POVERTY & EQUITY POVERTY AND EQUITY EQUITABLE GROWTH, FINANCE & INSTITUTIONS INSIGHT The Impact of COVID-19 on the welfare of households with children An overview based on High Frequency Phone Surveys KEY HIGHLIGHTS This analysis focuses on the socioeconomic impacts of the Covid-19 in 35 developing countries on three types of households: Those with many (3 or more) children, those with few (1 or 2) children, and those with no children. INITIAL IMPACT At the onset of the pandemic, households with many children were more likely than households with no children to: • Suffer from income loss, with 76% of households with many children reported experiencing total income loss, versus 55% of households with no children. • Suffer from food insecurity. 24% for households with many children reported an adult member who went without eating for the whole day due to lack of money or other resources, versus 14% of households with no children. • Receive social assistance. 26% of households with many children reported receiving government assistance, versus 12% of households with no children. • Not use technology for education. Only 4% of households with many children reported accessing mobile learning applications, compared with 11% of households with few children (one or two). • Overall, participation in educational activities since school closure due to Covid-19 was low. Among households with children who attended school before school closure, less than 60% of households reported children participating in any educational activities after the school closure due to the Covid-19 outbreak EVOLUTION OF IMPACT Exploring the evolution of impact shows that households with many children are: • Experiencing higher rates of income loss than households with no children, although the differences between groups are not statistically significant. • Experiencing higher rates of food insecurity than households with no children, although both moderate and severe food security indicators show decreasing trends for all households. Cover photo: Yaw Niel / Shutterstock • Are more likely to receive government assistance, as the share of households receiving government assistance increased for all households 6 to 9 months after the peak stringency of the government response. Copyright © 2022 International Bank for Reconstruction and Development / The World Bank and UNICEF The findings, interpretations, and conclusions expressed in this work do not necessarily reflect the views of The World Bank, Contents its Board of Executive Directors, or the governments they represent, or those of UNICEF. The World Bank and UNICEF do not guarantee the accuracy, completeness, or currency of the data included in this work and Introduction 5 do not assume responsibility for any errors, omissions, or discrepancies in the information, or liability with respect to the use of or failure to use the information, methods, processes, or conclusions set forth. The boundaries, colors, denominations, and Data 7 other information shown on any map in this work do not imply any judgment on the part of The World Bank, UNDP, or UNICEF Initial Impact of COVID-19 on Children’s Welfare 9 concerning the legal status of any territory or the endorsement or acceptance of such boundaries. Initial Impact of COVID-19 on Food Insecurity 11 Nothing herein shall constitute or be considered to be a limitation upon or waiver of the privileges and immunities of The World Bank, all of which are specifically reserved. Social Protection at the Onset of the Pandemic 12 Rights and Permissions Education at the onset of the crisis 13 Evolution of the Impact of the Pandemic on Children’s Welfare 17 Evolution of Impact on Income Loss and Job Loss 18 This work is available under the Creative Commons Attribution 3.0 IGO license (CC BY 3.0 IGO), http://creativecommons.org/ licenses/by/3.0/igo. Under the Creative Commons Attribution license, you are free to copy, distribute, transmit, and adapt this Evolution of Food Insecurity 18 work, including for commercial purposes, under the following conditions: Evolution of Receipt of Social Protection 19 Attribution—Please cite the work as follows: World Bank, UNICEF 2022. The Impact of COVID-19 on the Welfare of Households with Children. Washington, DC: World Bank. License: Creative Commons Attribution CC BY 3.0 IGO. Conclusion 20 Translations—If you create a translation of this work, please add the following disclaimer along with the attribution: This Appendix 1. List of Available Indicators 21 translation was not created by The World Bank and should not be considered an official World Bank translation. The World Appendix 2. List of Indicator Topics Available in the COVID-19 Bank shall not be liable for any content or error in this translation. Monitoring Dashboard 22 Adaptations—If you create an adaptation of this work, please add the following disclaimer along with the attribution: This is an adaptation of an original work by The World Bank. Views and opinions expressed in the adaptation are the sole responsibility Appendix 3. List of Available Survey Included in the Analysis 23 of the author or authors of the adaptation and are not endorsed by The World Bank. Appendix 4. Summary Statistics for Key Indicators in Section 1 Third-party content—The World Bank does not necessarily own each component of the content contained within the work. The (unconditional results) 24 World Bank therefore does not warrant that the use of any third-party-owned individual component or part contained in the work will not infringe on the rights of those third parties. The risk of claims resulting from such infringement rests solely with Appendix 5. Predicted Margins for Section 1 Robustness Check you. If you wish to reuse a component of the work, it is your responsibility to determine whether permission is needed for that (controlling for education of the respondent, urban/rural, reuse and to obtain permission from the copyright owner. Examples of components can include, but are not limited to, tables, and state/province, standard errors in parentheses) 25 figures, or images. References 26 All queries on rights and licenses should be addressed to World Bank Publications, The World Bank Group, 1818 H Street NW, Washington, DC 20433, USA; e-mail: pubrights@worldbank.org. Acknowledgements—This report was produced by a joint team from