Policy Research Working Paper 10395 Candle in the Wind? Insights from COVID-19 Emergency Cash Transfers to Informal Sector Workers in Sierra Leone Samik Adhikari Suneha Seetahul Social Protection and Jobs Global Practice April 2023 Policy Research Working Paper 10395 Abstract This paper takes stock of the insights and learnings from relationship between a one-off US$135 cash transfer and a COVID-19 emergency cash transfer program that was various labor market, food security, human capital, and sub- administered to vulnerable informal sector workers in jective well-being outcomes for recipient and nonrecipient Sierra Leone. It starts by reviewing relevant examples of households of the emergency cash transfer. The analysis cash transfer programs that were instituted in response to finds a positive potential impact of the transfer and the the COVID-19 crisis. It then describes the context, inter- number of hours worked as well as employment in the vention, and data of the emergency cash transfer program, medium term. It also finds that program beneficiaries report before presenting a quasi-experimental analysis of the emer- higher chances of their main income increasing or staying gency cash transfer’s potential impacts on various measures the same compared to nonbeneficiaries. The positive cor- of economic security and subjective well-being of house- relation between the transfer and income disappears over holds with urban informal sector workers. The analysis is the medium term, perhaps suggesting that one-off transfers conducted by matching administrative data to survey data work best to cushion vulnerable self-employed households and using program eligibility criteria and inverse proba- and informal wage workers in the short term but do not bility weights to identify the short- and medium-term impact medium-term employment or income. This paper is a product of the Social Protection and Jobs 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 sadhikari2@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 Candle in the Wind? Insights from COVID-19 Emergency Cash Transfers to Informal Sector Workers in Sierra Leone1 Samik Adhikari Suneha Seetahul Keywords: Unconditional Cash Transfer, Covid-19, Sierra Leone JEL codes: I38 1 The authors would like to thank the National Commission for Social Action (NaCSA), Statistics Sierra Leone (Stats SL), and other stakeholders involved in implementing the ECT Program. We would also like to thank Elizabeth Mary Foster for timely collaboration on collecting data for ECT Program Beneficiaries through the CIMS high-frequency phone surveys and Wendy Cunningham and Yuko Okamura for their valuable comments. We are indebted to Shreya Chatterjee for her role in conducting the preparatory data analysis for the design of the ECT and in preliminary data analysis for this paper. Finally, we would like to thank the World Bank Sierra Leone Social Protection and Jobs team: Junko Onishi, Abu Kargbo, Steisianasari Milieva, Judith Sanford, Sumati Rajput, Thomas Vaughn Bowen, Mpumelelo Nxumalo and Hannah Buya Kamara for their input and feedback on various drafts of this paper. Data files available upon request. 1. Introduction Because of their well-documented, positive short-term effect on welfare (see, for instance, Bastagli et al., 2016), cash transfers have constituted the majority of safety net interventions implemented to mitigate the effects of COVID-19 on households. Since the start of the pandemic, cash transfers have represented 42 percent of total social assistance programs and 24 percent of all global social protection measures to respond to COVID-19 (Gentilini et al. 2021).2 The initial design of many cash transfer programs reflected an objective to shield vulnerable households and individuals from the negative impacts of the unfolding COVID-19 pandemic. However, and increasingly, cash transfer measures are also being used to support workers in low- and middle-income countries to offset losses of incomes and revenues induced by COVID-19 (De La Flor et al., 2021; Gentilini et al., 2021). This paper takes stock of the Emergency Cash Transfer (ECT) provided to informal workers in urban Sierra Leone and implemented shortly after the onset of the pandemic, from June to August 2020. Our aim is to contribute to the evolving literature on the relationship between cash transfers and labor market outcomes for informal sector workers. While most studies that assess the impact of cash transfers look at outcomes related to consumption and subjective well-being, the targeting parameters and transfer size of the ECT allow us to also look at the relationship between the transfers and labor market outcomes of self-employed households in the informal sector, thus adding to the emerging strand of literature that explores the linkages between social assistance and employment-related outcomes in low-income settings (Bassier et al., 2021; Bhorat et al., 2021; Gerard et al., 2020). 2 Data until May 14, 2021. 2 The informal sector accounts for around a third of the GDP and 70 percent of all employment in emerging markets and developing economies (Ohnsorge & Yu, 2022). There are several reasons why workers, firms, and households operating in the informal economy have been impacted more severely by the COVID-19 crisis. First, informal sector workers are largely excluded from work-related social protection measures, which makes it costlier for them to stop working to adhere to social distancing and other restriction measures (Bassier et al., 2021; Gerard et al., 2020). Second, emerging evidence suggests that workers most affected by the COVID-19 crisis are not necessarily at the bottom of the income distribution and therefore have no recourse to safety nets which target the poorest and have limited coverage in low-income settings.3 Third, informal sector workers are less likely to be able to work from home (Garrote Sanchez et al., 2021). The evidence on the impact of social assistance measures implemented by governments to face the COVID-19 crisis, namely the ones directed at informal workers, remains to be assessed. Two general lessons can however be observed in the emerging literature on the impact of social assistance. First, the implemented social assistance schemes had important positive impacts. In low- and middle-income countries, 80 percent of people – who would have fallen below the US$1.90 poverty line – did not because of social assistance measures (Fajardo-Gonzalez et al., 2021). The expansion of existing social assistance schemes introduced by the governments in the form of cash transfers had a large offsetting effect on poverty in Brazil and Argentina where the social assistance programs were more ambitious compared to countries like Colombia where smaller effects were observed (Lustig et al., 2020). In India, an expansion of pre-existing social assistance schemes, which consisted in providing cash transfers and 3 For example, Lustig et al. (2020) find that in Argentina, Brazil, Colombia and Mexico, the COVID-19 crisis mostly affected those in the middle of the income distribution. 3 in-kind support to farmers, alleviated credit constraints, increased agricultural investments, and facilitated the procurement of seeds, fertilizers, and pesticides (Varshney et al., 2021). In Ethiopia, households participating in the Productive Social Safety Net Program experienced an increase in food insecurity of only 2.4 percentage points compared to 11.7 percentage points for other households. While previous research has shown that transfers are an effective means to stop individuals from falling into poverty traps (Balboni et al., 2021), these findings provide evidence that social assistance measures have been successful in shielding households from the economic fallout caused by the COVID-19 pandemic. Second, because the COVID-19 crisis has impacted households and workers across the income distribution and not only the poorest, the social assistance measures implemented can be considered more as mitigation tools rather than purely a poverty alleviation mechanism (Bhorat et al., 2021). This paper uses the COVID-19 Impact Monitoring Survey (CIMS, 2020) matched to administrative data to analyze a range of outcomes. The analysis is motivated by the question: does an ECT mitigate the microeconomic impact of COVID-19 on poor households that rely on informal self-employment for subsistence? The existing data allows us to draw short-term and medium-term correlations between the ECT and various objective economic security outcomes pertaining to labor market participation, food consumption and children’s human capital and subjective indicators of psychological wellbeing and satisfaction with the government’s response to the pandemic. Round 1 of data collection overlaps with the rollout of the program, allowing to analyze a cohort of beneficiaries as early as June-August 2020. The Round 2 of data collection (November-December 2020) allows to identify a second cohort of late beneficiaries, thus providing information on their outcomes, but also on the outcomes of the first cohort of beneficiaries. We identify correlations between receiving the ECT and the various outcomes for Round 4 1 and Round 2 cohorts by identifying beneficiary and comparable non-beneficiary households and by applying inverse probability weights which allow a more robust comparison between the two groups. The results show a positive potential impact of the ECT on labor market participation and income, child nutrition, psychological wellbeing and government satisfaction. However, these results also suggest that timing matters in terms of how long the positive effect of this type of one-off transfer can last and when the transfer is received. The positive relationship between receiving the ECT and income, for instance, is visible in the short-term for the recipients who received the transfer only a few months after the pandemic. These beneficiaries no longer benefit from the positive impact of the ECT 3 to 7 months after the reception date. Moreover, the positive correlation between receiving the ECT and psychological wellbeing also disappeared in the medium-term. These results suggest that for income stability, receiving the ECT early in the pandemic as opposed to a few months later might have proved beneficial, allowing households to maintain labor market activity and income in comparison to control households. While for other factors such as the ability to buy staple foods, children’s nutrition, decreasing worries about children being out of school and government satisfaction, the opposite is true. We provide a discussion on these results and directions for further work. 2. Description of the Context and ECT Intervention The Government of Sierra Leone declared a public emergency on March 24, 2020, and implemented a series of restrictions including lockdowns and limited inter-district movement. By May 15, 2020, the unfolding COVID-19-related restrictions had immediate impacts on private sector workers with 68 percent of businesses reporting a decrease in weekly income of on average half of their pre-pandemic income (Meriggi et al., 2020). This decline was accompanied by disruptions in employment, with a 5 reduction of 2.5 working hours on average per day for 37 percent of business owners. The main reasons reported by business owners for these economic difficulties are the COVID-19 restrictions leading to a drop in demand and a lack of access to suppliers (Meriggi et al., 2020). In a context where more than 40 percent of the population lived in extreme poverty before the pandemic,4 the lockdowns and restrictions threatened to further derail the economy and push more Sierra Leoneans into poverty. To combat the adverse economic impact of the COVID-19 induced lockdowns, the government launched the Quick Action Economic Recovery Program which aimed to maintain macroeconomic and financial stability and mitigate the effect of the pandemic on households and businesses (Showers and Ganson, 2020). The government had prior experience in the provision of cash transfers in responding to natural disasters and health shocks. The Government of Sierra Leone’s National Commission for Social Action (NaCSA) had implemented the Social Safety Net Cash Transfers, locally known as Ep Fet Po, since 2014. This program had already been used as a vehicle to scale up social protection coverage following the Ebola disease outbreak in 2015 and during the flood and mudslide catastrophe in 2017. Building on these previous experiences, and with the support of the World Bank, NaCSA provided a one-off COVID-19 Emergency Cash Transfer in the four provincial capitals of Sierra Leone and in Freetown in June 2020 to support 29,000 vulnerable informal sector households (Sanford, 2022). The amount of the transfer was Le1,309,000 (approximately US$135), which is equivalent to approximately two months of minimum wage in Freetown. This amount is also equivalent to one month of consumption expenditure of the bottom welfare quantile of Freetown households. This cash transfer was unconditional, and the benefit was provided in a single installment to prevent opportunities of 4 In 2018, 3.3 million individuals in Sierra Leone (43 percent of the population) were under the international poverty line of $1.90 per day. Poverty is expected to rise up to 44% following the COVID-19 crisis. 