Policy Research Working Paper 9959 Gender Differences in Household Coping Strategies for COVID-19 in Kenya Yuanwei Xu Antonia Delius Utz Pape Poverty and Equity Global Practice March 2022 Policy Research Working Paper 9959 Abstract Understanding how different households cope with networks to cope. No difference in coping by reducing COVID-19 among a vulnerable population is important meals is observed across these two types of households. This for the policy design aiming at relieving hunger and poverty paper documents that the reasons behind the gender dif- in a low income setting. This paper uses original household ference include that female-headed households are poorer, data from five waves of a phone survey conducted between and they are more likely to rely on friends and family to May 2020 and June 2021 in Kenya (sample size 31,715) cope with shocks even prior to the COVID-19 shock. The and investigates the gender differences in household coping findings suggest that widowed and divorced women are strategies during the COVID-19 shock. It finds that female- in high need of relief programs, and governments should headed households are less likely to cope by selling assets provide easily accessible loans to avoid negative impacts in or taking loans, compared with male-headed households. the long term from households selling assets. Instead, femaleheaded households rely more on social This paper is a product of the Poverty and Equity Global Practice. It is part of a larger effort by the World Bank to provide open access to its research and make a contribution to development policy discussions around the world. Policy Research Working Papers are also posted on the Web at http://www.worldbank.org/prwp. The authors may be contacted at yuanweii.xu@gmail.com, adelius@worldbank.org, and upape@worldbank.org. The Policy Research Working Paper Series disseminates the findings of work in progress to encourage the exchange of ideas about development issues. An objective of the series is to get the findings out quickly, even if the presentations are less than fully polished. The papers carry the names of the authors and should be cited accordingly. The findings, interpretations, and conclusions expressed in this paper are entirely those of the authors. They do not necessarily represent the views of the International Bank for Reconstruction and Development/World Bank and its affiliated organizations, or those of the Executive Directors of the World Bank or the governments they represent. Produced by the Research Support Team Gender Differences in Household Coping Strategies for COVID-19 in Kenya Yuanwei Xu *, Antonia Delius †, Utz Pape ‡ JEL-Codes: I31 (General Welfare, Well-Being), J16 (Economics of Gender), Z13 (Eco- nomic Sociology, Economic Anthropology, Social and Economic Stratification ) Keywords: Coping Strategy, Poverty, Social Network, COVID-19 * Xu: Leibniz University Hannover, Ruhr University Bochum, and the World Bank. Email: yuanweii.xu@gmail.com. † The World Bank. Email: adelius@worldbank.org. ‡ The World Bank and Georg-August-Universit¨ ¨ at Gottingen. Email: upape@worldbank.org. 1 Introduction The spread of COVID-19 has caused huge negative shocks on the economy (Altig et al., 2020), households (Hanspal, Weber and Wohlfart, 2020; Kansiime et al., 2021), as well as businesses (Bartik et al., 2020; Meyer, Prescott and Sheng, 2021) across the globe. Apart from the sharp increase in poverty and inequality (Han, Meyer and Sullivan, 2020; Deaton, 2021), other negative impacts have been documented, including rising armed conflict (Bloem and Salemi, 2021; Ide, 2021), inter-personal violence (Evans, Lindauer ¨ and Farrell, 2020; Aguero, 2021), and elevated mental stress (Pfefferbaum and North, 2020). Several studies have pointed out that the impact of COVID-19 varies across different socio-demographic groups. For example, poor households and countries are more vulnerable than richer ones (Schellekens and Sourrouille, 2020; Mena et al., 2021), as these groups do not have as much resources to cope with such a profound negative shock. Moreover, the current COVID-19 crisis disproportionately affected women compared to men (Alon et al., 2020). Women are more likely to lose employment or income (Etheridge and Spantig, 2020; Kugler et al., 2021; Dang and Nguyen, 2021; Xue and McMunn, 2021), which is due to the fact that they are more likely to work in the service sector, and they have a higher burden of childcare compared to men. Gender inequality is a challenge across the globe, especially in low-income economies (Sachs, 2012). However, the gender difference is likely to be elevated after the COVID- 19 pandemic (Alon et al., 2020). A large strand of literature shows that men and women have different preferences (Andersen et al., 2013), also during the COVID-19 time (Fan, Orhun and Turjeman, 2020). For example, women have been found to be more risk averse (Dohmen et al., 2011; Eckel and Grossman, 2008), more pro-social (Eckel and Grossman, 1998), more patient (Dittrich and Leipold, 2014), and more conducive to family welfare (Ashraf, 2009; Rubalcava, Teruel and Thomas, 2009). During the COVID- 19 crisis, women are more risk-averse, worry more about their and others’ health, and take more preventive actions (Fan, Orhun and Turjeman, 2020). In Kenya, women predominate in agriculture but their contributions are mostly unpaid. Moreover, they are less predominant in formal employment and women’s property rights are often through men (Ellis et al., 2007). It is important to understand the gender differences in the COVID-19 impact for the future policy design attempting at alleviating gender inequality as well as for post-COVID relief interventions. This study pools the first five waves from a nationally representative survey conducted 1 with Kenyan households via phone interviews between May 2020 and June 2021. By controlling for a rich set of household characteristics, which may affect the household head’s gender, we plausibly analyze the differences in household coping strategies by household head gender. We show that female-headed households in Kenya are more likely to rely on transfers or assistance from family members and friends, while male-headed ones have a higher probability of selling assets or taking loans to cope. However, no gender difference in reducing meals is observed. The gender difference in coping could be explained by differences in social preference, as well as poverty levels. Female-headed households are more likely to receive transfers from family members prior to the COVID-19 pandemic, and this habit may shape their coping strategies in response to COVID-19. Male-headed households are wealthier and thus they might have more assets to sell and be qualified to apply for loans to cope. This paper is related to two strands of literature. Firstly, it adds directly to the literature on the gender differences in the COVID-19 impacts. It has been found that women would experience higher mental stress (Olaseni et al., 2020; Garc´ ıa-Fern´andez et al., 2021), and are more likely to comply to the COVID-19 regulation (Galasso et al., 2020). In addition, during school closures, women undertake more childcare burden than men, and they were more likely to lose jobs and income (Etheridge and Spantig, 2020; Dang and Nguyen, 2021; Xue and McMunn, 2021). This paper complements this strand of literature by showing that female-headed households are relatively poorer, more likely to suffer from hunger, and thus have smaller assets to cope with the negative shock compared to the male-headed households. Moreover, this paper adds directly to the literature studying female headship and empowerment, especially in low income economies. For example, it has been found that female-headed households would have lower income but receive larger family transfers (Quisumbing, Haddad and Pena, ˜ 2001; Horrell and Krishnan, 2007). When women are empowered, they would spend more on children’s nutrition, food consumption, and education (Ashraf, 2009; Rubalcava, Teruel and Thomas, 2009; De Brauw et al., 2014; Armand et al., 2020). In this paper, we find evidence that female-headed households are more vulnerable during a profound negative economic shock, and further suggest that specific policies need to target female-headed households to further enhance female empowerment, reduce gender inequality, as well as achieve higher effectiveness in alleviating hunger and COVID-19 impacts. 2 The remainder of this paper is organized as follows. Section 2 provides an overview of the spread of COVID-19 and the government measures in Kenya. Next Section 3 describes the dataset used in this paper and discusses the empirical approach. Section 4 presents empirical results on the gender differences in household coping strategies, including the main results, potential reasons for the difference, as well as sensitivity analyses. Finally, Section 5 concludes. 2 Background In Kenya, it is estimated that about one-third of the population (20 million people) live under the poverty line (The World Bank, 2021b). The COVID-19 pandemic reached Kenya in early 2020, with the first case reported on March 13, 2020. Since then, there have been more than 210,000 confirmed cases, with 3,595 deaths as of June 29, 2021. The Kenyan government adopted several containment measures following the outbreak of COVID-19. These included promotion of social distancing practices, restrictions on public gatherings, night curfews and limits on public transport passenger capacities. Schools were closed from March 15, 2020, and the re-opening was done in phases, with schools fully re-opening on January 4, 2021. Another lockdown and renewed school closures were put in place in late March 2021. Based on the Oxford Stringency Index (Hale et al., 2021), a composite measure of the severity of policy response in nine areas, Kenya has consistently had more stringent policies in place than other countries in Sub- Saharan Africa. 3 Data and Empirical Strategy We use the first five waves of the data from a rapid response phone survey conducted with Kenyan households by the World Bank from May 2020 to June 2021 (The World Bank, 2021a). It includes samples of Kenyan nationals as well as refugees. The Kenyan national sample was drawn from two sources: (i) all households that were part of the 2015/16 Kenya Integrated Household Budget Survey Computer Assisted Personal Interview pilot and provided a phone number, and (ii) households selected using 3 the Random Digit Dialing method, using a random number generator from the 2020 Numbering Frame produced by the Kenya Communications Authority. The refugee samples were drawn from the UNHCR’s proGres database, which covers all registered refugees in Kenya. Two national lockdowns were put in place during the survey periods. The first one started in mid-March 2020 and covered the first months of the first wave of the survey. A second lockdown was introduced in late March 2021, wrapping the fifth wave of the survey. The data was collected by phone interviews with a nationally representative sample of households using cell phones in Kenya, covering both urban and rural areas. The households were called every two months for five survey rounds, to track the impact of the pandemic over time. The data consists of a wide range of information including the knowledge of COVID- 19 and mitigation measures, changes in behaviors in response to the COVID-19 pan- demic, employment, income, food security, as well as perceptions of the government’s response. We use the household head gender as the main indicator of gender. Compared to the male-headed households, the female-headed ones are more likely to receive transfers prior to the COVID-19 crisis (see Table 4 and Figure 2). Moreover, they have a higher childcare burden and smaller livestock values. However, these two types of households are not significantly different with regards to their poverty status. In Kenya, about 37.4 percent of the households are female-headed (The World Bank, 2021a). The female household heads are more likely to be separated, divorced or widowed relative to the male household heads. This indicates that it is likely that those female heads became the household head after the male head left the household due to death or divorce. There are in total three categories of coping strategies by the household. The first was coping using the household internal wealth or resources. These include coping by reducing meals, selling assets or livestock, and taking loans. The second kind of coping would be coping based on help from the social network, including using transfers from family members and assistance from friends. Lastly, very few households in Kenya receive help externally, for example, from governments, NGOs, or politicians. 4 There are two sets of coping strategies in the survey data. One set implies the general household coping strategies used in the past 14 days, and the other indicates the household actions specifically in response to the COVID-19 shocks. We use the first set of coping strategies in the main analysis and cross check the gender differences in COVID-19 specific household actions in the sensitivity analysis. As shown in the summary statistics displayed in Table A.1 in the Appendix, overall, the most commonly used coping strategy in Kenya is to reduce food consumption. About 38 percent of the households reduced meals when hit by COVID-19. Apart from that, households also take loans and rely on family transfers to cope. Only about four percent of the households have received any help from governments or NGOs. We first perform descriptive analysis by estimating the probability of using various coping strategies by male-headed and female-headed households in each survey wave. Afterwards, we use linear probability models pooling all survey waves to formally test the gender differences in coping strategies. Specifically, the model takes the following form. P (Copehct = 1) = α + βF emale Headht + γ Xht + δc + θt + hct (1) Where on the right hand side, Copehct is a dummy, indicating whether the household h in county c, interviewed in wave t, has used the specific coping strategy in the past 14 days. On the right hand side, there is a constant α, as well as a dummy, F emale Headht , indicating whether the head of household h in wave t is female (= 1) or male (= 0). Moreover, several household characteristics might determine the household coping strategies, including the level of wealth, local social norms and institution. In light of this, a set of household controls, Xht , is included in the regression model. These include the household head age, household size, the child-adult ratio, a dummy indicating whether the household is located in urban or rural areas, and a dummy indicating the poverty status by examining whether the household’s dwelling space has a dirt floor or not. Lastly, county fixed effects (δc ) and survey wave fixed effects (θt ) are included in the regression to control for all time-invariant characteristics specific to each county as well as all macro shocks at different interview waves, which might affect the choice of coping strategies directly. These might include, for example, different social norms 5 in coping strategies in different counties. The standard errors, hct , are clustered within counties. The child-adult ratio is calculated by dividing the number of children who are under 18 years old by the total number of adults aged between 18 and 65 in the household. We use a dummy indicating whether the dwelling space of the household has a dirt floor or finished floor as an indicator for poverty, as this measure is highly correlated with other poverty indicators in Kenya and feasible to elicit in a phone interview. Moreover, we hypothesize that female-headed households are poorer and have a closer relationship with social networks, and these cause the gender differences in coping. To formally check whether the gender differences in coping are due to these two reasons, we interact the household head gender dummy with the two hypothetical channels: the household poverty indicator, and the status of receiving transfers from families prior to the COVID-19 crisis. A series of sensitivity analyses are performed. Firstly, Kenyan nationals might employ different coping strategies compared to refugees. Thus, we exclude the refugee sample from the analysis and cross check the results. Second, 17.72 percent of the households have reported different household head gender across the five waves, and we exclude those households whose head gender was changed. All analyses are conducted in Stata v. 15.0 (College Station, Texas, USA) and take into account the sample design, particularly the sample weights and stratification. 4 Gender Differences in Coping Strategies 4.1 Main Results As shown in the descriptive figures (Figure 1), male-headed households are more likely to cope using household internal resources but female-headed ones rely more on social networks, although the differences are not significantly different. For example, male- headed households are more likely to sell assets to cope with a negative economic shock (Figure 1a). On average, about 10.4 percent of the male-headed households have sold any asset or livestock within the last 14 days, but only 6.9 percent of the female-headed 6 households have sold any asset. However, households with a female head are more likely to receive transfers from family members (Figure 1c). Specifically, female-headed (male-headed) households have a probability of 18.6 (12.7) percent of receiving any transfer or remittance from family. Although both the male-headed and female-headed households tend to reduce meals, there is no significant difference in the probability of reducing meals between these two types of households. The regression analysis further confirms the differences in coping strategies between female-headed and male-headed households (Table 1). Compared to male-headed households, female-headed households are three percentage points less likely to sell assets or take loans in the past 14 days (columns (2) and (4)). Moreover, they have five (or two) percentage point higher probability of receiving transfers from families (or gifts and assistance from friends), relative to male-headed households (column (6) and (7)). Table 1: Gender Differences in General Coping Strategies Household Internal External - Family/Network Reduce Meal Sell Asset ln Loan ln Family Gift/Assistance (0/1) (0/1) Asset (0/1) Loan Transfer (0/1) (0/1) (1) (2) (3) (4) (5) (6) (7) Female Headed Household (0/1) 0.021 -0.032∗∗∗ -0.142 -0.028∗∗ -0.006 0.036∗∗ 0.009 (0.018) (0.012) (0.104) (0.012) (0.130) (0.014) (0.011) Household Head Age 0.002∗∗∗ -0.000 0.001 -0.002∗∗∗ 0.009 0.002∗∗∗ 0.001∗∗ (0.001) (0.000) (0.004) (0.001) (0.007) (0.001) (0.001) ln Household Size 0.006 0.022 0.200∗ -0.009 0.197 -0.092∗∗∗ -0.039∗ (0.028) (0.015) (0.116) (0.020) (0.178) (0.020) (0.023) Child-adult Ratio 0.067∗∗∗ 0.010 -0.215∗∗∗ 0.010 -0.053 0.043∗∗∗ 0.003 (0.019) (0.015) (0.079) (0.014) (0.114) (0.012) (0.013) Urban (0/1) -0.029∗∗ -0.004 -0.108 -0.013 0.132 -0.002 0.002 (0.014) (0.011) (0.105) (0.015) (0.133) (0.013) (0.010) Dirt Floor (0/1) 0.087∗∗∗ 0.008 -0.146 -0.055∗∗∗ -0.266 0.041∗∗∗ 0.036∗ (0.015) (0.014) (0.102) (0.018) (0.172) (0.015) (0.018) Network (0/1) 0.014 0.023 0.051 0.043∗ 0.132 0.159∗∗∗ 0.131∗∗∗ (0.022) (0.020) (0.160) (0.024) (0.162) (0.023) (0.020) County, Wave, Strata FE Observations 30,967 30,978 2,521 30,978 4,666 30,934 30,968 R2 0.073 0.027 0.273 0.042 0.179 0.078 0.083 This table shows OLS regression estimates using the first five waves from the Rapid Response Phone Survey data in Kenya. Dependent variables are different coping strategies utilized by the household. These include reducing meals, selling the assets, taking loans, receiving transfers from family members, and getting assistance from friends. In all regressions, the household age, household size, child-adult ratio, the urban dummy, a dummy indicating whether the floor is made of dirt, and county, wave, strata fixed effects are adjusted. Robust standard errors clustered within counties are reported in parentheses. *** p<0.01, ** p<0.05, * p<0.1. In addition, as shown in Table 2, female-headed households would have smaller expenditures on food consumption and services, and more children in female-headed 7 Figure 1: Differences in Coping Strategies by Household Head Gender (a) Sell Assets (b) Take Loans (c) Family Transfers (d) Assistance from Social Network (e) Reduce Meals Data Source: Kenya COVID-19 RRPS. Authors’ own calculation. households would skip meals. 8 Table 2: Gender Differences in Household Consumption Expenditures on ... # Days Hungry At Night # Days Skipped Meal # Days No Meal All Day Food Personal Durables Services Communication Adult Child Adult Child Adult Child (1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) Female Headed Household (0/1) -127.920∗∗∗ 11.330 12.742 -36.903∗∗∗ -32.174∗∗∗ 0.015 0.027 0.097 0.097∗ -0.031 -0.022 (39.799) (15.679) (36.370) (11.247) (8.736) (0.033) (0.029) (0.066) (0.056) (0.027) (0.024) Household Head Age -3.201 -0.427 -0.263 -1.627 -0.828∗ -0.000 0.000 0.006∗∗ 0.006∗∗∗ -0.000 -0.001∗ (2.240) (0.499) (1.673) (1.160) (0.442) (0.001) (0.001) (0.003) (0.002) (0.001) (0.001) ln Household Size 594.504∗∗∗ 141.558∗∗∗ -5.293 52.107∗∗∗ 44.311∗∗∗ 0.069∗ 0.095∗∗∗ -0.025 0.200∗∗ 0.033 0.049∗∗ (61.717) (21.376) (54.955) (13.782) (14.886) (0.035) (0.023) (0.128) (0.085) (0.032) (0.024) Child-adult Ratio -76.848 -65.192∗∗∗ -66.381∗∗∗ -49.983∗∗ -36.264∗∗∗ 0.065∗ 0.114∗∗∗ 0.382∗∗∗ 0.498∗∗∗ 0.017 0.032 (66.662) (18.201) (16.521) (21.458) (8.186) (0.034) (0.031) (0.077) (0.076) (0.032) (0.027) Urban (0/1) 71.933∗ 0.025 -60.108 -20.670 33.342∗∗∗ -0.014 -0.014 -0.014 -0.031 -0.010 -0.012 (37.714) (15.620) (52.100) (35.968) (9.677) (0.035) (0.018) (0.078) (0.052) (0.021) (0.015) Dirt Floor (0/1) -247.111∗∗∗ -92.461∗∗∗ -75.471∗∗∗ -45.094∗∗ -71.131∗∗∗ 0.152∗∗∗ 0.085∗∗∗ 0.322∗∗∗ 0.236∗∗∗ 0.065∗∗ 0.045∗ (44.332) (16.593) (25.237) (20.850) (9.668) (0.035) (0.025) (0.069) (0.061) (0.030) (0.025) Network (0/1) 7.281 35.702∗ -32.669 13.936 -13.611 0.010 -0.003 -0.141∗ -0.177∗∗∗ -0.028 -0.046∗∗∗ (51.534) (18.055) (35.492) (10.783) (8.153) (0.035) (0.029) (0.074) (0.054) (0.026) (0.012) County, Wave, Strata FE Observations 30,550 30,545 24,354 30,739 30,788 30,884 30,896 30,892 30,900 30,897 30,901 R2 0.108 0.042 0.018 0.087 0.068 0.099 0.092 0.079 0.151 0.061 0.063 This table shows OLS regression estimates using the first five waves from the Rapid Response Phone Survey data in Kenya. All regressions include the set of controls described in Table 1. Robust standard errors clustered within counties are reported in parentheses. *** p<0.01, ** p<0.05, * p<0.1. When it comes to household actions specific to COVID-19, similar gender differences are observed. Male-headed households have a higher probability of selling assets and female-headed households are more likely to receive help from the social network (Figure A.1a-A.1b). In addition, the regression analysis in Table 3 suggests that female-headed households are less likely to sell assets and take loans, but have higher probabilities of receiving assistance from family members in light of the COVID-19 shocks. 9 Table 3: Gender Differences in Household Action to Cope with COVID-19 Sell Use Generate Sell Take Buy w. Delay Reduce Consumption Friend/Family Assist No Asset Savings Income Harvest Loan Credit Pay Food Non-food Assist Borrow NGO Employer Gov Insurance Actions (1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12) (13) (14) (15) (16) Female Headed Household (0/1) -0.034∗∗∗ -0.019 0.008 -0.014 -0.028∗∗∗ 0.019 0.014 -0.013 -0.016 0.037∗∗∗ 0.009 -0.002 -0.005∗ 0.004 0.001 -0.009 (0.009) (0.020) (0.013) (0.009) (0.010) (0.018) (0.010) (0.018) (0.025) (0.012) (0.015) (0.002) (0.002) (0.004) (0.005) (0.015) Household Head Age -0.000 -0.001 -0.001∗ 0.001∗∗ -0.001∗∗∗ -0.000 -0.000 0.001 0.001 0.002∗∗∗ -0.001∗∗∗ 0.000 -0.000 0.001∗∗ 0.000 0.000 (0.000) (0.001) (0.001) (0.000) (0.000) (0.000) (0.000) (0.001) (0.001) (0.001) (0.000) (0.000) (0.000) (0.000) (0.000) (0.001) ln Household Size 0.027∗ 0.047∗∗ 0.031 0.014 -0.008 0.042∗∗ 0.024 0.082∗∗∗ 0.065∗∗ -0.046∗∗∗ 0.026∗∗ -0.003 -0.006 -0.009 -0.001 -0.019 (0.015) (0.019) (0.027) (0.011) (0.013) (0.016) (0.016) (0.024) (0.025) (0.015) (0.011) (0.004) (0.004) (0.009) (0.007) (0.025) Child-adult Ratio 0.003 -0.034∗∗ -0.015 -0.001 -0.008 -0.011 -0.015∗ 0.029 0.011 0.005 -0.006 0.004 0.001 0.007 -0.002 -0.015 (0.013) (0.013) (0.021) (0.008) (0.006) (0.014) (0.008) (0.019) (0.016) (0.015) (0.013) (0.004) (0.003) (0.010) (0.005) (0.013) Urban (0/1) -0.018 -0.010 -0.000 -0.008 -0.015 -0.009 0.012 0.014 0.021 -0.014 -0.032∗∗∗ -0.005 -0.003 -0.003 0.001 0.006 (0.012) (0.019) (0.024) (0.006) (0.015) (0.015) (0.010) (0.018) (0.014) (0.011) (0.009) (0.005) (0.005) (0.003) (0.004) (0.012) Dirt Floor (0/1) 0.008 0.030 -0.023 0.017∗ -0.033∗∗ 0.022 -0.011 0.017 -0.002 0.005 -0.027∗ 0.004 -0.007∗∗∗ -0.001 -0.014∗∗∗ -0.020 (0.013) (0.019) (0.019) (0.010) (0.013) (0.019) (0.012) (0.020) (0.018) (0.014) (0.015) (0.004) (0.002) (0.003) (0.005) (0.018) Network (0/1) 0.004 0.021 -0.079∗∗∗ 0.017 0.052∗ 0.028 0.001 0.056∗ 0.084∗∗∗ 0.142∗∗∗ 0.043∗∗ -0.001 -0.004∗∗ 0.010 -0.001 -0.020 (0.015) (0.022) (0.021) (0.016) (0.026) (0.017) (0.016) (0.031) (0.027) (0.026) (0.021) (0.002) (0.002) (0.010) (0.008) (0.014) County, Wave, Strata FE Observations 30947 30947 30947 30947 30947 30947 30947 30947 30947 30947 30947 30947 30947 30947 30947 30947 R2 0.026 0.055 0.041 0.040 0.034 0.029 0.021 0.061 0.069 0.059 0.045 0.091 0.025 0.041 0.028 0.035 This table shows OLS regression estimates using the first five waves from the Rapid Response Phone Survey data in Kenya. All regressions include the set of controls described in Table 1. Robust standard errors clustered within counties are reported in parentheses. *** p<0.01, ** p<0.05, * p<0.1. 10 4.2 Reasons for the Gender Difference We propose two potential reasons behind the gender differences in coping strategies. Firstly, the female-headed households are relatively poorer. Although the female- headed households are not more likely to have unfinished floors, they bear relatively higher childcare burden (Figure 2b, and Table 4, column (3)). The child-adult ratio is about 7.2 percent higher in female-headed compared to male-headed households. In male-headed ones, the average child-adult dependency ratio is 0.47, that is, there are 2.13 adults for every child in the household. However, the female-headed households have an average child-adult dependency ratio of 0.54, and there are only 1.85 adults for every child. In addition, the average livestock value in the male-headed households is about 60 percent higher than in the female-headed ones (Figure 2c, and Table 4, column (4)). Moreover, the female-headed households are less likely to engage in agricultural production or have wage employees to generate income (Table 4, column (5)-(7)). Secondly, female-headed households are more dependent on social networks prior to the COVID-19 outbreak (Figure 2d, and Table 4, column (1)). These two factors might explain why male-headed households are more likely to cope using household internal wealth (since they are relatively richer) but female-headed one rely more on social network (as they are more dependent on network even before the COVID-19 shock). Table 4: Channels - Gender Differences in Social Network, Poverty, and Source of Income Network (0/1) Dirt Floor (0/1) Child-Adult Ratio ln Livestock Agriculture (0/1) Wage Employment (0/1) Family Business (0/1) (1) (2) (3) (4) (5) (6) (7) Female Headed Household (0/1) 0.071∗∗∗ 0.006 0.136∗∗∗ -0.600∗∗∗ -0.046∗ -0.064∗∗∗ 0.003 (0.017) (0.025) (0.027) (0.160) (0.024) (0.020) (0.015) Household Head Age 0.001 0.004∗∗∗ -0.004∗∗∗ 0.001 0.005∗∗∗ 0.001∗ -0.002∗∗∗ (0.001) (0.001) (0.001) (0.003) (0.001) (0.001) (0.000) ln Household Size -0.051∗ 0.124∗∗∗ 0.827∗∗∗ 0.028 0.208∗∗∗ 0.105∗∗∗ 0.028 (0.027) (0.033) (0.027) (0.241) (0.031) (0.033) (0.022) Child-adult Ratio -0.000 0.040∗ -0.114 -0.026 -0.012 0.001 (0.020) (0.022) (0.196) (0.022) (0.016) (0.022) Urban (0/1) 0.002 -0.132∗∗∗ 0.009 -0.100 -0.119∗∗∗ -0.031∗ 0.006 (0.010) (0.023) (0.029) (0.188) (0.024) (0.016) (0.012) County, Wave, Strata FE Observations 31,187 31,481 31,703 6,095 31,703 28,930 28,389 R2 0.058 0.252 0.501 0.161 0.311 0.140 0.050 This table shows OLS regression estimates using the first five waves from the Rapid Response Phone Survey data in Kenya. All regressions include the set of controls described in Table 1. Robust standard errors clustered within counties are reported in parentheses. *** p<0.01, ** p<0.05, * p<0.1. When we interact the household head gender dummy with the two hypothetical channels, and include the channel variables into the regressions, we find further 11 Figure 2: Channels for the Gender Differences in Coping (a) Poverty (b) Child-adult Ratio (c) Livestock Value (d) Pre-COVID Receipt of Family Transfers Data Source: Kenya COVID-19 RRPS. Authors’ own calculation. evidence that these might cause the gender differences in coping strategies. For instance, in Table 5 column (1), we see that households having a female head, or those being poor, are more likely to reduce meals. However, those households with a female head and which are poor are less likely to reduce meals. This is probably because this type of households already have very low levels of food consumption and could not further reduce meals to cope with COVID-19. In columns (2)-(5), after including the interaction term into the regression, none of the coefficients of the household head gender dummy is significant. This suggests that having a female head but not being poor is not associated with a lower probability of selling assets or taking loans, but being poor and having a female head would associate to a lower probability of coping using household internal wealth. Moreover, 12 the coefficients of the poverty dummy in columns (4) and (5) show that those poor households are less likely to take loans, and would take smaller loans if they take any, to cope. Table 5: Gender Differences in General Coping Strategies - Poverty Channel Household Internal External - Family/Network Reduce Meal Sell Asset ln Loan ln Family Gift/Assistance (0/1) (0/1) Asset (0/1) Loan Transfer (0/1) (0/1) (1) (2) (3) (4) (5) (6) (7) Female Headed Household (0/1) 0.044∗∗∗ -0.031∗ -0.140 -0.024 -0.072 0.034∗∗ 0.011 (0.016) (0.016) (0.145) (0.016) (0.134) (0.016) (0.014) Female Head (0/1) × Dirt Floor (0/1) -0.064∗∗ -0.003 -0.006 -0.013 0.241 0.008 -0.006 (0.026) (0.025) (0.245) (0.027) (0.286) (0.028) (0.024) Dirt Floor (0/1) 0.110∗∗∗ 0.009 -0.144 -0.051∗∗ -0.348∗ 0.038∗ 0.038∗ (0.020) (0.019) (0.131) (0.022) (0.198) (0.020) (0.021) Household Head Age 0.002∗∗∗ -0.000 0.001 -0.002∗∗∗ 0.009 0.002∗∗∗ 0.001∗∗ (0.001) (0.000) (0.004) (0.001) (0.006) (0.001) (0.001) ln Household Size 0.005 0.022 0.200 -0.009 0.182 -0.092∗∗∗ -0.039∗ (0.028) (0.015) (0.124) (0.020) (0.171) (0.020) (0.022) Child-adult Ratio 0.068∗∗∗ 0.010 -0.214∗∗ 0.010 -0.044 0.043∗∗∗ 0.003 (0.019) (0.015) (0.083) (0.014) (0.110) (0.012) (0.013) Urban (0/1) -0.030∗∗ -0.004 -0.108 -0.014 0.138 -0.002 0.002 (0.014) (0.011) (0.105) (0.015) (0.131) (0.013) (0.010) Network (0/1) 0.014 0.023 0.051 0.043∗ 0.137 0.159∗∗∗ 0.131∗∗∗ (0.022) (0.020) (0.160) (0.024) (0.160) (0.023) (0.020) County, Wave, Strata FE Observations 30,967 30,978 2,521 30,978 4,666 30,934 30,968 R2 0.074 0.027 0.273 0.042 0.180 0.078 0.083 This table shows OLS regression estimates using the first five waves from the Rapid Response Phone Survey data in Kenya. All regressions include the set of controls described in Table 1. Robust standard errors clustered within counties are reported in parentheses. *** p<0.01, ** p<0.05, * p<0.1. Similarly, in Table 6 column (1), we also find that households with a female head or those which relied on family transfers prior to the COVID-19 crisis are more likely to reduce meals. However, those with a female head and relied on family transfers before are less likely to reduce meals. In columns (6)-(7), we observe that the magnitude of the positive correlation between the female head dummy and coping using family transfers or network assistance reduces after including the network dummy, compared to Table 1. This indicates that the prior receipt of family transfers is correlated with the female head dummy, i.e. households with a female head are more likely to receive family transfers prior to COVID-19. In addition, those households, which received transfers prior to the crisis, are more likely to cope using help from the social network, which further confirms this channel. When both channel variables as well as the interaction terms are included in the same regression (Table 7), we still observe that being poor and reliant on social network 13 Table 6: Gender Differences in General Coping Strategies - Social Network Channel Household Internal External - Family/Network Reduce Meal Sell Asset ln Loan ln Family Gift/Assistance (0/1) (0/1) Asset (0/1) Loan Transfer (0/1) (0/1) (1) (2) (3) (4) (5) (6) (7) Female Headed Household (0/1) 0.037∗ -0.031∗∗ -0.095 -0.022 -0.033 0.037∗∗ 0.014 (0.019) (0.013) (0.102) (0.014) (0.138) (0.014) (0.011) Female Head (0/1) × Network (0/1) -0.146∗∗∗ -0.015 -0.326 -0.057 0.206 -0.002 -0.054 (0.046) (0.047) (0.369) (0.062) (0.250) (0.046) (0.044) Network (0/1) 0.091∗∗∗ 0.031 0.186 0.073 0.034 0.161∗∗∗ 0.159∗∗∗ (0.027) (0.028) (0.146) (0.045) (0.183) (0.030) (0.030) Household Head Age 0.002∗∗∗ -0.000 0.001 -0.002∗∗∗ 0.009 0.002∗∗∗ 0.001∗∗ (0.001) (0.000) (0.004) (0.001) (0.007) (0.001) (0.001) ln Household Size 0.005 0.022 0.191 -0.009 0.197 -0.092∗∗∗ -0.039∗ (0.028) (0.015) (0.117) (0.020) (0.179) (0.020) (0.023) Child-adult Ratio 0.068∗∗∗ 0.010 -0.210∗∗∗ 0.010 -0.055 0.043∗∗∗ 0.003 (0.019) (0.015) (0.078) (0.015) (0.114) (0.012) (0.013) Urban (0/1) -0.029∗∗ -0.004 -0.109 -0.013 0.131 -0.002 0.002 (0.014) (0.011) (0.105) (0.015) (0.132) (0.013) (0.010) Dirt Floor (0/1) 0.087∗∗∗ 0.008 -0.139 -0.055∗∗∗ -0.267 0.041∗∗∗ 0.036∗∗ (0.015) (0.014) (0.103) (0.018) (0.171) (0.015) (0.018) County, Wave, Strata FE Observations 30,967 30,978 2,521 30,978 4,666 30,934 30,968 R2 0.075 0.027 0.275 0.042 0.179 0.078 0.084 This table shows OLS regression estimates using the first five waves from the Rapid Response Phone Survey data in Kenya. All regressions include the set of controls described in Table 1. Robust standard errors clustered within counties are reported in parentheses. *** p<0.01, ** p<0.05, * p<0.1. could shape the choice of coping strategies used by the households. Firstly, as seen in columns (6) and (7), those households, which have received transfers from social networks prior to the COVID-19 crisis, are more likely to rely on transfers and assistance from family and friends to cope. In addition, when partialling out the effect of the pre-COVID dependency on social network, female-headed households are still 3.7 percentage points more likely to cope using transfers from family members. Secondly, we could observe that poorer households are less likely to take a loan to cope (columns (4)-(5)). 4.3 Robustness Checks We conduct two sensitivity analyses and check the robustness of the results in Table 1. To begin with, the rapid response phone survey sample consists of Kenyan nationals as well as refugees. Refugees might have different coping strategies from nationals and thus we exclude the refugee sample and cross-check the gender differences in coping among the Kenyan nationals. 14 Table 7: Gender Differences in General Coping Strategies - Both Channels Household Internal External - Family/Network Reduce Meal Sell Asset ln Loan ln Family Gift/Assistance (0/1) (0/1) Asset (0/1) Loan Transfer (0/1) (0/1) (1) (2) (3) (4) (5) (6) (7) Female Headed Household (0/1) 0.059∗∗∗ -0.030∗ -0.102 -0.018 -0.091 0.034∗∗ 0.016 (0.016) (0.017) (0.136) (0.018) (0.141) (0.016) (0.013) Female Head (0/1) × Network (0/1) -0.142∗∗∗ -0.015 -0.329 -0.056 0.175 -0.003 -0.053 (0.046) (0.046) (0.383) (0.062) (0.243) (0.046) (0.044) Network (0/1) 0.089∗∗∗ 0.031 0.186 0.072 0.054 0.161∗∗∗ 0.159∗∗∗ (0.028) (0.028) (0.148) (0.045) (0.181) (0.030) (0.030) Female Head (0/1) × Dirt Floor (0/1) -0.060∗∗ -0.003 0.017 -0.011 0.227 0.008 -0.004 (0.027) (0.025) (0.257) (0.026) (0.285) (0.028) (0.024) Dirt Floor (0/1) 0.109∗∗∗ 0.009 -0.144 -0.051∗∗ -0.345∗ 0.038∗ 0.038∗ (0.020) (0.019) (0.131) (0.022) (0.199) (0.020) (0.021) Household Head Age 0.002∗∗∗ -0.000 0.001 -0.002∗∗∗ 0.009 0.002∗∗∗ 0.001∗∗ (0.001) (0.000) (0.004) (0.001) (0.006) (0.001) (0.001) ln Household Size 0.004 0.022 0.192 -0.009 0.183 -0.092∗∗∗ -0.039∗ (0.028) (0.015) (0.125) (0.020) (0.172) (0.020) (0.023) Child-adult Ratio 0.068∗∗∗ 0.010 -0.211∗∗ 0.010 -0.046 0.043∗∗∗ 0.003 (0.019) (0.015) (0.082) (0.015) (0.109) (0.012) (0.013) Urban (0/1) -0.029∗∗ -0.004 -0.109 -0.013 0.136 -0.002 0.002 (0.014) (0.011) (0.105) (0.015) (0.130) (0.013) (0.010) County, Wave, Strata FE Observations 30,967 30,978 2,521 30,978 4,666 30,934 30,968 R2 0.076 0.027 0.275 0.043 0.180 0.078 0.084 This table shows OLS regression estimates using the first five waves from the Rapid Response Phone Survey data in Kenya. All regressions include the set of controls described in Table 1. Robust standard errors clustered within counties are reported in parentheses. *** p<0.01, ** p<0.05, * p<0.1. 15 As set out in Table A.2 in the Appendix, female-headed households are three percentage points less likely to cope by selling assets and taking loans, but are five percentage points more likely to receive family transfers. The results are almost the same as those in Table 1. Therefore, the results are in line with the main analysis. Secondly, about 18 percent of the households report different household head gender. We exclude this type of households and analyze whether there is different household coping by different gender heads. As seen from Table A.3 in the Appendix, the female- headed households are less likely to sell assets, and more likely to receive transfers from family and assistance from friends. Overall, the sensitivity analyses here confirms the main results as shown in Table 1. 5 Conclusion In this paper, we study the differences in coping strategies by households with different head genders in Kenya. We find that female-headed households are more likely to cope using transfers from family members and male-headed ones would sell assets or take loans. However, no gender difference in reducing meals is observed. We also show that the difference is due to differences in household wealth and reliance on social networks. While a few studies have analyzed the household coping strategies in light of COVID- 19, no prior paper has investigated in the gender differences in coping. This paper has not only revealed the differences in household coping strategies, but also pointed out the potential reasons behind them. Moreover, the findings in this study are consistent with Quisumbing, Haddad and Pena ˜ (2001), who find that female-headed households in Africa are poorer but more likely to receive family transfers. The findings in this paper have several implications. Firstly, we point out that the female-headed households in Kenya are specifically vulnerable in light of a negative shock. Moreover, those female household heads are more likely to be divorced or widowed compared to the male household heads. This suggests that the relief programs need to specifically target certain types of households (which are female-headed, and poorer) or widowed and divorced women to achieve higher effectiveness. Secondly, we point out that female-headed households rely more on family transfers even before the COVID-19 pandemic. This indicates that those households might be in desperate need 16 of resources, and governments and NGOs could set up special cash transfer programs for this specific group. Finally, COVID-19 might have a long-lasting negative influence on household assets. The male-headed households are more likely to sell assets and livestock than the female-headed ones, but the female-headed ones have a smaller amount of assets. For the male-headed ones, while they might be able to cushion the shock in the short run, the sale of assets or livestock could have severe consequences for household welfare and productivity in the medium and long run. For the female- headed ones, there is need of supporting them in securing assets. 17 References Aguero, ¨ Jorge M. 2021. “COVID-19 and the rise of intimate partner violence.” World development, 137: 105217. Alon, Titan, Matthias Doepke, Jane Olmstead-Rumsey, and Mich` ele Tertilt. 2020. “The impact of COVID-19 on gender equality.” National Bureau of economic research. Altig, Dave, Scott Baker, Jose Maria Barrero, Nicholas Bloom, Philip Bunn, Scarlet Chen, Steven J Davis, Julia Leather, Brent Meyer, Emil Mihaylov, et al. 2020. “Economic uncertainty before and during the COVID-19 pandemic.” Journal of Public Economics, 191: 104274. Andersen, Steffen, Seda Ertac, Uri Gneezy, John A List, and Sandra Maximiano. 2013. “Gender, competitiveness, and socialization at a young age: Evidence from a matrilineal and a patriarchal society.” Review of Economics and Statistics, 95(4): 1438– 1443. Armand, Alex, Orazio Attanasio, Pedro Carneiro, and Val´ erie Lechene. 2020. “The effect of gender-targeted conditional cash transfers on household expenditures: Evidence from a randomized experiment.” The Economic Journal, 130(631): 1875–1897. Ashraf, Nava. 2009. “Spousal control and intra-household decision making: An experimental study in the Philippines.” American Economic Review, 99(4): 1245–77. Bartik, Alexander W, Marianne Bertrand, Zoe Cullen, Edward L Glaeser, Michael Luca, and Christopher Stanton. 2020. “The impact of COVID-19 on small business outcomes and expectations.” Proceedings of the National Academy of Sciences, 117(30): 17656–17666. Bloem, Jeffrey R, and Colette Salemi. 2021. “COVID-19 and Conflict.” World development, 140: 105294. Dang, Hai-Anh H, and Cuong Viet Nguyen. 2021. “Gender inequality during the COVID-19 pandemic: Income, expenditure, savings, and job loss.” World Development, 140: 105296. Deaton, Angus. 2021. “Covid-19 and global income inequality.” National Bureau of Economic Research. De Brauw, Alan, Daniel O Gilligan, John Hoddinott, and Shalini Roy. 2014. “The ılia on womens decision-making power.” World Development, impact of Bolsa Fam´ 59: 487–504. Dittrich, Marcus, and Kristina Leipold. 2014. “Gender differences in time preferences.” Economics Letters, 122(3): 413–415. 18 Dohmen, Thomas, Armin Falk, David Huffman, Uwe Sunde, Jurgen ¨ Schupp, and Gert G Wagner. 2011. “Individual risk attitudes: Measurement, determinants, and behavioral consequences.” Journal of the european economic association, 9(3): 522–550. Eckel, Catherine C, and Philip J Grossman. 1998. “Are women less selfish than men?: Evidence from dictator experiments.” The economic journal, 108(448): 726–735. Eckel, Catherine C, and Philip J Grossman. 2008. “Forecasting risk attitudes: An experimental study using actual and forecast gamble choices.” Journal of Economic Behavior & Organization, 68(1): 1–17. Ellis, Amanda, et al. 2007. Gender and economic growth in Kenya: Unleashing the power of women. World Bank Publications. Etheridge, Ben, and Lisa Spantig. 2020. “The gender gap in mental well-being during the Covid-19 outbreak: evidence from the UK.” ISER Working paper series. Evans, Megan L, Margo Lindauer, and Maureen E Farrell. 2020. “A pandemic within a pandemic: Intimate partner violence during Covid-19.” New England journal of medicine, 383(24): 2302–2304. sim Orhun, and Dana Turjeman. 2020. “Heterogeneous actions, beliefs, Fan, Ying, A Ye¸ constraints and risk tolerance during the COVID-19 pandemic.” National Bureau of Economic Research. Galasso, Vincenzo, Vincent Pons, Paola Profeta, Michael Becher, Sylvain Brouard, and Martial Foucault. 2020. “Gender differences in COVID-19 attitudes and behavior: Panel evidence from eight countries.” Proceedings of the National Academy of Sciences, 117(44): 27285–27291. Garc´ andez, Lorena, Veronica ıa-Fern´ ´ Romero-Ferreiro, Sergio Padilla, Pedro David Lopez-Rold´ ´ an, Mar´ ıa Monzo-Garc´ ´ ıa, and Roberto Rodriguez-Jimenez. 2021. “Gender differences in emotional response to the COVID-19 outbreak in Spain.” Brain and behavior, 11(1): e01934. Hale, Thomas, Noam Angrist, Rafael Goldszmidt, Beatriz Kira, Anna Petherick, Toby Phillips, Samuel Webster, Emily Cameron-Blake, Laura Hallas, Saptarshi Majumdar, et al. 2021. “A global panel database of pandemic policies (Oxford COVID-19 Government Response Tracker).” Nature Human Behaviour, 5(4): 529–538. Han, Jeehoon, Bruce D Meyer, and James X Sullivan. 2020. “Income and Poverty in the COVID-19 Pandemic.” National Bureau of Economic Research. Hanspal, Tobin, Annika Weber, and Johannes Wohlfart. 2020. “Exposure to the COVID-19 stock market crash and its effect on household expectations.” Review of Economics and Statistics, 1–45. 19 Horrell, Sara, and Pramila Krishnan. 2007. “Poverty and productivity in female- headed households in Zimbabwe.” The journal of development studies, 43(8): 1351–1380. Ide, Tobias. 2021. “COVID-19 and armed conflict.” World development, 140: 105355. Kansiime, Monica K, Justice A Tambo, Idah Mugambi, Mary Bundi, Augustine Kara, and Charles Owuor. 2021. “COVID-19 implications on household income and food security in Kenya and Uganda: Findings from a rapid assessment.” World development, 137: 105199. Kugler, Maurice David, Mariana Viollaz, Daniel Vasconcellos Archer Duque, Isis Gaddis, David Locke Newhouse, Amparo Palacios-Lopez, Michael Weber, et al. 2021. “How Did the COVID-19 Crisis Affect Different Types of Workers in the Developing World?” The World Bank. Mena, Gonzalo E, Pamela P Martinez, Ayesha S Mahmud, Pablo A Marquet, Caroline O Buckee, and Mauricio Santillana. 2021. “Socioeconomic status determines COVID-19 incidence and related mortality in Santiago, Chile.” Science, 372(6545). Meyer, Brent H, Brian Prescott, and Xuguang Simon Sheng. 2021. “The impact of the COVID-19 pandemic on business expectations.” International Journal of Forecasting. Olaseni, Abayomi O, Olusola S Akinsola, Samson F Agberotimi, and Rotimi Oguntayo. 2020. “Psychological distress experiences of Nigerians during Covid-19 pandemic; the gender difference.” Social Sciences & Humanities Open, 2(1): 100052. Pfefferbaum, Betty, and Carol S North. 2020. “Mental health and the Covid-19 pandemic.” New England Journal of Medicine, 383(6): 510–512. Quisumbing, Agnes R, Lawrence Haddad, and Christine Pena. ˜ 2001. “Are women overrepresented among the poor? An analysis of poverty in 10 developing countries.” Journal of Development Economics, 66(1): 225–269. Rubalcava, Luis, Graciela Teruel, and Duncan Thomas. 2009. “Investments, time preferences, and public transfers paid to women.” Economic Development and cultural change, 57(3): 507–538. Sachs, Jeffrey D. 2012. “From millennium development goals to sustainable development goals.” The lancet, 379(9832): 2206–2211. Schellekens, Philip, and Diego M Sourrouille. 2020. “COVID-19 mortality in rich and poor countries: a tale of two pandemics?” World Bank Policy Research Working Paper, , (9260). The World Bank. 2021a. “Kenya COVID-19 Rapid Response Phone Survey Households 2020-2021, Panel.” 20 The World Bank. 2021b. “Poverty and Equity Brief, Kenya.” the World Bank. Xue, Baowen, and Anne McMunn. 2021. “Gender differences in unpaid care work and psychological distress in the UK Covid-19 lockdown.” PloS one, 16(3): e0247959. 21 Appendix Figure A.1: Differences in COVID-19 Household Actions by Household Head Gender (a) Sell Assets (b) Family Transfer Data Source: Kenya COVID-19 RRPS. Authors’ own calculation. 22 Table A.1: Descriptive Statistics Mean s.d. Min Max Count Household Characteristics Female Headed Household (0/1) 0.37 0.48 0 1 31,715 No Change in Household Head Gender (0/1) 0.82 0.38 0 1 31,715 Child-adult Ratio 0.50 0.57 0 7 31,704 Household Size 3.73 2.33 1 37 31,715 ln Household Size 1.43 0.50 1 4 31,715 Household Head Age 38.48 13.81 18 100 31,715 Urban (0/1) 0.36 0.48 0 1 31,715 Dirt Floor (0/1) 0.37 0.48 0 1 31,493 Pre-COVID Network (0/1) 0.10 0.30 0 1 31,199 Livestock Value 53,197.90 71,264.80 0 1,000,000 6,098 ln Livestock Value 10.04 2.13 0 14 6,098 Agriculture (0/1) 0.44 0.50 0 1 31,715 Wage (0/1) 0.73 0.44 0 1 28,942 General Household Coping Strategies Reduce Meals (0/1) 0.38 0.49 0 1 31,702 Sell Asset (0/1) 0.09 0.29 0 1 31,715 Asset Sold 4,810.77 16,112.27 1 400,000 2,562 ln Asset Sold 7.77 1.14 1 13 2,562 Take Loan (0/1) 0.17 0.37 0 1 31,715 Loan Taken 6,207.00 24,264.42 3 2,000,000 4,739 ln Loan Taken 7.73 1.29 1 15 4,739 Family Transfer (0/1) 0.15 0.36 0 1 31,665 Friends Assistance (0/1) 0.10 0.30 0 1 31,705 Government/Institutions (0/1) 0.04 0.20 0 1 31,715 COVID-specific Household Coping Strategies Sell Asset (0/1) 0.09 0.29 0 1 31,679 Use Savings (0/1) 0.34 0.47 0 1 31,679 Generate Income (0/1) 0.26 0.44 0 1 31,679 Sell Harvest (0/1) 0.04 0.21 0 1 31,679 Take Loan (0/1) 0.09 0.28 0 1 31,679 Buy w. Credit (0/1) 0.18 0.38 0 1 31,679 Delay Repayment (0/1) 0.10 0.29 0 1 31,679 Reduce Food Consumption (0/1) 0.39 0.49 0 1 31,679 Reduce Non-food Consumption (0/1) 0.36 0.48 0 1 31679 Friends/Family Assistance (0/1) 0.10 0.30 0 1 31,679 Borrow from Friends/Family (0/1) 0.07 0.26 0 1 31,679 Assistance from NGO (0/1) 0.01 0.08 0 1 31,679 Assistance from Employer (0/1) 0.01 0.08 0 1 31,679 Assistance from Gov (0/1) 0.01 0.09 0 1 31,679 Covered by Insurance (0/1) 0.02 0.13 0 1 31,679 No Actions (0/1) 0.19 0.40 0 1 31,679 Household Vulnerability Household Expenditure on Food 1,248.87 1,228.05 0 75,000 31,250 ... on Household and Personal Items 344.58 452.25 0 30,000 31,239 ... on Assets / Durable 174.83 1,083.38 0 80,000 24,945 ... on Local Services 164.72 494.64 0 20,028 31,442 ... on Communication 194.45 262.62 0 12,000 31,504 # Days Adult Hungry 0.30 0.81 0 7 31,611 # Days Children Hungry 0.16 0.57 0 7 31,625 # Days Adult Skipped Meal 1.12 2.02 0 7 31,620 # Days Children Skipped Meal 0.60 1.64 0 7 31,629 # Days Adult All Day No Meal 0.12 0.58 0 7 31,626 # Days Children All Day No Meal 0.07 0.48 0 7 31,630 Observations 31,715 23 Table A.2: Gender Differences in General Coping Strategies - Excluding the Refugee Sample Household Internal External - Family/Network Reduce Meal Sell Asset ln Loan ln Family Gift/Assistance (0/1) (0/1) Asset (0/1) Loan Transfer (0/1) (0/1) (1) (2) (3) (4) (5) (6) (7) Female Headed Household (0/1) 0.022 -0.033∗∗∗ -0.140 -0.028∗∗ -0.007 0.037∗∗ 0.008 (0.018) (0.012) (0.105) (0.012) (0.131) (0.014) (0.011) Household Head Age 0.002∗∗∗ -0.000 0.001 -0.002∗∗∗ 0.009 0.003∗∗∗ 0.001∗∗ (0.001) (0.000) (0.004) (0.001) (0.007) (0.001) (0.001) ln Household Size 0.006 0.022 0.201∗ -0.009 0.195 -0.093∗∗∗ -0.039∗ (0.028) (0.015) (0.116) (0.021) (0.181) (0.020) (0.023) Child-adult Ratio 0.068∗∗∗ 0.010 -0.215∗∗∗ 0.010 -0.054 0.043∗∗∗ 0.003 (0.019) (0.015) (0.080) (0.015) (0.117) (0.012) (0.013) Urban (0/1) -0.029∗∗ -0.004 -0.109 -0.014 0.132 -0.002 0.002 (0.014) (0.011) (0.105) (0.015) (0.133) (0.013) (0.010) Dirt Floor (0/1) 0.087∗∗∗ 0.009 -0.146 -0.055∗∗∗ -0.270 0.041∗∗∗ 0.036∗∗ (0.015) (0.014) (0.103) (0.018) (0.174) (0.015) (0.018) Network (0/1) 0.014 0.024 0.051 0.044∗ 0.133 0.160∗∗∗ 0.132∗∗∗ (0.022) (0.020) (0.162) (0.025) (0.163) (0.023) (0.020) County, Wave, Strata FE Observations 23,872 23,875 2,126 23,875 3,464 23,854 23,871 R2 0.073 0.027 0.275 0.042 0.179 0.077 0.083 This table shows OLS regression estimates using the first five waves from the Rapid Response Phone Survey data in Kenya. All regressions include the set of controls described in Table 1. Robust standard errors clustered within counties are reported in parentheses. *** p<0.01, ** p<0.05, * p<0.1. 24 Table A.3: Gender Differences in General Coping Strategies - Excluding Households That Reported Different Household Head Gender Household Internal External - Family/Network Reduce Meal Sell Asset ln Loan ln Family Gift/Assistance (0/1) (0/1) Asset (0/1) Loan Transfer (0/1) (0/1) (1) (2) (3) (4) (5) (6) (7) Female Headed Household (0/1) 0.036∗ -0.047∗∗∗ -0.096 -0.019 0.053 0.048∗∗∗ 0.015 (0.018) (0.013) (0.123) (0.016) (0.124) (0.016) (0.014) Household Head Age 0.002∗∗∗ -0.000 0.003 -0.002∗∗∗ 0.011 0.003∗∗∗ 0.001∗ (0.001) (0.000) (0.005) (0.001) (0.007) (0.001) (0.001) ln Household Size 0.026 0.019 0.162 -0.016 0.119 -0.101∗∗∗ -0.036 (0.030) (0.016) (0.127) (0.022) (0.197) (0.025) (0.026) Child-adult Ratio 0.062∗∗ 0.016 -0.184∗∗ 0.012 -0.061 0.046∗∗∗ 0.000 (0.024) (0.015) (0.076) (0.016) (0.108) (0.015) (0.014) Urban (0/1) -0.023 -0.014 -0.053 -0.005 0.032 -0.012 -0.001 (0.018) (0.012) (0.100) (0.017) (0.141) (0.015) (0.013) Dirt Floor (0/1) 0.077∗∗∗ 0.016 -0.126 -0.053∗∗∗ -0.256 0.052∗∗ 0.045∗∗ (0.020) (0.017) (0.120) (0.019) (0.186) (0.020) (0.021) Network (0/1) 0.024 0.022 0.188 0.031 0.172 0.173∗∗∗ 0.123∗∗∗ (0.024) (0.021) (0.186) (0.027) (0.123) (0.026) (0.022) County, Wave, Strata FE Observations 24,711 24,722 2,043 24,722 3,683 24,689 24,713 R2 0.074 0.040 0.319 0.041 0.178 0.094 0.087 This table shows OLS regression estimates using the first five waves from the Rapid Response Phone Survey data in Kenya. All regressions include the set of controls described in Table 1. Robust standard errors clustered within counties are reported in parentheses. *** p<0.01, ** p<0.05, * p<0.1. 25