Labour Productivity  ·  Japan measured against four peer economies
2014–2024  ·  JP · US · UK · AU · CA

Japan’s productivity is falling behind. The three numbers most often quoted all say otherwise.

Output per hour, per worker, and against the OECD field. Each reassuring figure is real and correctly computed — and each is the wrong index for the question asked of it. This report pairs every flattering reading with the measure that settles the point.

The reading that flatters
Growth in output per hour, indexed to 20142nd of 5
OECD rank, output per hour28th, ▲2
Output per hour, 2024$60.10
The reading that decides
Dollars added per hour, 2014–2023$3.93
OECD rank, output per worker32nd (2023)
Real output per worker, 2024−0.6%
E1 · The starting position

Japan produces $56.80 of output per hour worked. The United States produces $97.00. That distance, not any growth rate, is the subject of this report.

Output per hour worked, 2023 · current PPP, US dollars

Japan is last in the G7 on this measure and has been throughout the decade, below an OECD average of roughly $71. The Japan Productivity Center's 2025 edition puts Japan at $60.10 per hour (¥5,720) in 2024, 28th of 38 OECD members.

That 28th place is a two-position improvement on the postwar-low ranks of 2021–22, and it is the first flattering reading. E4 shows why it does not mean what it appears to.

Excluded from the scale: Ireland at $151, and Luxembourg and Norway above $120. Multinational profit-shifting and energy rents inflate these figures and with them the OECD average against which Japan is judged. Any ranking including them overstates Japan's shortfall.

Given a gap this size, the question is whether Japan is closing it. Growth data appears to say yes — E2.
E2 · The flattering reading

Indexed to 2014, Japan's line is the second-steepest of the five. On growth alone Japan beats Canada, the UK and Australia.

Progress since 2014 · output per hour worked · 2014 = 100

Solid lines are observed annual values (OECD, constant 2015 PPP). Dashed lines are trend rays drawn from each country's published 2014–2023 CAGR to its 2023 endpoint; the endpoints are real, the intermediate path is not observed. No matched annual series on a single consistent PPP basis was obtainable for those three.

Read at face value this chart exonerates Japan. Compounding at roughly 0.8% a year it outpaces Canada at 0.7%, the UK at 0.5% and Australia at near zero, trailing only a United States running at 1.3%. The familiar story of uniquely Japanese stagnation does not survive contact with it.

The defect is structural to the chart, not to the data in it. Indexing every country to 100 discards the one fact E1 established: they did not start in the same place.

Equal percentages on unequal bases are unequal amounts. Re-read all five in dollars at E3.
E3 · The correct index — distance, not slope

All four peers lost ground to the United States over the decade. Japan lost the least in percentage terms — and still fell $6.71 further behind per hour, from by far the furthest back.

Output per hour, 2014 → 2023 · hollow marker is 2014, solid is 2023 · figure at right is the decade gain

The index chart at E2 ranks by slope. Ranked by distance covered, the order changes completely. The United States added $10.65 an hour. No peer added even half of that: Canada $4.44, Japan $3.93, the UK $3.50, Australia $0.38.

Japan’s 0.8% a year, the second-best growth rate in the group, converts into the third-largest absolute gain — behind Canada, which grew more slowly from a higher base. That is the whole defect of index numbers in one line.

Measured against the frontier, every peer slipped: Australia by 10.4 points of the US level, the UK by 6.1, Canada by 4.1, Japan by 2.7. Japan slipped least. It also started at 61% and ended at 59%, which is a different kind of problem from Australia falling from 99% to 88%.

Share of the US level, 2014 → 2023
Derived · 2023 current PPP frame2014 $/hr2023 $/hrGain% of US 2014% of US 2023Change
United States86.3597.00+10.65100.0100.0
Australia85.2285.60+0.3898.788.2−10.4
United Kingdom76.2079.70+3.5088.282.2−6.1
Canada68.5673.00+4.4479.475.3−4.1
Japan+3.93

Derivation. 2014 levels are back-cast from each country’s 2023 current-PPP level using its own published 2014–2023 real CAGR. This is the only frame in which all five are comparable, since no matched constant-PPP annual series exists for the UK, Canada and Australia. The frame shifts levels slightly — Japan reads 61.2% → 58.6% here against 63.5% → 61.0% on the observed constant-2015-PPP series below — but the change is the same to a tenth of a point (−2.6 vs −2.5). Directions and magnitudes are robust; absolute levels are base-dependent, per E5.

The one observed pair · Japan and the United States, constant 2015 PPP · shaded band is the gap

On the one pair with a fully observed annual series, Japan moved from 63.5% of the US level in 2014 to 61.0% in 2023. It grew and lost ground at once, which is what growth on a small base does against growth on a large one.

One refinement, and it matters: this is a step, not a slide. Through 2019 the gap drifted from $27.20 to $29.00, widening 6.6% over five years. Then it jumped $3.90 in 2020 alone, when US measured productivity spiked as low-productivity sectors shut down. Since 2020 the gap has been flat to slightly narrowing.