the World Bank's Poverty and Equity and Social Sustainability and Inclusion Global Practices and UNICEF's Programme Group and Division of Data, Analytics, Planning and Monitoring. The core team members were Siwei Tian, Solrun Engilbertsdottir, Jose Cuesta, Enrique Delamónica, David Newhouse, and David Stewart. The authors thank Carolina Sanchez-Paramo, Benu Bidani, and Natalia Winder-Rossi for helpful comments and support, and the Poverty Global Practice and the Data for Goals team for providing access to the harmonized high frequency phone survey data. Graphic design by Hanna Chang with support from Jeeyeon Seo (World Bank). Introduction Children are disproportionately affected by poverty, whether measured in monetary or multidimensional terms. Prior to COVID-19, 1 in 6 children, or 356 million children in total, lived in extreme poverty, struggling to survive on less than PPP $1.90 per day. Meanwhile, nearly 1 billion children in multidimensional poverty in developing countries, suffering from at least one severe deprivation in education, health, education, housing, nutrition sanitation and water.1 2 When looking at a slightly higher poverty threshold of PPP USD 3.20 per day per person, a staggering 841 million or 41.5 percent of children worldwide live in households with income/ consumption levels equivalent to moderate poverty, compared with 23.5 percent of adults aged 18 and over. Furthermore, households with young children are the most affected by extreme poverty (PPP $1.90/day), 19.7 percent of children aged 0-4 years were in extreme poverty before the pandemic, compared to 12.9 percent of children aged 15 to 17 and 7.9 percent of adults ages 18 years and above. At the same time the vast majority of children have no effective social protection coverage, three out of four children globally are not covered by any type of social protection.3 The effects of the Covid-19 pandemic have been widespread and disproportionately affected vulnerable segments of the population, already in poverty or vulnerable to falling into poverty, including children and their families. The modest progress made in reducing child poverty has been reversed in all parts of the world by COVID-19.4 The pandemic is projected to have pushed an additional 150 million children into multidimensional poverty at the height of the pandemic by end of 2020.5 Various studies have highlighted this impact of the crisis on children and child poverty, both monetary and multidimensional. These studies typically rely on household surveys conducted prior to the crisis and use these surveys for various projections and simulations, based on multiple assumptions of the duration and severity of the crisis. 1 Silwal,Ani Rudra; Engilbertsdottir,Solrun; Cuesta Leiva,Jose Antonio; Newhouse,David Locke; Stewart,David. Global Esti- Photo: World Bank / Sambrian Mbaabu mate of Children in Monetary Poverty : An Update (English). Poverty and Equity discussion paper Washington, D.C.: World Bank Group. (2020, October). 2 Impact of COVID-19 on multidimensional child poverty. UNICEF. (2020, September). Retrieved January 23, 2022, from https://data.unicef.org/resources/impact-of-covid-19-on-multidimensional-child-poverty/ 3 International Labour Office. World Social Protection Report 2020–22: Social Protection at the Crossroads – in Pursuit of a Better Future. Geneva: ILO, 2021. 4 Richardson, Dominic, Carraro, Alessandro, Cebotari, Victor, Gromada, Anna, and Rees, Gwyther. Supporting Families and Children Beyond COVID-19: Social protection in high-income countries. Innocenti, Florence: UNICEF Office of Research, 2020. 5 Impact of COVID-19 on multidimensional child poverty. UNICEF. (2020, September). Retrieved January 23, 2022, from https://data.unicef.org/resources/impact-of-covid-19-on-multidimensional-child-poverty/ 4 THE IMPACT OF COVID-19 ON THE WELFARE OF HOUSEHOLDS WITH CHILDREN AN OVERVIEW BASED ON HIGH FREQUENCY PHONE SURVEYS 5 The High Frequency Phone Surveys (HFPS) coordinated by the World Bank, however, offer an opportunity to analyze the This paper contributes to this growing literature on the impacts of the COVID-19 shock by focusing solely on the Data actual impact of the crisis on the welfare of households with impact of the crisis on children, drawing on information from children, providing real-time information to inform and guide a set of High Frequency Phone Surveys from 35 countries9. To analyze the impact of the crisis on children’s welfare, in biased and differ from other nationally representative surveys.12 policies and programs to address the socio-economic impacts We analyze the initial impact of the crisis (with survey data particular children in poverty, we used data from the High Across the full set of phone surveys with information on the of the crisis. The HFPS database is being used to explore the collected during the period April to September 2020) as well Frequency Phone Survey supported by the World Bank since respondent’s relation to head, 69% of the respondents were impact of the pandemic across a range of issues, for example as the subsequent evolution of the impact of the crisis (with the beginning of the Covid-19 pandemic. The field work of the household head, 16% of the respondents were the spouse how the pandemic initially had major impacts on labor markets survey data collected during the period October 2020 to May these data typically took 1-2 months and occurred between of household head, 10% were children of the household head, (Khamis et al, 2021), has impacted different types of workers 2021. Based on data availability, we focus on the following April to September 2020. Since each country adopted country- and 5% were other relatives or non-relatives of the household in developing countries – showing that larger shares of female, harmonized key indicators of children’s welfare covering both specific questionnaires (adjusted from the core questionnaire), head. The average age of respondents in existing surveys is young, less educated and urban workers stopped working their individual conditions as well as those of the household the collected data are harmonized by the World Bank and approximately 45 years old. (Kugler et al. 2021)6; and these groups (women, youth and they live : (i) Income loss and job loss; (ii) Food insecurity included in the Covid-19 Household Monitoring