Sampling procedure
The sample universe for the firm survey was firms that had 6-199 workers pre-COVID-19. Country-specific sample frames of firms were used. Stratified random samples were used (strata varied by country) to ensure adequate sample size in key strata. A target of 500 firms per country was set. The sampling strategy was incorporated into the weights.
Up to three attempts (five in Tunisia) were made to ensure response if a phone number was not picked up/answered, was disconnected or busy, or picked up but could not complete the interview at that time. After the third (or fifth) failed attempt, a firm was treated as a non-response and a random firm from the same stratum was used as an alternate.
Sampling frames
The sample frames varied by country as follows:
· Egypt: Yellow Pages
o Data on broad categories (e.g. gas stations)
o Coded into four strata: (1) services, (2) food & accommodation, (3) trade, manufacturing, and agriculture, (4) construction
o Restricted to firms with 6-199 workers in February 2020 based on an eligibility question during the phone interview
· Jordan: Kinz (a Jordanian corporate data mining website, which had a larger sample of firms than the Yellow Pages in Jordan).
o Data on broad categories (e.g. Industry, Marketing)
o Coded into five strata: (1) services, (2) food & accommodation, (3) trade and agriculture, (4) construction, (5) industry
o Initial frame restricted to firms with 5-250 workers. Further restricted to firms with 6-199 workers in February 2020 based on an eligibility question during the phone interview
· Morocco: Yellow Pages (no efficient digital copy available; a physical copy was used)
o Data organized geographically, not categorically
o Three geographic strata used: (1) Casa-Rabat, (2) North, (3) South
o The page ranges for the strata were provided. A random page within a stratum was selected, and then a random firm on that page (without replacement).
o The number of firms on the page was recorded and incorporated into the inverse probability weights.
o Restricted to firms with 6-199 workers in February 2020 based on an eligibility question during the phone interview
· Tunisia: National Institute of Statistics (INS) and Agency for the Promotion of Industry and Innovation (APII) databases
o Tunisia did not have a Yellow Pages or similar database, so administrative/statistics data sources had to be used
o The sample started with the INS frame with 1,238 firms with 6-200 wage employees
§ Firms were stratified into: (1) Agriculture (2) Industry (3) Construction (4) Trade (5) Accommodation (6) Service
§ Firms were also stratified by size in terms of 6-49 versus 50-200 employees
§ A random stratified sample (order) was selected
§ Further restricted to firms with 6-199 workers in February 2020 based on an eligibility question during the phone interview
§ This sample frame was eventually exhausted
o After the INS sample was exhausted, the APII sample was used
§ APII only covered firms with 10+ workers
§ APII only covered (1) services & transport, and (2) industry
o Weights are based on the underlying data on all firms from INS, specifically: Enterprises privées selon l'activité principale et la tranche de salariés (RNE 2019).
§ We ultimately stratify the Tunisia weights by industry and firms sized: 6-9 employees (since APII only covered 10+), 10-49, and 50-199 in wave one and combine 6-49 and in some cases 6-199 in subsequent waves.
Weighting
Inverse probability weighting was undertaken to account for the sampling strategy and non-response. In addition, We adjust the total number of enterprises in the sampling frame to account for the fact that not all enterprises were eligible.
Note: there are more details on the weights and sampling at the “COVID-19 MENA monitor enterprise weights” document in the documentation tab.