Why it matters
Productivity sets the ceiling on what an economy can pay itself. Wages that rise faster than output per hour must be funded either by compressing profit margins or by raising prices; wages that rise in line with it are affordable indefinitely. This is the arithmetic behind most arguments about whether a pay settlement is inflationary, usually without being stated.
It is also what separates growth that improves living standards from growth that merely reflects scale. An economy with more people working produces more in total and not necessarily more per person. Only the second changes what an average life can afford, which is why per-hour and per-capita measures answer questions that headline output figures do not.
Over long horizons small differences dominate everything else. A sustained gap of a fraction of a percentage point per year in productivity growth separates economies that converge from economies that fall behind, and it does so without any single visible event to point at. This is why productivity stagnation is described as a structural problem rather than a cyclical one.
What to watch for
The ratio improves when the denominator shrinks. Output per hour rises if output falls and hours fall further. Recessions frequently produce apparent productivity gains for exactly this reason, as the least productive activity is shed first. A rising figure needs the numerator checked before it counts as good news.
It is measured well in some sectors and badly in others. Counting units produced per hour in manufacturing is tractable. Measuring the output of healthcare, education or public administration requires assumptions about quality that the statistics largely cannot make, so improvements there tend to be invisible in the data.
Aggregate productivity moves partly through composition. If employment shifts from a low-productivity sector to a high-productivity one, the national average rises without any firm becoming more efficient. The reverse shift lowers it without any firm becoming less so.
Revisions are substantial. Productivity is a derived series, calculated from output and hours estimates that are themselves revised. Early readings should be treated as provisional, and the direction of a quarter can change once the underlying data settles.
Series are published by national statistical offices — the Bureau of Labor Statistics for the US, Eurostat and national offices for Europe — and by the OECD for cross-country comparison, which is the only form in which they are constructed on a consistent basis.