Dashboard, lower-educated workers) which were already disadvantaged (households reporting an adult member didn’t eat for a whole which as of December 2021 included 143 harmonized Out of the 72 countries for which surveys were fielded, 35 in the labor market before the COVID-19 shock—were day or skipped a meal due to lack of money/resource)10; indicators on 16 topics for 72 countries. contain information on the number of children in a household significantly more likely to lose their jobs and experience (iii) Social protection programs11 (whether households have necessary for this analysis. Of the 35 countries, 20 are from decreased incomes (Bundervoet et al. 2021)7. At the same received any government assistance since the beginning The High Frequency Phone Surveys were implemented by Sub-Saharan Africa, 6 from Europe and Central Asia, 6 from time the gendered effects of the crisis were less pronounced of the pandemic); and (iv) Education (participation in making phone calls to respondents, who answer on behalf East Asia and Pacific, 2 countries from Middle East and in some countries, in particular those with extremely stringent educational activities following closures due to COVID-19). of the household for indicators measured at the household North Africa, and one country from Latin America. For this lockdown measures, for example in Colombia analysis shows These measures are compared between households level, and on behalf of themselves for indicators measured analysis the focus is on these 35 countries with information that women fared similarly to men in terms of the share of without children, households with one or two children, and at the individual level (adult level, not child level except for on the number of children in a household, which account for occupied workers affected (Cuesta et al. 2020).8 households with three or more children. Finally, we explore education related questions). Two types of sampling methods a combined population of approximately 1.21 billion people. whether there are differences in the pace of the recovery were adopted: Taking samples from previous nationally Among the 35 countries, 13 are low-income countries, 13 are between households with and without children. representative surveys and random digit dialing, with the latter lower middle-income countries, 5 are upper middle-income typically employed in the Latin America and the Caribbean countries, and 4 are high income countries. region. The weights in the High Frequency Phone Survey were adjusted so that the weighted results of household indicators To explore the impact of the crisis on children’s welfare, are nationally representative of households, but the sampling the analysis compares households according to how many method is still subject to two major limitations in terms of data children they have and various welfare proxies. Comparing representativeness. First, the survey by design excludes the households with no children versus households with many portion of the population that does not have access to phones children is also a proxy for poverty status, because households and a stable phone network. Secondly, for countries that used with many children tend to be poorer than households with an existing nationally representative surveys as a survey no or few children. For the analysis on the evolution of the frame, the respondents tend to be household heads and are impact, a total of 132 waves of surveys from 32 countries were therefore more likely to be male and older in general. As a used, and were organized as four quarters after peak month result, indicators measured at individual level are likely to be of Covid-19 based on the Oxford Covid-19 Stringency Index. 6 Kugler, Maurice; Viollaz, Mariana; Duque, Daniel; Gaddis, Isis; Newhouse, David; Palacios-Lopez, Amparo; Weber, Michael. 2021. How Did the COVID-19 Crisis Affect Different Types of Workers in the Developing World?. World Bank, Washington, DC. © World Bank. https://openknowledge.worldbank.org/handle/10986/35823 License: CC BY 3.0 IGO. 7 Bundervoet, Tom; Davalos, Maria E.; Garcia, Natalia. 2021. The Short-Term Impacts of COVID-19 on Households in Developing Countries: An Overview Based on a Harmonized Data Set of High-Frequency Surveys. Policy Research Working Paper; No. 9582. World Bank, Washington, DC. © World Bank. https://openknowledge. worldbank.org/handle/10986/35290 License: CC BY 3.0 IGO. 8 Cuesta, Jose, and Pico, Julieth. The Gendered Poverty Effects of the COVID-19 Pandemic in Colombia. Eur J Dev Res 32, 1558–1591 (2020). https://doi.org/10.1057/ s41287-020-00328-2 9 High frequency phone surveys are available for a total of 83 countries. However, we were limited to using surveys which included information on households with and without children, including the number of children, as well as surveys with harmonized indicators on the topics and time period explored in this analysis. The 35 countries included in this analysis met these requirements. 10 The HFPS data does not include information on food intake at level of child, nor allows disaggregation by gender 12 Bundervoet, Tom; Davalos, Maria E.; Garcia, Natalia. 2021. The Short-Term Impacts of COVID-19 on Households in Developing Countries: An Overview Based on a 11 This does not include comprehensive overview of social protection, but various questions related to social assistance, for example whether received any form of govern- Harmonized Data Set of High-Frequency Surveys. Policy Research Working Paper; No. 9582. World Bank, Washington, DC. © World Bank. https://openknowledge. ment assistance since start of the pandemic, and/or after losing a job and/or after reducing food consumption. worldbank.org/handle/10986/35290 License: CC BY 3.0 IGO. 6 THE IMPACT OF COVID-19 ON THE WELFARE OF HOUSEHOLDS WITH CHILDREN AN OVERVIEW BASED ON HIGH FREQUENCY PHONE SURVEYS 7 1. Initial Impact of COVID-19 on Children’s Welfare In this section, we analyze the initial impact of Covid-19 on children by conducting summary statistics on key indicators of interest using the first round of data collected in 35 countries with available data on number of children in the household. To check the robustness of the results, we also estimate a logit regression model to try to better distinguish the effect of households with children after controlling for