Baseline Study of the Resilience 
Food Security Activity in Ethiopia
March 2023 | Volume I
IMPEL | Implementer-Led Evaluation & Learning Associate Award
ABOUT IMPEL
The Implementer-Led Evaluation & Learning (IMPEL) Associate Award works to improve the design and implementation of 
Bureau for Humanitarian Assistance-(BHA)-funded resilience food security activities (RFSAs) through implementer-led 
evaluations and knowledge sharing. Funded by the United States Agency for International Development (USAID) BHA, IMPEL 
will gather information and knowledge in order to measure performance of RFSAs, strengthen accountability, and improve 
guidance and policy. This information will help the food security community of practice and USAID to design projects and 
modify existing projects to bolster performance, efficiency, and effectiveness. IMPEL is a seven-year activity (2019—2026) 
implemented by Save the Children (lead), TANGO International, Tulane University, Causal Design, Innovations for Poverty 
Action, and International Food Policy Research Institute (IFPRI). 
RECOMMENDED CITATION
IMPEL. (2023). Baseline Study of the Ifaa Resilience Food Security Activity in Ethiopia (Vol. I). Washington, DC: The Implementer￾Led Evaluation & Learning Associate Award. 
PHOTO CREDITS
Melikte Tadesse / Catholic Relief Services Ethiopia
DISCLAIMER
This report is made possible by the generous support of the American people through the United States Agency for 
International Development (USAID). The contents are the responsibility of the Implementer-Led Evaluation & Learning (IMPEL) 
award and do not necessarily reflect the views of USAID or the United States Government.
CONTACT INFORMATION
IMPEL Associate Award
c/o Save the Children
899 North Capitol Street NE, Suite #900
Washington, DC 20002
www.fsnnetwork.org/IMPEL 
IMPEL@fsnnetwork.org 
PREPARED BY:
Baseline Study of the Ifaa RFSA in Ethiopia (Vol. I)
Acknowledgments i 
ACKNOWLEDGMENTS 
The Causal Design research team would like to recognize several individuals who were instrumental in 
creating and refining this document. This includes Shane Lennon, Gregory Makibila with Ifaa, and 
Mohamoud Ahmed, Corena Sharp, and Tom Spangler with IMPEL. Special thanks also to Mara Mordini, 
Takawira Kapikinyu, Hilary Cook, Benita O’Colmain, Arif Rashid, and Adam Trowbridge with USAID for 
their ongoing technical support.
IMPEL | Implementer-Led Evaluation and Learning
ii Table of Contents
TABLE OF CONTENTS
Acknowledgments ...............................................................................................i
List of Tables......................................................................................................iv
List of Figures.....................................................................................................vi
Acronyms .......................................................................................................... vii
Executive Summary ........................................................................................... ix
1. Introduction ................................................................................................... 1
2. Ifaa Activities and Services ............................................................................. 3
2.1 Activities Under the Impact Evaluation .......................................................................................... 3 
3. Methodology and Limitations ........................................................................ 6
3.1 Evaluation Design ........................................................................................................................... 6 
3.1.1 Randomization Strategy ...................................................................................................... 6 
3.1.2 Sampling Strategy ............................................................................................................... 6 
3.1.3 Questionnaire Development ............................................................................................ 10 
3.2 Baseline Data Collection ............................................................................................................... 10 
3.3 Challenges and Limitations .......................................................................................................... 12 
3.3.1 Survey Challenges ............................................................................................................. 12 
3.3.2 Limitations ........................................................................................................................ 12 
3.4 Characteristics of the Study Population ....................................................................................... 13 
3.5 Food Security ................................................................................................................................ 18 
3.6 Child Nutrition and Feeding Practices .......................................................................................... 21 
3.7 Women’s Health, Maternal Nutrition, and Reproductive Health ................................................ 24 
3.8 Water, Sanitation, and Hygiene Practices .................................................................................... 29 
3.9 Agriculture .................................................................................................................................... 33 
3.9.1 Crops ................................................................................................................................. 35 
3.9.2 Livestock ........................................................................................................................... 37 
3.10 Poverty Measurement ................................................................................................................. 39 
3.11 Gender Dynamics ......................................................................................................................... 42 
3.11.1 Use of Financial Resources ............................................................................................... 42 
3.11.2 Credit................................................................................................................................. 44 
