I'm ten years burning down the road
Nowhere to run
Ain't got nowhere to go
Productivity has been making headlines recently. Specifically, there have been concerns about the data coming out of the Bureau of Labor Statistics (the BLS is the 'gatekeeper' of national productivity data) that indicates slowing productivity growth and perhaps even productivity decline. Several commentators, such as Bill Gross, have been commenting on the corrosive effects of productivity decline and why it is occurring.
I track the BLS productivity series for use in my operations management class. I recently updated my spreadsheet with an eye on recent trends. The graph below shows manufacturing and nonfarm business (essentially manufacturing + service sectors together). The nonfarm business series goes back further because, well, the BLS mysteriously truncated all years in the manufacturing series prior to 1987 a couple of years back. Why it did this I do not know.
The data points represent annual percentage changes in productivity from the year prior. Positive numbers reflect increasing productivity; negative numbers indicate productivity decline. Focusing on the non-farm business series for which we have more data (and likely to be more representative of the overall economy), note that, with few exceptions, productivity has increased each year since 1950 with the average being about 2.1%.
Until recently, there has been no noticeable trend in the data. While there have been short term excursions downward, they were followed by movement higher that made the entire series appear stationary from a statistical standpoint.
That has been changing, however. The eye is drawn to a more persistent downtrend that commenced in 2000. The graph below focuses on this period. It is difficult to ignore the series of lower highs and lower lows that define the downtrend pattern. The most recent five year period (2011-2016) is particularly noteworthy, as never in the history of this BLS series has there been a period of five successive years of productivity increases of 1% or less.
What about statistical tests of significance? Well, if we split the nonfarm business series into essentially two halves--a 1950-1980 period and a 1981-present period-- the averages for the two periods are 2.41 and 1.84 respectively. Although the average in the second half is lower, a t-test suggest that those differences are not statistically significant (p = .146). More lower data points ahead will obviously change that, of course.
I also broke the 2000-2016 period into 2000-2010 bucket and a 2011-2016 bucket. The average productivity changes of 2.62 and 0.40 in these two recent periods were, as the eye suggests, highly significant (p = .000).
Conclusion: evidence suggests that US productivity is in fact trending lower in a manner that we haven't seen in nearly 70 years.