NGDP targeting could change the way we manage inflation

The original goal of this post was to answer the following questions:

1. What was Mark Carney talking about in his December speech entitled “Guidance“?
2. What is NGDP targeting and why are people talking about it?

It turns out that that before I could figure out what to say about those two questions, I had to talk about two other things first: inflation targeting and price-level targeting.

There’s really no great mystery about what the Bank of Canada does and why: it has an explicit mandate from the government to conduct monetary policy in such a way as to ensure that inflation rates stay between one and three per cent per year, with two per cent being the de facto target. It’s been doing this for more than twenty years, and successfully, too:

An important feature of Canada’s inflation-targeting regime is that it is forward-looking. Milton Friedman famously remarked that monetary policy takes effect with “long and variable lags”; current decisions about the conduct of monetary policy are made with an eye on what is likely to happen six to eight quarters from now. Having a forward-looking target also means that past mistakes are ignored when planning for the future. The most recent data show Consumer Price Index (CPI) inflation running below target, but that is only of concern to the Bank of Canada insofar as it affects its assessment about what is likely to happen in mid-2014. The fact that year-over-year inflation in was only 0.8 per cent in November does not mean that the Bank will try to compensate by attempting to engineer an inflation rate above two per cent in the future.

So even though inflation targeting makes it relatively easy to make forecasts about inflation—two per cent a year would be the safest bet—the fact the the Bank treats past deviations from target as bygones to be forgotten means that forecast errors accumulate with longer horizons.

Now, uncertainty about anything is costly, so it’s worth revisiting the current setup with an eye to reducing uncertainty.

Suppose the Bank of Canada’s mandate was revised so that if inflation did run below target, it would “correct” this mistake by running higher-than average inflation so that long-run forecasts of inflation can be made with more certainty. This is the idea behind price-level targeting (PLT). Instead of targeting the rate of CPI inflation, the Bank would target the actual level of the CPI. If the growth of CPI fall below trend, the Bank would be expected to generate above-trend inflation rates in order to return to the target.

Surprisingly enough, it looks as though the Bank of Canada has in fact been conducting this policy for the better part of two decades:

Periods of below-trend inflation have been almost exactly offset by periods of above-trend inflation: average annual inflation between January 1995 and November 2012 was 1.94 per cent. The people at the Bank of Canada will tell you that this is simply a happy coincidence, but it’s a handy thing to remember: if the Bank’s mandate were changed to a price-level target, the path of CPI wouldn’t look much different than the current one.

There are a couple of reasons why price-level targeting might be a good idea:

1. More aggressive counter-cyclical policy. Inflationary pressures are reduced during recessions, and the Bank’s response is to bring inflation back up to target. But under PLT, the Bank’s goal would be to generate inflation that is above target in order to offset the effects of the recession. This extra monetary stimulus would also have the effect of making recessions shorter and less severe under PLT than under inflation targeting.
2. More effective monetary policy. One of the most important ways in which monetary policy works is through its effects on peoples’ expectations of inflation—this was the subject of Mark Carney’s recent “Guidance” speech. For example, the Bank of Canada’s policy rate is now at one per cent, so two per cent expected inflation implies a real interest rate of minus one per cent. But if it were known that the Bank was aiming at (say) four per cent inflation in the near term in order to return the price level to its trend, then the same nominal policy rate of one per cent would imply a real rate of minus three per cent, and would have a correspondingly larger effect.

A problem that PLT may have but which inflation targeting does not is dynamic inconsistency. Monetary policy may become more effective if people believe that inflation will go above trend after the economy has recovered. But once the recession is over, inflation returns to target and the economy returns to full employment, the central bank would be sorely tempted to simply declare victory and forget about trying to engineer a period of higher-than-average inflation. And if the situation were reversed—if there had been a temporary spike in inflation—the central bank may not be inclined to adopt a contractionary stance and risk a recession in order to return the price level to target.

Still, the Bank of of Canada has spent a lot of time researching the matter, and most people involved—both inside and outside the bank—seem persuaded of the theoretical merits of PLT. When the inflation targeting mandate came up for renewal in 2011, there was some speculation that the Bank would switch to PLT.

It didn’t, for what is to me a very compelling reason: inflation targeting was not broken in Canada, and the improvements offered by PLT had to be very large indeed to justify tinkering with a policy regime that had served us well. Moreover, it was not at all clear that the communication issues raised by PLT would work smoothly in practice. Under inflation targeting, everyone knows the inflation rate that the Bank of Canada intends to deliver: two per cent. But under PLT, these inflation objectives would vary continually. Depending on whether or not the CPI was above or below target, the Bank would announce inflation targets above or below two per cent, and converging to trend over the medium term. In theory, people would continually revise their inflation forecasts as the Bank revised its inflation objectives. We don’t really know how things would actually work in practice, but it seems clear that communications from the central bank—or guidance—would become even more important than it is now.

This is the part where I finally get to talk about NGDP targeting. I spent a lot of time talking about price-level targeting for three reasons. The first is that PLT has been the subject of much debate in Canada, and we may see it come up again. Another reason is that the research on PLT is—so far—more extensive than on NGDP targeting. But the most important is that the arguments for PLT are similar to those in favour of NGDP targeting: if you can understand why PLT might be a good idea, you can understand why some people are so keen on NGDP targeting.

An NGDP targeting regime would work in much the same way in which PLT does: the central bank targets nominal GDP and adjusts monetary policy accordingly. Nominal GDP is not adjusted for inflation, and can be roughly interpreted as being the product of the price level and real GDP. The advantages of NGDP targeting are the same as PLT, only more so. In a recession, the central bank wouldn’t be satisfied with getting inflation or even the price level back to trend; it would try to increase prices even higher in order to return NGDP to its trend. Monetary policy during a recession would be even more aggressive than under a PLT regime and much more aggressive than under inflation targeting.

Suppose that instead of a two per cent inflation target, the Bank of Canada had a NGDP target that was growing by (say) five per cent per year—two per cent inflation plus three per cent real growth:

The Bank of Canada believes that its current stance is consistent with its inflation target. If it were following a PLT, Canadian monetary policy might be more loose, but the current deviation of CPI from a two per cent trend isn’t all that large by historical standards. But if we had an NGDP target, the Bank of Canada would be much more aggressive than it is now: the policy interest rate would probably be still at zero, and it may have already resorted to quantitative easing. It is this feature that makes NGDP targeting such an attractive idea for places like the U.S. and the U.K., whose economies are still in deep recession. A mandate to do whatever it took to return NGDP back to trend could go a long way to accelerating their recoveries.

As with PLT, the biggest challenge with implementing NGDP targeting is communications. As Carney noted in his speech, central banks would be required to provide clear, credible and continual guidance to private sector forecasters in a complicated environment, and this task is especially difficult now that policy rates are at or very close to their lower bounds. Moreover, the credibility of the central bank’s guidance becomes even more important, because it will have to contend with the dynamic consistency problem. The private sector may not always believe a central bank that has a strong incentive to go back on its word.

The case for NGDP targeting is a difficult one to make for Canada in the near term: it’s hard to justify fixing inflation targeting when it’s not obviously broken. As I said, it is much easier to make the case for NGDP targeting in the U.S. and the U.K. But in the event that the Federal Reserve and/or the Bank of England adopt NGDP targeting and figure out how to meet the communications challenges, then when the Bank of Canada’s mandate comes up for renewal in 2016, it might no longer be a question of abandoning something that works for something that’s not been tried before: it might be a question of abandoning something that works for something that demonstrably works even better.