3.12 Resilience...................................................................................................................................... 45 
3.12.1 Ability to Recover from Shocks and Stresses Index .......................................................... 46 
3.12.2 Absorptive Capacity Index ................................................................................................ 49 
3.12.3 Social Capital Index ........................................................................................................... 52 
3.12.4 Adaptive Capacity Index ................................................................................................... 54 
3.12.5 Transformative Capacity Index ......................................................................................... 58 
Baseline Study of the Ifaa RFSA in Ethiopia (Vol. I)
Table of Contents iii 
4. Comparison of Treatment and Control Groups.............................................63
4.1 Characteristics of the Study Population ....................................................................................... 63 
4.2 Food Security ................................................................................................................................ 65 
4.3 Child Nutrition and Feeding Practices .......................................................................................... 66 
4.4 Women’s Health, Maternal Nutrition, and Reproductive Health ................................................ 68 
4.5 Water, Sanitation, and Hygiene Practices .................................................................................... 69 
4.6 Agriculture .................................................................................................................................... 70 
4.7 Poverty Measurement ................................................................................................................. 71 
4.8 Gender Dynamics ......................................................................................................................... 72 
4.8.1 Use of Financial Resources ............................................................................................... 72 
4.8.2 Credit................................................................................................................................. 73 
4.9 Resilience...................................................................................................................................... 74 
4.9.1 Ability to Recover from Shocks and Stresses Index .......................................................... 74 
4.9.2 Absorptive Capacity Index ................................................................................................ 74 
4.9.3 Social Capital Index ........................................................................................................... 75 
4.9.4 Adaptive Capacity index.................................................................................................... 76 
4.9.5 Transformative Capacity Index ......................................................................................... 77 
5. Comparison of Livelihood and Non-Livelihood Kebeles ................................ 79
5.1 Characteristics of the Study Population ....................................................................................... 83 
5.2 Food Security................................................................................................................................ 85 
5.3 Child Nutrition and Feeding Practices .......................................................................................... 86 
5.4 Women’s Health, Maternal Nutrition and Reproductive Health ................................................. 87 
5.5 Water, Sanitation and Hygiene Practices ..................................................................................... 89 
5.6 Agriculture .................................................................................................................................... 90 
5.7 Poverty Measurement ................................................................................................................. 91 
5.8 Gender Dynamics ......................................................................................................................... 92 
5.8.1 Use of Financial Resources ............................................................................................... 92 
5.8.2 Credit................................................................................................................................. 92 
5.9 Resilience...................................................................................................................................... 93 
5.9.1 Ability to Recover from Shocks and Stresses Index .......................................................... 93 
5.9.2 Absorptive Capacity Index ................................................................................................ 94 
5.9.3 Social Capital Index ........................................................................................................... 95 
5.9.4 Adaptive Capacity Index ................................................................................................... 95 
5.9.5 Transformative Capacity Index ......................................................................................... 96 
6. Conclusion ................................................................................................... 98
7. References ................................................................................................. 100
VOLUME II 
Annex A: Pre-Analysis Plan 
Annex B: List of Interventions 
Annex C: List of Kebeles with Treatment 
Assignment and Livelihood Status 
IMPEL | Implementer-Led Evaluation and Learning
iv List of Tables and Figures
LIST OF TABLES 
Table 1. List of interventions across the two evaluation packages .............................................................. 4 
Table 2. Sampling strategy ............................................................................................................................ 9 
Table 3. Number of households per woreda .............................................................................................. 11 
