How Well Does an Asset-Based Wealth Index Rank Households? Construction and Validation against Consumption in Bangladesh's Household Income and Expenditure Survey 2022

Abstract Asset-based wealth indices are the standard measure of household economic status in surveys without expenditure data, yet they are seldom built and tested inside surveys that also measure consumption well. This paper constructs a wealth index from Bangladesh's Household Income and Expenditure Survey 2022 (14,262 households) using the composite urban–rural principal component method of the Demographic and Health Surveys, and validates it against a consumption aggregate that reproduces the official national mean to within 1%. The index has a clear first component and yields steep, almost monotonic ownership gradients. Mean consumption per person rises from 4,260 taka a month in the poorest population quintile to 12,969 taka in the richest, while the food share of consumption falls from 60 to 40%. The Spearman correlation between the index and per-capita consumption is 0.61 (0.49 rural, 0.66 urban), and 76% of the population is placed in the same or an adjacent quintile by both measures. Household rankings barely change when survey weights, item imputation, multiple correspondence analysis or an alternative coding of categories are used (rank correlations 0.98 to 1.00), whereas an equal-weight count of durable goods performs worse. The paper also shows that the low Kaiser–Meyer–Olkin values often seen for such indices come from the full-dummy coding of housing variables rather than from weak common structure, and it documents a mistranslated questionnaire item that would otherwise distort the index. The index is a sound relative measure of long-run economic status in Bangladesh, but it ranks rural households less precisely.