Documentation

Python packages / tossd-reader / Understanding the data

Pillars and aggregate rows

Total Official Support for Sustainable Development (TOSSD) is an international statistical framework established by the International Forum on TOSSD (IFT) to track all officially supported financial resources for the Sustainable Development Goals (SDGs). The framework structures development finance into two distinct pillars based on delivery mechanisms and beneficiary scope, with Pillar I spanning cross-border flows to developing countries and Pillar II covering regional and global expenditures addressing international public goods.

Pillar I cross-border flows to developing countries

Pillar I comprises resource transfers provided directly to developing countries and territories on the official TOSSD recipient list. Every transaction in Pillar I associates with a designated recipient partner country or territory.

These cross-border flows encompass bilateral development projects, official grants, concessional and non-concessional loans, equity investments, and technical cooperation.

Pillar II global public goods and regional expenditures

Pillar II covers expenditures that generate shared regional or global benefits across multiple countries, and expenditures that provider countries incur within their own territories. These activities support sustainable development through two operational sub-pillars recorded in tossd_subpillar, as defined by the IFT at tossd.online:

  • Pillar II.A (coded as 21) covers issues specific to developing countries and/or their populations. This includes research and development for malaria, support to refugees and protected persons, and earmarked contributions to multilateral organisations addressing developing-country-specific issues.
  • Pillar II.B (coded as 22) covers issues of a global nature. This includes transnational climate change mitigation, biodiversity conservation, research and development for infectious diseases, pandemic preparedness, and international peacekeeping.

tossd_subpillar carries exactly those two categories, "21" and "22". A row reads NA unless it carries one of those tags, including every Pillar I row, every Pillar 0 row, and any Pillar II row left untagged by the reporting provider. Filtering with .notna() on tossd_subpillar selects rows with a sub-pillar tag, reflecting the rollout timeline described below. get_tossd_raw() returns the published sentinel codes verbatim, including the raw "1" and "2" values that get_tossd() maps to NA.

Sub-pillar implementation timeline

Sub-pillar classification phased in gradually across reporting cycles. In the 2022 dataset, sub-pillar tagging appeared in 24 rows out of 128,923 Pillar II rows (0.02% coverage). In 2023, reporting providers tagged 50.6% of Pillar II rows with sub-pillar codes. In 2024, coverage reached 99.1% across 155,908 Pillar II rows.

Because sub-pillar tagging was established incrementally, longitudinal analysis by sub-pillar is fully consistent from 2024 onward. Data for 2023 provides partial sub-pillar attribution, while data from 2019 through 2022 treats Pillar II as a unified total.

get_tossd() reflects this reporting history. Supplying a sub-pillar filter with an explicit year prior to 2023 raises InvalidPillarError. When querying without year constraints (years=None), sub-pillar filters select records from 2023 onward and emit an informational warning. Queries that examine total Pillar II volume can pass pillars=2 to include all Pillar II activities regardless of sub-pillar classification.

Aggregate provider records and double-counting protection

Published annual datasets combine activity-level project transactions from reporting providers with pre-computed summary rows. These summary records carry provider_code == 0. provider_name on those rows reads Aggregate.

The TOSSD Secretariat includes aggregate records for two reasons. Some non-concessional flows are disclosed only partially, for data-sensitivity reasons, and the Secretariat reports them as semi-aggregates worth USD 22 billion in 2024. The Secretariat also estimates resources that providers report to the OECD DAC Creditor Reporting System outside TOSSD, as with Germany and the World Bank Group, and publishes them as semi-aggregates worth USD 77 billion in 2024. In the 2024 dataset, aggregate records total 5,626 rows and USD 99.4 billion in disbursements, representing 20.0% of total reported disbursements (USD 497.7 billion).

The boolean is_aggregate column is present in every DataFrame returned by get_tossd(). The choice to include or exclude aggregate rows depends on the analytical objective. Calculating global headline volumes matching official IFT statistical publications requires retaining aggregate rows to capture all reported funding. Provider-level rankings, recipient analyses, or sector-level aggregations require excluding them (~df["is_aggregate"]) to prevent double-counting and isolate individual reporting institutions.

get_tossd(include_aggregates=False) applies that same exclusion at query time, as an alternative to filtering on is_aggregate afterward, and the tossd_reader.verbs aggregation functions default to include_aggregates=False.

import tossd_reader as tossd

df = tossd.get_tossd(years=2024, columns="minimal", units="usd_million")

# Exclude aggregate rows for provider rankings
df.loc[~df["is_aggregate"]].groupby("provider_name", observed=True)[
    "usd_disbursement"
].sum()

# Include all rows for headline totals matching official releases
df.groupby("provider_name", observed=True)["usd_disbursement"].sum()

Bilateral core contributions and multilateral double-counting

Beyond publisher aggregate rows (is_aggregate), development finance analysis involves a double-counting risk when combining bilateral and multilateral providers. The provider perspective measures a donor country's total financial effort, which includes bilateral cross-border transfers (Pillar I) plus core unearmarked contributions to multilateral organisations (aid modality B02). The recipient perspective measures resources received by developing countries, which includes bilateral cross-border transfers from donors plus the multilateral institutions' subsequent cross-border project disbursements (Pillar I).

Summing all bilateral providers and all multilateral institutions across both pillars counts the same funding twice, first as a bilateral core contribution to a multilateral fund, and second as a multilateral project disbursement in a partner country. When analysing cross-border finance received by partner countries, query Pillar I (pillars=1) and exclude core contributions (modality_code != "B02").

Provider-country expenditures in Pillar II

Pillar II includes expenditures incurred within provider territories that contribute to global sustainable development frameworks. The helper filter_provider_costs() isolates these domestic outlays by selecting Pillar II activities under sector 910 (administrative costs of donors) and sector 930 (domestic expenditures for refugees and asylum seekers in the host country). In the 2024 dataset, domestic provider costs total USD 47.5 billion across 27,275 rows, representing 35.6% of Pillar II gross disbursements.

Sector 700 records, named Humanitarian Assistance in the published data, cover in-country humanitarian assistance delivered in recipient territories and remain distinct from domestic provider expenditures.

Transitional Pillar 0 classifications

Datasets from 2021 through 2023 contain several hundred transactions recorded with pillar 0, plus a handful (10 rows) in 2020. These records are early submissions from provider entities prior to the uniform adoption of the two-pillar structure. Default queries (pillars=None) retain these rows to preserve the exact record count of the published source files. pillars="standard" selects pillars 1 and 2 together, excluding these placeholder rows.

Concessionality criteria

TOSSD spans both concessional and non-concessional resource flows across all eligible instruments. The concessionality_flag field records provider-reported concessionality status.

For debt instruments, the TOSSD methodology applies a uniform concessionality benchmark requiring a minimum 35% grant element calculated at a fixed 5% discount rate across loans and equity. By comparison, the OECD DAC grant-equivalent methodology for Official Development Assistance applies variable discount rates based on recipient income classifications (least developed countries, lower-middle-income countries, and upper-middle-income countries).

Related

View as markdown