Because a minimum level of consumption cannot be defined by commodity, there was the “bottom outliers” (for example, a household consuming so littleįood that survival could not be possible) are difficult to detect and fix. There are two types of such outliers: those that indicate “big spenders” (rich households), which are valid values and should not be removed or changed,Īnd those that are a result of errors in data collection (reporting), data coding, or data entry, which need to be fixed.Įxpenditure data can be unrealistically high or low. All datasets obtained from countries contain “outlying” expenditure values. The second step was to detect and fix outliers. Measurement Study Working Paper 135, World Bank, Washington, DC, 2002). In these cases an annual use value of the durable goods isĬalculated through the use of depreciation rates-a process that requires information, not always available, on the date and price of purchase and theĮstimated resale value (see Angus Deaton and Salam Zaidi, “Guidelines for Constructing Consumption Aggregates for Welfare Analysis,” Living Standards Many other surveys seek to measure consumption, not expenditure.

In these cases noĪnnualization is needed. Some surveys are intended to measure expenditure and collect data on purchases of durable goods with a recall period of one year. Very limited in developing countries, especially in rural areas, this process can be difficult and often produces unrealistic estimates. Two consumption items are typically problematic: imputed rents and use value of durable goods.Ī rental value may be imputed for owner-occupied dwellings, through a process using hedonic regression models. Products and regularly purchased nonfood products and services.Īnnualization becomes more complex for home-produced and received goods and services, for which consumption values have to be calculated on the basis ofĭata on quantities consumed and local (farm-gate or factory-gate) prices.Īnother challenge in annualizing data is that quantities are not always reported in metric units and not all countries provide conversion factors for (For example, food data collected “for the past 7ĭays” would be divided by 7, then multiplied by 365 monthly values would simply be multiplied by 12.) This is the method used for most purchased food (the period on which households are asked to report by recalling their expenditure during that period). In some cases annualization is straightforward and consists simply of applying a multiplying factor to the data, which is determined by the “recall period” No regional price deflators were applied). The first step consisted of annualizing each household’s consumption or expenditure data for each commodity (the data are nominal values in local currency Step 1: Annualizing consumption or expenditure data The data were standardized to the extent possible, through the six-step process The Global Consumption Database draws on a diverse set of surveys. Links are also provided to the microdata when available. Survey questionnaires are provided below as an important metadata component. Household survey datasets are complemented by data on population, purchasing power parity (PPP) conversion factors, and average exchange rates obtained from the World Bank’s World Development Indicators database.īecause of the diversity of methods and instruments used by the surveys, comparability across countries is limited.

When based on a survey conducted before 2010, the estimates were obtained by extrapolation,Īs described in the notes on the standardization of data (see Step 4). All data presented in the Global Consumption Database are as of 2010. The surveys used in the database were conducted between 20 (except the one for Djibouti, which was conducted in 1996) most were conducted during the period 2007–10. International organizations recommend conducting such surveys every three or four years.

Homeless and nomadic populations and visitors present in a country during a survey are also excluded from the sample.įew developing countries conduct household consumption or expenditure surveys on an annual basis. The universe of each survey is composed of ordinary households only “institutional households” (prisons, military barracks, hospitals, convents, and others) are not covered by household surveys. Their sample size ranges from less than 2,000 households to more than 100,000. Data presented in the tables and charts are World Bank estimates based on the national household consumption or expenditure survey datasets listed below.Īll surveys used have a nationwide coverage.

The consumption