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

In the study of human well-being, researchers often face a difficult choice: how to measure a family's economic standing without asking them to keep a detailed ledger of every rupee spent. Counting money is hard; people forget small purchases, hide income, or simply do not know their exact monthly totals. For decades, social scientists have relied on a clever shortcut: counting what a family owns. The logic is straightforward. A household that possesses a refrigerator, a television, or a concrete roof likely has more resources than one that relies on a tin roof and a wood fire. By tallying these possessions, researchers can build a "wealth index" that sorts families from poorest to richest. This method has become the standard for health and development surveys around the world, allowing experts to track inequalities in child health, education, and nutrition even when they lack spending data. However, a lingering question remains: does this list of possessions actually rank families in the same order as their real spending habits? If the index says a family is in the top tier, are they truly living like the top tier, or is the list of goods misleading?

A recent study by Atikur Rahman, based on the massive 2022 Household Income and Expenditure Survey in Bangladesh, sets out to answer this question with a high degree of precision. Bangladesh offers a rare opportunity for this test because its national survey collects both the detailed list of household assets and a rigorous record of what families actually spend on food and other needs. The researcher took the standard method used by international health surveys—assigning points to items like fans, computers, and types of toilets—and applied it to over 14,000 households. He then compared the resulting rankings against the actual spending data to see if the two measures agreed. The findings confirm that the asset-based index is a reliable tool for sorting households, particularly in cities, but it reveals a specific weakness when applied to rural areas where the variety of possessions is more limited.

The study began by constructing a wealth score for every household using a statistical technique that weighs different items based on how rare or common they are. In this system, owning a sofa or a computer adds significantly more to a family's score than owning a fan, because fewer people have those items. The researcher also had to navigate a few practical hurdles unique to the survey. For instance, a question in the English version of the questionnaire asked about "motor cars," but the version used by the enumerators in the field actually asked about "motors" in general. Without checking the resale values reported by the families, the researcher might have mistakenly counted thousands of households as car owners, when in reality they only owned small electric pumps. By correcting this translation error and carefully handling missing data, the study produced a clean, national ranking of economic status.

When the researcher compared these asset rankings to the families' actual monthly spending, the results were encouraging. The index successfully separated the poor from the wealthy. Families ranked in the poorest fifth of the population spent an average of 4,260 taka per person each month, while those in the richest fifth spent 12,969 taka. The index also captured a fundamental economic truth known as Engel's law: as families get richer, the share of their budget spent on food drops. In the poorest group, nearly 60% of spending went to food, whereas in the richest group, that share fell to 40%. This pattern emerged purely from the list of possessions, proving that the index reflects real economic gradients. The statistical agreement between the asset list and the spending records was moderate, with a correlation that suggests the two methods generally point in the same direction.

However, the study uncovered a crucial difference between city and country. The index worked very well for urban households, where the variety of goods—such as air conditioners, computers, and different types of toilets—provided a clear ladder of wealth. In rural areas, the ranking was less precise. Many rural families shared the same basic set of assets: a tin roof, a pit latrine, and a wood stove. Because so many households looked the same in terms of what they owned, the index struggled to distinguish between the slightly better-off and the slightly worse-off in the countryside. The study found that while the index correctly identified the poorest and richest groups overall, it was less effective at sorting the middle ranks in rural villages. This suggests that while the tool is excellent for broad national comparisons, researchers should be cautious when trying to measure small differences in wealth within rural communities.

The paper also addressed a technical concern that often confuses researchers: a statistical test called the Kaiser–Meyer–Olkin measure, which is used to check if data is suitable for this kind of analysis. In previous studies using the standard way of coding survey questions, this test often produced low scores, leading some to believe the data was too weak to use. This study demonstrated that those low scores were an artifact of how the questions were coded, not a sign of weak data. When the researcher adjusted the coding method, the test scores jumped to a level that indicated excellent data quality, yet the final ranking of the families remained almost exactly the same. This finding offers a practical lesson for anyone building similar indices: do not discard a dataset just because of a low score on this specific test if the questions were coded in the standard way.

Ultimately, the research confirms that counting what people own is a sound way to understand long-term economic status in Bangladesh. The index provides a stable picture of who is rich and who is poor, matching the spending data closely enough to be useful for policy and health research. It is not a perfect replacement for knowing exactly how much a family spends, but it is a robust tool for ranking households, especially in urban settings. For rural areas, the study suggests that while the index is still useful, it should be used with the understanding that it may blur the lines between families who are only slightly different in their means. The work also highlights the importance of careful data checking, showing how a single mistranslated question could have distorted the entire picture of wealth in the country. By validating the method against real spending and correcting technical misunderstandings, the study strengthens the confidence researchers can place in these asset-based rankings when they are used to guide decisions about poverty and development.