The 2018 Mozambique ISBS uses an innovative technique to survey these businesses. The survey follows an area-based sampling methodology with geographic area rather than an establishment or a business unit as a primary sampling unit. To account for potential clustering of informal business, the survey uses an area-based sampling called (stratified) Adaptive Cluster Sampling (ACS), whereby one selects a sample of starting squares and adaptively samples surrounding squares based on the number of informal firms discovered in the enumerated squares. All informal business in selected squares will be enumerated using a 2 to 3-minutes questionnaire, referred to in this document as the short-form questionnaire. The short form questionnaire is a listing questionnaire where basic information about the business is collected. A randomly selected subset of the enumerated businesses will be given a 20-minutes questionnaire, referred to in this document as the long-form questionnaire.
The survey is adaptive in the sense that if the number of informal units in a square exceeds a predefined threshold, all the squares surrounding the starting square are surveyed, following the same approach of enumeration and randomly conducting the main interview. If one of the surrounding squares exceed the threshold, then the squares surrounding that square in turn are also surveyed. This process continues until either the network is exhausted, or an arbitrary cut-off point is defined.
The first step in the sampling approach is the construction of a spatial grid as the Primary Sampling Units (PSU) frame. The grid covered the total of municipal areas and each cell had a size of 150 by 150 meters. This produced a total of about 24,000 squares between the three cities, excluding squares that are considered inaccessible. The second step was to stratify each grid, with in each city, based on likely concentration of informal business units. The grids were categorized into four strata: three strata of low, medium, and high concentration of informal sector activity, and a market centre. The stratification was based on local knowledge of the survey implementing contractor with approval from the WBG task team leader. The third step in the sampling process was to select a pre-defined number of starting squares from each stratum for enumeration and main data collection.