6 COVID-19 transmission. Moreover, households owning a mobile phone were provided the transfer through e-cheques and households without a mobile phone were provided a paper copy of the e-cheque. The aim of the ECT was to target informal workers and the number of beneficiaries per city was determined by a quota system based on the estimated size of the informal sector in each city. The identification of poverty for informal workers is complicated. In comparison to developed countries where cash transfers are targeted at low-income individuals, the income of a large share of developing country workers is not observable because they work in the informal economy (Hanna & Olken, 2018). Household targeting was therefore implemented in two steps. The first one was a community-based targeting. City councils assisted NaCSA in compiling prelists of potential beneficiary households with the help of market leaders and traders’ associations. A Light Proxy Means Test (LPMT) was then used to identify those needing support the most in the city-specific lists. The LPMT comprised a set of household characteristics, assets, work status and disability status (see Appendix 1). The rationale of a Proxy Means Test is to use observable characteristics to determine a predicted income or consumption level (Hanna & Olken, 2018) and households or individuals who are deemed eligible by the LPMT are provided the transfer. Figure 1 presents the timeline of ECT implementation and data collection. The transfers were first implemented in four regional headquarter towns of Sierra Leone, namely Bo, Kenema, Makeni, and Port Loko before the bulk of the transfers were provided to informal sector workers in Freetown (see Table 1). Following complaints on the transparency of beneficiary selection in Freetown, the government created a pre-listing review committee composed of public and private sector representatives to review community-based targeting conducted by local councils in association with market traders. Most of the 7 transfers in Freetown were therefore delayed by 2 months, with the last beneficiaries receiving the transfers in late August 2020. Figure 1: Timeline of ECT implementation and data collection Source: Authors’ calculations from CIMS (2020). Table 1: Numbers of beneficiaries enrolled and paid by city5 Quota Pre-listing LPMT LPMT Enrolled Enrolled Paid and applied passed6 (NaCSA) (RCB)7 reconciled Bo 2,500 3,010 2,698 2,477 2,430 2,417 2,383 Freetown 19,000 n/a 21,909 21,464 20,631 18,769 18,649 Kenema 2,500 3,185 2,692 2,538 2,465 2,527 2,501 Makeni 2,500 2,997 2,733 2,534 2,350 2,476 2,361 Port Loko 2,500 2,700 2,417 2,364 2,336 2,350 2,324 TOTAL 29,000 n/a 32,499 31,377 30,212 28,539 28,218 Source: Sanford (2023) 3. Data and Descriptive Statistics This study uses the Sierra Leone COVID-19 Impact Monitoring Survey (CIMS, 2020). This phone survey panel data contains two waves and follows 5,685 households (with one respondent per household) who are interviewed in both rounds. The first wave of data (Round 1) was collected between June and August 5 Project administrative data. 6 Households with an LPMT score ≥7.0. 7 Targeting, enrollment by NACSA and enrollment by RCB were three separate exercises using different and independent tools. As a consequence, a slight attrition in the numbers of households prelisted, targeted and enrolled can be observed. This is both the result of checks performed (for example to eliminate duplicates) and because of difficulties tracing households. 8 2020 and the second wave of data (Round 2) was collected between November and December 2020. The CIMS database sampled households through three different sample frames: the Sierra Leone Integrated Household Survey (SLIHS, 2018), a nation-wide random sampling and a random sample of beneficiaries from the ECT. The intent of including the beneficiaries of the ECT in the sampling frame was to allow an analysis of how the ECT recipients fared throughout the pandemic. Using this feature of the CIMS survey, as a secondary step, we matched the CIMS survey data to administrative data with information on the 32,499 individuals enrolled in the ECT. This step allowed to retrieve the LPMT scores of beneficiaries of the ECT, which are not available in the CIMS data. This paper analyzes two types of socioeconomic outcomes that are likely to be affected by the pandemic and its induced policy responses: (1). Objective economic security outcomes which pertain to the labor market, food security and children’s human capital and include: individual employment and hours worked, household income change, inability to buy staple foods (rice, dried fish and palm oil), ability to support children’s return to school, child identified as malnour ished and child given vitamin A supplementation; (2). Subjective indicators of psychological wellbeing and satisfaction with government measures (concern about children being out of school, food shortages, price increases, being sick with COVID-19, quarantine, lack of other health care and satisfaction with government response to COVID- 19). Although the CIMS panel data has 7,369 respondents in Round 1 and 5,685 respondents in Round 2, our analysis only includes a small share of this data for the following reasons. First, given our aim to analyze the ECT transfer, we restrict the analysis to the districts in which the transfer was provided. Second, 9 because of a high occurrence of missing data and the data collection design,8 our analysis is restricted to outcomes for which there are sufficient observations in terms of the ECT treatment. The structure of the survey and the issue of missing data also lead to highly unequal sample sizes across the different outcomes.9 Therefore, our active sample ranges from 187 to 913 observations depending on the outcome of interest and the specification (i.e., cross-sectional or pooled cross-sectional estimation). Moreover, some of the child human capital outcomes were exclusively collected in Round 1 whereas others in Round 2, limiting the scope of the analysis to only one of the two rounds. These characteristics of the CIMS data limit the scope of the analysis presented in this paper. A discussion on the data and future efforts for data collection that could make the analysis more robust are discussed in Section 6. The timing of the implementation of the ECT and of the CIMS data collection allow us to analyze two cohorts of beneficiaries: households that received the ECT before the first round of CIMS (hereafter referred to as ECT-R1 beneficiaries) and households that received the ECT after the first round of CIMS (hereafter referred to as ECT-R2 beneficiaries). The timing of the survey and the data collection (see Figure 1) allows us to analyze short-term ([0;3] months lapse between receiving the ECT and the survey) and medium-term ([3;7] months lapse) impacts of the ECT for ECT-R1 beneficiaries. We consider the impact of the ECT on ECT-R2 beneficiaries as a medium-term impact ([3;4] months lapse). Table 2 shows that the ECT was mostly used for food (76 and 80 percent respectively in Rounds 1 and 2) and investing in an existing business (73 and 86 percent respectively in Rounds 1 and 2). The other 8 Part of the questionnaire was administered to all respondents (i.e., Basic information, information about school-aged children and income), but only one of the four following sets of questions was asked to each respondent: “Education, Child Welfare”; “Employment, KAB, Lockdown”; “Health, Access”; “Agriculture, Food Security, Social Protection”. 9 As a robustness check, we compare the characteristics of the full sample to a sample restricted to observations for whom we have information on labor market participation (i.e. the sample which corresponds to a large set of the estimations presented in this paper). The results presented in Supplementary materials 1 show no significant differences between the two samples. 10 important items of expenditure of the ECT are medical care (32 and 24 percent respectively) and non- food consumption (14 and 31 percent respectively). The fact that the main use of the ECT transfer was for food consumption is in line with the fact that the COVID-19 pandemic affected households’ food insecurity (Londoño-Vélez & Querubin, 2020), especially in Sub-Saharan Africa (Aborode et al., 2021). Food consumption and investment in businesses being the two main items of expenditure of the ECT also points out the porosity of the household and labor spheres for informal workers who use this type of transfer to respond to immediate consumption needs and to alleviate credit constraints. Moreover, the increase in non-food consumption and investment in an existing business between the two rounds may reflect the fact that by the time of the second round of data collection, some of the government- mandated COVID-19 restrictions had started to be lifted, allowing individuals to earn labor income and use the transfer for a larger set of expenses. Table 2: Use of ECT-R1 and ECT-R2 Transfer used for Share of ECT- N (ECT-R1) Share of ECT- N (ECT-R) Two-sample R1 R2 t-test P-value beneficiaries beneficiaries Food 75.56% 311 80.20% 197 0.224 (0.024) (0.028) Invest in existing business 72.99% 311 86.29% 197 0.000 (0.025) (0.025) Medical care 31.51% 311 24.37% 197 0.083 (0.026) (0.031) Non-food consumable items 14.47% 311 31.47% 197 0.000 (0.020) (0.033) Paid debt 11.25% 311 3.05% 197 0.000 (0.018) (0.030) Start new business 9.97% 311 3.55% 197 0.007 (0.017) (0.036) House rent 8.04% 311 4.57% 197 0.128 (0.015) (0.015) Home improvement 2.89% 311 4.57% 197 0.321 (0.009) (0.015) Source: Authors’ calculations from CIMS (2020). Note: Standard errors in parentheses. 11 4. Methodology 4.1. Defining a Control Group In the absence of experimental data which would allow to identify the impact of the ECT, we implement a quasi-experimental approach which consists in defining a control group using the LPMT targeting tool and an Inverse Probability Weight (IPW) estimator. This method allows us to identify a potential impact of the ECT. The treated households are identified as those matched to the administrative data who declare that they have received a transfer from NaCSA and are residing in one of the five districts in which the program was implemented (Bo, Bombali, Kenema, Port Loko and Western Area Urban). These restrictions eliminate the possibility of including households before they received the ECT (i.e., a household listed as a beneficiary in the administrative data that declares it hasn’t received a cash transfer in the CIMS) and capturing the effect of other cash transfers. The sample restriction to the districts of interest excludes households that migrated outside of these cities or were not captured in Round 2 because of attrition. Moreover, as described in section 2, for a household to be considered as a beneficiary, they must be deemed eligible by the LPMT. The main challenge in the implementation of the quasi-experimental method is to identify a relevant counterfactual for the treated households. We exploit the fact that beneficiaries of the ECT were chosen based on LPMT but that the scope of the ECT program was not sufficient to cover all informal sector households in need in the five districts in which the program was implemented. We therefore identify the control group as households that passed the LPMT in the CIMS data but did not receive the transfer 12 as they were not in the initial pool of pre-listed beneficiaries to receive the transfer. We assume that these households are comparable to the treated households. The Proxy Means Test has been used in previous studies as a criteria to identify treated and control groups of a cash transfer (for example see Stoeffler et al. (2020)). The challenge with this methodology of identification is that we do not have the true LPMT score of the control households. Using the CIMS data, it is possible to recreate a proxy LPMT score, but only partly. Indeed, as described in Appendix 1, the LPMT score is calculated as a sum of 10 different characteristics which include household characteristics, assets, work status and disability status. Out of the 10 characteristics, 3 are unavailable in the CIMS data, two of which we are able to proxy. We do not have any information on whether the household uses bottled or sachet water for drinking. We also do not have the information on (1) whether the head of the household is literate in English or Sierra Leonian language and (2) whether the head of the household has an informal contract. For these two characteristics, we use information on the education level and work with a formal contract of the respondent (as opposed to the household head) as proxies. Among the Round 1 beneficiaries, 32 percent are non-household heads and among the R2 beneficiaries, 45 percent are non-household-heads. We are therefore able to recreate an LPMT score that can take a total of 9 instead of 10, but with potential errors because of the two proxies. The original LPMT was calculated using a simplified tool collected by program administrators.10 The LPMT threshold used by NaCSA to determine the eligibility of a household for the ECT was 7 out of 10. As described in Figure 2, households that scored 7 or above have necessarily passed the original LPMT because these households have a true score of 7 to 10. However, there are potentially households that 10 The data for the LPMT tool was collected by Statistics Sierra Leone (Stats SL) in presence of NaCSA and the Anti-Corruption Commission (ACC). Independent spot checks of the ECT confirm that most participants of the exercise thought the process was fair and transparent (Sanford 2023). 