Table 4. Household replacement cases ...................................................................................................... 11 
Table 5. Kebele replacement cases ............................................................................................................. 12 
Table 6. Individual-level sample characteristics ......................................................................................... 13 
Table 7. Household-level sample characteristics ........................................................................................ 15 
Table 8. Food insecurity experience scale .................................................................................................. 19 
Table 9. Food consumption score ............................................................................................................... 20 
Table 10. Number of days a household consumed food groups ................................................................ 20 
Table 11. Infant and young children feeding practices ............................................................................... 22 
Table 12. Food groups consumed by children 6–23 months in the last 24 hours ...................................... 22 
Table 13. Young children (0–59 months) diarrhea incidence and treatment ............................................. 23 
Table 14. Indicators for women's nutrition ................................................................................................ 24 
Table 15. GWRA consumption .................................................................................................................... 25 
Table 16. Indicators for reproductive health .............................................................................................. 27 
Table 17. Water indicators .......................................................................................................................... 29 
Table 18 Sanitation and hygiene indicators ................................................................................................ 30 
Table 19. Financial services, value chain activities, and natural resource management practices ............ 34 
Table 20. Main crops grown ....................................................................................................................... 36 
Table 21. Improved management practices or technologies in target crops ............................................. 36 
Table 22. Improved management practices or technologies in target animals ......................................... 38 
Table 23. Poverty measurements ............................................................................................................... 40 
Table 24. Gender (cash) indicators ............................................................................................................. 42 
Table 25. Access to credit and group participation indicators, by gender ................................................. 44 
Table 26. Ability to recover from shocks and stresses index ...................................................................... 46 
Table 27. Absorptive capacity index and related indexes .......................................................................... 49 
Table 28. Social capital index and related indexes ..................................................................................... 52 
Table 29. Adaptive capacity index and related indexes .............................................................................. 54 
Table 30. Transformative capacity index and indexes ................................................................................ 59 
Table 31. Individual-level sample characteristics ....................................................................................... 63 
Table 32. Household-level sample characteristics ...................................................................................... 64 
Table 33. Food Insecurity Experience Scale ................................................................................................ 65 
Table 34. Food consumption score ............................................................................................................. 66 
Table 35. Infant and young children feeding practices ............................................................................... 66 
Table 36. Food groups consumed by children (6–23 months) in the last 24 hours ................................... 66 
Table 37. Young children (0–59 months) diarrhea incidence and treatment ............................................. 67 
Table 38. Indicators for women's health, maternal nutrition, and reproductive health ........................... 68 
Table 39. Woman of reproductive age’s consumption............................................................................... 68 
Table 40. Water, sanitation, and hygiene indicators .................................................................................. 69 
Baseline Study of the Ifaa RFSA in Ethiopia (Vol. I)
List of Tables and Figures v 
Table 41. Improved management practices or technologies in target animals .........................................71
Table 42. Poverty measurement indicators ................................................................................................ 71 
Table 43. Gender (cash) indicators ............................................................................................................. 72 
Table 44. Absorptive capacity index and subindexes ................................................................................. 74 
Table 45. Social capital index and subindexes ............................................................................................ 75 
Table 46. Adaptive capacity index and indexes .......................................................................................... 76 
Table 47. Transformative capacity index and indexes ................................................................................ 77 
Table 48. Characteristics of the study population, variables with a significant difference across livelihood 
status ........................................................................................................................................................... 80 
