Winner of the New Statesman SPERI Prize in Political Economy 2016

Tuesday 29 January 2013

When formal monetary policy targets are useful

Stephanie Flanders makes a very perceptive point in a post today. She says:

“Does the chancellor want to set a new target for the Bank of England? The answer is no. But does he want have a debate about it? The answer is an emphatic yes - for political reasons as well as economic ones.”

Essentially the more it is debated what the Bank of England can do to get a recovery, the less it appears as if the Chancellor is responsible for the current state of the economy. I think in practical political terms this is correct. However it should not be so: the Chancellor and the Treasury set the MPC mandate. There are three key points here.

First, the inflation target matters. Some commentators suggest that the MPC in effect has the ability to set its own target, and that the formal inflation target is no constraint. I think this is completely wrong. As Adam Posen recently said (22nd January) to the Treasury Select Committee

“anyone who was on the committee basically took the equivalent, in my opinion, of an oath of office. They were serving on the committee under the terms of the given inflation target.”   

My conversations with other MPC members are fully consistent with this view.

Second, because we have had a series of positive shocks to the Phillips curve, UK monetary policy has not been as expansionary as it should have been because it was, and is, constrained by the 2% target for consumer price inflation. That has become increasingly clear over the last few years. Interest rates were almost raised in early 2011.

Third, the Chancellor had, and has, both the formal power, and the political ability, to change this. He might have been cautious about moving to a nominal GDP levels target, but there were less radical options that would have helped the UK economy. The most obvious was to move to a dual mandate: following what the US Fed does can hardly be thought of as dangerously radical. He could, as I have suggested before, change the inflation measure being targeted from consumer price inflation at 2% to nominal earnings at 4%. Controversial, yes, but part of being a good Chancellor is to make bold moves when conventional policy is not working.

OK, rant over. What I really want to say concerns nominal GDP targets. In her post, Stephanie Flanders gives the following reason why she thinks a nominal GDP levels target will not be adopted by the Chancellor.

“will [voters] really ever believe that a government is going to withstand years of well above, or well below, target inflation, simply to get back to a particular, fairly arbitrary path for the cash value of the economy that was laid down by the folk who were in power before them? The general view in Number 11 is that the answer to that question is no.

Politicians are always going to let bygones be bygones - for the very good reason that voters expect them to think and talk about the future, not the past.”

This puts into ordinary language what macroeconomists call the problem of time inconsistency. Suppose, for example, we had before the recession a target for the path of consumer prices, and not its rate of change. As everyone knows, for most of the last few years UK inflation has been above 2%. With a price level target, the Bank would now be raising interest rates for sure, because it would have to achieve years where inflation was below 2% to get back to target. Bygones would not be bygones. Is it credible that policy makers would do that? More important, would voters allow them to do that?

Now a nominal GDP target would not suffer from that specific problem, because the recession goes the other way. But in other circumstances a similar problem could arise. Suppose we had an unexpected boom, where the central bank cannot prevent nominal GDP exceeding its target path. It then is required not just to get the growth rate of nominal GDP back to desired levels, but to reduce it further to get back to the target path. I think Stephanie Flanders is right that selling that policy would be difficult.

However that does not mean we should not do it. Committing to doing it may bring significant benefits, as Michael Woodford has consistently argued. What it does mean is that having a formal target could be very useful. Not, as macroeconomists typically put it, to stop central banks reneging on the policy. Instead, to protect central banks from politicians and voters when the bank sticks to the policy.

I think in this respect formal inflation targets are rather different from levels targets (whether its price level targets or nominal GDP levels targets). In a recent post I was skeptical of the arguments for why formal inflation targets might be useful, particularly if you had a credible central bank. I prefer the less formal dual mandate of the Fed to the formal inflation target of the MPC. As I noted at the start of this post, formal inflation targets can do considerable harm, whereas their benefits in preventing inflation bias are I think overrated. But the time consistency problem with levels targets is in my view much greater, and so the usefulness of formal targets also becomes greater as a result.

