Britain’s Productivity Puzzle Looks Less Severe Now

I used to treat Britain’s productivity numbers the way you treat a creaky floorboard. You know it is there, you step around it, and you stop expecting it to get better. For more than a decade the story was grim and strangely tidy: output per hour barely moved after the financial crisis, living standards stalled, and every Budget seemed to rest on a hope that the next year would finally be the one. Then the official statisticians quietly changed the ruler. Same output. Fewer hours. A less embarrassing growth rate. If you have followed this debate for years, that revision feels less like a miracle and more like someone finally checking the tape measure.

Why The Productivity Puzzle Was Partly A Counting Problem

The headline shift is simple enough to remember and slippery enough to misuse. For the decade after the crisis, from 2009 to 2019, average annual productivity growth is now put at about 1.3 percent, not the 0.7 percent that had settled into speeches, charts and fiscal models. That is not a rounding error. It is the difference between a country that looked uniquely broken and a country that looked merely disappointing, sitting somewhere in the upper half of the big advanced economies rather than out on its own in the cold.

Here is the part that matters, and the part people skip. The revision does not mean the economy produced more goods and services than we thought. Output was not quietly upgraded. What changed was the estimate of how long people were actually at work. Divide the same pie by fewer hours and each hour looks more valuable. Britain did not suddenly discover a hidden factory. The clock was wrong.

I have found that this distinction gets lost within a news cycle. A better productivity ratio is welcome. It is not the same thing as a richer country. If you earned the same salary but your timesheet was corrected downward, your hourly rate looks healthier. Your bank balance does not.

How Hours Were Counted, And Why That Count Drifted

For years the main gauge of total hours came from a household labour survey. People were asked, in essence, how much they had worked in a reference week. On paper that is straightforward. In practice it frays. Response rates fell. The people who still answered were not a perfect mirror of the people who did not. And memory is a poor accountant.

Labour-market researchers have pointed out two biases that push hours up, and therefore push productivity down. Respondents often report contracted or “usual” hours rather than the hours they actually put in that week. A full-time contract of 37.5 hours survives in the answer even if the week included a dentist appointment, a school inset day, or an afternoon that simply evaporated. Second, when someone drops out of the survey for a spell, earlier answers can be carried forward. If the reason they missed the interview was a holiday, the carried-forward week still looks like a normal working week. Absences vanish. Hours inflate.

As fewer households replied, those upward biases had more room to grow. The survey did not collapse overnight. It degraded, which is worse for anyone trying to spot a trend. A noisy series can still be useful. A series that slowly leans in one direction will invent a puzzle.

Productivity is output divided by hours. If the hours are too high, the country looks lazier than it is, even when the workshops, wards and warehouses did exactly what they did.

The newer approach, often called a component method and already used in several peer economies, builds hours from several sources rather than one questionnaire. It leans on the annual earnings and hours survey, strips out leave and other absences, adjusts for overtime, and cross-checks business surveys. The picture that emerges is of falling average hours per job after 2008, and much slower growth in total hours than the old survey implied.

That is not an exotic statistical trick. It is closer to how a careful payroll manager would reconstruct a year: contracted time, minus holiday, minus sickness, plus the overtime that actually hit the system. Less romance. Better arithmetic.

What The Revised Path Actually Shows

Over the stretch from 2008 to 2024, the gap between the old and new readings is stark if you look at output per hour. On the previous method, productivity rose only modestly. On the component method, the cumulative gain is closer to the mid-teens in percentage terms, against something under 9 percent on the old path. People were more productive in the hours they worked. They also worked fewer of those hours.

Since the late 1990s, output per hour is now estimated a little over 40 percent higher, against roughly a third higher on the old series. Useful. Not transformative. If you were hoping the revision would rewind the last fifteen years and hand the Treasury a free growth miracle, it will not.