Table 49. Child nutrition and feeding practices, variables with a significant difference across livelihood 
status ........................................................................................................................................................... 81 
Table 50. GWRA age consumption, variables with a significant difference across livelihood status ......... 82 
Table 51. Gender dynamics, variables with a significant difference across livelihood status .................... 82 
Table 52. Individual-level sample characteristics across livelihood status ................................................. 83 
Table 53. Household-level sample characteristics across livelihood status ............................................... 84 
Table 54. Food insecurity experience scale across livelihood status .......................................................... 85 
Table 55. Food consumption score across livelihood status ...................................................................... 85 
Table 56. Infant and young children feeding practices across livelihood status ........................................ 86 
Table 57. Food groups consumed by children (6–23 months) in the last 24 hours across livelihood status
 .................................................................................................................................................................... 86 
Table 58. Young children diarrhea incidence and treatment (children 0–59 months) across livelihood 
status ........................................................................................................................................................... 87 
Table 59. Indicators for women's health, maternal nutrition, and reproductive health across livelihood 
status ........................................................................................................................................................... 87 
Table 60. Woman of reproductive age’s consumption across livelihood status ........................................ 88 
Table 61. Water, Sanitation, and Hygiene Indicators across livelihood status ........................................... 89 
Table 62. Financial services, value chain activities and natural resource management practices across 
livelihood status .......................................................................................................................................... 90 
Table 63. Improved management practices or technologies in target crops across livelihood status ...... 90 
Table 64. Improved management practices or technologies in target animals across livelihood status ... 91 
Table 65. Poverty measurement indicators across livelihood status ......................................................... 91 
Table 66. Gender cash indicators across livelihood status ......................................................................... 92 
Table 67. Gender access to credit and group participation indicators across livelihood status ................ 92
Table 68. Ability to recover from shocks and stresses index across livelihood status ............................... 93 
Table 69. Absorptive capacity index and subindexes across livelihood status ........................................... 94 
Table 70. Social capital index and subindexes across livelihood status ..................................................... 95 
Table 71. Adaptive capacity index and indexes across livelihood status .................................................... 95 
Table 72. Transformative capacity index and indexes across livelihood status ......................................... 96 
 
IMPEL | Implementer-Led Evaluation and Learning
vi List of Tables and Figures
LIST OF FIGURES 
Figure 1. Selection of 120 kebeles part of the IE .......................................................................................... 7 
Figure 2. Map of the Ifaa woredas and kebeles participating in the Impact Evaluation .............................. 8 
Figure 3. Age and gender composition of the sample ................................................................................ 14 
Figure 4. Selected household- and individual-level sample characteristics, by woreda ............................ 16 
Figure 5. Severe food insecurity, by woreda .............................................................................................. 19 
Figure 6. Poor food consumption score, by woreda ................................................................................... 21 
Figure 7. Minimum acceptable diet, by woreda ......................................................................................... 23 
Figure 8. Percentage of children under 5 who had diarrhea in the prior 2 weeks, by woreda .................. 24 
Figure 9. Dietary diversity score for GWRA ................................................................................................ 25 
Figure 10. Percentage of women with a minimum dietary diversity score, by woreda ............................. 26 
Figure 11. Contraceptive prevalence rate and number of contraceptive methods women in a union 
know, by woreda ......................................................................................................................................... 28 
Figure 12. Percentage of households with access to water services, by woreda ....................................... 31 
Figure 13. Use of improved sanitation facilities and open defecation by woreda ..................................... 32 
Figure 14. Percentage of households practicing correct use of recommended household water treatment 
technologies, by woreda ............................................................................................................................. 33 
Figure 15. Percent of farming households who used financial services in the past 12 months, by woreda
 .................................................................................................................................................................... 35 
Figure 16. Share of farming households using different value chain activities .......................................... 35 
Figure 17. Percent of farming households raising goats, by woreda.......................................................... 39 
