The monthly Employment Situation Report will always be one of our favorite releases not just because it’s so economically relevant, but because of how many interesting details there are in the data. This month, the BLS reported a very strong 31,000 net new jobs created in the manufacturing sector, with upward revisions bringing January’s total to 25,000. Over the last three months, 96,000 manufacturing jobs have been added, a far cry from the sector’s small job losses in 2016. The current ramp up in manufacturing jobs suggests that something is very different about this economic cycle.

Below we show the cumulative change in manufacturing jobs over the course of each economic expansion and contraction since the 1948 recession. Prior to the 1980s, there was a familiar pattern of huge additions to manufacturing payrolls in expansions, with big job losses in recessions. At the end of the 1970s, total payroll employment peaked out at 19.55mm. But cycles since have been different. While the mid-1990s expansion saw reasonably robust payrolls additions to US manufacturing, they were relatively small versus prior expansions. Then, in the 1990s cycle, cumulative manufacturing payrolls declined over the course of the full expansion, a result without precedent post-WW2. That was just a prelude, though. Over the full course of the 2000s expansion (6 years, from November 2001 to November 2007), manufacturing payrolls fell by more than 13% or more than 2 million jobs. That was in spite of robust expansions in aggregate employment, GDP, and stronger inflation. In fact more manufacturing jobs were lost in the 2000s expansion than any post-WW2 recession including 2007-2009! So why is this expansion different? Almost one million net new manufacturing jobs have been created, with a bit more than a quarter of those coming in the last 12 months. Manufacturing payrolls’ secular decline from 1990-2010 now appears to be over.

To see this, in the chart below we show the cumulative change in manufacturing payrolls in recessions (dark blue lines) and expansions (light blue lines). While total manufacturing payrolls are still a shadow of their 1970s self (35% below the record level from 1979), their solid gains this expansion is a big change from the last few cycles. The irony, of course, is that return to normal cyclical behavior comes just as US policymakers have shifted towards a policy that was more applicable in the last few cycles but no longer looks as necessary. We should note, of course, that US manufacturing output using monthly Fed data on real output is only 2.5% below 2008 record highs. In other words, while manufacturing payrolls went into secular decline, manufacturing output never did, as factories became much more productive.