Perhaps the most interesting aspect is what did not change. The pre-crisis pace, around 2 percent a year, is still the benchmark Britain missed. A revised 1.3 percent is better than 0.7. It remains a slowdown by historical standards, the sort of deceleration that official economists still describe as the weakest run since the Second World War, and arguably since the early nineteenth century if you squint at the very long record. I would not hang a policy on the Napoleonic comparison. The post-war comparison is enough.

Three Ways To Measure The Same Unease

Output per hour is the cleanest measure of how effectively time is turned into value. It is not the only one that households and chancellors feel. Output per worker and output per job move the public finances and real wages more directly, because pay packets and tax receipts follow people and posts, not abstract hours.

On those two gauges the new method changes the story far less. There is still a sharp break after the financial crisis. The jobs market that once looked remarkably resilient starts to look a bit less heroic once you accept that hours per job were drifting down. Employment can rise while intensity falls. Both things can be true, and both showed up in the data once the survey bias was trimmed back.

Read that table slowly. The flattering revision is concentrated in the hourly metric. The measures that map more cleanly onto wages and tax are still telling a story of a long, grinding deceleration. Anyone selling the revision as “problem solved” is selling a fraction of the page.

A Slower Country, Not A Uniquely Broken One

The productivity slowdown after 2008 was not a British monopoly. Across advanced economies, investment weakened, the easy gains from the earlier information-technology wave faded, and a set of common drags showed up in the numbers. Britain looked worse than peers on the old hourly series. On the new one, it looks ordinary-to-middling. That is good news of a limited kind. Being average is better than being the cautionary tale.

It also rearranges the blame. For years the debate hunted a uniquely British villain: planning rules, managerial quality, a finance-heavy economy, weak vocational training, the long tail of low-productivity firms. Some of those suspects still deserve a file. They just do not have to explain a gap that was partly manufactured by a survey. In my experience, policy arguments harden around a chart and then refuse to soften when the chart is redrawn. This is one of those moments.

• The level of output was not revised up. The country is not secretly richer.
• Total hours grew more slowly than the household survey suggested.
• Average hours per job fell after 2008 once absences were properly deducted.
• Hourly productivity growth from 2009 to 2019 looks nearer 1.3 percent than 0.7 percent.
• Output per worker and output per job still show a post-crisis break.
• The pre-crisis pace near 2 percent a year remains out of reach on the new series.

Hold those six points together and the mood shifts from scandal to something duller and more useful: a measurement correction inside a real slowdown. Dull is fine. Dull is how you stop designing policy for a phantom.

Why Fewer Hours Can Look Like Progress And Still Hurt

There is a temptation to treat shorter hours as an unmixed blessing. Sometimes they are. A society that produces the same output in less time has, in a real sense, gained leisure. If that leisure is chosen, paid, and shared, it can be a mark of success. The trouble is the mix. Some of the decline in average hours is annual leave properly counted at last. Some is sickness, caring responsibilities, underemployment dressed up as flexibility, and sectors where demand never quite returned to the old rhythm.

Shorter hours with rising hourly productivity can still leave weekly pay flat if the cut in time outweighs the gain in the rate. That is the household version of the statistical paradox. The country looks more efficient per hour. The Friday pay packet does not necessarily agree.

Think of a small professional firm that used to bill 1,800 hours a year per person and now bills 1,650, with a higher fee per hour. The hourly figure flatters the partners in any productivity league table. Cash in the year depends on whether the fee rose enough to cover the missing 150 hours. Britain is that firm, scaled up, with the added complication that public services do not bill by the hour at all.

The Jobs Market Was Less Miraculous Than It Looked

One corollary rarely gets the airtime it deserves. If productivity was less disastrous, the employment story was less impressive. Britain’s post-crisis labour market was praised for keeping people in work even while output crawled. Part of that resilience was real. Unemployment did not spiral the way many feared in 2009. Part of the glow came from a survey that overstated hours, which made each job look less productive and the headcount look more heroic by comparison.