Japan did not fall in 2020. The United States jumped. Australia’s 10-point collapse against the frontier has the same cause. That distinction routes the enquiry two ways — to Japan’s own measure of itself E4, and to what the US did E7.

Observed · constant 2015 PPP’14’16’18’20’22’23
United States74.675.577.482.082.383.6
Japan
Gap27.227.128.332.931.732.6
The per-hour measure has a second defect, independent of the base problem — its denominator is shrinking. E4.
E4 · The correct index — per worker, not per hour

Japanese hours per worker fell about 7%. Output per hour rises partly because the hours fell. Output per worker, which cannot be flattered this way, sank to 32nd of 38.

Average annual hours per worker, 2023
Japan's OECD rank of 38 · plotted downward

Japanese hours fell from roughly 1,730 a decade ago to 1,607 in 2023, driven by work-style reform (働き方改革) and a rising share of part-time, non-regular and elderly workers. Productivity per hour is output divided by hours. Shrink the divisor and the quotient rises without a single extra unit of output.

Strip that out and the picture inverts. Per worker, Japan fell from about 21st in 2010 to 32nd of 38 in 2023, its lowest position since 1970, at $92,663 against an OECD average near $129,600. The 2024 recovery to 29th arrived in the same year real output per worker fell 0.6%, the first decline in four years.

This resolves the tension in E2. Japan's creditable per-hour growth is substantially borrowed from the denominator. The per-worker series is what the economy actually did.

Why this is the deciding measure. Per-hour productivity answers how efficient an hour of Japanese labour is. Per-worker productivity answers how much the economy gets from a person. For an economy with a shrinking workforce the second determines output, tax base and living standards. The first can improve indefinitely while the second declines.

Forward test. If Japanese hours stop falling, per-hour and per-worker growth converge and the per-hour figure loses its cushion. That is the cleanest single thing to watch.

A third flattering reading operates on the currency rather than the measure — E5.
E5 · The deception that runs the other way

Germany overtaking Japan's economy in 2023 was mostly the yen, not Japanese output. Here the headline overstates the decline rather than concealing it.

Japan's 2023 nominal GDP of about $4.2tn (¥591.48tn) was passed by Germany's $4.5tn, dropping Japan to fourth in the world. The yen had gone from roughly ¥110 to over ¥150 to the dollar. As Fitch's Brian Coulton observed, the overtaking in dollar terms owes a great deal to the yen's collapse — Japan's real GDP has outperformed Germany's since 2019.

PPP conversion cushions this substantially: the JPC used ¥94.68 to the dollar for 2023, not the market rate. This is why every level figure in E1 and every series in E3 is PPP-converted, and why market-rate comparisons should be read as commentary on the currency rather than on productivity.

With the measurement settled, the decline can be located. It is not where most people assume — E6.
E6 · Where the deficit actually sits

Japanese manufacturing is unremarkable but competitive. The entire shortfall is in services — roughly three-quarters of the economy, running at about two-thirds of the US level.

On a per-worker basis the JPC ranks Japanese manufacturing around 19th to 20th of the 34–35 major OECD countries — middling, and as high as 3rd in the mid-2000s per RIETI. That is not an economy that has forgotten how to produce.

Services are the whole story. At 70–80% of GDP and employment, they run at roughly two-thirds of US service-sector productivity, a gap exceeding 30% that has barely narrowed since the early 1990s. The weakness concentrates in small firms and in retail, food service and healthcare. McKinsey put Japan at 50–70% of its productivity potential, a third of the upside sitting in advanced manufacturing, retail, financial services and health care.

These are the sectors that absorb the most labour and the least capital, which is why the deficit shows hardest in the per-worker measure of E4 rather than the per-hour one.

Causal chain, assembled. An ageing and shrinking workforce, thin capital deepening and chronic ICT and software under-investment, in an economy where low job mobility prevents reallocation, concentrated in a large low-productivity SME and service base — while statutory reform cuts hours. The falling hours flatter the per-hour index E2; the stagnant services and thin capital show through in per-worker output E4; the low base means adequate growth still loses ground E3.

Japan's growth rate looks acceptable only against a peer group that is itself struggling — E7.
E7 · The peer group

Only one of these five economies accelerated. Japan's growth looks respectable mainly because it is measured against three stalled comparators.

Output per hour · compound annual growth, 2014–2023
United States

Acceleration

Heavy tech and AI capital expenditure, high firm entry, and unusually large post-pandemic labour churn moving workers toward more productive firms. Running above its pre-COVID trend, and the direct cause of the 2020 step at E3.

BLS nonfarm business: +1.2% (2023) → +2.7% (2024)
Canada

Investment failure

The Bank of Canada called it an emergency in March 2024 — time to break the glass. Business investment per worker fell from about $18,363 in 2014 to $14,687 in 2021. Output per hour is now near 60% of the US level, down from 67%.

