In the first of a two-part blog unpacking the above question, Oxfam GB’s Impact Evaluation Advisers explain their approaches to measuring household consumption. In Part 2 next week: strategies for measuring household wealth.For this reason, in our impact evaluations of livelihoods projects (Effectiveness Reviews) we do not to attempt directly collecting data on total household income. Instead, in order to assess if our projects have had an impact on the welfare of the households we worked with, we rely on two different measures as proxies for income: household consumption and a household asset wealth index which is based on household assets, goods and home conditions.
While household consumption and household asset wealth indexare frequently used in conjunction (and sometimes interchangeably), they refer to two different concepts.
There is nothing particularly innovative about this approach – many of you will recognise that in a low-income context, there is a strong association between household income and consumption, and common practice in micro-level socio-economic analysis is to consider household consumption and expenditure as an indicator of income. But because we get the question so often, we thought it might be worth explaining the survey module and analysis strategy we use to measure consumption, for both food and non-food items.
We begin by asking respondents what types of food they consumed during the previous seven-day period, from a pre-populated list of food items. The list needs to be pre-populated, and include all the items that constitute a standard diet in the context under analysis. The questionnaire asks about the total consumption of each item, but also about the particular quantities consumed using three different sources; purchases, home production, and gifts or in-kind payments.
The quantities of each food item consumed are then converted into a monetary value by asking the respondent how much was paid for that item or (if it was from the household’s own production or from gifts) how much it would have been worth if it had been purchased from the local market.
In order to measure non-food expenses, respondents are asked how much they spent over the past four weeks on regular non-food items and services, from a comprehensive list.
Finally, respondents are asked to estimate the value of other occasional and seasonal types of expenditure that they incurred over the last 12 months. (This usually includes school fees, medical expenses, home repair, etc.)
Total household consumption is then calculated by converting each of the expenditure types into a per-day figure and adding them all together. Next, we need to adjust to reflect the existence of economies of scale within households, and the lower consumption needs of children.
Rather than just dividing by the number of people within the household and creating a per-capita measure, we divide by a factor representing household size, to generate an adult-equivalent expenditure per day. The formula used for calculating household size is (A + αK)θ where A is the number of adults in the household; K is the number of children; α is the consumption of a child relative to an adult; and θ captures the extent of economies of scale. For poor economies, common practice is to set α equal to 0.33 and θ equal to 0.9.
For more information, please refer to:
Designing Household Survey Questionnaires for Developing Countries: Lessons from 15 years of the Living Standards Measurement Study, Vols 1-3, in particular Vol 1 Ch 5: ‘Consumption; by Deaton and Grosh; and Guidelines for Constructing Consumption Aggregates for Welfare Analysis by Deaton and Zaidi.
Here you can find an example of Stata Do-file showing how to construct the household consumption indicator described above, and download examples of questionnaires by following the link to the UK Data Service web page from the following evaluations: Colombia, Ethiopia, Honduras, Philippines and Somalia.