predetermined proxies of welfare. These proxies include the level of education of the respondent, urban/rural location, and state/province. We use the regression results to examine the average predicted value of different outcomes according to the number of children in the household (see appendix 4). The main finding from this analysis on the initial impact of the crisis is that total income declines were more prevalent among households with many children during the early onset of the pandemic. INIT IAL IMPACT OF COVID-19 O N I NCO M E LO SS AND JOB LOSS The analysis on the initial impact on income and job loss shows that households with few children or households with many children were more likely than households with no children to suffer from total income loss since the onset of the pandemic. When asked how total household income changed since the start of the Covid-19 outbreak, the average share of households experiencing total income loss is 55% among households with no children, 68% among households with few children, and 76% among households with many children. The share of households reporting total income loss is significantly higher in households with children, compared to households with no children13. When controlling for welfare proxies including level of education of the respondent, urban/rural location and state/province, the predicted margins indicate that the share of households reporting total income loss is 5 to 7 percentage points higher among households with children compared to households with no children. However, the Photo: World Bank / Henitsoa Rafalia difference between households with few children and many children is only about 2 percentage points and is not statistically significant. 13 The standard error used to estimate the significance of the difference is grouped by country. This is the case for all the subsequent statements regarding the significance of the difference between groups in this note. 8 THE IMPACT OF COVID-19 ON THE WELFARE OF HOUSEHOLDS WITH CHILDREN AN OVERVIEW BASED ON HIGH FREQUENCY PHONE SURVEYS 9 Graph 1 Share of households reporting total Similarly, the average share of households experiencing wage the beginning of the pandemic, while the number is 24% for income loss Graph 2 Predicted margins for total income loss income loss is 56% for households with no children, 60% for households with many children. When controlling for welfare households with few children, and 63% for households with proxies, the difference is still not statistically significant. Total income decreased many children. The difference, however, is not statistically 80 .7 In sum, during the initial phase of the crisis, households with Predicted margins when controlling welfare proxies significant, both unconditionally and after controlling for predetermined welfare proxies. children were more likely to report total income declines .68 compared to households with no children. There is also no 60 Total income decreased Shifting to work stoppage (whether the respondent to the statistically significant difference in terms of wage income .66 phone survey stopped working after the pandemic), there loss and labor stoppage among the three groups. Finally, 40 is no significant difference between the three groups: 25% differences in the rate of work stoppage of the respondent are .64 of respondents from households with no children and similar for households with no and many children. households with few children reported stopped working since .62 20 .6 INITIAL IMPACT OF COVID-19 ON FOOD INSECURIT Y 0 No children Few children Many children No children Few children Many children To monitor food insecurity, the High Frequency Phone between households with many children and no children Survey questionnaire uses FAO’s Food Insecurity is approximately 4 percentage points and is statistically Graph 3 Share of households reporting wage Experience Scale survey module, which asks respondents significant. When asked whether an adult member in the income loss Graph 4 Predicted margins for wage income loss 8 questions regarding food security during the past 30 household skipped a meal due to lack of money or other days prior to the interview.14 The results indicate that resources, The percentage of households that responded Wage income decreased 80 households with children are more likely to suffer from food “yes” is 41% for households with no children, 47% for Predicted margins when controlling welfare proxies insecurity. On average, 14% of households with no children households with few children, and 50% for households with .62 reported an adult member who went without eating for the many children. The difference between groups, however, is not 60 whole day due to lack of money or other resources, but this statistically significant. When controlling for welfare proxies, Wage income decreased .6 rises to 18% for households with few children, and 24% for the results are similar and the differences are statistically households with many children, which is significantly higher significant, except for the difference between households with 40 .58 compared to households with no children. When controlling few and many children. This is consistent with the fact that for predetermined welfare proxies, the differences between households with more children are more likely to be poor and 20 the three groups are smaller. Nonetheless, the difference are more likely to report food insecurity. .56 Graph 8 Predicted margin for households with adult .54 0 No children Few children Many children No children Few children Many children Graph 7 Share of households with adult member member who did not eat for a whole day, controlling who did not eat for a whole day welfare proxies Adult member did not eat for a whole day 80 Graph 5 Share of respondents reporting labor .24 Predicted margins when controlling welfare proxies stoppage Graph 6 Predicted margins for labor stoppage Adult member did not eat for a whole day 60 Stopped working since the pandemic 80 .22 .27 Predicted margins when controlling welfare proxies Stopped working since the pandemic 40 .2 60 .26 .18 20 40 .25 .16 0 20 .24 No children Few children Many children No children Few children Many children .23 0 No children Few children Many children No children Few children Many children 14 These questions were only asked for