Figure 18. Distribution of log food consumption expenditure per adult equivalent per day in purchasing 
power parity USD ........................................................................................................................................ 41 
Figure 19. Poverty rate and depth of poverty per adult equivalent measure, by woreda ......................... 41 
Figure 20. Men/women in a union who reported earning cash, by woreda .............................................. 43 
Figure 21. Share of men and women in a union with access to credit, by woreda .................................... 45 
Figure 22. Most common shocks reported by households ......................................................................... 47 
Figure 23. Mean ability to recover from shocks and stresses, by woreda ................................................. 47 
Figure 24. Shock indicators ......................................................................................................................... 48 
Figure 25. Absorptive capacity index, by woreda ....................................................................................... 50 
Figure 26. Bonding social capital index, shock preparedness and responsiveness index and access to 
humanitarian assistance index, by woreda ................................................................................................ 51 
Figure 27. Bonding and bridging indexes, by woreda ................................................................................. 53 
Figure 28. Adjusted adaptive capacity index, by woreda ........................................................................... 57 
Figure 29. Education/training index by woreda.......................................................................................... 57 
Figure 30. Social capital related indexes, by woreda .................................................................................. 58 
Figure 31. Adjusted transformative capacity index, by woreda ................................................................. 61 
Figure 32. Access to agricultural services index, by woreda....................................................................... 62 
Baseline Study of the Ifaa RFSA in Ethiopia (Vol. I)
Acronyms vii 
ACRONYMS
ANC Antenatal Care
BHA Bureau for Humanitarian Assistance
CAPI Computer-Assisted Personal Interviews
CMAM Community-Based Management of Acute Malnutrition
CPR Contraceptive Prevalence Rate
CRS Catholic Relief Services
DFSA1 Development Food Security Activity 1
DS Direct Support
ESMF Environment and Social Management Framework
FAO Food and Agriculture Organization of the United Nations
FCS Food Consumption Score 
FDP Food Distribution Points
FIES Food Insecurity Experience Scale
FMNR Farmer Managed Natural Regeneration
FSTF Food Security Task Force 
GBV Gender-Based Violence 
GoE Government of Ethiopia
GWRA Girls and Women of Reproductive Age (15-49)
HEW Health Extension Workers
HH Household Head 
IE Impact Evaluation 
IGA Income Generation Activities
IMPEL Implementer-Led Evaluation and Learning Associate Award
IYCF Infant and Young Child Feeding 
LF Lead Farmer 
MAD Minimum Acceptable Diet
MDD Minimum Dietary Diversity
NGO Non-Governmental Organization
NRM Natural Resource Management
ORT Oral Rehydration Therapy
PIM Program Implementation Manual
PLW Pregnant and Lactating Women 
PPP Purchasing Power Parity 
PSNP Productive Safety Net Programme 
PSNP5 Productive Safety Net Programme Phase 5
IMPEL | Implementer-Led Evaluation and Learning
viii Acronyms
PW Public Work
RCT Randomized Controlled Trial 
RFSA Resilience Food Security Activity
RL Religious Leader
SBC Social and Behavioral Change
SPIR Strengthen PSNP Institutions and Resilience
TDS Temporary Direct Support
USAID United States Agency for International Development
USD United States Dollar
WASH Water, Sanitation, and Hygiene
WASHCO WASH Community Organization
WRA Woman of Reproductive Age
Baseline Study of the Ifaa RFSA in Ethiopia (Vol. I)
Executive Summary ix
EXECUTIVE SUMMARY
This report captures baseline round data collected from May to July 2022 of the Ifaa resilience food 
security activity (RFSA) implemented by Catholic Relief Services (CRS) in the region of Oromia in 
Ethiopia. Oromia is Ethiopia’s largest, most populous region and one of its poorest, with complex and 
interlinked causes of poverty and food insecurity. The goal of the Ifaa RFSA is to improve the food 
security of vulnerable households in targeted Productive Safety Net Programme (PSNP) communities, 
contributing to a sustained reduction in rural poverty in Oromia. Per the goals and objectives of reducing 
food insecurity and poverty, Ifaa aims to provide a range of support interventions to more than 60,000 
households. These interventions aim to strengthen and improve government services; agriculture and 
livelihood opportunities; health and nutrition; water, sanitation, and hygiene (WASH); gender and youth 
empowerment; and natural resource management and the environment. 
Under the Implementer-Led Evaluation and Learning (IMPEL) Associate Award funded by the United 
States Agency for International Development (USAID) Bureau for Humanitarian Assistance (BHA), Causal 
Design is conducting an Impact Evaluation (IE) of the Ifaa RFSA using baseline and endline survey data in 
the target areas in Oromia, Ethiopia. The IE relies on a clustered randomized controlled trial (RCT) design 
to estimate its impact on study indicators that include food security, child nutrition, and health; 
women’s maternal nutrition; WASH practice; agricultural practice and production; poverty 
measurement; gender dynamics; and resilience. In the case of the Ifaa RFSA IE, the interventions 
allowed for randomization at the kebele level (the smallest administrative unit in Ethiopia) to create a 
set of treatment and control kebeles. The treatment kebeles will receive the Ifaa Enhanced Package of 
interventions, while the control kebeles will receive the Basic PSNP package of interventions. As a result, 
the analysis will focus on the effect of the Ifaa Enhanced Package for PSNP Phase 5 (PSNP5) households 
on food security and other development outcomes. Overall, analysis at the baseline suggests that the IE 
is well-placed to estimate these differences at the endline, and treatment and control groups have 
similar key characteristics. 