13 have a real score of 7 which only have a proxy score of 4 to 6 with our scoring mechanism. To minimize exclusion errors, we set the threshold to LPMT>=4. We assume that households that declared receiving a transfer from NaCSA and score 4 or above in the proxy threshold are likely beneficiaries, as they have little incentive to report receiving a transfer if they are not beneficiaries. The likelihood of the inclusion error in the control group is therefore low. We exclude all households below the threshold of 4 from the analysis because if our proxies are correct, no households should be under the threshold of 4. It is likely that our proxies do not perfectly overlap with the original criteria, but the chances of excluding treated individuals is relatively low and restricting to LPMT>=4 ensures better comparability with control households. Figure 2: Proxy LPMT score calculation and potential errors Proxy threshold Original threshold 0 1 2 3 4 5 6 7 8 9 10 potential inclusion error Source: Authors 14 Figure 3: Comparison of Proxy LPMT and True LPMT for treated households Source: Author’s calculations from NaCSA administrative Data and CIMS (2020) The comparison of true LPMT scores and the proxy LPMT for treated households provides a further justification of the relevance of the chosen threshold of 4. In Figure 3, we match administrative data from NaCSA to the CIMS (2020) data to compare the distribution of the true LPMT and proxy LPMT for treated households. The figure shows that for households’ whose true LPMT are between 7 and 10, we can see that a large majority have a proxy LPMT equal to 4 or more. We argue that this threshold allows us to strike the balance of minimizing exclusion and inclusion errors. 4.2. Identification Method Since we do not have experimental data and although we implement a cautious identification of the control group, there remain significant differences between the treatment and control groups (see Appendices 3 and 4). For instance, 60.7 percent of ECT-R1 beneficiaries were working the week prior to the Round 1 of survey in comparison to only 50.6 percent of the control group. For this reason, we use an Inverse Probability Weighting method to rebalance treated and control observations along a set of 15 observable characteristics. The probability of treatment (propensity score) is calculated using a logistic model. These scores are then used to compute weights allowing to control for covariate imbalance prior to the treatment. This step combined with a regression analysis ensures a more robust identification of the impact of the ECT on the various outcomes and, as recommended by Imbens & Wooldridge (2009), this combination allows a better identification of treatment effects than matching estimators. We identify 4 types of results: • Short-term relationship between ECT-R1 and the outcome. Short-term refers to a lapse of 1 to 3 months between the reception of the ECT and the survey. • Medium-term relationship betweenECT-R1 and the outcome. Medium-term refers to a lapse of 3 to 7 months between the reception of the ECT and the survey. • Medium-term relationship between ECT-R2 and the outcome. Medium-term refers to a lapse of 3 to 4 months between the reception of the ECT and the survey. • Pooled relationship between both ECT-R1 and ECT-R2 and the outcome. This overall effect is detected after a 1 to 4 months lapse between the reception of the ECT and the survey. The IPW estimator allows to identify Average Treatment Effects as defined in Equation 1: 1 (1− ) ,=1,2 = ∑ =1 { ̂ − ̂( ) } [Equation 1] 1− For each and period t (t=1 for Round 1 of CIMS and t=2 for Round 2), we estimate an Average ̂( ) is the Treatment Effect ,=1,2 of (t=1 for Round 1 of CIMS and t=2 for Round 2). estimated propensity score conditional on a set of covariates (gender, age, household dependency ratio, head of household, completion of secondary school, main income source, asset score and location). 16 measures the different outcomes k for each household i in t=1 and t=2. identifies the treated households in each round t=1 and t=2. This method allows to estimate the Average Treatment Effect of the ECT at the two periods 1 and 2. In the pooled estimation, observations from t=1 and t=2 are pooled together. Despite our efforts to implement a robust identification method by selecting an adequate control group and using the IPW estimator, we cannot verify that there are no unobservable factors affecting the treated and control households differently given the non-experimental nature of the data. We implement balance and overidentification tests to test for comparability of the treated and control groups after reweighting. Although the tests are encouraging, as discussed in Section 6, we interpret the results as correlations (i.e. potential impact) instead of causality (i.e. impact) as a precautionary measure. 5. Results Figure 4 shows the distribution of control and beneficiary (treatment) households for the two rounds of the ECT. In total, the treatment groups contain 311 households that have received the transfer before the Round 1 of CIMS and 197 households that have received the transfer after the Round 1 of CIMS. 17 Figure 4: Treated and control groups Source: Authors’ calculations from CIMS (2020). Note: These shares represent the total number of beneficiaries and control households in the observation. Note that none of the estimations presented in the paper include all treated and control households in either rounds because of the structure of the survey and missing observations. The number of treated and control households for each outcome is presented in 2 to 4. Among households included in the analysis, 52 percent of respondents worked in Round 1 and 60 percent in Round 2 (see Appendix 3). These statistics suggest that although the share of inactivity/unemployment is high in the sample of analysis, there are signs of employment stability following the onset of the COVID-19 crisis. The number of hours worked suggests that time-related under-employment was not a salient issue following COVID-19. Indeed, excluding those who worked zero hours, eligible households in Round 1 and Round 2 worked respectively 40 and 42 hours weekly. However, further explorations of the distribution of hours show a bimodal distribution in both Rounds 1 and 2 (see Appendix 5). Fourteen percent of the eligible households in Round 1 and 11 percent in Round 2 worked under 10 hours in the week prior to the survey. 18 Table 3 shows results from the IPW estimations of the correlations between the ECT and objective economic security outcomes.11 The results show that in the short-term and medium-term, receiving the ECT is positively correlated to employment outcomes both at the extensive and intensive margin. At the extensive margin, there is no significant short-term correlation for ECT-R1 beneficiaries, but a significant medium-term positive coefficient of 13% is visible for ECT-R1 beneficiaries and of 18% for ECT-R2 beneficiaries. The pooled specification (Round 1 and 2 beneficiaries) which includes a larger set of observations also shows a significant positive correlation (12%) between receiving the ECT and the probability of working in the week preceding the survey. At the intensive margin, the ECT had a significant positive relationship for ECT-R1 beneficiaries with approximately 6 hours of additional weekly working hours in the short-term and 9 hours in the medium-term. For R2 beneficiaries, the relationship is also significant with 15 additional working hours. The pooled specification shows a significant coefficient of 10 hours, showing that all ECT beneficiaries worked more hours than their non- beneficiary counterparts. The results in Table 3 also show that ECT-R1 beneficiaries had a significantly higher likelihood of being in a household in which income increased or remained stable and a lower chance of being in a household in which income decreased or stopped in the short-term. In the medium-term, no significant relationship is found for the main income and the reverse relationship is visible for ECT-R1 and ECT-R2 beneficiaries for other sources of income. These results suggest that the ECT allowed households to maintain a stable income at the onset of the pandemic with a visible effect for Round 1 beneficiaries, but that this effect did not last. In the medium-term, although there is a larger potential effect on labor market participation, 11 Full table available upon request. 19 the potential effect on main income is no longer visible. Moreover, the negative coefficient of income stability in the medium-term may indicate a relative deprivation in comparison to the months in which they used the ECT and considered it as an “Other source of income”. Given the fact that the ECT was a one-time transfer and that its amount was equivalent to two months minimum wage in Freetown, it is not surprising to find a negative effect on income in the medium-term specifications which represent households’ income 3 to 7 months post-transfer. Treated households, whose other source of income stopped altogether was 22.5 percent in Round 1 and 32.9 percent in Round 2. Among these households, it is possible that some refer to the ECT transfer amount running out and others refer to another source of income stopping or decreasing. Concerning food consumption, Table 3 shows that the ECT did not have a significant association with the reduction of inability of buying staple foods (rice, dried fish and palm oil) for Round 1 beneficiaries but had a significant relationship for Round 2 beneficiaries. Indeed, Round 2 beneficiaries were 7 percent less likely to be unable to buy staple foods compared to non-beneficiaries. We argue that there are two possible explanations for this result. First, it is possible that in the early months following the pandemic, households may have had stocks of staple foods, which is why ECT-R1 beneficiaries did not need to use the ECT for this type of consumption as opposed to ECT-R2 beneficiaries. Second, 76 percent of treated households declare that they used the ECT to buy food (Table 2), however these types of expenses represent basic needs food consumption which have been found to be relatively more income inelastic, especially in contexts of poverty and urban areas (Melo et al., 2015). It is therefore likely that control households managed to acquire similar quantities of staple foods as treated ones by prioritizing these expenses even if they did not receive cash assistance. Moreover, the data on how the ECT was used does not allow to calculate the share of the transfer used for various purposes. Even if 76 percent of ECT-R1 20 beneficiaries used the ECT to buy staple foods, the estimation result suggests that this share might have been low. Concerning child human capital outcomes, Table 3 shows that the ECT does not have a significant relationship with the confidence of households in their ability to financially support children’s return to school for ECT-R1 beneficiaries. However, ECT-R2 beneficiaries are less likely to live in a household in which a child was identified as malnourished. The results show no significant association between receiving the ECT in Round 1 and malnourishment in the medium-term. No significant relationship between receiving the ECT in Round 2 and vitamin A supplementation was found which is expected given that it is generally advised to give children Vitamin A supplementation twice per year (World Bank, 2014) and that the short-term nature of the intervention and the context of COVID-19 restrictions may not be suited for this type of health expense. 