COVID-19 IN LAC HIGH FREQUENCY PHONE SURVEYS TECHNICAL NOTE* April 2021 L atin American and the Caribbean is one of the regions in the world most affected by the COVID-19 pandem- ic, and the welfare impacts for households have been national Labor Organization, 2020); a trend likely to continue in 2021. severe. At the macroeconomic level, the World Bank es- (ii) Remittances and other non-labor income: Remit- timates a contraction of 6.9 percent of the region’s GDP tance inflows also plummeted in most Latin-Ameri- in 2020, due to pandemic-control measures and the can countries as soon as the pandemic hit. However, deceleration of the global economy (World Bank, 2021). they recovered in the second half of 2020 in several Regional export prices significantly dropped in the first countries, resulting in a regional decrease of 0.2 per- semester of 2020 (5.2%) (Inter-American Development cent compared to the previous year (López-Calva, Bank, 2020), and although they began to recover in the 2020; World Bank, 2020a). At the same time, gov- second half of the year, the volume of goods-exports ernments have tried to compensate income losses dropped by 8 points by the third quarter of 2020 (World mainly through social assistance programs, such Bank, 2021). This crisis hit households through at least as cash and in-kind (food) transfers (World Bank, three transmission channels: 2021). (i) Impact on labor income: the largest total work- (iii) Long-term patterns of human capital due to service ing-hour loss (16.2 percent), a reduction in employ- disruption: over 170 million children in the region are ment of 28 million people in 2020 (International out of school, and the region has experienced sig- Labor Organization, 2021) and a labor income drop nificant reductions in the use of health care services of 19.3 percent by the third quarter of 2020 (Inter- (Benveniste & Linnea, 2020). Figure 1. Selected LAC Countries for the 2020 HFPS Source: LAC HFPS Team * This note was prepared by Carolina Mejía-Mantilla, Sergio Olivieri, Ana Rivadeneira, Gabriel Lara-Ibarra and Javier Romero, under the guidance of Ximena del Carpio, with the financial support from the Latin American and Caribbean Regional Vice Presidency. 1 www.worldbank.org /BancoMundial @BancoMundialLAC COVID-19 IN LAC HIGH FREQUENCY PHONE SURVEYS TECHNICAL NOTE April 2021 In this context, the World Bank conducted a series of prevalence of food insecurity, access to health, educa- High-Frequency Phone Surveys (HFPS) to assess the tion and financing services, coping mechanisms result- impact of the coronavirus pandemic on the welfare ing from the lockdown, and safe-distance measures of Latin American and Caribbean households. Three stated by governments to mitigate the spread of the waves of surveys were collected between May and Au- disease. The surveys also inquired about households’ gust 2020 in the following thirteen countries: Argen- quarantine compliance and their knowledge of the tina, Bolivia, Chile, Colombia, Costa Rica, Dominican disease. Republic, Ecuador, El Salvador, Guatemala, Honduras, Mexico, Paraguay, and Peru (Figure 1). These countries The HFPS followed a panel format over three waves represent almost 60 percent of the total regional pop- of data collection for twelve countries and four waves ulation (World Bank, 2020b) and depict a comprehen- for Ecuador (Figure 2). The first wave was conducted sive outlook of the consequences of COVID-19 in terms between May 8 and June 14, 2020, the second wave of wellbeing. from June 5 until July 16, and the third one from July 5 until August 25, 2020. For Ecuador, the fourth wave The surveys gather information on multiple dimen- was collected between August 15 and 25, 2020.1 sions such as changes in employment and income, Figure 2. Data collection dates of the HFPS in 2020 March April May June July August Country 3/9 3/16 3/23 3/30 4/6 4/13 4/20 4/27 5/4 5/11 5/18 5/25 6/1 6/8 6/15 6/22 6/29 7/6 7/13 7/20 7/27 8/3 8/10 8/17 8/24 Argentina Bolivia Chile Colombia Costa Rica Dominican Republic Ecuador El Salvador Guatemala Honduras Mexico Paraguay Peru Quarantine start First wave Second wave Third wave Fourth wave Source: LAC HFPS Team 1 Surveys were implemented by Opinión Pública Ecuador in Ecuador and by Sistemas Integrales in Bolivia. For the eleven remaining countries, data collection was performed by IPSOS. 2 www.worldbank.org /BancoMundial @BancoMundialLAC COVID-19 IN LAC HIGH FREQUENCY PHONE SURVEYS TECHNICAL NOTE* April 2021 Eligible respondents for the HFPS were adults 18 years Survey estimates for each country are representative old and above. Only one respondent per household of households with a landline and households for was interviewed, and she answered both individual which at least one member has a cellphone. Similarly, and household-level questions. The same respon- the survey is representative of individuals of 18 years of dent was contacted and interviewed in all waves. An age or above who have an active cellphone number or average of 1,000 interviews was completed in the first a landline at home. When comparing the HPFS sam- wave. The following waves were affected by attrition, ple to nationally representative samples from official and average response rates declined to 73% in