now? And wouldnt this create a situation where there is no incentive to save and everyone consumes all their capital?

[Nick Rowe]

jesse: I disagree with Paul Krugman on that point. It all has to do with the slope of the IS curve. See my post from about a week back.

The BoC has indeed done quite a good job of controlling inflation. But recently, doing quite a good job of controlling inflation didn't seem to be good enough to prevent the recession. If the BoC had instead been asked to control NGDP, and had done an equally good job at that, we might not have had as much of a recession.

Andy: agreed. It's very tempting to say "well, let's keep doing IT until we find it doesn't work". But the time to change is when everything is going smoothly, and before there's trouble.

jesse: the BoC does look at real GDP, relative to its estimate of potential GDP, as an intermediate target, or indicator, but only in order to hit it's final target, inflation. It does not trade off inflation against real GDP growth at the 2 year targeting horizon.

brian: If the BoC stuck with a 5% NGDP target, and RGDP growth trended below 3% (because of lower working-age population growth, for example), then inflation would trend higher than 2%. That's all.

Ian: in the long run, it's the other way around. The real rate of interest and real GDP growth rate are independent of inflation. It's the nominal interest rate and nominal GDP growth that must adjust.

It would be unwise to target NGDP growth less than 5%, IMHO.

[Lars Christensen]

Nick, interesting...I agree with you. Canada is not the US so the optimal institutional set-up (within a central banking framework) might not be the same.

A couple of observations...

First argument to me is that BoC should move to level targeting - it has done it anyway so why not make it formal.

Second, historically there has basically been no difference between NGDPLT and PLT - other than of course during the Great Recession.

Third, is NGDPLT really what we what to target in an open economy? Wouldn't it be better to target for example the domestic demand level? I am still thinking about what would be the appropriate measure for a small open economy. NGDPLT is fine for the US, but not necessarily for small open economies - especially not a commodity exporter like Canada.

[jesse]

"If the BoC had instead been asked to control NGDP, and had done an equally good job at that, we might not have had as much of a recession"

I don't disagree -- we may never know -- but here's the thing: there were deflationary shocks that they weren't entirely prepared for and it's hard to know if NGDP would provide enough certainty for them to properly react in time. I'm not convinced GDP is measurable/certain enough on a coincident basis for them to react to it. That said, there were clear coincidental indicators of a recession starting in early 2008 -- a deflationary shock should have been on the radar quarters before it happened.

All's I'm saying is that the Bank can invoke NGDP or other coincident measures into their feedback and still be completely consistent with their stated inflation mandate. That is, if they see NGDP falling off the rails, perhaps bolstered by other indicators, they can use that as a leading indicator of CPI deflation and recommend fiscal stimulus plans in advance (ready the shovels or what have you), as well as starting to tip overnight rates lower and the usual barking at gala lunch speeches hinting at future moves. That CPI blipped as it did is an indication they were unable to react to the economic shock fast enough, but this may be in part because their overnight rate mechanism is incapable of producing a faster reaction time to near step changes in output, regardless what mechanism they are tracking.

[jesse]

Nick, I read the post, it's arguable that the IS curve slopes the way you say, but it depends on what country. I can believe a scenario where one country has negative sloping IS another positive sloping IS. In China, for example, empirical evidence suggests that when real rates were higher consumption went higher too.

[Nick Rowe]

Lars: "why not make it formal."

Yep, but that argument cuts both ways. You could argue that there are no benefits from PLPT, since IT gives the same results in practice.

It all depends on whether that was just a fluke in the data, and whether people expected the price level to return to trend.

In an open economy, if net exports fluctuate, it would not be good to target domestic demand. It think NGDP would be better. But there are possible effects of fluctuations in the terms of trade that I haven't thought through fully.

[Nick Rowe]

jesse: agreed, the BoC can use NGDP as an indicator even under IT. But:

1. NGDPLPT does give an automatic negative feedback stabilising force even if the BoC does nothing.

2. Suppose, just suppose, we forget about that little up and down blip in the CPI. That says that BoC policy was (nearly) perfect, even in hindsight. And we know it wasn't perfect in hindsight. CPI is giving a false signal of what the BoC got wrong.

A change in r has both income effects and substitution effects. For the representative agent in a closed economy, there is no income effect. (I did a post on that once). The higher wealth of the lenders cancels the lower wealth of the borrowers. But maybe, under some circumstances, or in an open economy, it doesn't all wash out.