Arnold Kling  

Measured Productivity and Actual Productivity

What Just Happened?... Hmmm...

Is measured aggregate productivity accurate? Probably not, but I can think of important reasons that it would be understated as well as overstated.

1. For goods where there is a long history of measurement, productivity growth tends to be higher than aggregate productivity growth. Nordhaus' classic study of light, cited a couple of times in FP2P is one example. In general, if you measure the amount of hours one must work at the median wage to obtain goods in the category of food, clothing, and durable goods, it has gone way down.

2. However, for services, productivity growth is notoriously difficult to measure. With so much of our economy devoted to health care, education, finance, and government services, it is hard to gauge productivity growth. It is doubtful that the measured productivity growth in financial services earlier this decade is indicative of higher quality output.

3. Speaking of health care, if one looks at longevity and health overall, the steady improvement must reflect some tremendous increased productivity. Again, in FP2P, we cite studies showing that gains in health among Americans are as significant as the gains in GDP per capita, and those gains are not counted in productivity measures. The improvements in longevity and health may or may not be due to health care per se. However, if they are due to better nutrition, or work that is safer and less physically stressful, in some deep sense the health improvements come from productivity growth.

4. The greater variety of goods and services contributes a great deal to well-being, but greater variety does not enter measured productivity.

5. Along the same lines, I do not how to measure the added value of new computing and communication technology. The ability to post this entry using free wireless away from home does not enter measured productivity.

So, even though one can raise objections to some of what is counted as a productivity increase in the standard data, I think that one can also raise issues that go in the opposite direction. Overall, I am inclined to believe that productivity growth has been high over the past one hundred years and that in recent years it has accelerated.

Comments and Sharing

CATEGORIES: Growth: Consequences

COMMENTS (8 to date)
P Allen writes:

Great post Arnold! This is what I have alluded to over the last couple of years here as the current productivity surplus. Since 2000, measured productivity has outstripped that of GDP by an average of 1.00 percent per quarter. Over the post war period, productivity has lagged that of GDP by a similar rate. Because there is a lot of productivity, a tremendous amount of pressure is on the corresponding output gap to widen in the absence of leverage. As the Fed deleveraged the economy shortly after the conundrum of early 2005, the financial sector could not provide enough leverage to keep the output gap in an economically acceptable range.
As the economy has releveraged itself through lower rates, the gap has closed back in. This is why a lower FFTR has “worked” to date. It should have never been higher than 2.50 percent in the first place!
In essence the problem wasn’t a self-fulfilling one of too much leverage waiting to explode at anytime; it was a classic case of not enough leverage to handle the needs of the economy based on the productivity inherent in the landscape, plain and simple.

Elvin writes:

My guess is that productivity in services has been pretty high. I'm in consulting and I can put together a presentation in a matter of hours: download data from the internet, analyze it in spreadsheets, create tables and graphs, and then summarize it a PowerPoint presentation. I can even make changes minutes before the presentation. Then I send out to the client over the internet.

This was impossible 20 years ago when I first started consulting. I would have had to go to the library to find the data, photocopy it, punch it into a spreadsheet, give the analysis to someone in word processing (remember Wang computers), no color, limited graphics, etc.

Also, I can manage my work while sitting at my son's baseball game through my BlackBerry. Again, impossible twenty years ago.

Dave writes:

Don't forget about imputed rents...

E writes:


A microeconomic tangent on this subject, particularly relating to points 2 and 5: I'm curious, professionally and personally, about your thoughts about measurability of software development productivity, and what a viable methodology for doing so (if any) might look like. (I'm assuming this question has been answered in #5, in which I believe you meant to say "know how": "Along the same lines, I do not how to measure the added value of new computing and communication technology." But I'm encouraging you to openly consider the problem, even if a useful solution is unavailable.)

Martin Fowler, easily one of the profession's currently most-respected practitioners, wrote a 2003 essay titled "CannotMeasureProductivity".

I realize you're not a professional developer, but from previous entries, I know you're familiar with some programming languages. You have a JavaScript Bible review, for example, and given your work experience and career (and my own experience with econometrics), I wouldn't doubt you have experience with e.g. SQL, SAS, R, and probably C/C++. If true, IMO, it's fair to say you probably at least understand the fundamentals of developer work, and so would have some useful insights to contribute.

If nothing else, if you agree with Fowler's premise (as it seems), then I'm sure you'd still have some interesting supporting thoughts... But perhaps they are better for another post. :)

Dirtyrottenvarmint writes:

This is excellent stuff! Thank you Arnold.

If as you say productivity growth is accelerating, wouldn't this be deflationary? (I am speaking of price deflation not monetary deflation, though of course someone can probably make the logical link between the two.) P Allen seems to imply that leverage should be increased in order to counteract the deflationary pressure of productivity increases. This seems like unnecessary market intervention to me - why not just let prices fall?

Like "E", I vote for a post on your opinion re: micro productivity measurement, Arnold, if you are so inclined.

Dirtyrottenvarmint writes:

To P Allen:

(Or anyone else who cares to shed some light)

I can only find a very few comments by a "P Allen" regarding surplus production or an output gap. Perhaps I will play around with the search tool and try to find more. In the meantime, can you explain further? Or cite some references? A couple points of confusion:

1) I understand "surplus production" to be a term referring to land rents in a 3-input (Land, Labor, Capital) production model. Land and Capital are productive and have associated costs. Whatever is left over ("surplus production") is equal to the value of land rent. You seem to be using this term to reference something else?

2) On the concept of an "output gap". I think what you are saying, P Allen, is that increasing productivity plus sticky labor markets results in unemployment. This results in a decline of AD relative to AS. "Actual" GDP is then less than "Potential" GDP would be in a full-employment scenario. You seem to favor an increase in money supply/leverage in order to stabilize AD at existing levels in an attempt to return to (or preserve) full employment. Is this correct?

If so, I question why such intervention is preferable to simply allowing prices (and wages) to naturally fall to the new equilibrium level? Surely credit/monetary policy lags are in general much "stickier" than market wages and prices. Does the natural decline in prices resulting from backward-shifting demand not result in a natural closing of the "output gap" without the added complication and risk of monetary intervention? Are you of the opinion that borrowing from future production to pay for present consumption entails no increase in risk, or is it that you think there is a benefit which outweighs the cost?

Perhaps I am wrong but from my standpoint it looks like you are advocating wage/price controls, just enacted through monetary policy and credit markets rather than as outright political fiat. I am guessing you do not see things quite this way...or do you?

Robby Slaughter writes:

It's almost an unstated secret among productivity consultants that you really can't measure productivity.

The phenomenon is sort of a mix of the Hawthorne Effect and the Heisenberg Uncertainty Principle. When we get more interested in measurement than in people, we tend to break the human spirit toward innovation and creativity. That's while smarter business process improvement focuses on individual psychology, not gross evaluation of inputs and outputs.

Dirtyrottenvarmint writes:

Clearly I did not check spelling closely enough - I meant to say in a 3-input model LABOR and Capital are productive and have associated costs, and whatever surplus production is left over is Land rent. Oops.

Comments for this entry have been closed
Return to top