the sec- surveys (restricted to those with access to landline and ond wave and 70% in the third wave (see Table 1). In cellphone when possible2), we observe that the gender the fourth wave in Ecuador, 971 households completed composition is similar but individuals in the HFPS sam- the interview, translating into a 79.1% retention rate. ple are more educated (larger proportion of people Attrition was addressed in the estimation of the survey with tertiary education or higher), more likely to reside sample weights for the second and third waves (see in urban areas and slightly younger. These differences Sampling section below). are partially explained by the fact that phone surveys are not strictly comparable to face-to-face surveys, Table 1. Number of completed interviews per wave due to phone coverage and non-response biases. Country W1 W2 W3 Questionnaires Argentina 987 694 629 Bolivia 1,075 670 711 The questionnaire for the first wave followed closely Chile 1,000 622 684 the World Bank’s HFPS Global Core Questionnaire but Colombia 1,000 730 638 had some critical variations. There were also some Costa Rica 801 636 658 modifications in the subsequent waves, as shown in Dominican Republic 807 673 667 Table 2. All questionnaires are available in Spanish. Ecuador 1,227 1,027 853 El Salvador 804 625 604 Guatemala 806 625 636 Honduras 807 550 521 Mexico 2,109 1,245 1,152 Paraguay 715 486 457 Peru 1,000 841 821 Source: LAC HFPS Team 2 Only in 2 countries was this not possible. Note that household surveys used for this comparability exercise lag almost two years with the HFPS. However, individual characteristics compared are mostly structural and are not expected to be affected by this difference. 3 www.worldbank.org /BancoMundial @BancoMundialLAC COVID-19 IN LAC HIGH FREQUENCY PHONE SURVEYS TECHNICAL NOTE* April 2021 Table 2. Topics covered in the LAC COVID-19 HFPS No. Modules Description (based on Wave 1) WAVE 1 WAVE 2 WAVE 3 Questions about the propagation of COVID-19, preventive 1 Knowledge 7 questions Not included. Not included. measures, and satisfaction with government actions. Questions on quarantine compliance, reasons for non-com- 7 questions. 7 questions. 2 Behavior pliance and adoption of other recommendations to prevent 9 questions No relevant changes. No relevant changes. infection. Information on access to staple foods, medical assistance and 21 questions. 21 questions. 3 Access medicines, banks and ATMs. For households with school-age 21 questions No relevant changes. No relevant changes. children, questions about homeschooling activities. 30 questions. Questions about the respondent to assess current and pre Additional questions 30 questions. 4 Employment pandemic employment status, type, industry. Family business 21 questions to assess changes Questions on changes compared to wave 1. and agricultural activities of respondent. compared to previous A question on job search is also added. wave. Questions to assess changes in income of different sources: 3 questions with 4 questions. labor (wage, self-employed or agricultural work), non-labor 14 sub-questions 5 Income loss Not included Additional questions to ask for new sources (remittances, unemployment insurance, government trans- for each income of income since wave 1. fers, pensions, other private transfers). source. Food Questions based on Food and Agriculture Organization Food 4 questions. 4 questions. 6 4 questions insecurity Insecurity Experience Scale for the household. No relevant changes. No relevant changes. Questions on respondents’ perception about the impact 2 questions. 2 questions. 7 Concerns of COVID-19 on household finances and their own or their 2 questions No relevant changes. No relevant changes. family’s health. Information about shocks households might have experienced 2 questions with and their coping mechanisms. Shocks comprise job loss or 9 sub-questions 2 questions. 8 Coping Not included. business closures, theft, price increases, illness or death of for different No relevant changes. income recipients, excessive health expenses. types of shocks. 6 questions. Questions to assess coverage of both private and government Country-specific 10 questions. in-kind and cash transfers since the beginning of quarantine 9 Safety nets 7 questions emergency programs Additional questions ask about applying for measures. It includes a question on country-specific COVID-19 not included in this government programs. emergency and other important government programs. wave. 6 questions 8 questions. General household characteristics such as: size, location and Basic Update household Update household information if the respon- 10 area, number of rooms and access to internet. Respondent 12 questions information information if the dent moved. Questions on home ownership characteristics: sex, education level, age. respondent moved. and ability to pay rent are added Questions about trust on governments capacity to handle the 11 Governance COVID-19 crisis and about equality of access to public medical Not included Not included 3 questions services and economic assistance. Source: LAC HFPS Team 4 www.worldbank.org /BancoMundial @BancoMundialLAC COVID-19 IN LAC HIGH FREQUENCY PHONE SURVEYS TECHNICAL NOTE* April 2021 Sampling3 The