21 Table 3: Correlations between the ECT and socioeconomic outcomes Worked Hours Main Main Other Other Inability Able to Child Small children last week worked source of source of source of source of to buy financially identified given vitamin A last week income income income income staple support as mal supplementation increased decreased increased decreased foods child's nourished or stayed or stopped or stayed or stopped return to same same school 1. ECT-R1: Round 1 (short-term) outcomes ECT-R1 0.050 5.72* 0.135** -0.135** 0.192*** -0.396*** 0.000 0.025 (0.058) (2.947) (0.060) (0.060) (0.066) (0.061) (0.050) (0.053) Observations 569 569 562 562 392 392 533 512 2. ECT-R1: Round 2 (medium-term) outcomes (ECT-R2 beneficiaries excluded) ECT-R1 0.126* 8.576** 0.025 -0.025 -0.164** 0.064 0.030 - 0.024 - (0.065) (3.690) (0.056) (0.056) (0.077) (0.071) (0.039) 0.071 Observations 702 702 682 682 503 503 640 208 3. ECT-R2: Round 2 (medium-term) outcomes (ECT-R1 beneficiaries excluded) ECT-R2 0.180*** 15.340*** -0.145 0.145 -0.265*** 0.290*** -0.068*** - -0.059* -0.063 (0.060) (4.461) (0.094) (0.094) (0.077) (0.092) (0.021) (0.034) 0.066 Observations 459 459 442 442 328 328 433 187 305 4. Pooled outcomes of ECT-R1 and ECT-R2 ECT-R1 and 0.121** 10.448*** 0.173*** -0.173*** 0.156** -0.294*** -0.031 ECT-R2 (0.056) (3.270) (0.062) (0.062) (0.061) (0.070) (0.034) Observations 908 908 893 893 627 627 810 Source: Authors’ calculations from CIMS (2020) Note: Standard errors in parentheses: *** p<0.01, ** p<0.05, * p<0.1. Inverse probability weighted regressions. For each estimation, overidentification tests shows that the IPW model balances the covariates. Full set of results available in Online supplementary materials S2 to S6. 22 Table 4 shows results from the IPW estimations of psychological wellbeing and government measures satisfaction outcomes. The results indicate positive relationships for ECT-R1 beneficiaries in the short- term, who experienced significantly lower levels of concerns about food shortages, being sick with COVID-19, quarantine, and the lack of health care (besides COVID-19). These short-term relationships do not however last in the medium-term for R1 beneficiaries. ECT-R2 beneficiaries report experiencing lower levels of concerns concerning children being out of school and food shortages, however they are significantly more concerned about price increases, which may indicate a sense of relative deprivation. Receiving the ECT is also significantly correlated with the satisfaction with the government response in all estimations except for the medium term ECT-R1 beneficiaries. Table 4: Correlations between the ECT and subjective well-being measures, satisfaction with government response Not Not Not Not Not Not Satisfaction Concerned Concerned Concerned Concerned Concerned Concerned with about about food about price about about about lack government children shortages increases being sick quarantine of other response out of with healthcare school COVID-19 1. ECT-R1: Round 1 (short-term) outcomes ECT-1 0.146 0.466*** 0.062 0.414** 0.387*** 0.222* 0.041*** 0.096 0.127 0.045 0.114 0.118 0.120 0.009 Observations 568 569 569 569 567 567 555 2. ECT-R1: Round 2 (medium-term) outcomes (ECT-R2 beneficiaries excluded) ECT-1 -0.305** -0.064 -0.015 -0.028 0.217 0.122 0.017 0.055 0.108 0.040 0.195 0.184 0.199 0.015 Observations 648 649 649 649 649 648 647 3. ECT-R2: Round 2 (medium-term) outcomes (ECT-R1 beneficiaries excluded) ECT-2 0.594** 0.567** -0.040** -0.120 -0.045 0.177 0.035*** 0.244 0.231 0.020 0.140 0.140 0.163 0.009 Observations 438 438 438 438 438 437 435 4. Pooled outcomes of ECT-R1 and ECT-R2 ECT-1 and ECT-2 0.319*** 0.473*** 0.012 0.389*** 0.389*** 0.221** 0.046*** 0.114 0.123 0.031 0.098 0.100 0.098 0.008 Observations 911 913 913 913 911 910 892 Source: Authors’ calculations from CIMS (2020) Note: Standard errors in parentheses: *** p<0.01, ** p<0.05, * p<0.1. Inverse probability weighted regressions. For each estimation, overidentification tests shows that the IPW model balances the covariates. Psychological wellbeing outcomes are originally in a 5-point Likert scale format and are considered as continuous in the IPW estimation. The coefficients should 23 therefore be interpreted based on a score of 1 (largest level of concern) to 5 (lowest level of concern). Full set of results available upon request. 6. Discussion and Next Steps Cash transfers have been established as a consumption support measure in traditional social protection programming. The COVID-19 pandemic has shifted that paradigm to a certain degree by introducing cash-based measures to support vulnerable workers in the informal sector on a one-off or punctual basis. This paper assessed the potential impact of one-such measure on subsequent labor market, economic security, and subjective psychological well-being for vulnerable self-employed households in Sierra Leone. While the ECT was found to have positive association with various measures of well-being in the short run, some of the positive relationship dissipated in the medium run. In this section, we discuss four areas that have implications on the analysis conducted in this paper and suggest areas for future research. Because of lack of experimental data, we use a quasi-experimental approach to create a control group for our analysis. This approach uses two main steps to allow the identification of a potential impact of the ECT on households. First, we construct a proxy LPMT and uses administrative data to choose the threshold that will best allow to identify a relevant control group. The 10 LPMT questions used to assess eligibility for the ECT program, and the questions captured in the CIMS surveys had some notable differences. We were able to proxy for two out three of the questions that were missing from the CIMS survey. Moreover, we use administrative data to determine the eligibility cutoff that strikes the balance between minimizing both the exclusion and inclusion errors. We use the cutoff of 4 out of 10 to identify the treated and the control group and note the margin of inclusion error given this identification design. While we show that the cutoff of 4 successfully identifies program recipients from non-recipients, there 24 could still be notable differences between the two comparison groups at baseline, which we address in a second step, by using an IPW method. The IPW allows to balance the treated and control samples on a set of covariates. A balance check table is presented in Appendix 2 and shows that the inverse probability reweighting improves the covariate balance in most cases. One notable exception is the household dependency ratio which remains different and slightly higher after matching. However, overidentification tests conducted after each estimation confirm that the IPW successfully rebalanced the covariates in each estimation. We therefore consider the matching to be satisfactory. However, we only control for characteristics that are observed in the data and there could be other omitted variables driving the short- and medium-term impact that is not captured in the surveys. Furthermore, the timing of the ECT implementation and the two rounds of CIMS surveys create both opportunities and challenges for our analysis. While it allows us to estimate impact on beneficiaries who received the ECT before and after the first round of data collection, the long-time lag between the end of the ECT implementation (August 2020) and the start of Round 2 of the CIMS survey (November 2020) presents difficulties in interpreting the results for beneficiaries who received the transfers after the conclusion of the first round. Descriptive data also shows that compared to June-August 2020, the economic recovery was well on-track by November-December 2020 and could introduce sources of bias in the findings, especially on Round 2 beneficiaries. Finally, while the CIMS data provided a timely opportunity to track the outcomes of ECT beneficiaries, the survey itself had limitations because of the way it was designed. In particular, part of the questionnaire was administered to the full set of respondents available in the sample, but other sections 25 of the survey was only administered to one-third of the sample. This significantly reduces the number of identifiable ECT beneficiaries when analyzing specific outcomes and reduces the overall sample size. Several recommendations arise from this discussion: (1) collecting experimental data by aligning the purposes of policy and research in future implementations of ECTs would allow to collect stronger evidence; (2) survey data such as the CIMS (2020) are extremely valuable in the context of missing experimental data and they should pursue efforts to minimize data issues (e.g. missing responses, small sample size); (3) qualitative data would be a complementary source of information for our study and would allow us to gain further insight into how households allocate the amount of the ECT across their various needs. 7. Conclusion This paper took stock of lessons from a COVID-19 ECT program that was administered to vulnerable informal sector workers in Sierra Leone. It discussed how cash transfers are increasingly being instituted as a policy measure to support informal sector workers in the wake of the pandemic and reviewed emerging evidence on the impact of these measures in other countries. It also described the setting and context of the ECT program and presented analysis on the relationship between receiving the ECT and various socioeconomic outcomes and measures of subjective well-being for households with urban informal sector workers in Sierra Leone. While the ECT was found to be positively associated with effectively shielding vulnerable informal sector workers from income and job losses in the short-term, the analysis was inconclusive on the medium-term outcomes related to income, suggesting the need for further experimental data. The ECT was also found to be positively associated with measures of subjective well-being, especially on households receiving the transfers being less concerned about 26 children being out of school, food shortages, being sick with COVID-19, and lack of health care. Receiving the ECT was also associated with positive views of the government response to the pandemic. The paper contributes to the literature on this new form of punctual social protection measures and calls for further research allowing to further understand how emergency cash transfers can have a lasting impact. 27 References Aborode, A. T., Ogunsola, S. O., & Adeyemo, A. O. (2021). Perspective Piece A Crisis within a Crisis : COVID-19 and Hunger in African Children. 104(1), 30–31. https://doi.org/10.4269/ajtmh.20-1213 Balboni, C., Bandiera, O., Burgess, R., Ghatak, M., & Heil, A. (2021). Why do people stay poor? https://doi.org/10.3386/w29340 Bassier, I., Budlender, J., Zizzamia, R., Leibbrandt, M., & Ranchhod, V. (2021). Locked down and locked out: Repurposing social assistance as emergency relief to informal workers. World Development, 139, 105271. https://doi.org/10.1016/j.worlddev.2020.105271 Bastagli, F., Hagen-Zanker, J., Harman, L., Barca, V., Sturge, G., Schmidt, T., & Pellerano, L. (2016). Cash transfers: what does the evidence say? A rigorous review of programme impact and of the role of design and implementation features. ODI Report, July, 1–300. www.odi.org/twitter%0Awww.odi.org/twitter%0Awww.odi.org/twitter%0Ahttps://www.odi.org/s ites/odi.org.uk/files/resource-documents/10749.pdf Bhorat, H., Oosthuizen, M., & Stanwix, B. (2021). Social Assistance Amidst the COVID‐19 Epidemic in South Africa: A Policy Assessment. South African Journal of Economics, 89(1), 63–81. https://doi.org/10.1111/saje.12277 De La Flor, L., Mujica, I., Fontenez, M. B., Newhouse, D., Rodriguez Alas, C., Sabharwal, G., & Weber, M. (2021). Taking Stock of COVID-19 Labor Policy Responses in Developing Countries. https://openknowledge.worldbank.org/handle/10986/35331 Fajardo-Gonzalez, J., Gray Molina, G., Mantoya-Aguirre, M., & Ortiz-Juarez, E. (2021). Mitigating Poverty : Global Estimates of the Impact of Income Support during the Pandemic (Development 28 Futures Series Working Papers). https://www.undp.org/publications/mitigating-poverty-global- estimates-impact-income-support-during-pandemic Garrote Sanchez, D., Gomez Parra, N., Ozden, C., Rijkers, B., Viollaz, M., & Winkler, H. (2021). Who on Earth Can Work from Home? The World Bank Research Observer, 36(1), 67–100. https://doi.org/10.1093/wbro/lkab002 Gentilini, U., Almenfi, M., Blomquist, J., Dale, P., De, L., Giuffra, F., Desai, V., Fontenez, M. B., Galicia, G., Lopez, V., Marin, G., Mujica, I. V., Newhouse, D., Palacios, R., Quiroz, A. P., Alas, C. R., Sabharwal, G., Weber, M., Acar, A., … Grun, R. (2021). Social Protection and Jobs Responses to COVID-19 : A Real-Time Review of Country Measures. https://doi.org/https://openknowledge.worldbank.org/handle/10986/33635 Gerard, F., Imbert, C., & Orkin, K. (2020). Social protection response to the COVID-19 crisis: options for developing countries. Oxford Review of Economic Policy, 36(Supplement_1), S281–S296. https://doi.org/10.1093/oxrep/graa026 Hanna, R., & Olken, B. A. (2018). Universal Basic Incomes versus Targeted Transfers: Anti-Poverty Programs in Developing Countries. Journal of Economic Perspectives, 32(4), 201–226. https://doi.org/10.1257/jep.32.4.201 Imbens, G. W., & Wooldridge, J. M. (2009). Recent Developments in the Econometrics of Program Evaluation. Journal of Economic Literature, 47(1), 5–86. https://doi.org/10.1257/jel.47.1.5 Londoño-Vélez, J., & Querubin, P. (2020). The Impact of Emergency Cash Assistance in a Pandemic : Experimental Evidence from Colombia. Lustig, N., Pabon, V. M., Sanz, F., Younger, S. D., & Younger, S. (2020). The Impact of COVID-19 Lockdowns Inequality, Poverty and Mobility. Center for Global Development, October 2020. 