Development is multidimensional — income, health, education, and more — so there is a strong pull toward combining many indicators into a single index that captures it all and ranks countries. This module covers composite indicators like the Human Development Index, their genuine usefulness, and the crucial critique that they bury contestable value judgments (especially the WEIGHTS) inside an apparently objective number.
What composite indicators are and why
A composite indicator combines multiple individual indicators into a single index, to capture a multidimensional concept (development, poverty, governance, competitiveness) in one comparable number. The appeal is real: a single number is easy to communicate, compare across countries, and track over time — it cuts through complexity and grabs attention (a country's HDI rank makes headlines in a way a table of separate indicators never would). Composite indicators have been powerful tools for ADVOCACY and AGENDA-SETTING — the HDI deliberately challenged GDP-per-capita as the measure of development (the Development course's beyond-GDP theme), and the message landed precisely because it was a single rival number. So composites serve a genuine communicative and political purpose. The question is whether that single number is meaningful, given how it's built.
The HDI and the MPI
Two influential composites
The Human Development Index (HDI), created by the UNDP (inspired by Amartya Sen's capability approach — the Development course), combines three dimensions: a long and healthy LIFE (life expectancy), KNOWLEDGE (schooling), and a decent STANDARD OF LIVING (income per capita). It deliberately broadened 'development' beyond income, and its country rankings are widely cited. The Multidimensional Poverty Index (MPI), developed by Alkire and Foster (OPHI/UNDP), measures poverty as DEPRIVATIONS across multiple dimensions — health, education, and living standards (e.g., nutrition, schooling, sanitation, water, electricity) — counting a person as multidimensionally poor if they're deprived in enough weighted dimensions. The MPI captures that poverty is more than low income (a household can have income above a poverty line but lack water, sanitation, and schooling), and it can be decomposed by dimension and group (showing WHICH deprivations and WHERE), making it useful for targeting. Both are serious, influential attempts to measure multidimensional welfare in a single framework — and both illustrate the construction problem below.
The construction choices that drive results
Where the value judgments hide
Building a composite requires a chain of choices, each of which affects the result and embeds value judgments — usually hidden behind the apparently objective final number: • Indicator SELECTION — which dimensions and indicators to include (and which to leave out) shapes what the index measures. • NORMALISATION — the indicators are in different units (years of life, dollars, percentages), so they must be made comparable (rescaled to 0-1, standardised, etc.); the normalisation method affects the result. • WEIGHTING — how much each dimension COUNTS in the total. This is the crux: the HDI weights its three dimensions EQUALLY, and the MPI uses particular weights — but there is no objective basis for any weighting (is health worth the same as income? twice as much?), so the weights are a VALUE JUDGMENT presented as a technical choice. Equal weighting is itself a contestable assumption, not a neutral default. • AGGREGATION — how to combine the normalised, weighted dimensions (add them? multiply? — and whether dimensions can SUBSTITUTE for each other: additive aggregation lets high income compensate for low health, which may be ethically dubious; the HDI switched to a geometric mean partly to limit this substitutability). Each choice can change the index value AND the country RANKINGS — and because the choices (especially the weights) are buried in the methodology, the headline rank looks objective while resting on contestable judgments. This is the same lesson as the Worldwide Governance Indicators (the Governance course): a composite's ranking is only as solid as its (often arbitrary) construction choices, and small changes in weighting or aggregation can reshuffle the rankings.
Using composites critically
So are composite indicators useful? The balanced verdict: they are valuable for COMMUNICATION and ADVOCACY (the HDI successfully broadened the development conversation) and, when decomposable (the MPI), for TARGETING (showing which deprivations and where). But they should be read CRITICALLY: the single number buries contestable choices (especially weights and substitutability), the rankings are sensitive to those choices (a different defensible weighting can reorder countries), and the composite can obscure more than it reveals (a single index hides the underlying dimensions — a country could rise in the index while one dimension worsens). Good practice (and the OECD's composite-indicator handbook): be transparent about the construction choices, conduct SENSITIVITY ANALYSIS (show how much the rankings change under alternative weights/aggregation — if they're robust, more trustworthy; if they swing wildly, the rank is arbitrary), and ALWAYS report the underlying component indicators alongside the composite (so users can see what's driving the index and judge for themselves). The composite is a useful summary and advocacy tool, not an objective truth — treat the rank as a starting point for inquiry, not a verdict, and look underneath it. This humility about composites mirrors the course's broader lesson: every summary number embeds choices, and the responsible analyst makes them visible.
Exercise
A new 'Development Index' combining health, education, income, and environmental quality (equally weighted, added together) ranks Country A above Country B, and A's government trumpets the result. (1) Explain the appeal of such a single index. (2) Explain how the equal-weighting and additive-aggregation choices could be driving the ranking. (3) Explain how a different defensible construction could reverse the A-vs-B ranking. (4) Advise how the index should be built and reported to be credible.