When should NGDP be unstable?
By Scott Sumner
Over time, the labor force (employed plus unemployed) usually tends to grow at a pretty stable rate. In addition, hourly wage rates are sticky, or slow to adjust to shocks. As a result, a healthy economy requires a relatively slow but steady growth in nominal labor compensation. One way to do that is to have the central bank target NGDP growth at 4% or 5%/year.
In a recent post, I suggested that the Covid recession was one case where NGDP targeting might not have worked very well. That’s because the labor force plunged much lower in March and April 2020. Given the slow adjustment in nominal wages, it was appropriate to have some slowdown in NGDP growth. For the same reason, it’s probably appropriate that French NGDP dips a bit each year in August. Thus central banks probably shouldn’t try to target current NGDP, rather they should set policy at a position expected to produce on target future NGDP. How far in the future? I’m not sure, but a year or two seems reasonable.
Here’s an example. Suppose the Fed is targeting NGDP growth at 4%/year. Also suppose that in the 4th quarter of 2024, the previous 12-month growth rate has been only 3%. In that case they might aim for a total growth of 9% by the 4th quarter of 2026. This would represent growth of 4%/year for 2 years, plus 1% more of catch up growth. I believe that sort of policy regime would have worked fine during the Covid recession, although I can imagine situations where even that approach might not be appropriate. (Say an epidemic kills 50% of the labor force.)
I’d also emphasize that the Covid recession was highly unusual. Even the Spanish flu of 1918-19 did not cause a big recession or a major fall in the labor force. (The big recession in 1920-21 was unrelated to the Spanish flu—it was caused by tight money.) So the Covid recession was very unusual, and monetary policy going forward would do better to focus on preventing ordinary recessions.
PS. Stable NGDP growth also helps with financial stability. That slightly modifies the analysis provided here, but doesn’t overturn the result.