Productivity growth determines whether living standards can rise without working longer. Understanding what drives it clarifies why the rate is hard for policy to move quickly.
Capital deepening is the mechanical channel
Giving workers more or better equipment raises what each hour of work produces, and this is the most direct route to higher output per hour.
It requires investment, which requires firms expecting returns, so the channel depends on financing conditions and on confidence about future demand.
Its effects appear with a lag, because equipment must be installed, integrated and learned before the productivity gain shows up in the figures.
Technology needs complementary change
A new technology raises productivity only once work is reorganised around it, which historically has taken far longer than the invention itself.
Firms must redesign processes, retrain staff and often restructure management before the gains materialise, and many attempts fail at that stage.
This is why periods of rapid technological change can show disappointing measured productivity for years before the effects arrive.
Reallocation matters as much as improvement
Aggregate productivity rises when resources move from less efficient firms to more efficient ones, even if no individual firm improves at all.
That requires unproductive firms to shrink or close and productive ones to expand, which depends on competition, credit access and how easily workers can change jobs.
Policies that keep failing firms operating raise employment in the short run and suppress this channel, which is a genuine trade-off rather than a straightforward error.
Skills constrain the ceiling
Equipment and methods deliver less where workers lack the training to use them fully, so education and vocational systems set an upper bound on what investment can achieve.
The lag here is the longest of any channel, since improvements in schooling reach the workforce over decades rather than years.
Employer training operates faster but is undersupplied, because a firm that trains a worker cannot prevent a competitor from hiring them afterwards at no training cost.
Measurement is genuinely difficult
Output is straightforward to count in manufacturing and hard to define in services, where quality changes and unpriced digital goods resist measurement.
Some of the apparent slowdown in recent decades may reflect these difficulties rather than a real deterioration, though most researchers think the slowdown is largely real.
The uncertainty matters, because the policy response differs depending on whether the problem is fewer gains or gains that the statistics fail to capture.