sample is based on a dual frame of cellphone and A smaller second-phase sample was selected from landline numbers generated through a Random Digit the active non-business numbers from the first-phase Dialing (RDD) process. The RDD methodology produc- sample and was delivered to the country survey teams es all possible phone numbers in the country under the to be called by the enumerators. The reason for select- national phone numbering plan and draws a random ing a second-phase sample was that delivering a large sample of numbers. This method ensures coverage of first-phase sample of active numbers to the country all landline and cellphone numbers active at the time teams at once could facilitate the “misuse” of the sam- of the survey. ple, raise non-response rates, and increase potential non-response biases.4 In the first phase, a large sample was selected in both frames, with an allocation ranging from 0 percent The HFPS has two sampling units: households and landlines and 100 percent cellphones, to 20 percent individuals. Sampling weights were computed for landlines and 80 percent cellphones. The landline each unit and should be used accordingly to obtain frame was geographically stratified by first-level ad- estimates of interest for each unit. The estimation of ministrative unit (department, province or state), and weights involves four steps: the sample of landlines was selected with a propor- tionate allocation among strata. Geographic propor- 1. Calculation of the inclusion probabilities of landline tionate stratification was also done for cellphones in and cellphone numbers. Argentina, Bolivia and Mexico, where cellphone num- bers are linked to the area where they were issued. It 2. Computation of base weights for households and is important to underline that the HFPS sample design individuals. allows for obtaining precise estimates only at country level. 3. Adjustment for non-response The first-phase samples of landline and cellphone 4. Calibration of individual and household weights, us- numbers were then screened through an automated ing external data from official sources (adjusted by process to identify the active numbers. These were the national phone coverage). then cross-checked with business registries to identify Note that household and individual weights were ad- business numbers not eligible for the survey. justed for attrition in the second and third HFPS waves. 3 A detailed description of sampling and weighing is presented in Flores Cruz, Ramiro. 2020. COVID-19 High-Frequency Survey (HFS) in Latin American Countries Tech- nical Note: Sampling Design, Weighting and Estimation. Mimeo. Available upon request. 4 For Ecuador, the dual sampling frame was different: a part of the sample came from the active phone numbers collected in the 2019 Survey for Venezuelan Migrants and Host Communities in Ecuador (EPEC). The sample was completed with phone numbers using RDD and the methodology explained above. To know more about the sampling methodology for the 2019 EPEC Survey, refer to: Muñoz, Juan; Muñoz, Jose; Olivieri, Sergio. 2020. Big Data for Sampling Design: The Venezuelan Migration Crisis in Ecuador. Policy Research Working Paper No. 9329. World Bank, Washington, DC. Available at: https://openknowledge.worldbank.org/ handle/10986/34175 License: CC BY 3.0 IGO. 5 www.worldbank.org /BancoMundial @BancoMundialLAC COVID-19 IN LAC HIGH FREQUENCY PHONE SURVEYS TECHNICAL NOTE* April 2021 References Benveniste, L. & M. Linnea. 2020. Rethinking the future: Human capital in times of pandemic. 23 September 2020. Available at: https://blogs.worldbank.org/latinamerica/rethinking-future-human-capital-times-pandemic Inter-American Development Bank. 2020. Trade and Integration Monitor 2020: The COVID-19 Shock: Building Trade Resilience for After the Pandemic. Giordano, Paolo; Campos, Rosario; Michalczewsky, Kathia. http://dx.doi. org/10.18235/0002844 International Labour Organization. 2021. ILO Monitor: COVID-19 and the world of work. Seventh edition. 25 January 2021. Available at: https://www.ilo.org/global/topics/coronavirus/impacts-and-responses/WCMS_767028/lang-- en/index.htm International Labour Organization. 2020. ILO Monitor: COVID-19 and the world of work. Sixth edition. 23 September 2020. Available at: https://www.ilo.org/global/topics/coronavirus/impacts-and-responses/WCMS_755910/lang-- en/index.htm López-Calva, Luis Felipe. 2020. Stand by me: COVID-19 and the Resilience of Remittance Flows to LAC. UNDP DIREC- TOR’S BLOG: GRAPH FOR THOUGHT. December 2, 2020. Available at: https://www.latinamerica.undp.org/content/ rblac/en/home/presscenter/director-s-graph-for-thought.html World Bank. 2021. Global Economic Prospects. January 2021. Washington, DC: World Bank. doi: 10.1596/978-1-4648- 1612-3. License: Creative Commons Attribution CC BY 3.0 IGO. World Bank. 2020 (a). COVID-19: Remittance Flows to Shrink 14% by 2021. Press Release. October 29, 2020. Avail- able at: https://www.worldbank.org/en/news/press-release/2020/10/29/covid-19-remittance-flows-to-shrink-14- by-2021 World Bank. 2020 (b). World Development Indicators. Total population estimates 2019. Available at: https://data. worldbank.org/indicator/SP.POP.TOTL?name_desc=false 6 www.worldbank.org /BancoMundial @BancoMundialLAC