As Jim Manzi explains, there are many different ways to achieve productivity growth:

In somewhat simplified terms, if I’m doing stuff that earns returns below my cost of capital, or if I can get someone else to do it for me at lower cost than I’m doing it, it makes sense to cut out the activity. These cut activities will tend to be those with lower productivity. Cutting activities for shareholder value reasons will therefore strongly tend to cut low-productivity activities, and increase my firm’s average productivity through pure “high-grading.” But this ignores at least a couple of important questions. First, did I fail to uncover economically achievable improvements in operational efficiency that would have allowed me to conduct these activities at higher returns and cheaper than alternatives? Second, are the cut activities linked in some non-obvious way, and potentially only over time, to the other more profitable activities, such that I have fooled myself into putting the profitable parts of the business at risk?
A business culture that ignores these questions can tend to get into a death spiral of endless high-grading against an ever-rising tide of competition that eats the business one bite at a time. The fear of many critics of American business (or “Anglo-Saxon financial capitalism”) has for a long time been that this is what is happening to the American economy on a grand scale.
And further, at the level of the entire society, while a firm can get more productive by high-grading, if the alternative employment for the people who used to work at the closed factory is collecting unemployment checks, can’t this become a society with an ever-shrinking base of people with high-paying jobs? This is the nightmare scenario of an ever shrinking number wealthy financiers, who are increasingly detached from a broader society all around them living off a combination of table scraps and handouts.
There is something to this fear. But on the other hand, the failure to allocate capital and labor from kinds of activities where there are inherent limitations to how productive they can be to those where they have greater inherent productivity will also hurt productivity growth in the long run. The key word in that sentence is “inherent.” The more we can take what is currently viewed as inherent productivity by analysts, economists and others, and improve it by unanticipated innovations, the more we can have allocative efficiency without giving up as many manufacturing jobs.

The post is lucid and well worth your time.