29 https://www.cgdev.org/publication/impact-covid-19-lockdowns-and-expanded-social-assistance- inequality-poverty-and-mobility Melo, P., Abdul-Salam, Y., Roberts, D., Gilbert, A., Matthews, R., Colen, L., Mary, S., Gomez Y Paloma, S. (2015) ; Income Elasticities of Food Demand in Africa: A Meta-Analysis; EUR 27650 EN; doi:10.2791/661366 Meriggi, N., Humphreys, M., Kamara, A. B., Krupoff, M., Levine, M., Mcleod, H., Mobarak, M., Prichard, W., Shridhar, A., Van Der Windt, P., & Voors, M. (2020). Tracking the economic consequences and responses to COVID-19 in Sierra Leone (Working Paper - IGC, Issue May). https://sl- dashboard.github.io/corona/ Ohnsorge, F., Yu, S. (2022). The Long Shadow of Informality : Challenges and Policies. World Bank, Washington, DC. https://openknowledge.worldbank.org/handle/10986/35782 Sanford, J. 2023. Expanding safety nets in response to shocks: The Experience of the Government of Sierra Leone’s Ep Fet Po Programme’s COVID-19 Response. Forthcoming. Stoeffler, Q., Mills, B., & Premand, P. (2020). Poor households’ productive investments of cash transfers: Quasi-experimental evidence from Niger. Journal of African Economies, 29(1), 63–89. https://doi.org/10.1093/jae/ejz017 Varshney, D., Kumar, A., Mishra, A. K., Rashid, S., & Joshi, P. K. (2021). India’s COVID -19 social assistance package and its impact on the agriculture sector. Agricultural Systems, 189(July 2020). https://doi.org/10.1016/j.agsy.2021.103049 World Bank (2014). Vitamin A at a glance. World Bank, Washington, DC. http://web.worldbank.org/archive/website01213/WEB/IMAGES/AAGENGVI.PDF. Accessed in March 2023. 30 31 Appendix Appendix 1: LPMT scoring mechanism and CIMS proxy LPMT scoring method Category LPMT Question LPMT Scoring Mechanism Availability in CIMS survey and Proxy variables Household What is the gender of the head of the 1 if female, 0 if male Available characteristics household? How many children below 18 years of age 1 if above 3, 0 if below 3 Available are in the household? How many members above 18 years of age 1 if above 6, 0 if below 6 Available are in the household? Can the head of the household read or write 1 if illiterate, 0 if literate Not available. Education level (i.e., in English or Sierra Leonean language? below primary dummy) of respondent is used as a proxy Household Assets How many rooms does your house have? 1 if number of rooms divided Available by household size is less than 1, 0 if number of rooms divided by household size is more than 1 Does your household have cement walls? 1 if no, 0 if yes Available Does the household use a flush to pit 1 if the answer is no, 0 if the Available latrine? answer is yes Does the household use a bottle/sachet 1 if the answer is no, 0 if the Not Available water for drinking? answer is yes Work Status Does the head of the household/main 1 if the answer is no, 0 if the Not available. We use information income-earner have a contract with his/her answer is yes) on whether respondent has a formal employer? contract as a proxy What is the occupation of the household Validation question Available head/main income-earner of the family? Illness/Disability Does anyone in your household suffer from 1 if yes, 0 if no Available illness or disability? Total possible score 10 9 Source: Authors’ calculations based on administrative data from NaCSA (2020) and CIMS(2020). Note: LPMT variables and threshold were calibrated by analyzing the overlap the LPMT score and poverty level in the SLIHS (2018) household survey data. 32 Appendix 2: Balance check table Standardized differences Variance ratio Raw Weighted Raw Weighted Female 0.111 0.037 0.964 0.986 Age 0.051 0.013 0.778 0.695 Household dependency ratio 0.100 0.112 0.728 0.715 Head of household 0.129 -0.038 0.964 1.008 Completed secondary school -0.019 0.001 0.989 1.000 Main income source: Agriculture -0.033 -0.037 0.989 1.000 Main income source: Business 0.283 0.029 0.785 0.980 Main income source: -0.314 0.016 0.544 1.024 Wage/Salary Main income source: Other -0.029 -0.055 0.841 0.697 Asset score -0.401 -0.135 0.560 0.818 Bo -0.031 -0.068 0.954 0.877 Bombali -0.176 0.027 0.681 1.052 Kenema -0.336 -0.095 0.356 0.790 Port Loko 0.227 -0.007 1.892 0.979 Western Area Urban 0.199 0.093 0.999 0.999 Source: Authors’ calculations from CIMS (2020). Note: This Balance check table includes Round 1 beneficiaries in the largest estimation sample of our ECT-R1 analysis. “Raw” columns show the values prior to the inverse-probability weighting and “Weighted” columns show the values post weighting. The standardized differences should be equal to zero and the variance ratio should be equal to 1 if the samples are perfectly balanced. 33 Appendix 3: Objective economic security outcomes: descriptive statistics and sample sizes N N Mean Mean Mean Differenc Control Treated Total Control Treated e p-value Round 1 outcomes R1 Beneficiary Households (ECT-1) Worked last week 480 89 0.522 0.506 0.607 0.082 Hours worked last week 480 89 21.098 19.907 25.528 0.013 Main source of income increased or stayed the same 475 87 0.336 0.314 0.460 0.008 Main source of income decreased or stopped 475 87 0.663 0.686 0.540 0.008 Other source of income increased or stayed same 303 89 0.349 0.284 0.573 0.000 Other source of income decreased or stopped 303 89 0.543 0.637 0.225 0.000 Inability to buy staple foods 492 55 0.117 0.116 0.128 0.803 Able to financially support child's return to school 446 66 0.863 0.857 0.909 0.246 Round 2 outcomes R1 Beneficiary (ECT-1, R2 Beneficiary households excluded from Control) Worked last week 623 78 0.603 0.588 0.731 0.015 Hours worked last week 623 78 25.471 24.416 33.898 0.003 Main source of income increased or stayed the same 606 76 0.683 0.690 0.632 0.304 Main source of income decreased or stopped 606 76 0.317 0.310 0.368 0.304 Other source of income increased or stayed same 450 67 0.603 0.625 0.463 0.011 Other source of income decreased or stopped 450 67 0.265 0.256 0.329 0.208 Inability to buy staple foods 566 76 0.081 0.077 0.105 0.41 Child identified as malnourished 214 24 0.080 0.080 0.084 0.947 Child given vitamin A supplementation 348 35 0.389 0.394 0.343 0.558 Round 2 outcomes R2 Beneficiary (ECT-2, R1 Beneficiary households excluded from Control) Worked last week 623 40 0.594 0.588 0.700 0.161 Hours worked last week 623 40 25.185 24.416 37.175 0.004 Main source of income increased or stayed the same 606 38 0.688 0.690 0.658 0.681 Main source of income decreased or stopped 606 38 0.312 0.310 0.342 0.681 Other source of income increased or stayed same 450 26 0.617 0.625 0.500 0.205 Other source of income decreased or stopped 450 26 0.265 0.256 0.423 0.06 Inability to buy staple foods 566 40 0.074 0.077 0.025 0.22 Child identified as malnourished 214 24 0.080 0.080 0.084 0.947 Child given vitamin A supplementation 348 38 0.388 0.394 0.342 0.537 Round 2 outcomes R1 and R2 Beneficiary Households (ECT-1 and ECT- 2) Worked last week 623 118 0.609 0.588 0.721 0.007 Hours worked last week 623 118 24.416 24.416 35.008 0.000 Main source of income increased or stayed the same 606 114 0.682 0.690 0.640 0.299 Main source of income decreased or stopped 606 114 0.318 0.310 0.359 0.299 Other source of income increased or stayed same 450 93 0.599 0.625 0.473 0.007 Other source of income decreased or stopped 450 93 0.273 0.256 0.355 0.051 Inability to buy staple foods 566 116 0.078 0.077 0.077 0.996 Source: Authors’ calculations from CIMS (2020). 34 Appendix 4: Subjective indicators of psychological wellbeing and satisfaction with government measures: descriptive statistics and sample sizes N N Mean Mean Mean Difference Control Treated Total Control Treated p-value Round 1 outcomes R1 Beneficiary Households (ECT-1) Concerned about children out of 479 89 1.445 1.405 1.596 0.039 school Concerned about food shortages 480 89 1.561 1.486 1.967 0.000 Concerned about price increases 480 89 1.103 1.089 1.180 0.041 Concerned about being sick with 480 89 1.528 1.438 1.978 0.000 COVID-19 Concerned about quarantine 480 89 1.606 1.519 2.045 0.000 Concerned about lack of other 478 89 1.659 1.619 1.877 0.018 healthcare Satisfaction with government 468 87 0.966 0.960 1.000 0.056 measures Round 2 outcomes R1 Beneficiary (ECT-1, R2 Beneficiary households excluded from Control) Concerned about children out of 572 78 1.432 1.451 1.321 0.206 school Concerned about food shortages 573 78 1.570 1.571 1.603 0.782 Concerned about price increases 573 78 1.065 1.068 1.052 0.663 Concerned about being sick with 573 78 1.851 1.850 1.872 0.873 COVID-19 Concerned about quarantine 573 78 1.948 1.941 2.013 0.61 Concerned about lack of other 572 78 1.759 1.752 1.821 0.572 healthcare Satisfaction with government 569 77 0.958 0.960 .948 0.636 measures Round 2 outcomes R2 Beneficiary (ECT-2, R1 Beneficiary households excluded from Control) Concerned about children out of 572 40 1.466 1.451 1.725 0.036 school Concerned about food shortages 573 40 1.589 1.571 1.900 0.73 Concerned about price increases 573 40 1.068 1.068 1.050 0.412 Concerned about being sick with 573 40 1.837 1.850 1.700 0.632 COVID-19 Concerned about quarantine 573 40 1.931 1.941 1.850 0.87 Concerned about lack of other 572 40 1.742 1.752 1.725 0.196 healthcare Satisfaction with government 569 40 0.962 0.960 1.000 0.036 measures Round 2 outcomes R1 and R2 Beneficiary Households (ECT-1 and ECT-2) Concerned about children out of 572 118 1.452 1.451 1.458 0.941 school 35 Concerned about food shortages 573 118 1.593 1.571 1.704 0.174 Concerned about price increases 573 118 1.065 1.068 1.051 0.594 Concerned about being sick with 573 118 1.844 1.850 1.814 0.748 COVID-19 Concerned about quarantine 573 118 1.944 1.958 1.958 0.885 Concerned about lack of other 572 118 1.758 1.788 1.788 0.719 healthcare Satisfaction with government 569 117 0.961 0.966 .966 0.753 measures Source: Authors’ calculations from CIMS (2020) 36 Appendix 5: Kernel density plot of the number of hours worked prior to the survey Source: Authors’ calculations from CIMS (2020) 37 Supplementary Materials 38 Supplementary materials S1. Comparison of full sample and sample with labor market information Variable Mean full sample N full sample Mean sample N sample with Two-sample t- with labor market labor market test P-value information information Female 0.486 3078 0.488 770 0.897 (0.009) (0.018) Age 36.748 3082 36.499 770 0.609 (0.219) (0.433) 1.090 3080 1.090 771 0.818 Household dependency ratio (0.033) (0.016) Head of household 0.590 3080 0.585 772 0.825 (0.008) (0.018) 0.337 3080 0.334 772 0.896 Completed secondary school (0.009) (0.017) Main source of income: Business 0.583 3080 0.588 772 0.812 (0.028) (0.018) Asset score 0.355 3080 0.357 772 0.853 (0.003) (0.006) Source: Authors’ calculations from CIMS (2020). Note: Standard errors in parentheses. 39 Supplementary materials S2. Full regression table of ECT-R1: Round 1 (short-term) outcomes Able to Main source Main source Other source Other source Inability to financially Worked last Hours worked of income of income of income of income buy staple support week last week increased or decreased or increased or decreased or foods child's return stayed same stopped stayed same stopped to school ATE of ECT-R1 0.050 5.724* 0.135** -0.135** 0.192*** -0.396*** 0.001 0.025 (0.058) (2.947) (0.060) (0.060) (0.066) (0.061) (0.050) (0.053) POM (Control) 0.502*** 19.598*** 0.307*** 0.693*** 0.284*** 0.640*** 0.118*** 0.856*** (0.023) (1.199) (0.021) (0.021) (0.026) (0.028) (0.015) (0.017) Female 0.260 0.260 0.252 0.252 0.223 0.223 0.008 0.406 (0.272) (0.272) (0.278) (0.278) (0.296) (0.296) (0.355) (0.306) Age 0.002 0.002 0.003 0.003 -0.003 -0.003 0.011 0.013 (0.010) (0.010) (0.010) (0.010) (0.011) (0.011) (0.012) (0.011) Household dependency ratio 0.035 0.035 0.050 0.050 0.223 0.223 0.431*** 0.268** (0.101) (0.101) (0.101) (0.101) (0.145) (0.145) (0.138) (0.132) Head of household 0.354 0.354 0.342 0.342 0.408 0.408 0.119 0.701** (0.294) (0.294) (0.297) (0.297) (0.321) (0.321) (0.369) (0.353) Completed secondary school 0.087 0.087 0.057 0.057 -0.102 -0.102 0.078 -0.534 (0.276) (0.276) (0.281) (0.281) (0.305) (0.305) (0.346) (0.361) Main income source: Business 0.036 0.036 0.235 0.235 0.268 0.268 0.748* 0.893* (0.412) (0.412) (0.447) (0.447) (0.432) (0.432) (0.444) (0.541) Main income source: Wage/Salary -0.758 -0.758 -0.561 -0.561 -0.850 -0.850 -0.694 0.021 (0.527) (0.527) (0.554) (0.554) (0.553) (0.553) (0.674) (0.616) Main income source: Other -0.362 -0.362 -0.163 -0.163 -0.171 -0.171 0.335 (0.831) (0.831) (0.846) (0.846) (0.924) (0.924) (1.181) Asset score -2.749*** -2.749*** -2.707*** -2.707*** -3.663*** -3.663*** -3.715*** -0.732 (0.848) (0.848) (0.865) (0.865) (0.948) (0.948) (1.048) (0.863) Bombali -0.517 -0.517 -0.558 -0.558 -0.723 -0.723 0.997* -0.213 (0.470) (0.470) (0.488) (0.488) (0.513) (0.513) (0.554) (0.484) Kenema -1.191** -1.191** -1.092* -1.092* -1.340** -1.340** -0.608 -0.206 (0.589) (0.589) (0.591) (0.591) (0.592) (0.592) (0.744) (0.536) Port Loko 0.602 0.602 0.696 0.696 0.481 0.481 0.901 0.593 (0.455) (0.455) (0.457) (0.457) (0.487) (0.487) (0.617) (0.511) Western Area Urban 0.271 0.271 0.311 0.311 0.470 0.470 0.275 -0.054 (0.343) (0.343) (0.350) (0.350) (0.362) (0.362) (0.457) (0.352) Constant -1.266** -1.266** -1.539** -1.539** -0.623 -0.623 -2.864*** -3.722*** (0.615) (0.615) (0.629) (0.629) (0.664) (0.664) (0.811) (0.884) Observations 569 569 562 562 392 392 533 512 Source: Authors’ calculations from CIMS (2020). Note: Standard errors in parentheses: *** p<0.01, ** p<0.05, * p<0.1. Inverse probability weighted regressions. For each estimation, overidentification tests shows that the IPW model balances the covariates. 