A labour market can absorb people into posts with thinning hours. That shows up as high employment and weak pay growth at the same time. It is not a failure of character. It is a composition effect. Retail, hospitality, care and parts of the public sector can add posts without adding much output per post. Finance and professional services can shed hours at the margin and still look efficient on an hourly basis. Mix those patterns and you get a jobs miracle that does not pay like one.

A resilient jobs market and a weak productivity record were two sides of the same mismeasured coin. Correct the hours, and both stories lose a little of their drama.

Labour-market economists, in various notes on the revision

None of this means employment does not matter. It means we should stop treating a high employment rate as proof that the growth model is fine. Jobs without hours, and hours without rising value, do not fund schools, pensions or a credible path for debt.

Recent Payroll Evidence Points To A Broader Uptick

There is a second, more current strand, and it does not rest on the old household survey at all. An independent analysis built from real-time payroll submissions and self-employment tax returns put average productivity growth near 1.1 percent since the autumn of 2024, after a fall of about 0.7 percent a year over the two years before that. The incomplete older official series, still tied to the labour survey, was still showing small declines over a similar window.

That uptick is easy to dismiss as a base effect or a sector shuffle. The authors of the payroll study argue it is not. They do not see it as a story of low-productivity retail and hospitality simply shrinking, nor as a story of a narrow technology boom or a sudden wave of automation doing the work. Twelve of nineteen broad sectors showed better productivity growth over the latest two-year stretch, including information and communications, retail, science, transport and health.

Broad-based is the phrase that should stick. A recovery confined to a handful of software firms would be real and still too small to move the national wage bill. A recovery that reaches transport, health and the shop floor is the kind that can, eventually, show up in ordinary pay. Early, yes. Worth more attention than another lecture about a puzzle we may have overstated.

1. Payroll and tax-return data avoid the worst biases of a thinning household survey.
2. The latest two-year patch looks like a turn from contraction toward modest growth.
3. The gain is spread across most major sectors, not a single glamour industry.
4. It is too soon to call a new trend, and too grounded to ignore.

I would not build a forecast on four or five quarters. I would stop writing as if nothing has moved since 2022. Those are different mistakes, and the second one is currently more common.

What This Does To Wages And The Public Purse

Productivity is not a trophy. It is the speed limit on sustainable pay rises and the silent partner in every fiscal forecast. If output per hour runs near 1.3 percent and output per worker runs slower than that, real wage growth has a low ceiling unless labour’s share of income rises, which is a fight with its own losers. The revision makes the ceiling a little higher on the hourly measure. It does not remove the ceiling.

There is an awkward implication that fiscal hawks and growth optimists should both sit with. If productivity is already growing at something like 1.3 percent, the easy “we will catch up by fixing a broken number” story gets harder. You cannot close a gap that was partly a mirage and then claim the same gap as future upside. The remaining shortfall versus the old 2 percent norm is still large. It is just no longer a story of unique collapse.

Public finances care about nominal tax bases: wages, profits, spending. A higher hourly productivity rate that comes with fewer hours does less for income tax and national insurance than a rise in output per worker. That is why the milder revision to output per job should sober anyone treating the 1.3 percent figure as a Budget gift. Debt interest does not care how elegant your hourly ratio looks.

A rough way to hold the revision in your head:
Same output
Fewer hours
Higher output per hour
Little change in output per worker
Still well short of the pre-crisis pace
Pay and tax follow the worker more than the hour

If that block feels blunt, good. The debate has had enough elegance.

Investment, Management And The Part The Revision Cannot Explain

Measurement error does not retire the deeper questions. Britain still invests less, as a share of output, than many peers. Business investment took a long time to recover after 2008, and another knock around the years of exit from the European trading bloc and the pandemic. Weak capital deepening shows up as weak productivity with a lag, whether your hours series is perfect or not.

Management quality is the other durable suspect. Studies of firm-level practice keep finding a long tail of businesses that do not track performance, do not train, and do not adopt tools that competitors treat as ordinary. A revised national average can hide that tail. The component method improves the denominator. It does not reorganise a warehouse.

