FAOSTAT - Agricultural Machinery

FAOSTAT - Agricultural Machinery is the catalog's historical anchor for farm equipment: the Food and Agriculture Organization's machinery series inside the Inputs family, holding about 10,700 rows - one per country x machine category x measure - with annual observation columns running Y1961 through Y2009 across roughly 258 area labels.

Two things separate it from the bigger FAO flagship domains. It measures capital stock rather than output: the headline element is In Use, counting installed machines rather than tonnes harvested. And it ends: nothing extends past 2009, which makes it simultaneously the longest machinery time series in the agricultural-farm-machinery slice and the least current record on the shelf. Underneath, figures compile from national questionnaires, official publications and international sources, each observation flagged with how it arrived. Get a sample of this dataset cut to the countries and categories you follow.

One row per country per machine category, year columns carrying the stock counts, blanks marking years nobody reported. Five verified rows exactly as they arrive:

dataset      : FAOSTAT Inputs - Agricultural Machinery        element : In Use
unit         : No (number of machines)

area         : United States of America        item : Agricultural tractors
Y1961        : 4690000      Y1980 : 4726000      Y2000 : 4503625
# four decades, a flat line - the fleet peaked before the series began

area         : China, mainland                 item : Agricultural tractors
Y1961        : 52239        Y1990 : 813512       Y2009 : 3515757
# sixty-seven-fold growth across the panel

Read the five rows side by side and the panel's analytical value shows itself. America's tractor count drifts from 4,690,000 to 4,503,625 across forty years - a saturated market. China, mainland climbs from 52,239 tractors in 1961 to 3,515,757 by 2009, and separately reports 17,509,031 pedestrian-controlled single-axle units that final year: mechanisation arriving on foot before it arrives on four wheels. Germany falls from 833,200 tractors in 2005 to 681,200 in 2009 and Japan from 3,075,463 in 2000 to 1,910,724 in 2005 - consolidation wearing the numbers openly. None of these curves need cleaning; they need only the countries you care about swapped in.

Seven documented fields define every row, each definition verified against the source's own file structure during research rather than inferred from prose documentation. Three identify the observation - area code, area name, machine category - one states the measure, one states the unit, and the wide Y1961-through-Y2009 block carries the values themselves. A seventh surface holds per-observation provenance flags, present in the full extraction so reported and compiled figures stay separable downstream.

Everything beyond that spine folds under additional fields on request: the flag decode joined as a readable dimension, the trade block reshaped as tidy element-year-value rows, and the pivot from wide year columns to one row per country-category-year that most panel workflows want anyway. Say which when you request the sample.

Geography - roughly 258 countries, territories and regional aggregates worldwide, from major producers to the smallest territories, with aggregates such as China, mainland carried beside their parent groupings. Country is the analytical grain; the aggregates are conveniences built on top, which is why joins should bind on the numeric area code.

Temporal - annual observations from 1961 through 2009. Depth is unevenly distributed the way real reporting always is: 2007 is the densest single year at 8,204 observations, and 146 areas report stock at least once during the 2000s. Some country-years stay blank even for large producers - India's tractor stock being the famous case - so coverage deserves a per-country check before any model leans on it.

Granularity - national totals per machine category per year. There is no sub-national, holding-level, model-level or ownership resolution anywhere in the domain; anyone promising province-level tractor counts from this source is selling a different dataset.

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

The cadence is yours regardless of the rhythm of the underlying series - these are annual observations, so a weekly warehouse load keeps dashboards current between refreshes and nobody babysits a pipeline. Every delivery ships the complete field dictionary above, the verified sample rows and the coverage profile mapped to the countries, categories and years you named, wide or tidied to one row per year.

• Cross-country mechanisation benchmarking - six decades of tractor, harvester and implement stocks per country make like-for-like comparisons possible across markets at every income level, the comparison no sales ledger can supply.
• Emerging-market mechanisation studies - China, mainland climbing from 52,239 tractors in 1961 to 3,515,757 by 2009, alongside 17,509,031 walk-behind single-axle units that year, is the canonical adoption curve the literature cites.
• Installed-base sizing for parts and services - fleet counts by category size the aftermarket that spares, attachments and service contracts sell into, with enough history to weight cohort turnover.
• Historical demand models and backtests - country-year panels that join onto crop-output, land-use and trade series give mechanisation-intensity features the depth short windows cannot.
• Trade-context pairing - import and export values for four machinery lines sit beside the stock counts in one schema, so stock-versus-flow gaps become computable rather than anecdotal.
• Citation-grade baselines - a UN-agency compilation with documented provenance per observation gives papers, policy work and market sizings a defensible number.

Market researchers and consultants rank highest in the catalog's tagging - relevance 3 of 3 - because the cross-country stock matrix is the denominator behind any credible equipment-market sizing or entry thesis; see market researchers use cases. Data scientists and ML engineers score 2: tidy numeric panels that drop straight into feature engineering, provided models respect the 2009 ceiling; see data scientists use cases. Investors and quant researchers, also 2, mine the long mechanisation history for emerging-market agriculture theses with staleness stated up front; see investors quants use cases. Journalists, academics and students cite UN-published figures with provenance attached; see journalists academics use cases. Developers building data products sit at 1 - a clean flat schema undermined by a frozen series. Sales-growth teams and competitive-intel product teams score zero: country-year totals name no companies, carry no prices and signal no competitor moves. Everything above ships daily, weekly, or hourly.

Within agricultural and farm machinery data, this record owns breadth-plus-history: roughly 258 area labels back to 1961 across twelve machine categories. The neighbors own different jobs. World Bank WDI - Agricultural Machinery, Tractors stretches further across economies - about 292 - but tracks tractors only. CEMA European Agricultural Machinery Industry Association reports the European industry's own economic indicators, closer to the trade press than to a census. FRED's farm machinery production index is the opposite measurement: American manufacturing output by month - flow against this stock. Marketplace listings sample today's dealer inventory, which is a different decade entirely.

The stock-versus-flow head-to-head is worked through on vs FRED - Industrial Production: Farm Machinery and Equipment: installed machines worldwide by country-year versus American farm-equipment output by month, and the pairing explains more than either alone.

Four things worth knowing upfront. First, the boundary: observations stop at 2009, so anything requiring a current-year read needs a companion source, with this panel supplying the structural baseline underneath.

Second, blanks are real: cells stay empty where nothing was reported, even for large producers - India's tractor stock is the standing example - so run the per-country coverage check your sample provides before modelling.

Third, aggregates ride beside countries among the 258 area labels; binding on names rather than codes risks double-counting China twice over.

Fourth, shape: upstream the values sit in wide year columns, one row per country-category-measure. If your pipeline wants tidy long format - one row per country, category and year - that reshape ships with the sample rather than landing on your desk.

Because the raw artifact is a wide, flag-coded historical table wearing six decades of inconsistent national reporting, and most questions want one clean panel. Datadory normalizes the schema, decodes the provenance flags, separates aggregates from countries, and pairs the frozen stock series with whatever current source answers the live half of your question - all scoped to the countries, categories and years you actually model. Start with a sample of this dataset, then browse the rest of the shelf on the agricultural farm machinery data hub or the best agricultural farm machinery datasets ranking.