Commenters Patrick Sullivan and gofx directed me to a post on NGDP targeting in developing countries, by Pranjul Bhandari and Jeffrey Frankel:

Targeting Nominal GDP has been proposed in the context of major industrialised countries. (Frankel, 2012, gives other references to the literature.) But a good case can be made that the idea is in fact more applicable to countries further down the income ladder. The reason is that emerging market and developing countries tend to experience bigger terms of trade shocks and supply shocks than industrialised countries do. NGDP targets are robust with respect to precisely these kinds of shocks.

I’ve argued that NGDP targeting is less likely to be optimal for developing countries than developed countries. Does that mean I disagree with Bhandari and Frankel? Not entirely. I think they are right that developing countries are more prone to supply shocks, and that NGDP targeting is superior to inflation targeting when there are supply shocks. In that case the gain from switching from inflation targeting to NGDP targeting might well be larger for developing countries.

However I’ve also argued that NGDP targeting is not optimal when countries depend heavily on the production of commodities with very volatile prices. Suppose 50% of Kuwait’s NGDP were oil production. If global oil prices doubled, and Kuwaiti oil output remained unchanged in physical terms, then the central bank of Kuwait would have to reduce non-oil production to zero in order to keep NGDP stable. Obviously that would not be optimal.

My hunch is that the optimal monetary policy target is something slightly different from NGDP, perhaps total labor compensation. If nominal hourly wages are sticky, then targeting total nominal labor compensation will tend to minimize employment volatility, while still keeping inflation low on average.

For a big diversified economy like the US, Japan, or the eurozone, there isn’t much difference between targeting NGDP and targeting nominal aggregate labor compensation. Indeed that may also be true of large developing countries like China (and increasingly true of India, as agriculture declines as a share of GDP.) But for a country like Kuwait the two series might diverge sharply. In that case, target total labor compensation. One implication of this claim is that Australia might want to let NGDP growth slow slightly if the mining boom peters out.