Arnold Kling  

Key Labor Market Indicator

PRINT
Best Comment Thread Ever... Morality is No Smokescreen...

While the media focus on job creation--a classic case of make-work bias, I remain focused on the five-year average gain in productivity. Based on the revised second quarter figures, this five-year gain was at an average annual rate of 2.4 2.2 (I mistyped the first time) percent from the second quarter of 2002 to the second quarter of 2007. That is low for the 21st century, but higher than the 1970's-1990's. I believe that it is higher than what is needed to achieve actuarial balance in Social Security [maybe I am wrong about that--see the comments].

I should note that productivity growth gets harder and harder to measure as the variety of goods and services increases and as the service sector becomes a larger share of output.


Comments and Sharing


CATEGORIES: Labor Market



COMMENTS (7 to date)
Tom Myers writes:

I'd really appreciate a discussion of "singularity"-related hypotheses about productivity growth; not so much Moore's Law in itself, but Kurzweil's generalizations as a possible framework for factual claims like the World Robotics 2006 (PDF)'s saying

the relative prices of robots fell from 100 in 1990 to 23 in 2005 without quality adjustment, and to 10 when taking into account quality improvements in robots. Other major robot using countries had similar developments...
. I often argue that we might refrain from creating hyper-intelligence, but not from the hyper-productivity of fully-automatic factories produced by fully-automatic factories, i.e. self-reproducing machines. But I would like to see a good discussion which tries to fit in (or rebut) claims like the above. Any pointers?

General Specific writes:

Measuring job creation is not a make-work bias. It is not a bias to create jobs. It is a measure of what the economy is doing. Productivity is another measure. Both are useful.

Economists use both measures equally and regularly. They are--quite likely--closely related at times, with stagnant job growth leading to increased productivity as hiring slows and those still working take on additional work tasks. Not a given, but a strong possibility.

Mike writes:

Arnold, you made a very interesting comment in passing that I feel needs some follow-up in a subsequent posting. Specifically

"... this five-year gain was at an average annual rate of 2.4 percent from the second quarter of 2002 to the second quarter of 2007. That is low for the 21st century, but higher than the 1970's-1990's. I believe that it is higher than what is needed to achieve actuarial balance in Social Security."

Could you put it on your to do list to more fully explain the theory behind this issue and more fully expand on this topic at a separate post.

If eliminating the social security problem is just a matter of keeping our productivity growth at a minimum level, that might be a good stimulant to further discussion of how policy can influence productivity growth.

Matt writes:

Job creation is unbiased if the total economy is converted to job equivalents, othrwise you likely have left a chunk of the economy on the floor that is highly covariant to some cycle in job statistics.

But, man, a continuing 2.4% productivity growth is still amazing after all these (100? years), and after the length of this cycle. That productivity growth, in human terms [as De Long would readily show in his economic history] puts us in a vey unique period of extremely rapid innovation, almost semi-automated innovation that we ha never seen before. So, our economy, in historical terms, is akin to a highly energized atom undergoing successive transformation to higher energy configuration, where we stay for just moments, then move to higher states.

Something happened, and the debate is what was it. I still put forward my theory that since the mid 19th century, the rapid rise in lifespan has doubled the length, in time, of the economic outlook. The effect was to double the precision that we can measure volatility in the economy, and respond to it.

Arnold Kling writes:

Tom,
I have talked about this in the past. See Nonlinear thinking.

Victor writes:

" I believe that it is higher than what is needed to achieve actuarial balance in Social Security."

On what do you base that belief? Table VI.D4 in the latest Trustee report suggests that a 0.5% increase in real wages over the intermediate assumption improves actuarial balance by 0.55%.

Their intermediate productivity assumption is 1.7%, corresponding to real wage growth of 1.1%. Linearly extrapolating their sensitivity analysis and holding all else equal suggests that a "break-even" level of productivity would be 3.5%.

Now, all else is not equal. If productivity were to be that high, we would likely take more leisure and work a bit less, pushing the needed productivity increase higher.

Assuming 75 years of 2.4% productivity growth rate, taken in isolation, seems woefully insufficient to restore solvency. It would need to be combined with other improved assumptions, perhaps longer work hours, more immigration, increased fertility, or increased mortality.

Tom Myers writes:

Thanks, and ouch; I had not seen the 2004 essay you link, and have pulled "Learning Economics" off the shelf and reread Chapter 16. A pleasant break from writing javascript for synchronized video.

(But I want More!) Incidentally, back in 2005, Andrew Samwick at Vox Baby seemed to be saying that you needed rather higher productivity growth,

long-term growth rates of 3.3 percent for productivity and 3.5 percent for real GDP. That productivity growth rate strikes me as too high.
It strikes me and evidently you as too low, but it is higher than the 2.4% of this post -- unless I'm misreading something again, which is very possible.
You might be amused that Samwick links to a justificatory Brookings productivity report by Robert Gordon (PDF), which says
The size of human fingers and the ability of human eyes to absorb information from tiny screens set limits to miniaturization.

But is the 2.4% adequate?

Comments for this entry have been closed
Return to top