Wednesday, June 18, 2014

Uptrend to 1966, then stable to 1982, then uptrend again.


On 11 June I showed this graph:

Graph #1: Total Factor Productivity (TFP)
I wrote:

Uptrend to 1966, then stable to 1982, then uptrend again. I know I've seen that pattern before. I just can't remember where.

I'll find it eventually.

Browsing my test & development blog a few days later, I came upon these notes:
Saturday, May 3, 2014
It turns a corner in 1966

I have to think about what this graph shows:


Dunno, but look at the second graph at SRW's "Not a monetary phenomenon"

This one, the second graph today, I'm subtracting number-of-employees from gross-domestic-product. You can't do that. It's vaguely, vaguely related to Okun's law. But you can't subtract people from GDP. It's mixed units. (So, that graph never made it to this blog till now.)


Here's the second graph from Interfluidity #4561:

Graph #3: "RGDP divided by the number of workers in the labor force" -- SRW
(See? You can DIVIDE GDP by number-of-workers. But you can't SUBTRACT people from GDP. Remarkably, though, similar patterns emerge both ways.)

Here's Total Factor Productivity (from Graph #1 above) in blue, left scale, and the Interfluidity graph in red, right scale:

Graph #4, Comparing Graphs #1 and #3
Now that's interesting.

Okun's law defines a relation between employment and output. SRW's graph tests that relation. Looks like variations in that relation are related to Total Factor Productivity.

SRW says the baby boom's entry into the workforce outpaced the advance of technology, and is responsible for the flat spot there from 1966 to 1982. Seems to make sense. The problem I have with the idea is, when I look at that flat spot I can't help but think "that's not my fault. I didn't cause the flat spot." Subjective, yes.

I'm wondering now how they calculate TFP. They probably back into it, plugging results into some calculation and coming up with an immaculate number. The results that SRW's graph shows could easily be responsible for the pattern that TFP shows.

I'm wondering now how they calculate TFP.

1 comment:

Jazzbumpa said...


"TFP cannot be measured directly. Instead it is a residual, often called the Solow residual, which accounts for effects in total output not caused by inputs."

I don't see any obvious reason why those graphs should track so closely.