adult members of households, not children 10 THE IMPACT OF COVID-19 ON THE WELFARE OF HOUSEHOLDS WITH CHILDREN AN OVERVIEW BASED ON HIGH FREQUENCY PHONE SURVEYS 11 Graph 9 Share of households with adult member Graph 10 Predicted margins for share of households Graph 13 Share of households received any Graph 14 Predicted margins for receiving any who skipped a meal with adult member who skipped a meal assistance assistance Adult member skipped a meal Received any kind of assistance 80 80 .55 .35 Predicted margins when controlling welfare proxies Predicted margins when controlling welfare proxies Received any kind of assistance Adult member skipped a meal 60 60 .5 .3 40 40 .45 .25 20 20 .4 .2 .35 .15 0 0 No children Few children Many children No children Few children Many children No children Few children Many children No children Few children Many children SO C IA L PROT EC T I O N AT T H E O NS ET O F THE PAND EM IC EDUCATION AT THE ONSET OF THE CRISIS The High Frequency Phone Surveys include multiple indicators assistance is 30% for households with many children, 27% for The phone survey includes a range of questions to analyze reported children listening to educational radio compared to related to social assistance, including whether households households with few children, and 19% for households with the impact of the pandemic on education. For the education- households with few children (8%). However, children from have received government assistance or any source of no children. In both cases, after controlling the predetermined related analysis, we cannot compare households without households with few children have more access to alternative assistance since the onset of the pandemic15. Households welfare proxies, households with children are about 7 to 9 children versus households with children, as education educational activities that require high technology, including with many children were more likely to receive government percentage points more likely to report having received any indicators only pertain to households with children. We educational TV program (20% for households with few assistance: The percentage of households reporting receiving social assistances, compared to households with no children therefore explore the difference between households with few children and 18% for households with many children) and government assistance since the beginning of the pandemic is (Graph 14). In addition, there is no discernable difference in children versus households with many children. mobile learning application (11% for households with few 26% for households with many children, 21% for households the share of households receiving government and social children and 4% for households with many children). Overall, with few children, and 12% for households with no children. assistance between households with few children and In general, participation in any educational activities since from the unconditional results, children in households with Similarly, the percentage of households receiving any kind of households with many children. school closure due to Covid-19 is low, for both households many children were more likely to listen to educational radio, with few and many children. Among households with children less likely to complete assignments, and much less likely to who attended school before school closure, around 53% of use a mobile app for learning. households reported children participating in any educational Graph 11 Share of households received government Graph 12 Predicted margins for government activities after the school closure due to the Covid-19 outbreak. Many of these differences disappear, however, after controlling assistance assistance The activities include completing school assignments, listening for predetermined welfare proxies including education level Received government assistance to educational radio, watching educational TV programs, of the respondent, urban/rural and state/province. There 80 using mobile application for learning, meeting with tutors or is less than half a percentage point difference in the share .3 Predicted margins when controlling welfare proxies teachers, and other educational activities. Because the data is of households with many and few children completing an Received government assistance 60 compiled at the household level, it overestimates the average assignment given by the teacher. In no case is the difference .25 participation rate across all children, especially for children greater than 2 percentage point, and only for listening to radio who come from households with many children. is the difference close to being statistically significant. Although 40 families with more children are less likely to be engaged in .2 Of the households with few children, 14% reported that educational activities than households with few children children completed a teacher-provided assignment, while for overall, these differences are mainly due to differences in 20 .15 households with many children, the number drops to 10%. A the location of residence and education of these two groups, higher percentage (11%) of households with many children rather than the number of children in the household per se. 0 .1 No children Few children Many children No children Few children Many children 15 The indicators on social assistances do not differentiate existing programs and new programs but only reflect whether households received assistances since the refer- ence period. The reference period in the first wave is “since the beginning of the pandemic”, and “since last round of interview” in subsequent waves. 