Evaluation of the Hunger Safety Net Programme Phase 3: COVID-19 Cash Transfer 2020-2021 - Kenya

Abstract
---------------------------

To support the urban poor during the COVID-19 crisis, the UK government provided a monthly Cash Transfer (CT) of 4,000 Kenyan Shillings (KSH) (or £27) to approximately 52,000 vulnerable people living in informal settlements in Nairobi and Mombasa over a period of three months. The COVID-19 CT was implemented by a consortium led by GiveDirectly, and the monthly stipend was paid using mobile money transfers, with the first transfers taking place from October 2020. The CT was designed to support beneficiaries to buy food or meet other high-priority needs-such as purchasing water, paying for medical care, or making rent payments as well as to reduce the use of negative coping strategies (e.g., selling assets, borrowing money).

OPM was contracted to conduct the monitoring and evaluation of the COVID-19 CT. The main objective of this evaluation was to determine whether, and to what extent, the emergency COVID-19 CT had a positive effect on its target population in informal urban settlements in Kenya. The evaluation also provided an assessment of the implementation parameters and mechanisms adopted as part of the design and delivery of the COVID-19 CT.

To fulfil these aims, the evaluation was structured around two separate components-an impact evaluation and a process review-and drew on multiple research methods through a mixed methods research framework. The objective of the quantitative impact evaluation was to assess whether the COVID-19 CT has had an impact on its beneficiaries, and to quantify the scale of any effect detected. This estimation of impact was based on a longitudinal non-experimental design, focusing on a panel of beneficiaries interviewed at three points in time (baseline - prior to the intervention, midline, and endline - post-intervention) over the course of the implementation period. All quantitative data collection took place remotely using Computer-Assisted Telephone Interview (CATI) software.

Geographic coverage
---------------------------

Nairobi and Mombasa in Kenya

Analysis unit
---------------------------

Individuals
Households

Universe
---------------------------

The study population consists of individuals included in the lists of enrolled beneficiaries covered by Give Directly for the COVID-19 CT.

Sampling procedure
---------------------------

The evaluation team implemented a stratified one-stage probability sampling strategy for the selection of survey respondents from the individuals included in the lists covered by Give Directly for the COVID-19 CT. The goal was to select at baseline a sample of 1,000 eligible individuals who would receive the COVID-19 CT, which would then be interviewed by the evaluation team at baseline, midline, and endline.

The sampling strategy considered the following process:

1) The sample was drawn once the COVID-19 CT beneficiaries were considered as enrolled into the intervention. After discussions with Give Directly, it was decided that an individual was considered a future COVID-19 CT recipient when he/she had responded to the short SMS-based survey delivered by Give Directly.

2) The sample was drawn in two separate batches. The first batch of recipients comprised 6,838 vulnerable individuals from informal settlements in Nairobi, while the second batch contained 1,596 vulnerable individuals from Mombasa. We sampled the same number of beneficiaries from the first and second batches (500 individuals from each batch).

3) Explicit stratification was first applied based on the geographical location of the COVID-19 CT recipient. This entailed that we sample 500 individuals from Nairobi from the first batch, and 500 from Mombasa from the second batch. This allowed us to disaggregate our quantitative findings between Nairobi and Mombasa, and produce informative descriptive and regression analyses for each of the two cities included in the intervention.

4) Implicit stratification was then applied based on the following categorical variables: i) local partner from which the eligible beneficiary was selected, and ii) gender of the COVID-19 CT recipient. The goal of this stratification process was to enhance the representativeness of our sample in terms of these variables, so that our evaluation sample resembled as much as possible the distribution of these characteristics in the target population (i.e. the list of beneficiaries of the COVID-19 CT used as sampling frame for our sample).

5) We did not cluster our survey respondents. Apart from spill-over effect issues, which were not a concern due to the lack of a counterfactual in our methodological approach, this is normally a logistical necessity for in-person surveys. This was not an issue either, given the remote nature of the data collection process.

