Development monitoring cannot work around country systems forever

The OECD and UNDP published Making Development Co-operation More Effective Progress Report 2026 on 15 September 2026, presenting findings from the Global Partnership’s latest monitoring round. The Global Partnership says the 2023–26 round drew on reporting from 44 partner country governments and 94 development partners. Its findings are not just a scorecard for aid administration. They pose a practical question for anyone producing evidence about development: can the people responsible for a national plan see how outside-funded work relates to that plan?

The OECD and UNDP report that development partners’ use of country results frameworks has stagnated, while their use of country data and systems is low and declining. A project can have a detailed monitoring plan and still be difficult to place alongside national priorities if its indicators, records and reporting cycle sit apart from those used by the country. The issue is not whether project-level evidence exists. It is whether that evidence remains useful beyond the project.

A budget gap is an evidence gap

Among countries recording development co-operation on their budgets, the OECD and UNDP find that the average share recorded fell from 61% in 2018 to 41% in 2026. The qualifier matters: this is an average for countries that record such funding, not a measure of every country’s budget coverage. The Development Policy Centre also notes a decline in the share of aid recorded on national budgets across almost all developing-country categories between 2014 and 2026.

For monitoring teams, budget visibility should be treated as part of the evidence chain, not merely a finance-office concern. If an activity is hard to find in a national budget, it may also be harder to connect spending, delivery and results in a form that public officials or local organisations can scrutinise. That does not mean a missing budget entry proves a project failed. It means the conditions for checking how it fits with public priorities are weaker.

Participation needs information people can use

The Global Partnership reports that 62% of 52 countries participating in the two most recent consecutive rounds registered a decline in multi-stakeholder development-effectiveness monitoring frameworks. That finding deserves attention alongside what CIVICUS reported: more than 70 per cent of civil society organisations surveyed rated consultations as limited in quality. Inviting organisations into a process is not the same as giving them a meaningful opportunity to test its evidence or influence what follows.

The problem is sharper when data cannot show whose circumstances are changing. CIVICUS reported that only 12 per cent of countries reported high availability of disaggregated data needed to track progress for people most at risk of being left behind. Meanwhile, the OECD and UNDP report that only around half of partner countries publish progress reports associated with their national development plans. A careful team should therefore ask not only whether it collected disaggregated data, but whether it can share findings responsibly and in a form that others can compare with public reporting.

Treat the findings as a design test

There are signs of capacity to build on. The Global Partnership reports that 45% of partner country governments had systems to track and publicly report budget allocations for gender equality and women’s empowerment, compared with 31% in 2018. The Global Partnership also says that, by September 2026, 25 partner country governments had convened development partners and non-state actors to discuss monitoring findings and agree commitments for change. Neither a tracking system nor a meeting guarantees that evidence will change decisions; both can provide places to test whether it does.

Before launching a study, a research team should identify the relevant national indicators, budget classifications and reporting commitments, then document where its own measures cannot align. During analysis, it should preserve those differences rather than present a project dataset as a substitute for a country system. When sharing results, it should make clear what was measured, whose experiences may be missing and what a local reviewer can verify. That is a more demanding standard than producing a polished report, but it gives evidence a better chance of being challenged and used.