12 THE IMPACT OF COVID-19 ON THE WELFARE OF HOUSEHOLDS WITH CHILDREN AN OVERVIEW BASED ON HIGH FREQUENCY PHONE SURVEYS 13 Graph 16 Predicted margins for share of households Graph 20 Predicted margins for share of households Graph 15 Share of households with children who with children who completed assignments after Graph 19 Share of households with children who with children who watched educational TV after completed assignment after school closure school closure watched educational TV after school closure school closure Completed assignment Watched educational tv 30 .135 .23 Predicted margins when controlling welfare proxies Predicted margins when controlling welfare proxies 30 .13 Watched educational tv Completed assignment 20 .22 20 .125 10 .21 10 .115 .12 .2 0 0 Few children Many children Few children Many children Few children Many children Few children Many children Graph 18 Predicted margins for share of households Graph 22 Predicted margins for share of households Graph 17 Share of households with children who with children who listened to educational radio after Graph 21 Share of households with children who with children who used mobile app for learning after listened to educational radio after school closure school closure used mobile app for learning after school closure school closure Listened to educational radio Used mobile app for learning 30 30 .105 .12 Predicted margins when controlling welfare proxies Predicted margins when controlling welfare proxies .1 Used mobile app for learning Listened to educational radio 20 20 .11 .09 .095 10 10 .1 .08 .085 .09 0 0 Few children Many children Few children Many children Few children Many children Few children Many children 14 THE IMPACT OF COVID-19 ON THE WELFARE OF HOUSEHOLDS WITH CHILDREN AN OVERVIEW BASED ON HIGH FREQUENCY PHONE SURVEYS 15 2. Evolution of the Impact of the Pandemic on Children’s Welfare The above analysis has focused on the cross-sectional analysis of the initial impact of the pandemic using the first wave of the survey for each country. To analyze the evolution of the impact of Covid-19, we used 132 surveys from the 3216 countries with information on the number of children in the households. The goal of the analysis is to investigate trends in income loss, employment, food insecurity, and social protection as the pandemic evolved. To examine the trends, we estimate the following model: y hct = αf(t) + βf(t) * Children h + θ c + ε ct Yh(c)t is the outcome of interest observed in household h, country c, and period t, and in this case is one of six indicators: “share of households receiving decreased total income”, “share of respondents currently employed”, “share of households with adult member skipped a meal due to lack of money or other resources”, “share of households with adult member who did not eat for a whole day”, “share of households receiving government assistance” and “share of households receiving any source of assistance”. On the right-hand side of the equation, αf(t) captures the time after the peak month of the pandemic, ranging from quarter 1 to 4. To account for the different time frame in terms of data collection and the evolution of pandemic in each country, we used the Oxford COVID-19 Government Response Stringency Index to determine the peak month of Covid-19 pandemic in 2020 and organize waves of available surveys according to quarters after the peak month17. The Stringency Index shows the aggregated scores of policy stringency based on the Oxford COVID-19 Government Response Tracker, which collected information on policy measure taken by governments in response to the COVID-19 pandemic.18 For example, if a country collected two surveys in June and October 2020, and the Oxford Stringency Index of that country peaked in May 2020, the survey collected in June would be assigned as T=1 because it was collected within the first three months after the peak month. Similarly, the survey collected in October would be assigned T=2 because it was collected in the second quarter after the peak month. The second term f( . ) Photo: World Bank / Henitsoa Rafalia 16 Three countries (Saint Lucia, Gabon, Lebanon) were included in section 1, but were excluded in section 2. For Saint Lucia, the Oxford Stringency Index used to benchmark quarters after the peak month was not available. For Gabon and Lebanon, the peak months of COVID-19 came after the data collection date. 17 Since data collection started in 2020 for most countries, peak months in this analysis were determined based on the Oxford Stringency Index in 2020. 18 Thomas Hale, Noam Angrist, Rafael Goldszmidt, Beatriz Kira, Anna Petherick, Toby Phillips, Samuel Webster, Emily Cam- eron-Blake, Laura Hallas, Saptarshi Majumdar, and Helen Tatlow. (2021). “A global panel database of pandemic policies (Oxford COVID-19 Government Response Tracker).” Nature Human Behaviour. https://doi.org/10.1038/s41562-021-01079-8 16 THE IMPACT OF COVID-19 ON THE WELFARE OF HOUSEHOLDS WITH CHILDREN AN OVERVIEW BASED ON HIGH FREQUENCY PHONE SURVEYS 17 is the cubic polynomial of time (in quarter) interacted with the for households with no children relative to the baseline period. Graph 25 Evolution of estimated margin for adult Graph 26 Evolution of estimated margin for adult number of children in the household. The term for number of The αf(t)+βf(t) * Children hr(c) term captures the same member did not eat for a whole day member skipped a meal children is organized as three categories: households with no information for households with few and many children. Finally, children, households with few children, and households with ε ct is the error term, clustered at the country level. Using this .45 .2 Adult member did not eat for a whole day many children. The term θ c captures the country fixed effect, model, we can examine the average predicted margins of the Adult member skipped a meal with each country defined as a binary variable. Thus, the coefficients for households with different number of children, .4 αf(t) term captures the evolution of the outcome of interest .15 across quarters after the peak month. .35 EVO LUT ION O F I MPAC T O N I NCO M E LO S S AND J O B LO S S .1 .3 Over time, the percentage of households reporting total children, and the lowest among households with no children income loss shows a decreasing trend, which signals a partial in the first three quarters. Although the differences between .05 .25 1 2 3 4 recovery from the crisis. However, there is no statistically households with few children and households with many Number of qurters after peak month 1 2 3 4 Number of qurters after peak month discernable difference in terms of recovery speed or pattern children are not statistically significant, the differences between No children Few children No children Few children between groups. households with children and households with no children Many children Many children are statistically significant in all four quarters. In general, the In terms of employment, the estimated share of employed employment rate increased from