6) Extensive replacement lists were created to maximise efficiency during survey implementation without sacrificing representativeness of the sample. A detailed replacement protocol was elaborated, which took into account the stratification process described above.
Given the longitudinal nature of the evaluation, the same baseline respondents were tracked and re-interviewed at midline and endline so as to create a panel of survey respondents.The final baseline quantitative survey sample achievement is shown below, including the distribution by county

Sample achievement
Baseline Survey
Nairobi 500
Mombasa 500
Total 1,000

Midline Survey
Nairobi 483
Mombasa 489
Total 972

Endline Survey
Nairobi 463
Mombasa 478
Total 941

Mode of data collection
---------------------------

Computer Assisted Telephone Interviewing (CATI)

摘要
---------------------------

为应对COVID-19疫情危机中城市贫困人口的需求，英国政府向居住在内罗毕和蒙巴萨非正式定居点的约52,000名弱势群体每月提供4,000肯尼亚先令（KSH）（约合27英镑）的现金转账（CT），为期三个月。该COVID-19 CT由GiveDirectly领导的财团实施，每月津贴通过移动货币转账支付，首笔转账于2020年10月开始。CT旨在支持受益人购买食品或满足其他高优先级需求，如购买水、支付医疗费用或缴纳租金，同时减少负面应对策略的使用（例如，出售资产，借款）。

监测与评估（OPM）被委托对COVID-19 CT进行监测与评估。本次评估的主要目标是确定紧急COVID-19 CT在肯尼亚非正式城市定居点目标人群中的正面影响及其程度。评估还提供了对设计及实施COVID-19 CT所采纳的实施参数和机制的评估。

为达成上述目标，评估围绕两个独立部分——影响评估和流程审查——展开，并采用混合方法研究框架中的多种研究方法。定量影响评估的目标是评估COVID-19 CT是否对受益人产生了影响，并量化所检测到的任何效果规模。该影响评估基于纵向非实验设计，关注在实施期间的三次访谈（基线-干预前，中期，以及干预后）中接受访谈的受益人群体。

地理覆盖范围
---------------------------

肯尼亚内罗毕和蒙巴萨

分析单元
---------------------------

个人
家庭

总体
---------------------------

本研究人群包括GiveDirectly为COVID-19 CT所覆盖的受益人名单中的个人。

抽样程序
---------------------------

评估团队对GiveDirectly为COVID-19 CT所覆盖的个人名单中的调查受访者实施了分层单阶段概率抽样策略。目标是基线时选取1,000名合格的受益人样本，他们将成为COVID-19 CT的接收者，并将在基线、中期和终期由评估团队进行访谈。

抽样策略考虑了以下过程：

1) 在考虑COVID-19 CT受益人作为干预措施注册后，从受益人名单中抽取样本。在与GiveDirectly讨论后，决定当个人对GiveDirectly通过短信进行的简短调查做出回应时，该个人被视为未来的COVID-19 CT接收者。

2) 样本分为两个独立的批次。第一批次的接收者包括来自内罗毕非正式定居点的6,838名脆弱个体，而第二批次包含来自蒙巴萨的1,596名脆弱个体。从第一批和第二批中抽取相同数量的受益人（每批500人）。

3) 首先基于COVID-19 CT接收者的地理位置进行了明确的分层。这意味着我们从第一批中抽取500名内罗毕的个人，从第二批中抽取500名蒙巴萨的个人。这使我们能够在内罗毕和蒙巴萨之间分解我们的定量发现，并为干预措施中包含的两个城市提供有信息性的描述性和回归分析。

4) 然后基于以下分类变量进行了隐含分层：i)选择合格受益人的本地合作伙伴，以及ii)COVID-19 CT接收者的性别。该分层过程的目的是提高我们样本在这些变量方面的代表性，以便我们的评估样本尽可能与目标人群（即作为样本抽样框架的COVID-19 CT受益人名单）中的这些特征分布相一致。

5) 我们没有对调查受访者进行聚类。除了溢出效应问题（由于我们的方法论方法中没有对照实验，因此这不是问题）之外，这通常是在现场调查中的物流必要性。鉴于数据收集过程的远程性质，这也不是问题。

6) 创建了详尽的替换名单，以在实施调查时最大化效率，同时不牺牲样本的代表性。制定了一个详细的替换方案，考虑了上述分层过程。

鉴于评估的纵向性质，相同的基线受访者被追踪并在中期和终期重新访谈，以创建调查受访者群体。以下显示了基线定量调查样本的实现情况，包括按县分布的情况。

样本实现
基线调查
内罗毕 500
蒙巴萨 500
总计 1,000

中期调查
内罗毕 483
蒙巴萨 489
总计 972

终期调查
内罗毕 463
蒙巴萨 478
总计 941

数据收集方式
---------------------------

计算机辅助电话访谈（CATI）”}