Let me put the point another way. Delegation works best when the objectives of the policy are clear. Keeping inflation and unemployment low are uncontentious, as is (nowadays) the idea that there are limits to how low unemployment can be pushed without compromising inflation. So delegating to a central bank that tries to keep inflation and unemployment low can work without hard wiring these objectives by using formal targets. Targeting the price level, when what you care about is inflation, is much less intuitive, and therefore potentially contentious. For that reason, it is a good idea to not leave this issue to the discretion of central bankers, but to formalise it as an instruction to central bankers - for their sake as much as ours.

So, if we are looking for the optimal monetary policy, I think the informal dual mandate of the Fed dominates a formal inflation target, but whether it dominates a formal price level or level of nominal GDP target remain open. Which means the new Bank of England governor should ask the Chancellor to change something. However, as I said at the start, the Chancellor really should have changed something already.


  1. Prof- Wren-Lewis,
    Have you seen the new book Monetary Policy by John Fender?
    It's especially suitable for advanced undergraduates.

  2. Are we always going to end up fighting the last war with our policy targets? Inflation targeting made sense to people because it was a way of killing off the inflation expectations problems that arose in the 70s. For a while it worked well, locking in the benefits of harsh policies from the early 80s without the need to raise interest rates to scary levels - the "Chuck Norris" effect, as it were.

    Now we find that there are circumstances where inflation targeting doesn't work so well, and we observe that NGDP targeting would be a better fit. But this only works so long as our situation remains within certain parameters. Maybe NGDP targeting works well for 25 years and then breaks down - after all, who knows what might happen in 25 years, given climate change and new technology. We could be extending human lives for decades, whizzing around in self-driving cars, having all of our work done by robots and our manufacturing done in local 3D-printing labs, and it's highly unlikely that we can currently conceive of a monetary policy that gives us the optimal outcome for today and the optimal outcome for all future scenarios.

    In light of that, we might say that inflation targeting did extraordinarily well to last as long as it did, and that the best hope of any replacement system would be to last for an approximately similar length of time before it no longer addresses the concerns of the day.

  3. I should add if what I just said is at all correct (and it might well not be, I'm no expert) then perhaps we should be biased towards whichever monetary policy does the most to encourage future technological improvements. If it's inevitable that any monetary policy will break down in 25 years, we'd rather it broke down with us living in a world of robots-doing-all-the-work than a world that's basically like now but with 25 years more wear and tear on our unreplaced infrastructure. If in doubt, take a punt on the option which gives us the most growth?

    1. This comment has been removed by the author.

    2. Good post Rob, with some very interesting points raised. Unfortunately it seems that most of the political class, largely for party political reasons perhaps or simple ideology (or both) are often too blinkered or inflexible to amend the policy when it starts not too be so effective. More than this, as is currently exemplified in government today, they can't even recognise a disasterous policy - even after it has been shown to be so. So what chance of them being able to identify, recognise and adopt a good, effective policy appropriate for the prevailing economic circumstances? If only they would/could!

  4. Here is something that I have wondered about concerning level targets that doesn't get mentioned (though it is very much related to the points you raise).

    All targets are forward looking and all central banks will give themselves a margin of error since they don't have perfect control over their objectives. For example, the Bank of Canada formally has a target band for inflation of 1-3 percent. With an inflation target, a target band is very workable. If inflation comes in at 2.75 percent this has no bearing on subsequent monetary policy. Bygones are bygones.

    With a price level target, however, the situation is more complicated. If the target allows for an ongoing two percent growth in the price level, any miss, even those within an acceptable margin of error, has to be addressed. Otherwise, the bank will move away from the target. So, for example, from an accepted starting point, policy might deliver 2.75 percent price growth. That is going to require a tightening of monetary policy even though it is within the band that would apply for an inflation target. It seems very difficult to allow for these typical margins of error with price level targets. The same applies of course to nominal GDP targets (in levels), more so since there are greater sources of error.

    On other matters, I greatly enjoy your blog Simon.


Unfortunately because of spam with embedded links (which then flag up warnings about the whole site on some browsers), I have to personally moderate all comments. As a result, your comment may not appear for some time. In addition, I cannot publish comments with links to websites because it takes too much time to check whether these sites are legitimate.