In the past I’ve been critical of Paul Krugman’s approach to austerity and business cycles. For instance, he argued that austerity in the UK led to slow growth in real GDP. I countered that RGDP is not the right variable for business cycle analysis; you need to look at employment (relative to trend.) The British economy did quite well in terms of job creation after the Cameron government took office, and the low RGDP growth was due to abysmal productivity numbers. But there is no mechanism in the Keynesian model by which austerity can reduce GDP without reducing jobs. Instead, supply-side factors seemed to explain Britain’s poor productivity numbers.

Regardless of who was right in that particular case, Krugman has now come around to my view that employment figures are more meaningful than RGDP in business cycle analysis. Here are his recent comments on Ireland and Iceland:

It’s true that Irish GDP per capita (in this case using GNI doesn’t make much difference) recovered to its pre-crisis level only a bit later than Iceland’s. But that’s not the only indicator, and it’s one that is arguably distorted by the nature of the Irish export sector, which held up fairly well and is highly capital-intensive (think pharmaceuticals) — that is, it contributes a lot to GDP but employs very few people.

If you look at employment instead, as in the chart, Iceland did far better than Ireland; and Icelandic unemployment similarly shows a much more favorable picture. Less formally, everyone I know who tracked both countries has the sense that the human toll in Iceland was much less than it was in Ireland.

I think this is exactly right, and am very happy to see Krugman making this sort of analysis. In Britain, output from oil and high-end finance dropped sharply, “but employs very few people.” It also helps explain why Abenomics has been a big success (in monetary policy terms, not the other two “arrows”.) Since the beginning of 2013, employment relative to trend in Japan has done far better than RGDP.

Let me try to head off a few possible comments:

1. Yes, I sometimes refer to RGDP data. In some cases, such as the famous 2013 “test” of market monetarism, it’s because I want to evaluate the test using the metric that Keynesians were focusing on at the time. But market monetarism also passed the 2013 austerity test with flying colors if you focus on employment rather than RGDP.

2. Yes, RGDP matters for living standards. But there is more to life that RGDP/person. During the Great Depression of the 1930s, RGDP/person in the US was not particularly low by developed country standards, but people were suffering. In the deep 2009 recession, RGDP/person in the US, RGDP/person exceeded the levels observed in Europe at the peak of a boom. Indeed if RGDP/person in the US were to fall to German levels it would be one of the great economic catastrophes in all of American history. But not because German living standards are low (they are very high), rather because 20% of Americans would be jobless.

3. Some people are surprised to hear that I am just as dismissive of RGDP as someone like Arnold Kling, given that I often talk about NGDP. But despite the similarity of names, NGDP has absolutely nothing to do with RGDP; they are completely unrelated in a conceptual sense, although obviously correlated in a statistical sense (in the US, not Zimbabwe). Here’s what RGDP looks like:

Screen Shot 2015-11-27 at 1.40.56 PM.png

In contrast, this is what NGDP looks like:

Screen Shot 2015-11-27 at 1.42.23 PM.png

The most useful way of thinking about NGDP is as the value of a dollar. Thus 1/NGDP is the share of total output that can be bought with a single dollar. It’s a way of thinking about monetary policy. It only matters for employment crises and financial crises because so many labor and debt contracts are priced in dollars, not tons of steel. Hence it really is appropriate to obsess over NGDP, even as we ignore RGDP

PS. Krugman also thinks nominal wage stickiness is a factor in unemployment. Now if he were to switch from inflation to NGDP growth as the relevant demand shock indicator, he’d be using the three components of the musical chairs model (sticky nominal wages, NGDP, and employment.)

HT: TravisV