55¢ of capital per worker per US dollar, 2024
United Kingdom

The puzzle

Growth collapsed after 2008 and never resumed: weak capital deepening, low R&D and infrastructure investment, Brexit uncertainty, and extreme regional dispersion — only London and the South East beat the national average.

0.5%/yr since 2008 against ≈2%/yr before
Australia

Bubble, burst

Measured productivity spiked when low-productivity sectors shut during COVID, then fully reversed as hours returned. The Productivity Commission's own framing: the pandemic productivity bubble has well and truly burst.

−3.7% in 2022–23, largest fall in the series
Every flattering reading has now been paired with its correction — the verdict.
Conclusion

Japan is declining relative to its peers. Slowly, from the lowest base in the G7, and behind a set of statistics that individually all look acceptable.

The three readings that reassure — second-best growth E2, a rank improving to 28th E1, and a headline per-hour figure of $60.10 — each measure something real and each answer the wrong question. The three that decide — $3.93 added per hour against the frontier's $10.65 E3, a per-worker rank of 32nd E4, and real per-worker output falling 0.6% in 2024 — point one way without exception.

Use these indices

Output per worker for the state of the economy. The absolute dollar gap, or Japan as a share of the US level, for the trajectory. Constant PPP for anything across time. Never an index number alone, and never a market exchange rate.

Discard the stagnation cliché

Japan's 0.8%/yr beat Canada, the UK and Australia, and it slipped least of the four against the US frontier. The problem is not momentum. It is a base so low that adequate momentum still converts into only the third-largest dollar gain, plus a per-worker measure sinking underneath it.

The real outlier is the US

All four peers lost ground to the frontier, and the US added more per hour than the four of them combined E3. This is an American acceleration story before it is a Japanese decline story. Whether that dividend holds above 2% is the largest open question here.

What would change the read

Japanese hours stabilising while per-worker output rises. Japanese service-sector and SME productivity improving E6. US quarterly productivity falling back below 2%. Absent those, the ranking is stable and the gap does not close.

Reference · deceptive readings and their corrections
ReadingAppears to sayWhy it misleadsUse instead
Per hour, indexed 2014=100Japan 2nd of 5 on growthIndexing discards the base. Japan's 2nd-fastest growth is only the 3rd-largest dollar gain — Canada grew slower and gained more.Dollars added per hour · % of US level
Per hour, level and rank28th of 38, up two placesHours per worker fell ≈7%. The divisor shrank; output did not correspondingly grow.Output per worker
GDP in USD, market ratesOvertaken by Germany; collapseYen went ¥110 → ¥150+. Real Japanese GDP outperformed Germany's since 2019.Constant-PPP series
Gap to OECD average ($71)Japan $14 below averageIreland, Luxembourg and Norway inflate the mean through profit-shifting and energy rents.Median ex-IE/LU · G7 set
2024 rank recoveryTurnaround underwayArrived in the same year real per-worker output fell 0.6%. One year, both directions.Multi-year per-worker trend
Japan's absolute growth rateMid-table, unremarkableMeasured against three comparators themselves stalled or reversing.Convergence rate to the leader

PPP base year

Levels here are current PPP; the time series is constant 2015 PPP. Values differ substantially by base — US 2023 is ≈$97 current, $83.6 constant-2015, ≈$79 on a constant-2020 decade average. Bases are never mixed within a chart.

Hours are not harmonised

The ONS warns explicitly that G7 hours-worked methods differ. Level comparisons of per-hour productivity carry more uncertainty than trend comparisons — a further reason to lean on the per-worker measure.

Revisions

BLS quarterly productivity moves ±1pt routinely; the 2024 annual figure was revised from +2.3% to +2.7%. JPC restates prior-year USD values each edition as OECD revises GDP and PPPs. Some 2024 figures rely on OECD nowcasts.

Mixed conversion inside JPC

JPC manufacturing rankings convert at market exchange rates while its whole-economy figures use PPP. The two are not comparable and are not plotted together here.

Incomplete annual series

No matched five-country annual series on one consistent PPP basis was obtainable. UK, Canadian and Australian paths in E2 are trend rays between real endpoints; their year-to-year volatility — Australia's COVID spike especially — is not shown.

Source framing

Underlying data is OECD, BLS, ONS, ABS and the Productivity Commission, StatCan, Bank of Canada, RIETI and JPC. Some interpretive commentary in circulation originates with advocacy think-tanks; the numbers are authoritative, the framing is not always.

Sources — OECD Productivity Statistics and Compendium of Productivity Indicators · 日本生産性本部 International Comparison of Labor Productivity (2025 ed.) · US BLS Nonfarm Business Productivity (Mar 2025 release) · UK ONS Productivity Statistics and International Comparisons · Australian Productivity Commission Annual Productivity Bulletin 2024 · Statistics Canada and Bank of Canada (Rogers, Halifax, 26 Mar 2024) · C.D. Howe Institute · RIETI · McKinsey Global Institute · Library of Parliament HillNotes (Sep 2025).
Note — figures rounded for display. Where a consistent matched series does not exist, growth is shown as CAGR rather than interpolated, and labelled as such at the point of use.