OECD Municipal Waste Statistics (SDMX API)

The OECD's answer to a deceptively simple question: what do we throw away, and where does it go? Compiled by the OECD Environment Directorate as the dataflow OECD.ENV.EPI:DSD_MUNW@DF_MUNW(1.0) - formally titled Waste - Municipal waste: generation and treatment - the collection holds roughly 3,300 active observations across 58 reference areas. Every OECD member sits in it, next to five aggregates (EU27_2020, OECD, OECDE, OECDA, OECDSO) and partner economies including China, Argentina, South Africa, Ukraine and Serbia.

What makes it analytically useful is the fork baked into the schema. Generation measures count what arrives in the system - total municipal waste, household and similar waste, bulky waste, WEEE. Treatment measures count where it goes - recovery through recycling, composting or incineration with energy recovery, disposal through landfill or incineration without energy recovery. One schema, both halves of the balance, and four unit families (tonnes, kilogrammes per person, kilogrammes per 1 000 US dollars, shares of treated waste) so no chart ever needs a homemade conversion. See where it sits in the environmental facilities services data hub.

One observation per row: a geography, a measure, a unit, a reference year and the reading, with its quality flag riding alongside:

Ten documented fields define every observation, and each is a coded dimension rather than free text - which is why a join across 58 areas and several decades never drifts. The dictionary below is the complete core schema.

Three fields deserve a second look. MEASURE is the analytical spine: a controlled codelist separating generation from recovery from disposal, so a mass balance is a filter away. UNIT_MEASURE travels on every row - the same measure arrives in tonnes, per person, per 1 000 US dollars or as a share of treated waste - letting you pick the denominator your argument needs instead of reverse-engineering one. And the pair of flag columns makes data quality a filterable field: OBS_STATUS grades the reading itself while OBS_STATUS_2 names the caveat, estimated value or time-series break, that would otherwise be buried in a footnote.

Derived columns ship as additional fields on request: computed recovery shares, joined mass balances, regional rollups and crosswalks to national reporting categories.

Geography - 58 reference areas: all OECD members, the EU27_2020 / OECD / OECDE / OECDA / OECDSO aggregates sitting beside the national rows, and partners including China, Argentina, South Africa, Ukraine, Serbia, Armenia, Azerbaijan, Bosnia and Herzegovina and Belarus. Aggregates ship in the same structure as countries, so a bloc-level chart and a per-country breakdown come from one query shape.

Temporal - annual reference years. The active window concentrates on 2022 through 2024 in the August 2026 extract, and series for some countries extend back to the mid-1980s. Treat it as confirmed history rather than a nowcast: these are reference-year statistics, not live telemetry.

Granularity - national-level observations, disaggregated by waste measure and unit. There is no sub-national or facility detail anywhere in the collection; for facility-level handling records, EPA ECHO hazardous waste data covers that job at a different altitude.

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

Pick the cadence and the landing zone; the same harmonised rows arrive either way, filtered to whichever measures, units and geographies your work names. The sample comes first - rows and field dictionary confirmed against your use case before any commitment.

• Circular-economy disclosure. Benchmark recycling and recovery shares across 58 economies on one definition - the version that survives due diligence.
• Policy evaluation. Watch whether landfill and incineration-without-recovery shares bend after directive deadlines, using operation-coded treatment series.
• Market sizing. Convert tonnage tables into addressable treatment volumes per economy, then overlay per-person trends to see which markets grow in waste versus merely in population.
• Intensity benchmarking. Kilogrammes per 1 000 US dollars adjusts for economic scale; per-person units adjust for population - pick the normalization before the ranking, not after.
• Forecasting. Train on a dimension-coded annual panel whose flags tell the model which points are estimates.
• Reporting and research. Cite one consistent international series instead of a stack of national releases with different boundaries.

For the European frame of the same question, read Eurostat waste statistics.

ESG and sustainability analysts get disclosure-grade harmonisation across OECD members and partners - see esg analysts use cases. Policy and regulatory researchers get operation-coded treatment series that can evaluate a directive rather than describe it. Waste-sector investors and operators get tonnage and per-capita views of every market they might enter - market researchers use cases. Data scientists and ML engineers get a typed panel with flags as columns, ready for feature work. Journalists and academics get one citable international source - journalists academics use cases.

Three things worth knowing upfront. First, this is a counting dataset, not a pricing one: tonnes, kilogrammes and shares - sizing a market in currency needs a price layer joined separately, which we can scope with you. Second, aggregates and countries share one structure, so compute weighted figures after removing OECD-family rows unless double-counting is the effect you are after. Third, the flags matter more here than in most collections: time-series breaks mark methodology changes that will otherwise masquerade as real trends in any long-run chart. Get a sample of this dataset scoped to the countries and measures you actually report on.