What was left out

The logic of omission

The Millennium Development Goals were built around things that could be measured at the household level and reported against a single 1990 baseline. That design principle was also a filter. Household surveys can track whether a child is in school or whether a family has access to clean water; they are far less equipped to track how power is distributed, how ecosystems are degrading or whether growth is reaching the poorest half of the population rather than just lifting the average.

Inequality within countries is the clearest casualty. The MDG framing used national rates — the share of children enrolled, the proportion living below a poverty line — and a national rate can improve while gaps between the richest and poorest households widen. A country could halve its headline poverty headcount, as measured by the World Bank's poverty-line methodology, while the bottom quintile barely moved, and still record the goal as met. The targets gave no purchase on distributional outcomes, because distributional outcomes require disaggregated data that most national statistical offices, particularly in sub-Saharan Africa, did not yet routinely produce.

What the list could not hold

Governance sat awkwardly from the beginning. There was no MDG on rule of law, corruption, property rights or political freedom — not because drafters were unaware of their importance, but because no short, quantifiable, internationally comparable indicator for any of them existed at the time. The World Bank's Worldwide Governance Indicators had only been running since the mid-1990s, were composite and contested, and carried confidence intervals wide enough to make year-on-year comparisons almost meaningless. Putting a contested composite into a framework that traded on the authority of precise numbers would have undermined both. Governance was, in practice, handled through conditionality in debt relief and aid programmes rather than tracked as a goal in its own right.

The environment received one goal — MDG 7, which covered water, sanitation and a target on biodiversity loss — but the framing was thin relative to the scale of the challenge. Carbon emissions appeared only as a single indicator, with no target attached. Soil degradation, ocean acidification and deforestation had no targets attached to them. The OECD's creditor reporting system tracked aid flows to environmental projects, but there was no mechanism connecting those flows to measurable environmental outcomes in the way that, say, child mortality statistics connected aid to health outcomes. The result was that environmental commitments remained largely narrative.

There is also a structural gap that the goal list could not bridge: the absence of indicators for domestic resource mobilisation, tax systems and the policy environment in which growth occurs. Reaching the goals required governments to spend, but the MDG architecture said almost nothing about whether countries could generate revenue, manage public finance or avoid illicit outflows. These gaps were noted by critics at the time, and they shaped the ambition of the Sustainable Development Goals agreed in 2015 — which expanded to seventeen goals and explicitly included inequality, institutions and climate, though at the cost of a focus that many analysts argued the MDGs possessed and the SDGs did not.

The harder question is whether these omissions were failures of design or rational concessions to measurability. A goal without a credible indicator is not a goal in any operational sense; what a target does to a programme depends entirely on whether the target can be tracked. The MDG architects made a bet that a short list of measurable commitments would drive more action than a comprehensive list of aspirations. What they could not avoid was that the boundary between the measurable and the important does not run neatly, and the things left outside it did not stop mattering.