40 Supplementary materials S3. Full regression table of ECT-R1: Round 2 (medium-term) outcomes Main source of Main source of Other source of Other source of Child identified Worked last Hours worked income income income income Inability to buy as mal week last week increased or decreased or increased or decreased or staple foods nourished stayed same stopped stayed same stopped ATE of ECT-R1 0.126* 8.576** 0.025 -0.025 -0.165** 0.064 0.030 0.024 (0.065) (3.690) (0.056) (0.056) (0.078) (0.071) (0.039) (0.071) POM (Control) 0.586*** 24.315*** 0.688*** 0.312*** 0.624*** 0.260*** 0.075*** 0.082*** (0.020) (1.066) (0.019) (0.019) (0.023) (0.021) (0.011) (0.020) Female 0.432 0.432 0.453 0.453 0.436 0.436 0.677** -0.142 (0.290) (0.290) (0.295) (0.295) (0.302) (0.302) (0.292) (0.585) Age -0.012 -0.012 -0.012 -0.012 -0.010 -0.010 -0.013 0.065*** (0.011) (0.011) (0.012) (0.012) (0.013) (0.013) (0.012) (0.024) Household dependency ratio 0.163 0.163 0.185 0.185 0.204 0.204 0.371** 0.470** (0.127) (0.127) (0.127) (0.127) (0.154) (0.154) (0.148) (0.204) Head of household 0.182 0.182 0.176 0.176 0.431 0.431 0.797** -0.448 (0.306) (0.306) (0.309) (0.309) (0.337) (0.337) (0.339) (0.575) Completed secondary school -0.292 -0.292 -0.381 -0.381 -0.439 -0.439 -0.352 -0.145 (0.295) (0.295) (0.305) (0.305) (0.338) (0.338) (0.328) (0.528) Main income source: Business 0.275 0.275 0.221 0.221 0.388 0.388 1.229** 1.220 (0.449) (0.449) (0.466) (0.466) (0.457) (0.457) (0.623) (0.980) Main income source: Wage/Salary -0.882 -0.882 -1.050* -1.050* -0.999* -0.999* 0.081 -0.557 (0.567) (0.567) (0.590) (0.590) (0.601) (0.601) (0.733) (1.232) Main income source: Other -0.865 -0.865 -0.918 -0.918 0.521 (1.063) (1.063) (1.067) (1.067) (0.933) Asset score -2.759*** -2.759*** -2.620*** -2.620*** -2.878*** -2.878*** -1.518* -5.273*** (0.790) (0.790) (0.803) (0.803) (0.899) (0.899) (0.833) (1.822) Bombali 0.460 0.460 0.511 0.511 0.379 0.379 0.460 1.375* (0.516) (0.516) (0.523) (0.523) (0.554) (0.554) (0.584) (0.830) Kenema -0.190 -0.190 -0.118 -0.118 -0.518 -0.518 -0.678 (0.588) (0.588) (0.592) (0.592) (0.670) (0.670) (0.832) Port Loko 1.410*** 1.410*** 1.507*** 1.507*** 1.283** 1.283** 0.995 1.838* (0.485) (0.485) (0.495) (0.495) (0.508) (0.508) (0.705) (1.070) Western Area Urban 1.231*** 1.231*** 1.236*** 1.236*** 1.252*** 1.252*** 1.299*** 0.817 (0.376) (0.376) (0.386) (0.386) (0.415) (0.415) (0.441) (0.718) Constant -1.980*** -1.980*** -2.019*** -2.019*** -1.980*** -1.980*** -3.979*** -4.514*** (0.670) (0.670) (0.671) (0.671) (0.733) (0.733) (0.719) (1.472) Observations 702 702 682 682 503 503 640 208 Source: Authors’ calculations from CIMS (2020). Note: Standard errors in parentheses: *** p<0.01, ** p<0.05, * p<0.1. Inverse probability weighted regressions. For each estimation, overidentification tests shows that the IPW model balances the covariates. 41 Supplementary materials S4. Full regression table of ECT-R2: Round 2 (medium-term) outcomes Other Other Main source Main source Child Hours source of source of Inability to Small children Worked last of income of income identified as worked last income income buy staple given vitamin A week increased or decreased mal week increased or decreased foods supplementation stayed same or stopped nourished stayed same or stopped ATE of ECT-R1 0.180*** 15.340*** -0.146 0.146 -0.266*** 0.290*** -0.068*** -0.059* -0.064 (0.060) (4.461) (0.094) (0.094) (0.077) (0.092) (0.022) (0.034) (0.066) POM (Control) 0.596*** 23.998*** 0.690*** 0.310*** 0.625*** 0.255*** 0.088*** 0.095*** 0.388 (0.024) (1.269) (0.023) (0.023) (0.028) (0.025) (0.014) (0.023) (0.030) Female 0.690 0.690 0.827* 0.827* 1.256** 1.256** 0.072 0.062 0.414 (0.423) (0.423) (0.429) (0.429) (0.540) (0.540) (0.366) (0.631) (0.449) Age 0.003 0.003 0.002 0.002 -0.007 -0.007 0.003 0.031 0.017 (0.016) (0.016) (0.016) (0.016) (0.020) (0.020) (0.014) (0.022) (0.015) Household dependency ratio 0.114 0.114 0.063 0.063 -0.487 -0.487 0.210 0.087 0.275* (0.150) (0.150) (0.169) (0.169) (0.308) (0.308) (0.160) (0.249) (0.167) Head of household 0.218 0.218 0.295 0.295 0.534 0.534 -0.593 -0.700 -0.222 (0.390) (0.390) (0.384) (0.384) (0.507) (0.507) (0.380) (0.504) (0.419) Completed secondary school -0.241 -0.241 -0.359 -0.359 -0.571 -0.571 0.009 -0.416 0.248 (0.439) (0.439) (0.463) (0.463) (0.536) (0.536) (0.414) (0.709) (0.435) Main income source: Business 0.261 0.261 0.471 0.471 0.497 0.497 0.922 0.568 0.734 (0.542) (0.542) (0.699) (0.699) (0.674) (0.674) (0.616) (0.776) (0.661) Main income source: Wage/Salary -0.233 -0.233 -0.007 -0.007 -0.084 -0.084 0.268 -0.877 -0.412 (0.709) (0.709) (0.834) (0.834) (0.863) (0.863) (0.734) (1.022) (0.830) Asset score -1.961 -1.961 -1.836 -1.836 -0.310 -0.310 -2.823** -2.866 -1.636 (1.243) (1.243) (1.253) (1.253) (1.408) (1.408) (1.366) (1.918) (1.551) Bombali 0.133 0.133 -0.162 -0.162 0.740 0.740 -0.253 -0.640 -1.224 (0.754) (0.754) (0.823) (0.823) (1.267) (1.267) (0.879) (1.345) (1.175) Kenema -0.066 -0.066 -0.033 -0.033 1.315 1.315 0.179 0.077 0.050 (0.815) (0.815) (0.819) (0.819) (1.188) (1.188) (0.767) (1.233) (0.809) Port Loko 0.858 0.858 0.878 0.878 1.870 1.870 0.859 1.941* 1.663** (0.790) (0.790) (0.792) (0.792) (1.150) (1.150) (0.872) (1.137) (0.815) Western Area Urban 0.761 0.761 0.663 0.663 1.736* 1.736* 0.744 1.049 0.850 (0.565) (0.565) (0.567) (0.567) (1.016) (1.016) (0.654) (0.803) (0.607) Constant -3.177*** -3.177*** -3.356*** -3.356*** -4.304*** -4.304*** -2.644*** -2.702 -3.584*** (1.001) (1.001) (1.078) (1.078) (1.509) (1.509) (0.869) (1.923) (1.241) Observations 459 459 442 442 328 328 433 187 305 Source: Authors’ calculations from CIMS (2020). Note: Standard errors in parentheses: *** p<0.01, ** p<0.05, * p<0.1. Inverse probability weighted regressions. For each estimation, overidentification tests shows that the IPW model balances the covariates. 42 Supplementary materials S5. Full regression table of pooled outcomes of ECT-R1 and ECT-R2 Main source of Main source of Other source of Other source of Worked last Hours worked income income income income Inability to buy week last week increased or decreased or increased or decreased or staple foods stayed same stopped stayed same stopped ATE of ECT-R1 0.121** 10.448*** 0.173*** -0.173*** 0.156** -0.294*** -0.031 (0.056) (3.270) (0.062) (0.062) (0.061) (0.070) (0.034) POM (Control) 0.509*** 19.517*** 0.347*** 0.653*** 0.284*** 0.614*** 0.120*** (0.018) (0.930) (0.017) (0.017) (0.020) (0.022) (0.012) Female 0.579** 0.579** 0.605** 0.605** 0.667*** 0.667*** 0.208 (0.238) (0.238) (0.241) (0.241) (0.259) (0.259) (0.274) Age -0.000 -0.000 0.001 0.001 -0.008 -0.008 0.015* (0.009) (0.009) (0.009) (0.009) (0.010) (0.010) (0.009) Household dependency ratio 0.126 0.126 0.145 0.145 0.194 0.194 0.420*** (0.106) (0.106) (0.107) (0.107) (0.133) (0.133) (0.118) Head of household 0.388 0.388 0.424* 0.424* 0.580** 0.580** -0.007 (0.253) (0.253) (0.254) (0.254) (0.281) (0.281) (0.270) Completed secondary school -0.066 -0.066 -0.143 -0.143 -0.168 -0.168 -0.018 (0.240) (0.240) (0.245) (0.245) (0.265) (0.265) (0.276) Main income source: Business 0.539 0.539 0.735* 0.735* 0.626* 0.626* 0.940** (0.356) (0.356) (0.386) (0.386) (0.376) (0.376) (0.378) Main income source: Wage/Salary -0.458 -0.458 -0.247 -0.247 -0.551 -0.551 -0.263 (0.442) (0.442) (0.465) (0.465) (0.476) (0.476) (0.486) Main income source: Other -0.439 -0.439 -0.247 -0.247 -0.446 -0.446 (0.772) (0.772) (0.785) (0.785) (0.801) (0.801) Asset score -2.464*** -2.464*** -2.423*** -2.423*** -2.589*** -2.589*** -3.473*** (0.665) (0.665) (0.675) (0.675) (0.699) (0.699) (0.800) Bombali -0.329 -0.329 -0.442 -0.442 -0.530 -0.530 0.345 (0.416) (0.416) (0.443) (0.443) (0.481) (0.481) (0.460) Kenema -0.915* -0.915* -0.824* -0.824* -0.867* -0.867* -0.104 (0.468) (0.468) (0.472) (0.472) (0.502) (0.502) (0.509) Port Loko 0.209 0.209 0.335 0.335 0.053 0.053 0.328 (0.385) (0.385) (0.386) (0.386) (0.430) (0.430) (0.504) Western Area Urban 0.518* 0.518* 0.548* 0.548* 0.634* 0.634* 0.622 (0.298) (0.298) (0.305) (0.305) (0.338) (0.338) (0.381) Constant -2.155*** -2.155*** -2.474*** -2.474*** -1.684*** -1.684*** -3.122*** (0.533) (0.533) (0.533) (0.533) (0.604) (0.604) (0.623) Observations 908 908 893 893 627 627 810 Source: Authors’ calculations from CIMS (2020) Note: Standard errors in parentheses: *** p<0.01, ** p<0.05, * p<0.1. Inverse probability weighted regressions. For each estimation, overidentification tests shows that the IPW model balances the covariates. 43 Supplementary Materials S6. Balance tests for ECT-R1 short-term, ECT-R1 medium-term, ECT-R2 short-term and pooled estimations Short-term ECT-R1 – Worked last week Standardized Standardized differences- Variance Variance Ratio- differences- Raw Weighted Ratio- Raw Weighted Female 0.110566 0.037604 0.963533 0.985948 Age 0.051401 0.013315 0.777923 0.694858 Household dependency ratio 0.100014 0.112127 0.72764 0.715065 Household head 0.129401 -0.03805 0.964385 1.008376 Completed Secondary School -0.01866 0.000524 0.989061 1.000697 Main income source: Business 0.282555 0.028862 0.785406 0.98044 Main income source: Wage/Salary -0.31419 0.015886 0.544325 1.024299 Main income source: Other -0.02958 -0.05497 0.841382 0.696539 Asset score -0.40076 -0.13508 0.559976 0.817707 Bombali -0.17604 0.02651 0.681147 1.051766 Kenema -0.33648 -0.09471 0.356268 0.790145 Port Loko 0.227008 -0.00678 1.892117 0.978709 Western Area Urban 0.198976 0.093012 0.999461 0.999398 Overidentification test for covariate balance Chi-square 7.698917 P-value 0.904415 Source: Authors’ calculations from CIMS (2020) 44 Short-term ECT-R1 - Hours worked last week Standardized Standardized differences- Variance Variance Ratio- differences- Raw Weighted Ratio- Raw Weighted Female 0.110566 0.037604 0.963533 0.985948 Age 0.051401 0.013315 0.777923 0.694858 Household dependency ratio 0.100014 0.112127 0.72764 0.715065 Household head 0.129401 -0.03805 0.964385 1.008376 Completed Secondary School -0.01866 0.000524 0.989061 1.000697 Main income source: Business 0.282555 0.028862 0.785406 0.98044 Main income source: Wage/Salary -0.31419 0.015886 0.544325 1.024299 Main income source: Other -0.02958 -0.05497 0.841382 0.696539 Asset score -0.40076 -0.13508 0.559976 0.817707 Bombali -0.17604 0.02651 0.681147 1.051766 Kenema -0.33648 -0.09471 0.356268 0.790145 Port Loko 0.227008 -0.00678 1.892117 0.978709 Western Area Urban 0.198976 0.093012 0.999461 0.999398 Overidentification test for covariate balance Chi-square 7.698917 P-value 0.904415 Source: Authors’ calculations from CIMS (2020) 45 Short-term ECT-R1 - Main source of income increased or stayed same Standardized Standardized differences- Variance Variance Ratio- differences- Raw Weighted Ratio- Raw Weighted Female 0.112985 0.026034 0.961657 0.990399 Age 0.059943 0.010385 0.78753 0.705493 Household dependency ratio 0.11486 0.114783 0.723936 0.696763 Household head 0.127472 -0.02303 0.9636 1.005808 Completed Secondary School -0.03293 0.00963 0.97366 1.010351 Main income source: Business 0.310788 0.026403 0.749839 0.981168 Main income source: Wage/Salary -0.3121 0.029918 0.551329 1.044856 Main income source: Other -0.02787 -0.05754 0.851761 0.685269 Asset score -0.38692 -0.13456 0.577666 0.844283 Bombali -0.20556 0.022463 0.627169 1.044344 Kenema -0.33023 -0.10418 0.365475 0.769362 Port Loko 0.233113 -0.00469 1.910482 0.985425 Western Area Urban 0.219652 0.101067 0.991745 0.996824 Overidentification test for covariate balance Chi-square 7.824087 P-value 0.898288 Source: Authors’ calculations from CIMS (2020) 46 Short-term ECT-R1 - Main source of income decreased or stopped Standardized Standardized differences- Variance Variance Ratio- differences- Raw Weighted Ratio- Raw Weighted Female 0.112985 0.026034 0.961657 0.990399 Age 0.059943 0.010385 0.78753 0.705493 Household dependency ratio 0.11486 0.114783 0.723936 0.696763 Household head 0.127472 -0.02303 0.9636 1.005808 Completed Secondary School -0.03293 0.00963 0.97366 1.010351 Main income source: Business 0.310788 0.026403 0.749839 0.981168 Main income source: Wage/Salary -0.3121 0.029918 0.551329 1.044856 Main income source: Other -0.02787 -0.05754 0.851761 0.685269 Asset score -0.38692 -0.13456 0.577666 0.844283 Bombali -0.20556 0.022463 0.627169 1.044344 Kenema -0.33023 -0.10418 0.365475 0.769362 Port Loko 0.233113 -0.00469 1.910482 0.985425 Western Area Urban 0.219652 0.101067 0.991745 0.996824 Overidentification test for covariate balance Chi-square 7.824087 P-value 0.898288 Source: Authors’ calculations from CIMS (2020) 47 Short-term ECT-R1 - Other source of income increased or stayed the same Standardized Standardized differences- Variance Variance Ratio- differences- Raw Weighted Ratio- Raw Weighted Female 0.138687 0.011981 0.954945 0.996544 Age 0.033693 -0.02077 0.739206 0.666668 Household dependency ratio 0.184073 0.056575 