Robin Brooks has a couple tweets that express frustration with IMF estimates of output gaps:
Campaign against Nonsense Output Gaps (CANOO): since 2007, German real GDP is up 12%, Spain is up 1% and Italy is down 8%. Yet the latest numbers from the IMF, published today, say these countries have the same output gaps: Italy (-1.0%), Spain (+0.7%) and Germany (+0.8%). What?!
In another tweet he illustrates his problem with IMF estimates in a graph:
I am not a big fan of output gaps and have argued that these estimates should play no role in monetary policy. Instead I favor NGDP targeting. Nonetheless, I can see both sides of this issue.
I’d argue that the basic problem here is that the term ‘output gap’ has multiple meanings. Among academic economists, the output gap is supposed to measure the difference between the current level of output and the output level that would occur once wages and prices had adjusted to previous nominal shocks. By that measure, Italy’s output gap might be relatively low, say minus 1%. You could view that definition as the gap between current output and output levels under appropriate monetary policy. (That’s not to say I have a strong opinion on the -1% figure, the output gap defined this way might well be considerably larger.)
Another definition of output gaps is the difference between the current level of output and the level of output that would occur with appropriate economic policy (including more efficient taxes, spending, regulations, etc.) By this definition, Italy’s output gap might be minus 10% or 20%, or even more. Indeed Italy’s per capita GDP is 20% to 30% lower than in most northern European countries, and that gap has widened considerably in recent decades. It’s hard to see how that reflects anything other than Italy’s dysfunctional economic policies. Economists label these problems “structural”, not demand related.
Output gaps are basically about policy counterfactuals. And because there are an almost infinite number of policy counterfactuals, there is no single true output gap, which we can precisely measure. (Just as there is no single “true” rate of inflation.) The real question is: What are you trying to measure?
PS. While I believe that Italy’s output gap is now relatively small if we use the “appropriate monetary policy” counterfactual, it was considerably larger in 2009, and also in 2013, using that same definition.
HT: David Beckworth
READER COMMENTS
Jesper
Apr 23 2019 at 1:31am
Prof Sumner,
I have one question I have wondered about for a while.
You favour NGDP level targeting. Yet, implicit in the targeting rule of some % increase in NGDP you also have to make a stand on what is the productive capacity of the economy is (i.e. potential GDP). Now, whats wrong with the GDP gaps is that they require a policy maker to “know” potential GDP, which is a uncertain estimate.
But isn’t the NGDP targets subject to the same critique? You’ve stated that a NGDP target would be somewhere around 4-5 % depending on the potential GDP of the country + inflation.
Why is this different from estimating the potential GDP and calculating a gap with current GDP (abstracting from revisions)?
Scott Sumner
Apr 23 2019 at 1:55am
Jesper, No you do not need to know anything about the potential of the economy to set an NGDP target. Perhaps what I might have said is that if for some silly reason you want to average roughly 2% inflation, then you’d set the NGDP target at 2% plus the estimated trend growth in RGDP. But there’s no reason to do that, and even if you do you no longer have to monitor the output gap once the regime is first put in place.
Thaomas
Apr 23 2019 at 11:06am
This reply seems to prove too much; it would imply that the optimum NGDP target is indeterminate, one trend percentage — +150 pa -15% pa — is just a good as another. Surely one wants to get as close as possible to an optimal inflation trend. Too low and some nominal prices can’t adjust quickly enough to clear the market. Too high and it makes prediction of future relative prices more uncertain. We may think that it should be at least 2% but then we’ve never seen how the economy would have behaved if the Fed had had a 2% PL target. So some idea of the trend increase in potential real output seems necessary in deciding on a NGDP target.
Scott Sumner
Apr 23 2019 at 11:38am
You said:
“Surely one wants to get as close as possible to an optimal inflation trend.”
Not at all, there is no optimal inflation trend. You are correct that you don’t want excessively high or excessively low NGDP growth—I favor something close to 4%. But the optimal NGDP growth rate depends on the welfare costs of high and low NGDP growth; inflation has nothing to do with it. If you really wanted to target inflation, you should target wage inflation, not price inflation.
Thaomas
Apr 24 2019 at 11:10am
What would the problem be with 15% NGDP growth if it ware all real growth? It seems that in saying that “too high” NGDP growth would be non-optimal, you must be thinking that real growth is unlikely to be more than 5-6 % so the other 7-8% would be inflation that much inflation is harmful.
Actually, I don’t see any theoretical advantage to NGDP targeting v making PL targeting the “stable prices” half of the Fed’s dual mandate. Politically I see the advantage that with NGDP targeting the Fed doesn’t have to “explain” to non-economists who do not understand why both too low and too high inflation is bad, why it is a good thing in some circumstances to stimulate more inflation.
Scott Sumner
Apr 24 2019 at 1:12pm
Thaomas, A 15% NGDP growth rate that was all real growth would be bad because it would raise the tax on capital income, which is correlated with NGDP growth.
Comments are closed.