Description of node <>

Subject Predicate Object <p>There is no sampling plan tracing all household consumption by item, retail outlets, etc.</p> <p>In the absence of a sampling plan, consumption is first broken down by product type: a sample of just over 1,600 product families, called consumption segments, is defined to take into account the heterogeneity  of purchased goods and services  as defined from the COICOP  classification. The choice of these consumption segments is based on data from National accounts, household budget survey and other sources. The sample is updated annually to take into account changes in consumer behavior and, in particular, to introduce new products or, conversely, no longer follow the prices of products whose consumption tends to decline, or even disappears. The list of consumption segments remains confidential and the CPI is not disseminated at this level.</p> <p>The sampling plan  depends on the type of product and on the available information.</p> <p>• For the prices collected in stores by price collectors (about 160,000 prices each month), the sample comes from a two-stage survey: (i) in addition to each oversea territory, 99 agglomerations of more than 2,000 inhabitants are randomly selected from the metropolitan territory so as to be representative in terms of size and geographic location of household consumption habits; (ii) the products for which the price will be tracked and the surveyed outlets are chosen according to a quota method, by asking the collectors to carry out a given number of surveys by form of sales.</p> <p>In total, about 30,000 outlets are surveyed each month.</p> <p>    • For the collection on the Internet, ad hoc information can sometimes be gathered on the market shares of the various providers or on the most purchased products (statistics on journeys for example for transport services, statistics on consumer profiles for telecommunications, insurance, etc.). In the absence of any information, the choice of products is made by the collector and the prices are equally weighted.</p> <p>    • Finally, for the products tracked by transaction data or scanner data, we have a sampling frame since the precise quantities consumed are known by articles and outlets. Completeness of data is used.</p>