Tuesday, October 25, 2011

The Skills Shortage, and Reading a Graph

I absolutely agree with Billy that the skills shortage is a "ruse".

Let me condense a piece of his post. Billy quotes the Skills Gap Report:

Think of it this way: as many as 600,000 well-paying jobs are going unfilled while the national unemployment rate hovers around 9%.

Well I would think about it this way, Billy says:

600,000 thousand workers at the current productivity level in manufacturing would create a lot of output. That output, apparently, is not being produced but could be if there were workers available if the Skills Gap Report is to be taken literally.

That would represent a very large volume of unfilled orders sitting on the companies order books... If there was that frustrated demand some firm would work out a way to satisfy it and thus gain market share.

The fact is that no survey reveals that level of unmet demand (that is, consumers or firms with cash to spend who systematically cannot purchase goods because the firms have said they cannot supply them because of labour shortages.

I thought that was really good.

On an unrelated note, Billy presents this graph of "US manufacturing employment (thousands) since 1939":

Graph #1

If you were to look for overall trends in his graph, what you see might look like this:

Graph #2
These trend lines clearly put the problem some time around 1980.

However, when I look for overall trends in Billy's graph, what I see looks like this:

Graph #3
I see the problem starting well before 1980, in the mid- or late-1960s.


Jazzbumpa said...

On properly reading a graph.

A trend is determined by observing a series of {higher,lower} highs and lows.

The correct way to construct trend lines is across the troughs if the trend is upward, and across peaks if the trend is downward. Each line, so constructed, points clearly to the time ca 1974-1980.

Your method of constructing a quasi-midline can lead to a false-flat, as you have demonstrated in your graph #3. Note how at the left end, you are hitting near bottoms, and at the right end you are hitting near tops. That cannot possibly be good methodology.

Alternatively, take a fairly long average (to smooth the variations) and look for rate of change in the average. Or wrap a +/- 1 standard deviation envelope around the average and see where the original data points fall in that envelope.

I see three regimes:
1) up to 1974/80 increasing
2) thence to 1999/2000 decreasing slowly
3) the Shrubian collapse


Anonymous said...

My first question to companies short of skilled workers is, "What did you do with the skilled workers you had?

This is as old as the Rust Belt days when companies moved South for the cheap labor. Some returned North when they found the taken for granted industrial infrastructure of tool makers, etc. was insufficient. Somehow they expected the skilled people they abandoned were going to be waiting by the phone for their call.

American manufacturers did it again sending jobs to China. They'll come back to what?

Two years ago a manufacturer was on NPR complaining he couldn't find skilled workers. I had to listen to the end of his speechifying to learn that he was offering $12/hr for these high skill jobs. I knew right then where his workers were. They were at companies paying $25/hr.

And so it goes.