Paris, and the twelve indicators it set

Principles became columns in a spreadsheet

The Paris Declaration on Aid Effectiveness, signed at an OECD-hosted high-level forum in February and March 2005, committed donors and partner countries to five principles: ownership, alignment, harmonisation, managing for results, and mutual accountability. Those words had appeared in aid debates before. What Paris added was a set of twelve indicators — specific, scored, tracked — against which both donors and recipients agreed to be measured at fixed points. The commitment mattered less than the mechanism for checking it.

Each indicator was assigned a target to be reached by 2010. Indicator 1 asked whether partner countries had operational development strategies — assessed on a four-point scale derived from World Bank and IMF diagnostic tools. Indicators 2a and 2b assessed the quality of partner countries' financial management and procurement systems: the alignment measures in their most operational form. Indicator 5a tracked the share of aid using partner countries' public financial management systems. Indicator 8 asked what share of aid was untied — procurement open to competitive tender regardless of donor nationality. Indicator 12 asked how many countries had put mutual accountability reviews in place.

The twelve were not symmetric. Some measured donor behaviour — how much was disbursed predictably, how much bypassed national systems. Others measured recipient-side capacity. The blending was deliberate: the architecture tried to make responsibility bilateral rather than one-directional, even if donor compliance was always the harder political variable.

The monitoring survey as the evidence base

The indicators would have been decorative without measurement. The OECD's Development Co-operation Directorate designed and administered a monitoring survey to collect them — three rounds in 2006, 2008, and 2011, the last coinciding with the Busan High Level Forum at which the framework was substantially redesigned. Each round asked the same questions of the same partner countries, producing a time-series thin enough to be fragile but real enough to be argued over.

Coverage was the first source of uncertainty. The 2011 survey, the most complete, covered around 78 countries and territories, but participation was voluntary and uneven. Smaller statistical offices — already under strain — found the reporting burden substantial. The figures that emerged were aggregates of self-reported data from partner governments and aid management databases of varying quality; the OECD directorate was explicit in its published results that cross-country comparability was limited.

Progress, where it appeared, was uneven. Alignment of aid to country financial systems improved in some regions and stagnated in others. Predictability — Indicator 7, the share of aid disbursed within the year for which it was scheduled — showed persistent shortfalls: donors routinely disbursed less than planned, and the gap rarely closed across the survey period. Most 2010 targets were not met.

What the architecture produced

The monitoring survey matters for a reason beyond whether targets were hit: it created a structured evidence base that had not previously existed for donor behaviour in aggregate. Before Paris, what donors did — how much went through which systems, how tightly tied to supplier nationality, how coordinated with other donors — was visible only in scattered bilateral reports. The survey made it comparable.

Busan in 2011 revised the framework, folding it into the Global Partnership for Effective Development Co-operation administered jointly by the OECD and the United Nations Development Programme, with a smaller indicator set and a wider range of development actors including south-south providers. The Paris indicators were effectively retired. Whether the replacement framework generates evidence of equivalent granularity remains, as of the mid-2020s, an open question.

The twelve indicators' lasting contribution is methodological: they showed that donor behaviour could be scored, compiled, and published on a fixed schedule. That is a narrow claim, but it is a real one.