For a while now, I've been arguing that we're going to reach a point at which productivity growth permanently exceeds output growth, which will cause the economy to shed jobs. (Productivity) = (output) / (hours worked), so (productivity) * (hours worked) = (output). If productivity increases faster than output, then hours worked has to get smaller. Fewer hours worked generally means fewer people employed.
Or does it? To find out, I graphed GDP growth, GDP per worker (computed from BLS historical stats), and productivity growth. Here's the result:
That's a terrible mess, so here's a 5-year moving average of the same data:
That's clear enough to draw some conclusions.
First, note that productivity growth never goes above GDP growth, even in the unsmoothed version, for more than a year until we get to about 2000. Then it's a fairly regular occurrence. However, note that productivity growth is always higher relative to GDP growth as we come out of a recession, so the trend isn't completely definitive yet. The we're-not-in-Kansas-any-more moment will come when productivity growth permanently breaks trend and exceeds GDP growth. We're not there yet, but things are looking interesting.
Next, note that GDP per worker growth never exceeds GDP growth until 2000, even in the unsmoothed version. The first time that GDP per worker growth breaks through GDP growth in the post-war period is 2009.
Finally, note that there are several places where GDP per worker growth significantly underperforms productivity growth. This always occurs during periods of declining productivity growth. If GDP per hours worked is declining slower than GDP per worker, I think that means that hours worked per worker has to be declining--I think.
Bottom line: nothing conclusive here, but the trend I expected to occur shows signs of emerging. Note that this is really bad news in the long run.