Arnold Kling  

Job Losses vs. Unemployment

PRINT
You Call this Stagnation?... The Libertarian Penumbra...

Christian Stucchio writes,


it looks like job losses in construction and manufacturing are huge! In contrast, job losses in finance or business are much smaller, government is flat, and health and education have actually gained jobs!

On his post, you can see two graphs. One graph, showing for each sector the ratio of unemployment now vs. unemployment before the crisis, shows no structural shift at all. The other graph, showing absolute job losses by sector, shows more of a structural shift.

I am not surprised by this, although I do not agree with Stuccio's explanation. I think it is a case of measuring an absolute number vs. measuring a ratio. Hypothetically, if before the crisis there were 400,000 unemployed manufacturing workers and 10,000 unemployed high-tech workers, while after the crisis there were 1 million unemployed manufacturing workers and 25,000 unemployed high-tech workers, would you say that there has been no structural shift? Depends on how you look at it, I suppose.

These attempts to shape data to make a point ("It's mostly aggregate demand!" "No, it's not!") do not impress me in the least. In the long run, if more people adopt the paradigm of Patterns of Sustainable Specialization and Trade, they will stop looking at static concepts like sectors and instead look at the dynamics of what I call discovery. I think if we are going to get anywhere with the new paradigm, we will have to come up with new metrics. I think we will have to get a handle on adjustment costs, including the cost of firm formation, the cost of expanding enterprises, and the cost of integrating new labor and capital into an existing enterprise.


Comments and Sharing





COMMENTS (6 to date)
Chris Stucchio writes:

Hi Arnold,

My graph measures a ratio as well (albeit shifted to zero):

1 - Jobs(2007)/Jobs(2010).

I would argue that this data bolsters the need for an alternative story, such as your PSST, ZMP, or perhaps the "hiring as an investment" theory (sorry, don't know a technical name for this one). PSST suggests the old patterns of trade involving construction are not sustainable, whereas the theories you are attempting to displace do not.

This is evidence in favor of PSST and related theories which distinguish between housing and computers, and against theories involving "GDP factories" and simple calculus.

I'm arguing that theories along the lines of what you propose are necessary, not that I know how to build them.


Chris

J Oxman writes:

I did some similar work here, although I think my pictures are nicer.

I also go back to 1995, and index the level of employment by sector (which I see you don't like!) to 2007, and then to 2000. There might be some general trends in there, but it looks like a lot of shocks occurred too.

Philo writes:

You call for economists to "stop looking at static concepts like sectors and instead look at the dynamics of what I call discovery." Why do you consider looking at sectors "static," and looking at discovery "dynamic"?

Dividing the economy into a relatively small number of sectors for purposes of analysis is intellectually manageable. Discovery, by contrast, seems different in each individual, and there are too many individuals. One might try to simplify matters by grouping individuals, and focusing on how discovery is carried out by a representative member of each group; but this looks a lot like the hated division into sectors.

"I think we will have to get a handle on adjustment costs . . . ." If you do this individual by individual, or even firm by firm, you will have an awfully unwieldy "handle." How about doing it *sector by sector*?

prometheefeu writes:

I don't think this is productive. The point of data should be to help us decide between stories. if PSST can't explain the data that Keynsians can explain, it's not more useful than Keynsian stories. There are clearly conflicts between the two stories and psst should show some sectorial differences here. Instead of dismissing the data as unimpressive show us that psst can explain this data. Psst is useful only if it explains more than the keynsians, not just different stuff.

prometheefeu writes:

I don't think this is productive. The point of data should be to help us decide between stories. if PSST can't explain the data that Keynsians can explain, it's not more useful than Keynsian stories. There are clearly conflicts between the two stories and psst should show some sectorial differences here. Instead of dismissing the data as unimpressive show us that psst can explain this data. Psst is useful only if it explains more than the keynsians, not just different stuff.

Steve Roth writes:

>I think if we are going to get anywhere with the new paradigm, we will have to come up with new metrics.

Go for it. Would truly love to see it.

Decided absence of graphics and numbers in posts here. Lots of theory. Show us the facts on the ground.

Comments for this entry have been closed
Return to top