Evan Soltas
Mar 15, 2012

Not So Shocking

How does an NGDP target perform under the stress of real shocks? Better, in a word.

One of the most persistent criticisms of NGDP targeting is that if inflation expectations become unanchored then an NGDP target would underperform a discretionary monetary policy or an inflation target.

Or so the reasoning goes. But in what environment are inflation expectations most likely to be destabilized? This seems to me the most promising way to either validate or dismiss the criticism.

My admittedly broad interpretation of the economic history of the postwar era is that the two most common risks to inflation expectation stability are (1) consistently over-expansionary monetary policy, especially in combination with a lack of central bank independence and (2) severe supply shocks, namely when the supply shock hits highly visible prices for food and gasoline.

I don't find (1) a plausible scenario today, given the global shift to independent central banks and the significant credibility that the central banks for global reserve currencies now have.

So let's talk about (2). The beauty of NGDP targeting is that you can all but eliminate nominal shocks and, through expectations channels, even the risk of nominal shocks.*

Back to the conversation about supply (or real) shocks. Any monetary policy has a rough time handling a supply shock -- but that's because, as Scott Sumner has written, supply shocks "reduce resources available to our society." Inflation targeting handles it particularly poorly, because any increase of inflation above target threatens the credibility of the target, and thus contractionary policy becomes necessary to defend inflation expectations at the worst possible moment for real output stabilization.

Not so with NGDP targeting, which by making the trade-offs between real and nominal variables explicit, allows for inflation to rise temporarily above trend without danger to credibility during real shocks. (See my illustration above, but note that I couldn't make AD a hyperbola, as it should be in my model.)

Therefore NGDP targeting outperforms inflation targeting because it prevents the supply shocks from being transmitted into core inflation and expectations without the need to contract real output further. In effect, if the central bank can convince people that, despite the increase in gas prices, they shouldn't expect lots of other prices to rise and thus shouldn't ask for a cost-of-living increase, then the central bank need not be contractionary.

Even critics of NGDP targeting, such as Bank of Canada Governor Mark Carney, acknowledge the response to supply shocks to be the strong suit for such policy (Note: Carney sees price level targeting and NGDP targeting as similar in this case.):

Based on simulations using the Bank’s main projection model, the benefits of this greater stabilization under PLT [price level targeting] are comparable to a permanent quarter-point reduction in the standard deviation of CPI inflation...Much of this benefit arises following shocks that create an explicit trade-off between output and inflation stabilization, such as supply shocks, since credible and well-understood PLT improves this trade-off.
David Beckworth has more on this topic, including research which suggests that NGDP targeting would avoid inappropriate stimulus or contraction in response to shifts in either direction of aggregate supply.

The only argument I could find that NGDP targeting would mishandle a supply shock is this paper by Laurence Ball, which Ip links to...I was about to write a response when I realized Sumner's already been there and done that. The short version of it is that Ball's dismissal of NGDP targeting relies on his modeling of how Y and P respond to monetary policy to derive a sort of "standing wave" instability, but that his conclusion of instability is not rigorous in that if you change his assumptions of how Y and P behave to something a little more realistic, the instability disappears.

Another way to understand the advantage is by recognizing that the sum of the variances of inflation and real output are likely smaller for an NGDP targeting than for inflation targeting, as the impact shock is split between the price level and real output. (This is the same framework Ball uses, with his assumptions changed.) This conclusion is especially strong when one considers expectations and central bank credibility.

I guess I don't really understand the argument that NGDP targeting is going to destabilize inflation expectations, which seems to be the main line of argument coming from the Fed about the policy, since we've just examined the main scenario where inflation expectations are jeopardized and concluded that NGDP would actually handle it better.

* Sidebar: Greg Ip of The Economist thinks we're overselling the capacity of the central bank to quickly ensure nominal stabilization in extreme events or the strength of expectations, and this doesn't feel like a totally unjustified comment to me. He characterizes our position, rather cartoonishly, "Ignore the asteroid strike. The Fed will ensure everything will be alright.” (I think we're assuming that the asteroid strike is a nominal shock here.) What strikes me as the best response to Ip's argument is that even if it takes a quarter or two to restore NGDP to its level track, which would be true if expectations are realistically a bit shaky, an NGDP level target would vastly outperform flexible inflation targeting in minimizing the effect on real output because it would support aggregate demand more fully. In that respect, an NGDP target would represent, in Ip's words, something "less imperfect than the alternatives."