1.067173 0.821725 Household head 0.181898 0.001425 0.95491 1.00005 Completed Secondary School -0.07875 0.060682 0.931204 1.053737 Main income source: Business 0.410442 -0.0036 0.743303 1.002055 Main income source: Wage/Salary -0.39976 0.067388 0.487607 1.090169 Main income source: Other -0.00421 -0.05723 0.981152 0.661064 Asset score -0.52842 -0.14462 0.512284 0.853838 Bombali -0.24007 0.078601 0.608169 1.13886 Kenema -0.38125 -0.19002 0.324545 0.599926 Port Loko 0.192007 -0.01153 1.676343 0.966917 Western Area Urban 0.328335 0.139717 1.034558 1.015501 Overidentification test for covariate balance Chi-square 8.124636 P-value 0.882732 Source: Authors’ calculations from CIMS (2020) 48 Short-term ECT-R1 - Other source of income decreased or stopped Standardized Standardized differences- Variance Variance Ratio- differences- Raw Weighted Ratio- Raw Weighted Female 0.138687 0.011981 0.954945 0.996544 Age 0.033693 -0.02077 0.739206 0.666668 Household dependency ratio 0.184073 0.056575 1.067173 0.821725 Household head 0.181898 0.001425 0.95491 1.00005 Completed Secondary School -0.07875 0.060682 0.931204 1.053737 Main income source: Business 0.410442 -0.0036 0.743303 1.002055 Main income source: Wage/Salary -0.39976 0.067388 0.487607 1.090169 Main income source: Other -0.00421 -0.05723 0.981152 0.661064 Asset score -0.52842 -0.14462 0.512284 0.853838 Bombali -0.24007 0.078601 0.608169 1.13886 Kenema -0.38125 -0.19002 0.324545 0.599926 Port Loko 0.192007 -0.01153 1.676343 0.966917 Western Area Urban 0.328335 0.139717 1.034558 1.015501 Overidentification test for covariate balance Chi-square 8.124636 P-value 0.882732 Source: Authors’ calculations from CIMS (2020) 49 Short-term ECT-R1 - Inability to buy staple foods Standardized Standardized differences- Variance Variance Ratio- differences- Raw Weighted Ratio- Raw Weighted Female 0.141445 0.181159 0.957962 0.913636 Age 0.0605 0.231494 0.601474 0.603174 Household dependency ratio 0.486281 0.163569 1.670146 0.962864 Household head 0.158556 -0.01786 0.929527 1.007684 Completed Secondary School -0.06725 -0.17698 0.947407 0.806425 Main income source: Business 0.482997 0.029173 0.632474 0.98426 Main income source: Wage/Salary -0.50698 -0.01256 0.360587 0.984655 Main income source: Other -0.60679 -0.14727 0.356585 0.578599 Asset score 0.200524 0.166385 1.53369 1.401642 Bombali -0.23662 -0.103 0.491629 0.756605 Kenema 0.164866 0.070931 1.679337 1.25162 Port Loko 0.020986 -0.07167 1.0145 1.000133 Western Area Urban Overidentification test for covariate balance Chi-square 5.734646 P-value 0.955279 Source: Authors’ calculations from CIMS (2020) 50 Short-term ECT-R1 - Able to financially support child's return to school Standardized Standardized differences- Variance Variance Ratio- differences- Raw Weighted Ratio- Raw Weighted Female 0.144492 0.197026 1.007653 0.97411 Age 0.232264 -0.0301 0.755094 0.618669 Household dependency ratio 0.395131 0.014127 1.270649 0.72094 Household head 0.312932 -0.09522 0.804875 1.036004 Completed Secondary School -0.3179 -0.07419 0.717916 0.937632 Main income source: Business 0.446019 -0.02242 0.69752 1.009794 Main income source: Wage/Salary -0.37333 -0.02929 0.551646 0.965149 Main income source: Other -0.08171 -0.01847 0.577404 0.891167 Asset score -0.20761 -0.09643 0.645923 0.629428 Bombali -0.12124 0.047648 0.788359 1.088685 Kenema -0.05569 -0.05387 0.871798 0.863822 Port Loko 0.142204 -0.04285 1.492264 0.873241 Western Area Urban -0.01466 0.111592 1.010608 1.004693 Overidentification test for covariate balance Chi-square 8.771452 P-value 0.845407 Source: Authors’ calculations from CIMS (2020) 51 Medium-term ECT-R1 - Worked last week Standardized Standardized differences- Variance Variance Ratio- differences- Raw Weighted Ratio- Raw Weighted Female 0.325107 0.007003 0.948769 1.000568 Age -0.07059 -0.04472 0.642092 0.696045 Household dependency ratio 0.23065 0.184583 0.97709 0.880408 Household head -0.05253 -0.02222 1.036986 1.011144 Completed Secondary School -0.22285 0.035711 0.837987 1.021944 Main income source: Business 0.510121 0.076759 0.695971 0.97134 Main income source: Wage/Salary -0.5132 -0.03366 0.440884 0.966429 Main income source: Other -0.13828 -0.12328 0.393588 0.427012 Asset score -0.45839 -0.15991 0.477536 0.713749 Bombali -0.09423 5.76E-05 0.790404 1.000385 Kenema -0.228 -0.15207 0.487973 0.630465 Port Loko 0.254439 0.028066 2.042123 1.093537 Western Area Urban 0.320215 0.085345 1.014534 1.015124 Overidentification test for covariate balance Chi-square 5.267087 P-value 0.981737 Source: Authors’ calculations from CIMS (2020) 52 Medium-term ECT-R1 - Hours worked last week Standardized Standardized differences- Variance Variance Ratio- differences- Raw Weighted Ratio- Raw Weighted Female 0.325107 0.007003 0.948769 1.000568 Age -0.07059 -0.04472 0.642092 0.696045 Household dependency ratio 0.23065 0.184583 0.97709 0.880408 Household head -0.05253 -0.02222 1.036986 1.011144 Completed Secondary School -0.22285 0.035711 0.837987 1.021944 Main income source: Business 0.510121 0.076759 0.695971 0.97134 Main income source: Wage/Salary -0.5132 -0.03366 0.440884 0.966429 Main income source: Other -0.13828 -0.12328 0.393588 0.427012 Asset score -0.45839 -0.15991 0.477536 0.713749 Bombali -0.09423 5.76E-05 0.790404 1.000385 Kenema -0.228 -0.15207 0.487973 0.630465 Port Loko 0.254439 0.028066 2.042123 1.093537 Western Area Urban 0.320215 0.085345 1.014534 1.015124 Overidentification test for covariate balance Chi-square 5.267087 P-value 0.981737 Source: Authors’ calculations from CIMS (2020) 53 Medium-term ECT-R1 - Main source of income increased or stayed the same Standardized Standardized differences- Variance Variance Ratio- differences- Raw Weighted Ratio- Raw Weighted Female 0.347752 -0.00649 0.947546 0.999685 Age -0.05166 -0.02583 0.647642 0.716448 Household dependency ratio 0.265708 0.225829 0.970369 0.824953 Household head -0.05489 -0.02432 1.03927 1.012513 Completed Secondary School -0.26668 0.023589 0.795508 1.015206 Main income source: Business 0.538904 0.083592 0.650224 0.963876 Main income source: Wage/Salary -0.56407 0.000718 0.395336 1.000913 Main income source: Other -0.14072 -0.12622 0.39268 0.422959 Asset score -0.45786 -0.11116 0.479889 0.749331 Bombali -0.08674 0.004311 0.808028 1.010608 Kenema -0.21222 -0.16091 0.512958 0.60416 Port Loko 0.270219 0.034098 2.130724 1.114496 Western Area Urban 0.302118 0.098466 1.006715 1.012144 Overidentification test for covariate balance Chi-square 6.364688 P-value 0.956446 Source: Authors’ calculations from CIMS (2020) 54 Medium-term ECT-R1 - Main source of income decreased or stopped Standardized Standardized differences- Variance Variance Ratio- differences- Raw Weighted Ratio- Raw Weighted Female 0.347752 -0.00649 0.947546 0.999685 Age -0.05166 -0.02583 0.647642 0.716448 Household dependency ratio 0.265708 0.225829 0.970369 0.824953 Household head -0.05489 -0.02432 1.03927 1.012513 Completed Secondary School -0.26668 0.023589 0.795508 1.015206 Main income source: Business 0.538904 0.083592 0.650224 0.963876 Main income source: Wage/Salary -0.56407 0.000718 0.395336 1.000913 Main income source: Other -0.14072 -0.12622 0.39268 0.422959 Asset score -0.45786 -0.11116 0.479889 0.749331 Bombali -0.08674 0.004311 0.808028 1.010608 Kenema -0.21222 -0.16091 0.512958 0.60416 Port Loko 0.270219 0.034098 2.130724 1.114496 Western Area Urban 0.302118 0.098466 1.006715 1.012144 Overidentification test for covariate balance Chi-square 6.364688 P-value 0.956446 Source: Authors’ calculations from CIMS (2020) 55 Medium-term ECT-R1 - Other source of income increased or stayed the same Standardized Standardized differences- Variance Variance Ratio- differences- Raw Weighted Ratio- Raw Weighted Female 0.273857 0.027878 0.980656 1.001761 Age 0.025975 -0.00871 0.681612 0.670358 Household dependency ratio 0.225126 0.157866 1.028576 0.995633 Household head 0.096459 0.038705 0.959858 0.980652 Completed Secondary School -0.28538 0.033886 0.790476 1.020189 Main income source: Business 0.553208 0.05126 0.705147 0.986437 Main income source: Wage/Salary -0.58263 -0.03943 0.424025 0.965869 Asset score -0.46745 -0.16371 0.497625 0.780001 Bombali -0.07445 0.064321 0.846515 1.146496 Kenema -0.29993 -0.17744 0.386978 0.596691 Port Loko 0.241668 0.029713 1.891039 1.091494 Western Area Urban 0.342601 0.062745 1.059769 1.020777 Overidentification test for covariate balance Chi-square 3.017594 P-value 0.99787 Source: Authors’ calculations from CIMS (2020) 56 Medium-term ECT-R1 - Other source of income decreased or stopped Standardized Standardized differences- Variance Variance Ratio- differences- Raw Weighted Ratio- Raw Weighted Female 0.273857 0.027878 0.980656 1.001761 Age 0.025975 -0.00871 0.681612 0.670358 Household dependency ratio 0.225126 0.157866 1.028576 0.995633 Household head 0.096459 0.038705 0.959858 0.980652 Completed Secondary School -0.28538 0.033886 0.790476 1.020189 Main income source: Business 0.553208 0.05126 0.705147 0.986437 Main income source: Wage/Salary -0.58263 -0.03943 0.424025 0.965869 Asset score -0.46745 -0.16371 0.497625 0.780001 Bombali -0.07445 0.064321 0.846515 1.146496 Kenema -0.29993 -0.17744 0.386978 0.596691 Port Loko 0.241668 0.029713 1.891039 1.091494 Western Area Urban 0.342601 0.062745 1.059769 1.020777 Overidentification test for covariate balance Chi-square 3.017594 P-value 0.99787 Source: Authors’ calculations from CIMS (2020) 57 Medium-term ECT-R1 - Inability to buy staple foods Standardized Standardized differences- Variance Variance Ratio- differences- Raw Weighted Ratio- Raw Weighted Female 0.270454 -0.1971 0.941497 0.97275 Age 0.006313 0.126964 0.783657 0.906111 Household dependency ratio 0.487109 0.303494 0.751922 0.408769 Household head 0.212845 0.060917 0.874476 0.966316 Completed Secondary School -0.26632 0.150894 0.762734 1.100262 Main income source: Business 0.662185 0.034682 0.576618 0.988823 Main income source: Wage/Salary -0.4893 0.130671 0.457093 1.119184 Main income source: Other -0.09739 -0.15081 0.611851 0.404139 Asset score -0.21558 -0.09474 0.633631 0.676225 Bombali -0.08121 0.171797 0.818628 1.423689 Kenema -0.36662 -0.11399 0.244446 0.7231 Port Loko -0.00249 0.122602 1.001481 1.526815 Western Area Urban 0.580684 -0.12337 0.788946 0.982667 Overidentification test for covariate balance Chi-square 16.11727 P-value 0.306265 Source: Authors’ calculations from CIMS (2020) 58 Medium-term ECT-R1 - Child identified as malnourished Standardized Standardized differences- Variance Variance Ratio- differences- Raw Weighted Ratio- Raw Weighted Female 0.198341 0.130961 1.010053 0.983851 Age 0.293792 0.491933 0.68682 0.900374 Household dependency ratio 0.554716 0.297804 2.493232 1.36491 Household head 0.109697 -0.02343 0.992767 1.00874 Completed Secondary School -0.2917 -0.07643 0.821139 0.953858 Main income source: Business 0.851924 0.143018 0.454255 0.951622 Main income source: Wage/Salary -0.72242 0.006718 0.340264 1.006764 Asset score -0.65286 -0.04445 0.332041 0.578509 Bombali 0.205946 0.266292 1.509115 1.538423 Port Loko 0.316544 0.049909 3.101695 1.20894 Western Area Urban 0.0645011 -0.21293 1.04224 0.927701 Overidentification test for covariate balance Chi-square 1.826647 P-value 0.999629 Source: Authors’ calculations from CIMS (2020) 59 Medium-term ECT-R2 - Worked last week Standardized Standardized differences- Variance Variance Ratio- differences- Raw Weighted Ratio- Raw Weighted Female 0.365174 -0.01629 0.824682 1.003837 Age 0.111013 0.090337 0.870673 1.052575 Household dependency ratio 0.194286 -0.1805 3.234074 1.336616 Household head 0.001935 -0.03197 1.022384 1.01169 Completed Secondary School -0.16927 0.055904 0.826611 1.056511 Main income source: Business 0.254888 -0.17184 0.825316 1.073115 Main income source: Wage/Salary -0.23946 0.068772 0.663537 1.096681 Asset score -0.31364 0.007725 0.67199 0.911279 Bombali -0.13839 0.267097 0.740313 1.52662 Kenema -0.1917 0.026184 0.613047 1.058022 Port Loko 0.128681 -0.03446 1.52749 0.883245 Western Area Urban 0.254694 -0.16893 0.959249 0.977487 Overidentification test for covariate balance Chi-square 8.025189 P-value 0.841955 Source: Authors’ calculations from CIMS (2020) 60 Medium-term ECT-R2 - Hours worked last week Standardized Standardized differences- Variance Variance Ratio- differences- Raw Weighted Ratio- Raw Weighted Female 0.365174 -0.01629 0.824682 1.003837 Age 0.111013 0.090337 0.870673 1.052575 Household dependency ratio 0.194286 -0.1805 3.234074 1.336616 Household head 0.001935 -0.03197 1.022384 1.01169 Completed Secondary School -0.16927 0.055904 0.826611 1.056511 Main income source: Business 0.254888 -0.17184 0.825316 1.073115 Main income source: Wage/Salary -0.23946 0.068772 0.663537 1.096681 Asset score -0.31364 0.007725 0.67199 0.911279 Bombali -0.13839 0.267097 0.740313 1.52662 Kenema -0.1917 0.026184 0.613047 1.058022 Port Loko 0.128681 -0.03446 1.52749 0.883245 Western Area Urban 0.254694 -0.16893 0.959249 0.977487 Overidentification test for covariate balance Chi-square 8.025189 P-value 0.841955 Source: Authors’ calculations from CIMS (2020) 61 Medium-term ECT-R2 - Main source of income increased or stayed the same Standardized Standardized differences- Variance Variance