The history of U.S. productivity, in a nutshell

by on February 10, 2011 at 3:48 pm in Data Source, Economics, History | Permalink

There have been some recent confusions in the comments about the historical record on productivity.  The excellent Alexander J. Field sets it straight, after noting that TFP (Total Factor Productivity) growth in the interwar years was remarkably strong:

…TFP persisted at high although more modest rates during the golden age (1948-73).  But then it ground to an almost complete halt between 1973 and 1995.  Output per hour continued to rise, albeit much more slowly, but this was almost entirely attributable to physical capital deepening.  Data are now available for the entire century, and it is no longer possible to interpret the high rate of TFP advance during the interwar years that prompted the Abramowitz/Solow generalization [TC: the generalization was about knowledge-based progress] as a defining characteristic of the century as a whole.

In this context, think of TFP as the growth due to new ideas, rather than just throwing capital or labor at a problem or production process.  Here is a related Field paper.  It's also wrong to think of the post-WWII period as the peak of progress, rather as Field shows high TFP growth starts post Civil War and the time after WWII is somewhat slower than many previous decades.  The early 19th century, by the way, was not so splendid for TFP.

The critical responses to The Great Stagnation prefer to attack median income measures and in general they are reluctant to talk about total factor productivity.  Yet we are pointed very much toward the same conclusion.  My first post on TGS also considered these issues and you will find some relevant Charles Jones papers here

1 Sergey Kurdakov February 10, 2011 at 12:06 pm

think of TFP as the growth due to new ideas

why not energy and it's utilization? I mentioned Ayres works
see also… – they recognised what you try to push now – quite a while ago and within bounds of their theory

2 Tomasz Wegrzanowski February 10, 2011 at 12:37 pm

Why is this whole discussion of "Great Stagnation" so highly US-centric? How different would be conclusions if larger part of the world was included?

3 OneEyedMan February 10, 2011 at 1:25 pm

In many ways it has all the problems of GDP and more. Why? Because TFP is a decomposition of GDP into parts from hours, capital and ideas. That decomposition is from a model where as GDP is much closer to a model free concept. Believing in their measurement of TFP is akin to accepting that their model is giving the proper decomposition.

4 Chris T February 10, 2011 at 2:15 pm

According to the Field paper TFP during the 2000-2005 period (1.83) was the highest since the 1948-1973 period (2.13). Yet during this timespan median income actually fell.

This seems to be in direct contradiction to your thesis.

5 astonerii February 10, 2011 at 2:34 pm

None of the productivity measures ever seem to account for the burden of government. Funny how all the slowdown occurred a decade after the government started taking environmentalists seriously, the costs of the safety net began to balloon, taxes began to hit more and more people and overall burdens of the government came to a critical mass triggering the stagflation of the 70's. But hey, we got other ideas on what caused it, so none of that matters.

6 Hieronymus Goat February 10, 2011 at 4:15 pm

"Why is this whole discussion of "Great Stagnation" so highly US-centric? How different would be conclusions if larger part of the world was included?"

For one thing, the U.S. is almost the only country in the world that was not flattened by war or uprooted by violent revolution in the last 70 years, which overrides all other long-term considerations.

7 Troy Camplin February 10, 2011 at 9:40 pm

Hmmm. I wonder what Richard Florida would have to say about all this.

8 Chris R February 11, 2011 at 6:56 am

Putting on my cranky seminar hat here.

I have a hard time taking the pre-1929 real income data way too seriously because of index number weighting issues. Kendrick's estimates are probably the best since he varies the weights across major categories, but he uses fixed 1929 weights within major categories. This will have the effect of overstating TFP growth toward the beginning of the sample. Splicing the Kendrick data to the post-1929 chain-weighted data will make the mid to late 19th century look really good in comparison with later periods–the question is by how much. If the effect persists after correcting for index number creep, then that would be really interesting.

9 Scott February 12, 2011 at 8:36 am

The dates for the reduction in TFP are interesting from a demographic standpoint. By 1973 half of the baby boom had finished high school and the bulge in the labor force had started. 22 years later the entire baby boom had been absorbed into the labor force.

Perhaps the decrease in TFP was really just a reaction to a substantial increase in labor relative to the other factors.

As the baby boom leaves the labor force over the next decade perhaps we will see the opposite effect.

10 Mark A. Sadowski February 15, 2011 at 11:20 am

The BLS maintains a database of annual TFP for the US back to 1948. A good source of information on TFP growth prior to that is a few of papers by Alexander J. Field: “US economic growth in the gilded age”, “The Most Technologically Progressive Decade of the Century” and “The origins of US total factor productivity growth in the golden age”.

Average annual TFP growth is as follows:











The first thing that should grab your attention is that TFP growth was at its most rapid during the Great Depression. The second thing you should observe is that TFP growth was at 1.9% or higher from the 1870s through 1973 with the exception of 1892-1919 and 1941-1948. TFP growth has picked up since 1995 (but has slowed since 2005). So this pretty much supports Tyler Cowen’s conjecture.

Now, why was TFP growth faster during the periods mentioned?

Well, Field analyzes the growth by sector and sector size and comes to some interesting conclusions. TFP growth was fast from the 1870s through 1892 because of railroads (which peaked in track mileage in 1916) and to a much lesser extent because of the telegraph. Almost all growth in TFP in the 1920s can be accounted for by manufacturing and that probably fed that decade’s stock market boom. Why did manufacturing TFP explode in the 1920s? According to Field it was due to the widespread electrification of factories (which had started in the 1880s). In the 1930s manufacturing TFP, although still relatively fast, slowed down. (He also points out that private R&D quintipled from 1929-1941.) But transportation TFP soared from 1929-1941 mainly due to the five fold increase in the share of tons-miles hauled by interstate trucking and its interaction with railroad transportation. (The US built its first interstate highway system in the 1930s.) And he argues that transportation TFP was largely responsible for the growth seen from 1948-1973, as manufacturing TFP actually went negative for part of that period. (And recall the Interstate Highway System, built on top of or paralleling the US Route system of the 1930s was largely completed from 1956-1973.) TFP growth was negative from 1855 to the 1870s primarily because of the Civil War.

Recent work by Bart van Ark shows that TFP in the distribution sector was the main source of the surge in growth from 1995-2005, and he argues that was due to the widespread adoption of ICT technology by that sector. (Think big box Walmarts.)

What's interesting is that Federal government money played a major role in all of those developments with the exception of factory electrification (urban areas were largely electrified with private money). This of course comes with the qualifier that much of public fixed investment is probably nonproductive. But evidently some of it mattered a great deal.

Comments on this entry are closed.

Previous post:

Next post: