To compute the IWI value of a household, information is needed on twelve assets. These assets include seven consumer durables (possession of a TV, fridge, phone, bike, car, a cheap utensil and an expensive utensil), access to two public services (water and electricity) and three housing characteristics (number of sleeping rooms, quality of floor material and of toilet facility).
For each of the seven consumer durables, an indicator must be created with two possible values: the value 'one' if the household (or one of its members) owns the specific durable and the value 'zero' if they do not own the durable. A similar two-category indicator has to be created to denote whether (1) or not (0) the household has access to electricity.
Quality of water supply, flooring material, and toilet facility is measured with three categories: low quality, intermediate quality, high quality. The number of sleeping rooms is also measured with three categories: zero or one sleeping room, two sleeping rooms, three or more sleeping rooms. For each of the categories of these variables an indicator is created that shows whether (1) or not (0) a household is in that category.
The total number of indicators needed for computing IWI is thus 20. Eight indicators to denote the possession of the consumer durables and access to electricity, three times three indicators to denote the quality of water supply, flooring material, and toilet facility, and three indicators to denote the number of sleeping rooms.
To rank a household on the IWI scale, its values on the 20 indicators must be included in the following equation:
IWI = constant + Σ(Wn ⋅ Xn)
where Xn are the indicator variables and Wn their IWI weights. The values of the IWI weights and of the constant are presented in Table 1 below. These weights were computed by performing principal component analysis (PCA) on a database with asset information on 2.1 million households. IWI weights are the PCA weights transposed to the 0-100 range.
If one or more indicators are missing for a household, it will still be possible to obtain an approximation of the household's true value on the IWI scale by using an adapted formula. Test analyses revealed very high correlations between IWI values based on all items and values for one item missing (about .98 for three-category items and .99 for two-category items).
As for each (combination of) missing variable(s) a separate formula has to be used, the number of these adapted formulas is rather high. They therefore will be provided in SPSS and Excel files that can be downloaded from the IWI download page.
|Access to electricity||8.056664|
|Number of sleeping rooms:|
|Zero or one||-3.699681|
|Three or more||3.445009|
How to measure the indicators?
Measurement of most consumer durable items is rather straightforward. Whether the household or one of its members owns a TV, fridge, phone, bicycle, or car can generally be answered without much problems, or observed by an interviewer. The same is true for access to electricity and the housing characteristics.
The way the categories of the 'quality' variables are defined requires some further explanation. The kind of floor material that is used may differ depending on local traditions and availability. Still the allocation of the specific flooring categories available in survey data to the three quality categories is generally not very difficult, because the quality differences between categories are often very obvious (e.g. sand, cement, or carpet as flooring material). The same is true for quality of water supply and toilet facilities; there is variation among contexts, but people generally have clear ideas of what is better or worse quality. For coding the wide range of categories available in the different surveys into the three quality categories, we used the following guiding criteria:
- high quality is private piped water or bottled water;
- middle quality is public tap or standpipe, tubewell or borehole, protected well or spring, or tanker truck;
- low quality is unprotected well or spring, cart with tank/drum, or surface water.
- high quality is private flush toilet (or flush toilet if private unknown);
- middle quality is public toilet, ventilated/improved pit latrine, pit latrine with slab, or composting toilet;
- low quality is pit latrine without slab, open pit, bucket, hanging toilet or no toilet.
- high quality is finished floor with parquet, carpet, tiles, linoleum, ceramic etc.;
- middle quality is cement, concrete, wood, bamboo etc.;
- low quality is none, earth, dung etc.
Cheap & expensive utensils
The variables 'cheap utensils' and 'expensive utensils' need further explanation, as these are constructed variables. Early experience with the use of wealth indices revealed a lack of discriminatory power at the lower end of the wealth distribution, as the number of households having none of the assets and lowest quality housing was substantial in some countries. To be able to better differentiate among these poorest groups, some cheaper assets, like having a chair, table, clock, watch, water cooker, radio, fan or mixer were included in later surveys.
As the kind and number of these cheaper assets varies considerably among surveys, it is however not possible to include them as separate items in a comparative wealth index. As an alternative, we have created the indicator 'cheap utensils' that is based on the information on any cheap (roughly under 50 US Dollar) item that is present in the data. This indicator can be created if information for one or more of such items is available. Household owning one or more cheap utensils get value 'one' and other households value 'zero' on this indicator.
The indicator 'expensive utensils' is constructed in a similar way, but with respect to the possession of expensive (over 250 US Dollar) items, like having a washer, dryer, computer, motorbike, motorboat, air conditioner, or generator. If information on the possession of at least one of these items is available, the indicator 'expensive utensils' can be created by giving households owning one or more expensive utensils the value 'one' and other households the value 'zero'.
Households owning a TV, fridge, phone, bicycle, expensive utensil, or car, and households having a high quality floor or toilet facility, are assumed to have also a cheap utensil and obtain value 1 on this variable. Household owning a car are assumed to have also an expensive utensil and obtain value 1 on this variable.
Data files for adding IWI to DHS and MICS
We have developed data files to add IWI to a large number of DHS and MICS surveys. These files contain the value of IWI for all included households plus the DHS/MICS household IDs. With these files you can easily add IWI to the households in the original datasets. This has the advantage that you do not need to compute IWI yourself and that the IWI values in your data are exactly the same as those of other users of the same data.
Given that the files contain information at the household level, we can only make them available to users after having obtained permission from DHS/MICS for this. Please contact us if you are interested in this option.