Ratio- differences- Raw Weighted Ratio- Raw Weighted Female 0.411515 0.039543 0.800282 0.9889 Age 0.110411 0.200938 0.866374 1.035344 Household dependency ratio 0.150199 -0.10502 2.640978 1.276105 Household head 0.002645 0.060737 1.023317 0.970422 Completed Secondary School -0.20963 -0.0349 0.777 0.961766 Main income source: Business 0.310255 -0.12346 0.749441 1.070476 Main income source: Wage/Salary -0.2391 0.114025 0.676103 1.150042 Asset score -0.30093 0.095959 0.68808 0.970241 Bombali -0.21232 0.176645 0.597337 1.361031 Kenema -0.16997 0.016914 0.653593 1.037888 Port Loko 0.145959 -0.05208 1.602414 0.825917 Western Area Urban 0.25594 -0.10566 0.953631 0.995148 Overidentification test for covariate balance Chi-square 5.687832 P-value 0.95678 Source: Authors’ calculations from CIMS (2020) 62 Medium-term ECT-R2 - Main source of income decreased or stopped Standardized Standardized differences- Variance Variance Ratio- differences- Raw Weighted Ratio- Raw Weighted Female 0.411515 0.039543 0.800282 0.9889 Age 0.110411 0.200938 0.866374 1.035344 Household dependency ratio 0.150199 -0.10502 2.640978 1.276105 Household head 0.002645 0.060737 1.023317 0.970422 Completed Secondary School -0.20963 -0.0349 0.777 0.961766 Main income source: Business 0.310255 -0.12346 0.749441 1.070476 Main income source: Wage/Salary -0.2391 0.114025 0.676103 1.150042 Asset score -0.30093 0.095959 0.68808 0.970241 Bombali -0.21232 0.176645 0.597337 1.361031 Kenema -0.16997 0.016914 0.653593 1.037888 Port Loko 0.145959 -0.05208 1.602414 0.825917 Western Area Urban 0.25594 -0.10566 0.953631 0.995148 Overidentification test for covariate balance Chi-square 5.687832 P-value 0.95678 Source: Authors’ calculations from CIMS (2020) 63 Medium-term ECT-R2 - Other source of income increased or stayed the same Standardized Standardized differences- Variance Variance Ratio- differences- Raw Weighted Ratio- Raw Weighted Female 0.513653 0.014403 0.73864 0.997515 Age 0.067122 0.161388 0.917287 0.859361 Household dependency ratio -0.314 -0.24329 0.664501 0.637824 Household head -0.02505 0.173392 1.045448 0.909966 Completed Secondary School -0.22348 -0.16097 0.78504 0.825713 Main income source: Business 0.319282 -0.04108 0.836893 1.014228 Main income source: Wage/Salary -0.21222 0.101187 0.74316 1.12261 Asset score -0.10245 0.04081 0.815513 0.92257 Bombali -0.24542 0.128724 0.560097 1.251391 Kenema -0.09843 -0.0285 0.834379 0.94338 Port Loko 0.144247 -0.1404 1.566493 0.580967 Western Area Urban 0.393254 0.137241 0.944385 0.994477 Overidentification test for covariate balance Chi-square 2.548511 P-value 0.999134 Source: Authors’ calculations from CIMS (2020) 64 Medium-term ECT-R2 - Other source of income decreased or stopped Standardized Standardized differences- Variance Variance Ratio- differences- Raw Weighted Ratio- Raw Weighted Female 0.513653 0.014403 0.73864 0.997515 Age 0.067122 0.161388 0.917287 0.859361 Household dependency ratio -0.314 -0.24329 0.664501 0.637824 Household head -0.02505 0.173392 1.045448 0.909966 Completed Secondary School -0.22348 -0.16097 0.78504 0.825713 Main income source: Business 0.319282 -0.04108 0.836893 1.014228 Main income source: Wage/Salary -0.21222 0.101187 0.74316 1.12261 Asset score -0.10245 0.04081 0.815513 0.92257 Bombali -0.24542 0.128724 0.560097 1.251391 Kenema -0.09843 -0.0285 0.834379 0.94338 Port Loko 0.144247 -0.1404 1.566493 0.580967 Western Area Urban 0.393254 0.137241 0.944385 0.994477 Overidentification test for covariate balance Chi-square 2.548511 P-value 0.999134 Source: Authors’ calculations from CIMS (2020) 65 Medium-term ECT-R2 - Inability to buy staple foods Standardized Standardized differences- Variance Variance Ratio- differences- Raw Weighted Ratio- Raw Weighted Female 0.199451 0.03436 0.941202 0.990411 Age 0.018971 0.047896 0.651603 0.580698 Household dependency ratio 0.266639 0.100594 1.109815 0.81727 Household head -0.24005 0.123391 1.033034 0.955766 Completed Secondary School -0.10596 0.176796 0.886679 1.191172 Main income source: Business 0.363563 0.069144 0.787511 0.968387 Main income source: Wage/Salary -0.25489 -0.16381 0.692152 0.789024 Asset score -0.31039 -0.08586 0.696983 0.944909 Bombali -0.23111 0.145755 0.564162 1.298586 Kenema -0.11412 0.005173 0.810597 1.009557 Port Loko 0.065539 -0.01225 1.288094 0.954945 Western Area Urban 0.297173 -0.00477 0.931013 1.000342 Overidentification test for covariate balance Chi-square 6.084817 P-value 0.943011 Source: Authors’ calculations from CIMS (2020) 66 Medium-term ECT-R2 - Child identifies as malnourished Standardized Standardized differences- Variance Variance Ratio- differences- Raw Weighted Ratio- Raw Weighted Female 0.437432 0.331687 0.857061 0.863702 Age 0.070508 0.24601 0.568777 0.672155 Household dependency ratio 0.187255 0.089587 1.630749 1.202679 Household head -0.27311 -0.2004 1.068523 1.02474 Completed Secondary School -0.37063 -0.15487 0.656425 0.851183 Main income source: Business 0.640397 0.301112 0.684208 0.871224 Main income source: Wage/Salary -0.63745 -0.285 0.357599 0.699253 Asset score -0.32767 -0.02131 0.57054 0.809347 Bombali -0.41243 -0.12747 0.299632 0.751353 Kenema -0.1991 -0.07075 0.630943 0.853942 Port Loko 0.351471 0.024098 3.814846 1.110667 Western Area Urban 0.345241 0.177521 0.921881 0.958517 Overidentification test for covariate balance Chi-square 2.725001 P-value 0.998758 Source: Authors’ calculations from CIMS (2020) 67 Medium-term ECT-R2 - Child given Vitamin A supplementation Standardized Standardized differences- Variance Variance Ratio- differences- Raw Weighted Ratio- Raw Weighted Female 0.42627 0.247279 0.841892 0.910028 Age -0.00046 0.195852 0.470924 0.489853 Household dependency ratio 0.386636 0.080619 2.099085 0.943181 Household head -0.09371 -0.09919 1.053348 1.029306 Completed Secondary School -0.07242 -0.20546 0.952315 0.781106 Main income source: Business 0.599141 0.221124 0.620482 0.883014 Main income source: Wage/Salary -0.44544 -0.13376 0.481014 0.839602 Asset score -0.20454 -0.03295 1.16488 1.298996 Bombali -0.52886 -0.20075 0.174967 0.636225 Kenema -0.17371 -0.09016 0.700147 0.829298 Port Loko 0.322578 -0.02005 2.959776 0.922085 Western Area Urban 0.368979 0.072075 0.962162 1.003674 Overidentification test for covariate balance Chi-square 5.623444 P-value 0.958791 Source: Authors’ calculations from CIMS (2020) 68 Pooled ECT-R1 and ECT-R2 - Worked last week Standardized Standardized differences- Variance Variance Ratio- differences- Raw Weighted Ratio- Raw Weighted Female 0.317476 0.055515 0.933596 0.996481 Age 0.03099 -0.01329 0.771258 0.790843 Household dependency ratio 0.195185 0.002332 0.958411 0.823199 Household head 0.03448 -0.02096 0.992484 1.007795 Completed Secondary School -0.1685 -0.02919 0.86412 0.977169 Main income source: Business 0.512727 -0.03453 0.751549 1.004984 Main income source: Wage/Salary -0.39952 0.065452 0.521977 1.072577 Main income source: Other -0.10614 -0.1117 0.512448 0.469018 Asset score -0.37695 -0.07308 0.525677 0.775805 Bombali -0.17486 0.161954 0.677803 1.310566 Kenema -0.30823 -0.02235 0.427671 0.953302 Port Loko 0.030698 -0.04316 1.078924 0.901675 Western Area Urban 0.329879 -0.00861 1.033772 0.997171 Overidentification test for covariate balance Chi-square 7.146145 P-value 0.928931 Source: Authors’ calculations from CIMS (2020) 69 Pooled ECT-R1 and ECT-R2 - Hours worked last week Standardized Standardized differences- Variance Variance Ratio- differences- Raw Weighted Ratio- Raw Weighted Female 0.317476 0.055515 0.933596 0.996481 Age 0.03099 -0.01329 0.771258 0.790843 Household dependency ratio 0.195185 0.002332 0.958411 0.823199 Household head 0.03448 -0.02096 0.992484 1.007795 Completed Secondary School -0.1685 -0.02919 0.86412 0.977169 Main income source: Business 0.512727 -0.03453 0.751549 1.004984 Main income source: Wage/Salary -0.39952 0.065452 0.521977 1.072577 Main income source: Other -0.10614 -0.1117 0.512448 0.469018 Asset score -0.37695 -0.07308 0.525677 0.775805 Bombali -0.17486 0.161954 0.677803 1.310566 Kenema -0.30823 -0.02235 0.427671 0.953302 Port Loko 0.030698 -0.04316 1.078924 0.901675 Western Area Urban 0.329879 -0.00861 1.033772 0.997171 Overidentification test for covariate balance Chi-square 7.146145 P-value 0.928931 Source: Authors’ calculations from CIMS (2020) 70 Pooled ECT-R1 and ECT-R2 - Main source of income increased or stayed the same Standardized Standardized differences- Variance Variance Ratio- differences- Raw Weighted Ratio- Raw Weighted Female 0.32657 0.065189 0.924584 0.99414 Age 0.052608 0.021551 0.775216 0.800541 Household dependency ratio 0.217035 0.01807 0.944594 0.778084 Household head 0.046006 0.012142 0.987015 0.99501 Completed Secondary School -0.20087 -0.05122 0.835909 0.959453 Main income source: Business 0.543058 -0.02729 0.712085 1.005178 Main income source: Wage/Salary -0.39996 0.099109 0.52847 1.10455 Main income source: Other -0.10576 -0.11615 0.517616 0.454917 Asset score -0.36986 -0.03777 0.53682 0.819484 Bombali -0.22022 0.144175 0.592974 1.283515 Kenema -0.29812 -0.02153 0.441624 0.954747 Port Loko 0.046599 -0.03521 1.117964 0.919109 Western Area Urban 0.34925 0.013336 1.021568 1.003815 Overidentification test for covariate balance Chi-square 7.132129 P-value 0.929498 Source: Authors’ calculations from CIMS (2020) 71 Pooled ECT-R1 and ECT-R2 - Main source of income decreased or stopped Standardized Standardized differences- Variance Variance Ratio- differences- Raw Weighted Ratio- Raw Weighted Female 0.32657 0.065189 0.924584 0.99414 Age 0.052608 0.021551 0.775216 0.800541 Household dependency ratio 0.217035 0.01807 0.944594 0.778084 Household head 0.046006 0.012142 0.987015 0.99501 Completed Secondary School -0.20087 -0.05122 0.835909 0.959453 Main income source: Business 0.543058 -0.02729 0.712085 1.005178 Main income source: Wage/Salary -0.39996 0.099109 0.52847 1.10455 Main income source: Other -0.10576 -0.11615 0.517616 0.454917 Asset score -0.36986 -0.03777 0.53682 0.819484 Bombali -0.22022 0.144175 0.592974 1.283515 Kenema -0.29812 -0.02153 0.441624 0.954747 Port Loko 0.046599 -0.03521 1.117964 0.919109 Western Area Urban 0.34925 0.013336 1.021568 1.003815 Overidentification test for covariate balance Chi-square 7.132129 P-value 0.929498 Source: Authors’ calculations from CIMS (2020) 72 Pooled ECT-R1 and ECT-R2 - Other source of income increased or stayed the same Standardized Standardized differences- Variance Variance Ratio- differences- Raw Weighted Ratio- Raw Weighted Female 0.363793 0.065311 0.916764 0.993696 Age -0.00601 0.019239 0.735458 0.770118 Household dependency ratio 0.207948 -0.06378 1.135738 0.877051 Household head 0.069648 0.011191 0.981626 0.996367 Completed Secondary School -0.21384 -0.06615 0.832673 0.948934 Main income source: Business 0.593785 -0.0395 0.75769 1.001865 Main income source: Wage/Salary -0.45342 0.12658 0.496658 1.122887 Main income source: Other -0.09614 -0.10607 0.553317 0.48674 Asset score -0.41852 -0.05135 0.502444 0.755212 Bombali -0.22384 0.223223 0.61846 1.386332 Kenema -0.30075 -0.04809 0.449715 0.89958 Port Loko -0.02262 -0.06539 0.960168 0.864506 Western Area Urban 0.413181 -0.02067 1.059287 0.991409 Overidentification test for covariate balance Chi-square 7.904167 P-value 0.894259 Source: Authors’ calculations from CIMS (2020) 73 Pooled ECT-R1 and ECT-R2 - Other source of income decreased or stopped Standardized Standardized differences- Variance Variance Ratio- differences- Raw Weighted Ratio- Raw Weighted Female 0.363793 0.065311 0.916764 0.993696 Age -0.00601 0.019239 0.735458 0.770118 Household dependency ratio 0.207948 -0.06378 1.135738 0.877051 Household head 0.069648 0.011191 0.981626 0.996367 Completed Secondary School -0.21384 -0.06615 0.832673 0.948934 Main income source: Business 0.593785 -0.0395 0.75769 1.001865 Main income source: Wage/Salary -0.45342 0.12658 0.496658 1.122887 Main income source: Other -0.09614 -0.10607 0.553317 0.48674 Asset score -0.41852 -0.05135 0.502444 0.755212 Bombali -0.22384 0.223223 0.61846 1.386332 Kenema -0.30075 -0.04809 0.449715 0.89958 Port Loko -0.02262 -0.06539 0.960168 0.864506 Western Area Urban 0.413181 -0.02067 1.059287 0.991409 Overidentification test for covariate balance Chi-square 7.904167 P-value 0.894259 Source: Authors’ calculations from CIMS (2020) 74 Pooled ECT-R1 and ECT-R2 - Inability to buy staple foods Standardized Standardized differences- Variance Variance Ratio- differences- Raw Weighted Ratio- Raw Weighted Female 0.273111 0.184681 0.940942 0.956988 Age 0.112318 0.087058 0.677163 0.693254 Household dependency ratio 0.477583 0.097305 1.558982 0.843397 Household head 0.014814 -0.09048 1.003843 1.025577 Completed Secondary School -0.205 -0.10764 0.828928 0.907788 Main income source: Business 0.578807 0.039674 0.687818 0.991156 Main income source: Wage/Salary -0.49227 -0.06195 0.4638 0.936657 Asset score -0.52741 -0.1024 0.439546 0.683311 Bombali -0.01795 0.070634 0.973481 1.141096 Kenema -0.17326 -0.08697 0.657662 0.816611 Port Loko -0.05307 0.094355 0.880773 1.234011 Western Area Urban 0.254263 0.020447 1.00754 1.003648 Overidentification test for covariate balance Chi-square 5.716287 P-value 0.955872 Source: Authors’ calculations from CIMS (2020) 75