This is the third in a series of posts on NGDP targets. Links to the earlier two are given at the appropriate points in the text.
At last we are getting a Bank of England led debate on alternatives to UK inflation targeting. Charlie Bean gave a speech on Wednesday on nominal GDP (NGDP) targeting which is well worth reading, not just because he is Deputy Governor for Monetary Policy at the Bank (and of course on the MPC), or because he is one of the UK’s top macroeconomists (he was at the LSE before going to the Bank), but also because he is something of a veteran on this issue, having written a chapter of his PhD on the topic.
Charlie quite rightly distinguishes between NGDP growth and NGDP levels targets. On the former, he argues that their superiority over inflation targets in the face of cost-push shocks (shocks to inflation, like increases in VAT)  is something of a red herring, because in practice inflation targets are flexible, and central banks have the discretion not to come down too hard on such shocks. That is the party line, which I called into question here.
What is interesting is that he goes on to say that one advantage of NGDP growth targets is that “it might lead to fewer instances where we have to resort to the use of our ‘constrained discretion’ to justify a temporary acceptance of inflation away from the target”. I think that may be a subtle way of saying that the MPC has felt inhibited from exercising this discretion in the recent past, and as a result it has not pushed stimulus measures as hard as it might over the last few years, and indeed on one occasion came close to raising interest rates. I’m also glad to see that the MPC are now forecasting inflation above 2% in two years time, yet they are also moving towards more QE, which suggests their appetite for discretion has returned. Perhaps what Charlie is saying here is that he did not vote in favour of more QE last month because he still felt constrained by the inflation target.
On NGDP levels targets, Charlie shows some simulations of a toy New Keynesian model which illustrate the advantages of NGDP level targets when interest rates hit the zero lower bound (ZLB). The model assumes that there are no alternative monetary policy instruments. Similar pictures can be found in Eggertsson and Woodford. Here I cannot help repeating a point I have made before. While NGDP targets get close to the optimal monetary policy, they still mean any demand shock has a significant negative impact on output and inflation while interest rates are at the ZLB. Expansionary fiscal policy still has an important role to play, and austerity will make things much worse.
I want to make one final point, which I do not think is mentioned often enough. Levels targets (for prices, or NGDP) are not just advantageous when we hit the ZLB. In the lectures I have just finished to second year undergraduates, I go through a little numerical example of how monetary policy might deal with a one-off cost-push shock (perhaps an increase in VAT), assuming for simplicity that monetary policy has perfect control over the output gap. Crucially inflation is forward looking (the Phillips curve is New Keynesian), and current inflation depends on inflation expected next period. If policy can only act in the same period as the cost-push shock (the ‘first period’), it reduces the output gap to get the best trade-off between higher inflation and lower output. However, I show the students that if monetary policymakers can also change output in the next period (the second period), they will lower output in that period as well. This will generate negative inflation in that second period, but bad though this is in itself, the anticipation of it reduces inflation further when the cost-push shock hits in the first period. If policy does it right, and expectations anticipate policy and its impact, by reducing output in both periods it can improve welfare compared to the case when it only acts in the first period.
Now I present this as an example of the problem of time inconsistency. By promising to reduce output in the second period, we get a better outcome. But once we get to that second period, the policymaker or the public may ask why do we need to actually go through with the plan? It was beneficial to promise to reduce future output and inflation when the shock hit, but now the shock is over it would be better still to do nothing. If the policymaker changes their policy as a result, then of course a rational public will anticipate this, and we will not get the benefits in the first period, because these benefits come through expectations effects. So to overcome this time inconsistency problem, we need some device to commit the policymaker to stick to the plan.
Although that is the main point, I also ask the students to note what happens to the price level in either policy. With the time inconsistent policy where output and inflation is reduced in the second period, the price level we end up with is a lot closer to its original level than the case where policy only acted in the first period. But why stop in period 1? Policy could also promise to reduce output a little bit in the third period as well, and so on, and that would indeed be the optimal thing to do (although it remains time inconsistent). If we were to compute that fully optimal policy, we would find that the price level would end up where it started. (The proof is not that difficult - maybe I’ll put it in the lectures next year.)
The implication is interesting. If the central bank had a price level target, and it had discretion about how quickly to achieve it, then it could implement the optimal, time inconsistent policy. The central bank would be committed to that policy, because it had to bring the price level back to its original level. A levels target can act as a commitment device. (The is an example of a general point I made here.) It uses the same principles as the ZLB case, but it applies even when interest rates are in no danger of hitting that ZLB. (A good paper to read on this is by David Vestin, but the point will also be familiar to students of Michael Woodford.)
So the case for levels targets (often called history dependent targets) is stronger than the particular issue of hitting the ZLB. However it depends crucially on the idea that agents are forward looking. Replace this New Keynesian Phillips curve with something more old fashioned and ‘backward looking’ in this particular example, and price level targets would just involve additional costs with no benefits. Other models would give different results (see, for example, a well known paper by Svensson .) However I think it is reasonable to make the following generalisation: this particular attraction of levels targets (of which NGDP levels targets are an example) depends crucially on how forward looking we think agents really are.
 Bean, C.R. (1983). ‘Targeting nominal income: an appraisal.’ Economic Journal, vol. 93, (December), pp. 806-19.
 In the face of such shocks, the central bank can only reduce their impact on inflation by reducing output. A strict inflation target means you ignore the output costs of reducing inflation, whereas a nominal income target will mean you have a chance at getting nearer the optimal trade-off between lower output and inflation.
 Gauti B. Eggertsson & Michael Woodford, 2003. "Optimal Monetary Policy in a Liquidity Trap," NBER Working Papers 9968
 Svensson, Lars E.O. (1999) “Price-Level Targeting versus Inflation Targeting: A Free Lunch?” Journal of Money, Credit, and Banking, vol. 31, no. 3, pp. 277-95.