A 71% access average can conceal a 9% bottleneck

Why averaging four services does not tell us who has all four.

Lao PDR’s captured 2023 figures show 92.6% access to basic drinking water, 87.0% to basic sanitation and 96.5% to electricity. Clean cooking access is just 9.0%. The arithmetic mean across the four is 71.3%.

That mean is a valid average of four service shares. It is a poor answer to the question “what share of people has all four?” Even if everyone with clean cooking also had the other three services, joint access could not exceed the smallest share: 9%.

Two calculations worth keeping separate

Essentials Floor names the weakest marginal service share. The mean-minus-floor gap is 62.3 percentage points in this case, flagging how uneven the four service indicators are. It does not imply a measured joint deprivation rate.

For a compatible common population, the four marginals imply an all-four intersection between max(0, their sum minus 300) and their minimum. Here that conditional interval is 0–9%. The width reflects missing information about overlap; it is not a survey confidence interval. We do not know where the true joint share lies within it.

What this means for our own ranking

Our experimental index averages the four access indicators into an essentials domain. This example identifies a limitation of that choice: a strong value in one service can compensate for a weak value in another. A dashboard should expose the weakest service alongside the average, especially when the question concerns a complete package of essentials.

Source series can use different reporting systems and estimation methods even at the same endpoint. The mathematical bounds are conditional on their shares describing compatible populations. Microdata or a harmonised joint-access source would be needed to turn the overlap question into a direct observation.

Explore the six diagnostic models →