quarter 1 to quarter 2, but respondents is the highest among households with many then started to show a decreasing trend afterwards. EVOLUTION OF RECEIPT OF SOCIAL PROT ECTION Graph 23 Evolution of estimated margin for total Graph 24 Evolution of estimated margin for current As mentioned in the previous section, households with many children are more likely to report receiving government assistance. income loss employment of respondent In general, there was a slight increase in the percentage of households receiving government assistance from quarter 2 to quarter .75 3, a reflection of the delay in expanding social protection programs following the pandemic, with no significant differences between households with many, few, or no children. When looking at any type of assistance, the graph shows similar trend as government .6 assistance from quarter 2 to quarter 4 but there is a decreasing trend from quarter 1 to quarter 2. Respondent currently employed Total income decreased .7 .5 Graph 27 Evolution of estimated margin for Graph 28 Evolution of estimated margin for .65 receiving government assistance receiving any kind of assistance .4 .6 .2 .3 Received government assistance Received any kind of assistance .55 .3 .25 1 2 3 4 1 2 3 4 Number of qurters after peak month Number of qurters after peak month .15 .2 No children Few children No children Few children Many children Many children .15 .1 .1 EVO LUT ION O F FO O D I N S EC URITY .05 .05 1 2 3 4 1 2 3 4 Number of qurters after peak month Number of qurters after peak month Households with many children were hit the hardest in terms trend until quarter 3. However, there is no discernable No children Few children No children Few children of food insecurity at the onset of the pandemic. Overall, both difference between groups. Many children Many children severe food insecurity indicator like “adult member did not eat for a whole day” and the more moderate food insecurity indicator “adult member skipped a meal” show a decreasing To test the robustness of the results, we estimated the following model with four additional control variables, including strictness of lockdown measured by the Oxford Stringency Index, regional fixed effect measured by a series of binary variables indicating state/ province the household resides in, urban-rural location, and education level of the respondent. y hr(c)t = αf(t) + βf(t) * Children hr(c) + Stringency ct +θ r(c) + Urban + Education + ε r(c)t The trends and key conclusions do not change after controlling for the additional variables. 18 THE IMPACT OF COVID-19 ON THE WELFARE OF HOUSEHOLDS WITH CHILDREN AN OVERVIEW BASED ON HIGH FREQUENCY PHONE SURVEYS 19 Conclusion Appendix 1. List of Available Indicators The primary objective of this analysis was to identify the impact of the crisis on households NUMBER OF without and with (few or many) children, focusing both on the initial impact in 2020 as well as INDICATOR COUNTRIES LIST OF COUNTRIES the subsequent evolution of this impact. The analysis is based on only 35 countries, with the Bulgaria, Ethiopia, Gabon, Ghana, Guinea, Gambia, Cambodia, Lao PDR, St. Lucia, majority of these countries in the Sub-Saharan Africa region. Therefore, the conclusions need Total income 18 Mozambique, Mauritius, Malawi, Nigeria, Poland, Romania, Senegal, Tajikistan, decreased to be interpreted with caution. Nonetheless, the results provide a clear indication that across Uzbekistan these indicators, households with many children fared worse during the initial onslaught of the pandemic (April to September 2020). At the initial onset of the crisis, households with many Ethiopia, Gabon, Ghana, Guinea, Gambia, Cambodia, Lao PDR, St. Lucia, Wage income 21 Madagascar, Mali, Mongolia, Mozambique, Mauritius, Malawi, Nigeria, Rwanda, children were substantially more likely than households with no children to report a decline in decreased Senegal, Sierra Leone, Uganda, Uzbekistan, Zimbabwe total income (76% versus 55%) and more likely to report an adult who did not eat for a full day (24% versus 14%). Both of these results are robust to controlling for residence and respondent Burkina Faso, Bulgaria, Congo, Dem. Rep., Congo, Rep., Djibouti, Ghana, Guinea, education. After the initial impact, trends appear to be broadly similar for households with many Stopped working Gambia, Croatia, Indonesia, Kenya, Cambodia, Lao PDR, Lebanon, St. Lucia, since the beginning 32 Madagascar, Mali, Myanmar, Mongolia, Mozambique, Mauritius, Malawi, Nigeria, and few children, after controlling for region of residence and education. There is insufficient of the pandemic Poland, Romania, Rwanda, Senegal, Solomon Islands, Sierra Leone, Uganda, data, however, to estimate these trends precisely. Uzbekistan, Zimbabwe On the positive side, the analysis also highlights that households with many children were more Congo, Dem. Rep., Ethiopia, Ghana, Kenya, St. Lucia, Madagascar, Mali, Myanmar, likely to receive some type of social assistance. The percentage of households reporting receiving Did not eat for a day 16 Mongolia, Mozambique, Malawi, Nigeria, Solomon Islands, Sierra Leone, Uganda, government assistance since the beginning of the pandemic was 26% of households with many Zimbabwe children, 21% of households with few children, and 12% for households with no children. These Adult member Congo, Dem. Rep., Gabon, Ghana, Lebanon, Madagascar, Mali, Mongolia, patterns also held when controlling for residence and education. These government assistance 13 skipped a meal Mozambique, Malawi, Nigeria, Sierra Leone, Uganda, Zimbabwe programs undoubtedly mitigated the adverse impact of the crisis on households with many children. This further reinforces UNICEF and World Bank's emphasis on sustaining these Bulgaria, Congo, Dem. Rep., Congo, Rep., Djibouti, Ethiopia, Gabon, Ghana, Guinea, government programs for an equitable and sustainable recovery. Received government Gambia, Croatia, Indonesia, Kenya, Cambodia, Lebanon, St. Lucia, Madagascar, 25 assistance Mongolia, Malawi, Nigeria, Poland, Romania, Rwanda, Sierra Leone, Uganda, Zimbabwe The analysis further reinforces UNICEF and other partners’ call to action to ensure that schools are opened.19 Both households with few and many children had low participation in any education Burkina Faso, Bulgaria, Congo, Dem. Rep., Congo, Rep., Djibouti, Ethiopia, Gabon, activities