How manufacturing employment has fallen

We all know that increasing commodity prices and the resulting appreciation of the CAD has resulted in a reduction in employment in the manufacturing sector:

The shift of productive resources out of the manufacturing sector is of course the appropriate response to the relative price movements we've seen over the past five years, and the outcome has been generally positive. But there has always been the question of the loss of 300,000 manufacturing jobs. Put that way - "the loss of 300,000 manufacturing jobs" - one is left with the impression that there are 300,000 people out there whose lives have been turned upside down by forces beyond their control. Even if the economy as a whole has benefited from the structural changes we've undergone over the past five years, we should be trying to compensate those who have paid the transition costs.

That said, those costs are hard to pick out from the macro evidence:

• Employment rates are at an all-time high.
• For the first time in decades, real wages are increasing.
• There appear to be no significant increases in inequality.

But there's still the question of the 300,000 fewer workers in the manufacturing sector: just how has that been accomplished?

Stephen Tapp, a PhD candidate from Queen's University, has been investigating these employment shifts and he has some surprising answers (pdf). He has also been kind enough to provide me with some of the data behind his analysis, and they tell a story that flatly contradicts the standard narrative.

(I should also add the disclaimer that this post is my interpretation of the stylised facts of Stephen's data; his paper has much more going on, including a search model that tracks the response to sectoral shocks, and an analysis of the winners and losers during transitions.)

Let's start with the basic dynamics of how employment levels evolve:

Change in employment = hires - separations

Here's a graph of the hiring and separation rates in the manufacturing sector. The data are pretty noisy, so everything that follows is expressed as a 12-month moving average:

Surprising fact #1: Separation rates in the manufacturing sector have been fairly constant since 2002, and at levels that are at the low end of their historical range of variation. The reduction in employment levels has been accomplished by a sharp drop in new hires.

Since separations include both voluntary departures (quits) and layoffs, it could be that layoffs have increased while worker quit rates have fallen. But this is not the case:

Surprising fact #2: Layoff rates - and in particular, permanent layoff rates - have remained fairly constant, and are generally lower than they were during the 1994-2002 run-up in manufacturing employment. A manufacturing worker was less likely to have lost her job during 2002-2006 than she would have been during the the glory years of the last half of the 1990's.

What about quit rates?

Surprising fact #3: Quit rates in the manufacturing sector are higher than what they were in the 1990's.

Of course, the scale of this graph is not comparable to that for layoffs, but it would appear that manufacturing workers have options outside the manufacturing sector. Here is a graph of the evolution of unemployment in the manufacturing sector (that is, those who are unemployed and whose last employment was with a firm in the manufacturing sector):

Surprising fact #4: Unemployment among manufacturing workers - expressed in both levels and as a percentage of the (falling!) number of those employed - has been decreasing since 2002 (albeit from a 2001-slowdown-induced spike).

Here is what I conclude from all this:

• The transition of employment from manufacturing to other sectors is going much more smoothly than what we might have predicted from such a sharp decline in employment: there has been no spike in layoffs or in unemployment.
• The most important factor leading to the decline in employment in the manufacturing sector is attrition:  workers who leave are not being replaced.

Comments

[Bill Conerly]

In the U.S., the all-time peak in manufacturing employment was 1979. The all-time peak in manufacturing output was last June.

[Simon van Norden]

Pierre Fortin has been arguing lately (including in a recent op-ed in La Presse) that Canada needs to fix the exchange rate (and perhaps adopt the US dollar) in part to protect the manufacturing sector from the damage being done by exchange rate swings. Do you have any views on this argument given the above analysis?

[Stephen Gordon]

Since the swings are being driven by the relative price of manufactured goods and commodities, it's hard to see how exchange rate policy could be effective. Regardless of what the exchange rate is doing, we'd still expect to see productive resources shifting out of manufacturing.

[btg]

i look at the graphs and i think "hey, that free trade agreement certainly did wonders for our economy in the early 90s, didn't it!"

[Tejvan Pettinger]

This is quite a common experience amongst developed economies. In the UK, the decline in manufacturing has been even sharper. The question is whether new jobs (and export revenue) can be created in other sectors like services.

[Robert Badaracco]

Its simple you don't make nothing and you aint got nothin this is why were in a financial crisis. We dont work because the left made it unprofitable to hire workers. So now workers don't have jobs and the economy is going nowhere.