Impact Evaluation of the FAO’s Cash(+) Project in Mali, 2017 - Mali

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

The dataset contains 17 modules of a household survey conducted between October and November 2017 to assess the impacts of the Cash plus (+) project in the Nioro Circle in the region of Kayes in Mali. The data look specifically at several outcome and output indicators related to the following areas: food and non-food expenditures, food insecurity, livestock production, farm and non-farm activities, dietary intake, housing condition, durable assets, decision making, and behavioural characteristics such as risk preference, locus of control, hope and aspirations.

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

Regional Coverage

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

Households

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

The survey covers the beneficiaries of the Cash+ project and the non-beneficiaries as part of a control group.

Kind of data
---------------------------

Sample survey data [ssd]

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

The sample contains 1151 households, among which 336 received Cash only, 344 received Cash plus (+), and 471 non-beneficiary households forming the comparison group. The beneficiaries were vulnerable households considered as "poor" and "very poor" according to a community targeting based on the Household Economy Analysis framework. The households in the control group are in neighbouring villages which did not receive the intervention and were chosen according to the same criteria for selecting the project's beneficiaries. In total, the impact evaluation dataset covers 58 villages, in which 34 are beneficiary and 25 are comparison villages.

Using this dataset, the impact evaluation study adopts a post-intervention single-difference approach. It distinguishes three treatment arms: households receiving Cash only (T1), households receiving Cash plus + (T2), and non-beneficiary households constituting the comparison/control group (C). This design allows calculating three types of impacts:
• The impact of the cash provided by the project by comparing the outcome for group T1 with the outcome of group C.
• The combined impact of the cash and the livelihood support given by the project, by comparing the outcome of group T2 with the outcome of group C.
• The different impacts of the two types of intervention, by comparing the outcome of group T2 with the outcome of group T1.

Having a non-experimental design, the project did not apply a random process to select its beneficiaries. They were chosen following a specific set of poverty and vulnerability criteria. Therefore, it is important to make sure that the treatment groups and the control groups are comparable. The impact evaluation started with estimating the probabilities of treatment, then applied their inversed values to the regressions in order to predict treatment-specific outcomes.

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

Computer Assisted Personal Interview [capi]

Cleaning operations
---------------------------

Data was cleaned and edited by the data provider. The datasets were made anonymous, by removing sensitive fields, such as names and surnames, GPS coordinates, village names, to avoid identification of respondents.

摘要
---------------------------
本数据集收录了2017年10月至11月间在马里凯耶斯地区尼约罗圈进行的 household survey 的17个模块，旨在评估 Cash plus (+) 项目在该地区的影响。数据特别关注以下领域的多个结果和输出指标：食品和非食品支出、粮食不安全、牲畜生产、农业和非农业活动、膳食摄入、住房条件、耐用资产、决策制定以及风险偏好、控制点、希望和抱负等行为特征。
地理覆盖范围
---------------------------
区域覆盖
分析单元
---------------------------
家庭
总体
---------------------------
调查涵盖了 Cash+ 项目的受益者和非受益者，作为对照组的一部分。
数据类型
---------------------------
样本调查数据 [ssd]
抽样程序
---------------------------
样本包含1151户家庭，其中336户仅接受现金，344户接受 Cash plus (+)，471户非受益户作为比较组。受益者是根据基于家庭经济分析框架的社区定位，被认定为“贫困”和“极度贫困”的脆弱家庭。对照组的家庭位于未接受干预的邻近村庄，其选择标准与项目受益者的选择标准相同。总体而言，影响评估数据集覆盖了58个村庄，其中34个为受益村，25个为比较村。
利用本数据集，影响评估研究采用干预后单一差异法。它区分了三个处理组：仅接受现金的家庭（T1）、接受 Cash plus + 的家庭（T2）以及构成比较/控制组的非受益户（C）。这种设计允许计算三种类型的影响：
• 通过比较T1组和C组的结果，评估项目提供的现金的影响。
• 通过比较T2组和C组的结果，评估现金和项目提供的生计支持的综合影响。
• 通过比较T2组和T1组的结果，评估两种干预类型的不同影响。
由于采用非实验设计，项目未采用随机过程选择受益者。受益者的选择遵循一系列特定的贫困和脆弱性标准。因此，确保处理组和控制组具有可比性至关重要。影响评估首先估计处理概率，然后应用其倒数值到回归中，以预测特定处理的结果。
数据收集方式
---------------------------
计算机辅助个人访谈 [capi]
数据清理操作
---------------------------
数据由数据提供方进行清理和编辑。数据集通过移除敏感字段，如姓名、姓氏、GPS坐标、村庄名称等，进行了匿名处理，以避免识别受访者。