Planning delays, grid connections, and a skills system that still struggles to produce technicians at scale belong in the same drawer. They were overworked as explanations for a uniquely awful hourly series. They remain plausible explanations for why 1.3 percent is hard to push toward 2. I would rather argue about those constraints with a cleaner baseline than with a survey artefact.

There is also the composition of the economy. A shift toward labour-intensive services can lower measured productivity growth even when each sector is doing fine on its own terms. Health and care absorb more labour as the population ages. Some of that is social choice, not failure. The national accounts are a poor judge of whether an extra hour with a patient was “productive.” They are a decent judge of whether the tax system can pay for it.

How To Read The Next Set Of Figures Without Getting Spun

The labour survey is still wounded. Response rates do not heal because a methodological paper has been published. Until the hourly series is fully anchored on the component approach, any single quarter will be a noisy guest. Treat large swings with suspicion. Look at payroll-based estimates alongside the official release. Ask which denominator moved.

A practical filter, the one I use when a chart is waved at me:

• Did output rise, or did hours fall?
• Is the gain in output per hour matched by output per worker?
• Is the move broad across sectors, or confined to one industry?
• Does average weekly pay agree with the productivity story?
• Are revisions larger than the headline change? If yes, wait.

That last point is easy to skip and expensive to ignore. When the revision is bigger than the news, the news is the revision. We just lived through one of those episodes. It will not be the last, because the underlying survey is still thin.

What Households Should Take From A Cleaner Number

For anyone watching a mortgage, a pension forecast or a public-sector pay round, the practical translation is modest. The country was not quite as unproductive, hour for hour, as the old charts claimed. It was working fewer hours than those charts assumed. Living standards still depend on whether weekly earnings outrun prices, and weekly earnings depend on hours as well as rates.

If the payroll-based uptick survives another year, wage negotiations get a slightly better backdrop. If it fades, the revision will stand as a historical correction rather than a new regime. Either way, the pre-crisis norm is not the default setting. Expecting 2 percent because that is what an older generation got is how forecasts go wrong and how political promises age badly.

There is a personal angle I keep coming back to. Productivity debates are usually conducted as if the only audience were the Treasury. The audience is also the person deciding whether a four-day pattern, a caring break, or a second job is rational. A statistics office that overstates hours will understate how hard the remaining hours are working, and it will misread why people are cutting back. Getting the clock right is not a academic nicety. It is how you stop lecturing households about effort they have already logged.

The Slowdown Is Still The Story

Official economists close to the new method have been careful, and they are right to be. The fundamental story of a slowdown after the global financial crisis survives the revision. Britain still had a weak run by its own modern history. The embarrassment of being a lonely outlier is what fades, not the deceleration itself.

That should change the tone of the argument more than the shopping list of reforms. Less national self-laceration. More attention to capital, skills, energy costs and the parts of the state that delay projects until the cost of capital has eaten the return. The revision removes an alibi for fatalism. It also removes an alibi for magical catch-up assumptions.

If productivity is already near 1.3 percent on the revised hourly path, significant further gains will have to be earned. They will not fall out of a corrected spreadsheet. Wage stagnation and tight public finances remain the default if investment stays timid and hours keep sliding for reasons households did not choose. The hopeful fragment is the broader sector uptick in the latest payroll reading. Early days. Not nothing.

A Clearer Baseline For The Next Decade

So where does that leave the famous puzzle? Smaller than advertised. Not imaginary. Britain spent years comparing itself with a distorted hourly series and concluding that something unique had broken. Something did slow, here and elsewhere. The unique part was inflated by a survey that lost respondents and kept old answers alive. Correcting that does not reprint the 1990s. It does let the next argument start from a less theatrical place.

I will take that. A country that knows its hours, its output and its sector mix can argue about investment and pay without first arguing about the ruler. The ruler was bent. The slowdown was real. Both sentences can sit on the same page, and both should.