JMTBA machine tool statistics Japan data

JMTBA Machine Tool Statistics (Japan) is the closest thing global capital-goods analysis has to a monthly starting gun. The Japan Machine Tool Builders' Association - the trade association of Japan's machine tool builders - counts what customers committed to buy in the month, and because machine tools sit upstream of nearly every manufactured good, the print is read as a forward indicator for manufacturing investment worldwide.

Three publication layers arrive each month. The flash report (jukyū tōkei sokuhō) is a single page with the preliminary figure, out roughly five weeks after the reference month. The confirmed report (kakuhō) revises it and adds breakdowns, landing about two months after month end. The main statistics compilation rolls the key series cumulatively through the latest confirmed month. Every layer reports total order value in millions of yen split domestic (uchinaiju) versus foreign (gaiju), with the momentum indices attached.

The scale of recent movement is why analysts watch it: the June 2026 confirmed month printed 203.38 billion yen, up 52.7 percent year on year - the twelfth consecutive month of growth, the sixteenth straight month above 120 billion yen, and the first month ever above 200 billion. Within Datadory's catalog of 1,744 datasets across 159 viable industries, this record is the only primary-demand series of its kind in the industrial-machinery slice: association-collected, revision-disciplined, and split by customer geography before anything else touches it. Get a sample of this dataset scoped to the months you model.

Two consecutive prints for mid-2026, exactly as they arrive, followed by the coverage check that anchors the archive: one row per document per reference period, carrying the order value, its domestic/foreign split and the momentum indices. Flash figures are preliminary; confirmed figures supersede them.

document     : jukyū tōkei sokuhō (flash report)   reference_month : 2026-07
total_orders : 193,102 million yen
uchinaiju    : <domestic demand component>          gaiju : <foreign demand component>
zengetsuhi   : 94.9   # month-on-month index, 100 = flat
dōgetsuhi    : 150.4  # year-on-year index
nenruikei    : 1,248,243 million yen cumulative 2026, ytd index 137.8

document     : English news release No. FY 26-4      issued : 5 August 2026
reference    : June 2026 confirmed
total_orders : 203.38 billion yen   +14.9% m/m   +52.7% y/y
note         : twelfth consecutive month of year-on-year growth; total above
               120 billion yen for the 16th consecutive month and above
               200 billion yen for the first time

archive check: earliest confirmed report January 2009 | latest flash July 2026

Read the two prints as one argument for the revision structure. The July flash lands first with the headline and the indices; the confirmed release then restates the prior month with the breakdowns and an English-language summary attached. A pipeline that ingests both can measure how much the first estimate moves - which is itself a signal, since a large gap between flash and confirmed says the month's demand shifted faster than usual. Rows arrive flattened to one observation per document per reference period, so a seventeen-year series assembles without reshaping.

Six fields carry every month, typed and defined below. Three carry the money - jusuke sōgaku total orders, uchinaiju domestic demand, gaiju foreign demand - and three turn levels into momentum: the month-on-month ratio zengetsuhi, the year-on-year ratio dōgetsuhi, and the calendar-year cumulative nenruikei with its year-to-date comparison. The spine is verified against the released documents during research - nothing inferred from labels alone.

Everything beyond the spine folds under additional fields on request: the deeper cuts tabulated in the multi-page confirmed releases, the main-statistics compilation aligned to the same reference periods, and the English-language summaries joined to their months as text. Name them when you request the sample and the delivered schema extends to match.

Geography - Japan, counted where the builder books the order, and already split into the two halves analysts actually use: domestic demand versus foreign demand. That split arrives on every month, so a China-led capex upswing or a home-market stall shows up in the ratio before any commentary explains it.

Temporal - monthly, with confirmed reports archived continuously back to January 2009: roughly seventeen years, about two hundred archived documents, each a few pages. The near edge sits within weeks of the present - July 2026's flash is the newest print - and earlier history remains available in the association's printed publications.

Granularity - deliberately blunt. Figures are national aggregates across the association's membership, never per-company, which keeps the series comparable across decades and immune to single-firm distortion. Depth comes in the confirmed releases' breakdowns rather than in company-level rows.

API, files, or your warehouse. Daily, weekly, or hourly.

Name the window and the fields when you request the sample - the full archive from January 2009 forward, or a cut starting at the months your models consume. Flash and confirmed layers arrive distinguished, so revisions update in place instead of overwriting the first estimate. The sample ships first either way; the ongoing feed lands on whatever cadence your models need.

• Capex-cycle timing - treat the domestic-versus-foreign split as the earliest regular read on where manufacturing investment is turning; a foreign-led surge flags export markets pulling before it reaches supplier order books elsewhere.
• Revision-aware forecasting - ingest flash and confirmed layers side by side and model the estimate gap itself, turning the publication process into information rather than noise.
• Competitive benchmarking - set Japanese order momentum beside German mechanical-engineering order intake for a two-market read on world equipment demand.
• Sales timing into the tooling chain - spot demand upswings at machine tool builders early enough to time outreach to their buyers and distributors while budgets are still forming.
• Citation-grade macro work - journalism, academic studies and analyst notes get an association-published series with named provenance instead of an anonymous estimate.

Market researchers and consultants score this highest of the eight personas in the catalog's tagging - relevance 3 - because quantifying Japanese machine-tool demand with confirmed monthly reports is the backbone of machinery market studies. Investors and quant researchers sit at relevance 2: the domestic-versus-foreign split works as a global capex leading indicator for cyclical positions. Sales and growth teams, also at relevance 2, time outreach to builders and their distribution channels off the same momentum indices. Competitive-intel product teams benchmark their own order intake against association-level totals. Journalists, academics and students cite a named institution behind every figure in coverage of Japan's manufacturing and capex cycle. All delivered daily, weekly, or hourly.

Against the rest of the shelf, this is the demand-side pulse. The NBER-CES Manufacturing Industry Database reaches back to 1958 but stops in 2018 - depth without freshness. VDMA Mechanical Engineering Industry Statistics tracks the same indicator family for German mechanical engineering, so the pair gives a two-country view of world equipment demand. Machine-tool order statistics for Europe (CECIMO) and India (IMTMA) extend the family to other producing nations. And the NASA Prognostics Data Set Repository is sensor-level engineering microdata - asset-side physics, not demand-side economics.

The head-to-head against the engineering corpus is worked through in Kaggle C-MAPSS turbofan vs JMTBA machine tool statistics: one supplies the failure physics a maintenance model learns from, the other the demand signal that decides when new machines get bought. Used together, predicted cycles-to-failure weights replacement timing while the order indices time the purchase.

Four honest caveats. First, resolution: these are national association-level totals, so company-level or prefecture-level questions need complementary sources rather than a deeper cut here. Second, revision discipline: the flash is preliminary and superseded by the confirmed print about two months later - decide whether your pipeline consumes first estimates, final ones, or both, and the delivery is shaped accordingly. Third, format lineage: the underlying documents publish as PDFs, so Datadory does the extraction once and delivers tidy rows keyed on document, reference period and field. Fourth, history boundary: confirmed archives run to January 2009 and earlier decades live in printed association publications, so pre-2009 panels come from those rather than the digital shelf.

Because the raw artifact is a monthly PDF cadence wearing two publication layers, and most questions want one tidy series. Datadory extracts the tables once, keys every figure to document type and reference period, distinguishes flash from confirmed so revisions update in place, and joins the English-language summaries to their months when you ask. Start with a sample scoped to the months and fields you actually model, then browse the rest of the industry on the industrial machinery supplies components data hub or the best industrial machinery datasets ranking.