since school closure due to COVID-19, with only 11% of households with few children Received any source Ghana, Guinea, Gambia, Croatia, Indonesia, Kenya, Cambodia, Lao PDR, Lebanon, St. 29 and 4% of households with many children had access to mobile learning applications. of assistance Lucia, Madagascar, Mali, Myanmar, Mongolia, Mauritius, Malawi, Nigeria, Rwanda, Solomon Islands, Sierra Leone, Uganda, Zimbabwe The data are insufficiently comprehensive to detect major differences in trends for households with many children and those with fewer and no children in the quarters following the initial crisis. Burkina Faso, Congo, Dem. Rep., Congo, Rep., Djibouti, Ethiopia, Ghana, Gambia, Children completed 19 Indonesia, Kenya, Cambodia, Lao PDR, Madagascar, Mali, Mongolia, Malawi, Nigeria, Respondents in households with many children were more likely to report being employed, but assignments Senegal, Uganda, Zimbabwe also more likely to report skipping a meal. Positively, there was an increase in the percentage of all households, with and without children, receiving government assistance from quarter 2 to Burkina Faso, Congo, Dem. Rep., Congo, Rep., Djibouti, Ethiopia, Ghana, Gambia, Children listened to quarter 3. Future analysis utilizing the HFPS data can shed light on the continued evolution of 17 Indonesia, Kenya, Cambodia, Madagascar, Mongolia, Malawi, Nigeria, Sierra Leone, educational radio Uganda, Zimbabwe these indicators (income/job loss, food security, education and social protection), and utilized to help ensure that children and their families are prioritized in the recovery, including in the scaling Children watched Burkina Faso, Congo, Dem. Rep., Congo, Rep., Djibouti, Ethiopia, Ghana, Gambia, up of social protection programs. educational TV 17 Indonesia, Kenya, Cambodia, Madagascar, Mali, Mongolia, Malawi, Nigeria, Senegal, program Uganda, Zimbabwe Burkina Faso, Bulgaria, Congo, Dem. Rep., Congo, Rep., Djibouti, Ethiopia, Ghana, Children used mobile 20 Gambia, Croatia, Indonesia, Kenya, Cambodia, Lao PDR, Madagascar, Mali, app for learning Mongolia, Malawi, Nigeria, Senegal, Uganda, Zimbabwe 19 Retrieved January 23, 2022, from https://www.unicef.org/coronavirus/reopen-schools 20 THE IMPACT OF COVID-19 ON THE WELFARE OF HOUSEHOLDS WITH CHILDREN AN OVERVIEW BASED ON HIGH FREQUENCY PHONE SURVEYS 21 Appendix 2. List of Indicator Topics Available Appendix 3. List of Available Survey Included in the COVID-19 Monitoring Dashboard in the Analysis INDICATOR TOPICS NUMBER OF INDICATORS COUNTRY WAVE COUNTRY WAVE Assets & Services 4 Burkina Faso WAVE1-8 Mali WAVE1-5 Coping 3 Bulgaria WAVE1-3 Myanmar WAVE1-4 Demographic 8 Congo, Dem. Rep. WAVE1,3,5 Mongolia WAVE3-4 Education 8 Congo, Rep. WAVE1,2,4,5,6 Mozambique WAVE2-5 Financial 4 Djibouti WAVE1-2 Mauritius WAVE1-3 Food Security 14 Ethiopia WAVE4-11 Malawi WAVE3-9,11 Health 11 Gabon WAVE1 (section 1 only) Nigeria WAVE2-11 Housing 4 Ghana WAVE1 Poland WAVE1 Income 24 Guinea WAVE1-3 Romania WAVE1 Safety Nets 14 Gambia, The WAVE1-4 Rwanda WAVE1-2 Knowledge 10 Croatia WAVE1-3 Senegal WAVE1 Labor 14 Indonesia WAVE1-5 Solomon Islands WAVE1 Preventive behaviors 8 Kenya WAVE1-4 Sierra Leone WAVE1-2 Subjective Wellbeing 4 Cambodia WAVE1-5 Tajikistan WAVE3-14 Vaccination 11 Lao PDR WAVE1 Uganda WAVE1-6 Vaccination (social media) 1 Lebanon WAVE1 (section 1 only) Uzbekistan WAVE1-12 St. Lucia WAVE1 (section 1 only) Zimbabwe WAVE1-2 Madagascar WAVE1 22 THE IMPACT OF COVID-19 ON THE WELFARE OF HOUSEHOLDS WITH CHILDREN AN OVERVIEW BASED ON HIGH FREQUENCY PHONE SURVEYS 23 Appendix 4. Summary Statistics for Key Appendix 5. Predicted Margins for Section 1 Indicators in Section 1 (unconditional results) Robustness Check (controlling for education of INDICATOR NO CHILDREN FEW CHILDREN MANY CHILDREN the respondent, urban/rural, and state/province, Total income decreased 0.546 (0.064) 0.683 (0.040) 0.756 (0.046) standard errors in parentheses) 0.557 0.597 0.626 Wage income decreased INDICATOR NO CHILDREN FEW CHILDREN MANY CHILDREN (0.051) (0.054) (0.072) 0.606 0.679 0.661 0.253 0.249 0.245 Total income decreased Stopped working since the beginning of the pandemic (0.008) (0.007) (0.009) (0.026) (0.026) (0.027) 0.572 0.607 0.599 0.136 0.180 0.238 Wage income decreased Did not eat for a day (0.015) (0.009) (0.008) (0.030) (0.036) (0.050) 0.245 0.253 0.252 0.410 0.472 0.505 Stopped working since the beginning of the pandemic Adult member skipped a meal (0.006) (0.005) (0.006) (0.078) (0.057) (0.081) 0.166 0.186 0.209 0.118 0.209 0.256 Did not eat for a day Received government assistance (0.006) (0.009) (0.009) (0.029) (0.064) (0.082) 0.397 0.483 0.507 0.194 0.267 0.295 Adult member skipped a meal Received any source of assistance (0.027) (0.012) (0.014) (0.035) (0.052) (0.056) 0.144 0.226 0.240 0.137 0.098 Received government assistance Children completed assignments (0.019) (0.014) (0.015) (0.023) (0.015) 0.206 0.279 0.292 0.075 0.112 Received any source of assistance Children listened to educational radio (0.021) (0.011) (0.012) (0.016) (0.025) 0.128 0.125 0.202 0.177 Children completed assignments Children watched educational TV program (0.003) (0.004) (0.063) (0.059) 0.098 0.113 0.107 0.041 Children listened to educational radio Children used mobile app for learning (0.004) (0.003) (0.041) (0.017) 0.221 0.208 Children watched educational TV program (0.004) (0.003) 0.090 0.091 Children used mobile app for learning (0.004) (0.006) 24 THE IMPACT OF COVID-19 ON THE WELFARE OF HOUSEHOLDS WITH CHILDREN AN OVERVIEW BASED ON HIGH FREQUENCY PHONE SURVEYS 25 References Bundervoet, T., Dávalos, M. E., & Garcia, N. (2021). The Short-Term Impacts of COVID-19 on Households in Developing Countries. Cuesta, J., & Pico, J. (2020). The gendered poverty effects of the COVID-19 pandemic in Colombia. The European journal of development research, 32(5), 1558-1591. Khamis, M., Prinz, D., Newhouse, D., Palacios-Lopez, A., Pape, U., & Weber, M. (2021). The Early Labor Market Impacts of COVID-19 in Developing Countries. Kugler, M. D., Viollaz, M., Vasconcellos Archer Duque, D., Gaddis, I., Newhouse, D. L., Palacios-Lopez, A., & Weber, M. (2021). How Did the COVID-19 Crisis Affect Different Types of Workers in the Developing World? (No. 9703). The World Bank. 26 THE IMPACT OF COVID-19 ON THE WELFARE OF HOUSEHOLDS WITH CHILDREN