The productivity puzzle was never a puzzle

There is a particular kind of question that survives not because it is hard but because answering it would oblige somebody to act. The productivity question has been one of those for most of a decade now. It gets called a puzzle in the papers and a conundrum in the speeches, and both words do the same quiet work: they suggest that the failure to solve it is intellectual rather than political.

It is worth being precise about what is actually being measured. Output per hour worked is not a measure of effort, and it has never been a measure of how hard anybody is trying. It is a measure of what a worker has been given to work with — the machine, the software, the training, the time. Countries that give their workers more of those things get more out per hour. This is not a contested finding. It is close to the oldest finding in the field.

Which makes the persistence of the puzzle framing quite remarkable. For ten years the explanation has been sought everywhere except the place where the evidence points. It has been blamed on measurement error, on the composition of the labour force, on the after-effects of a financial crisis that is now further away than the crisis it was compared to at the time.

What the numbers actually show

Strip out the sectors where measurement genuinely is hard — finance, professional services, anything where the output is a document — and the picture sharpens rather than blurs. Manufacturing productivity, which is about as countable as economic activity gets, has followed capital investment with the obedience of a shadow.

A decade of asking why nothing improved, from the people who decided not to spend anything on improving it.

The pattern holds across regions, and it holds across firm sizes, which is the part that should have ended the conversation. Where firms bought equipment, output per hour rose. Where they did not, it did not. The correlation is not subtle and it does not require a model to see.

Three things follow from this, and none of them are comfortable for the people who have spent the decade asking the question:

• The problem is an investment problem, and investment is a choice somebody made.
• The choice was rational for the firms making it, which is worse, not better.
• Nothing about the incentives that produced it has changed.

Why the framing survives

A puzzle has no villain. That is the entire appeal of the word. It converts a distributional question — who should have paid for the machines, and did not — into a technical one, and technical questions can be referred to a review. Reviews take eighteen months and produce recommendations, and recommendations do not appear on anybody’s balance sheet.

This is not a conspiracy. It is something more ordinary and harder to fix: an explanation that is useful to almost everyone in a position to challenge it. The firms prefer it because the alternative is an argument about capital expenditure. The government prefers it because the alternative is an argument about who funds that expenditure. And the commentary prefers it because a puzzle can be written about indefinitely, whereas a decision can only be reported once.

The honest version of the sentence is shorter and considerably less publishable. Output per hour did not rise because the things that raise it were not bought. Everything after that is a question about why, and every answer to that question has a name attached to it.