Baseline Survey Sample 
Baseline data were collected in 120 study kebeles across eight woredas in Oromia. Households included 
in the sample were PSNP5 households with a girl or woman 15–49 years old. The findings presented in 
this report are representative of this study population, which is the target population for Ifaa activities, 
but poorer than the general population. By the end of data collection efforts, 4,683 households were 
surveyed, 1,950 were in control kebeles, and 2,733 were in treatment kebeles. 
Key Findings
Demographic Profiles 
The average household size in the sample is six. On average, a household has 0.83 children under the 
age of 5. The level of schooling in the sample is low: around 26% of the adult population has at least 
some schooling, and this share drops to 13.7% among household heads. Regarding work-related 
activities, around 66% of adults are farmers, and 31.4% of people 10 years or older did any work and 
were paid in cash in the last 12 months. 
 
IMPEL | Implementer-Led Evaluation and Learning
x Executive Summary
Food Security
Food security was estimated using two standard measurement approaches, the Food Insecurity 
Experience Scale (FIES) and the Food Consumption Score (FCS) index. Based on the FIES, 77.6% of the 
population faces severe food insecurity, and over 96% are at least moderately food insecure. The FCS, 
which calculates overall consumption levels across food groups while accounting for cultural and 
regional weights for food preference and importance, finds that fewer than 7.8% of all households have 
an acceptable FCS, and almost 70% of households are considered to have a poor FCS. This reflects that 
households mainly consume staples daily and rarely consume meat, fish, milk, and dairy. Disaggregation 
by woreda (the third level of the administrative divisions of Ethiopia) shows large variability: while less 
than 40% of the households have a poor consumption score in Jarso, the share is more than 90% for 
Melka Belo. 
Child Nutrition and Health 
The overall diet quality for children 6 to 23 months appears to be poor across the survey population. 
Less than 2% of all children 6 to 23 months met minimum dietary diversity (MDD) criteria or minimum 
acceptable diet (MAD) standards. The low rates of dietary diversity drive the low rates of MAD. Most 
children consume either breastmilk or grains, roots, and tubers. Very few consume other food groups. 
Approximately 11% of children under 5 are reported to have experienced diarrhea within 2 weeks of the 
survey. Most children who experienced diarrhea were treated with oral rehydration therapy (ORT). 
Women’s Health, Maternal Nutrition, and Reproductive Health 
Observations around the health and reproductive decisions of girls and women of reproductive age 
(GWRA) among surveyed households suggest very poor food consumption diversity and low levels of 
access to health personnel during pregnancy. Altogether, less than 1% of women consume a diet that 
meets the MDD criteria. More than 80% of women consume two or fewer food groups, and the most 
common food groups are grain, roots, and other vitamin-A-rich fruits and vegetables. On average, only 
15.7% of women receive the recommended number of antenatal care (ANC) visits—at least four—
during their most recent pregnancies, with more than 64% of these ANC visits being with nurses or 
midwives. Additionally, less than 10% of women in a union report using a modern method of birth 
control. 
Water, Sanitation, and Hygiene Practices 
Based on indicator criteria, 13% of all households have access to basic drinking water services. While 
67.5% of households have water sources available all year round, only 30% have water sources within 30 
minutes of their residence. Regarding treatment, sanitation, and hygiene practice, less than 3% of 
households were observed to have handwashing facilities available in the home, and less than 8% use at 
least one form of evidence-based household water treatment technologies. Most households (59%) 
practice open defecation, and 16% have household-level improved sanitation facilities. 
Agriculture 
More than 90% of households with an available plot of land engage in raising crops. Very few 
households grow haricot beans, groundnuts, or potatoes—the reported crops of focus. The most 
common crops are maize and sorghum, each cultivated by 73% and 56% of households growing at least 
Baseline Study of the Ifaa RFSA in Ethiopia (Vol. I)
Executive Summary xi
one crop. Regarding other support practices, nearly 90% of households use at least one natural resource 
management (NRM) practice. A high share of households (76%) reported using at least one value chain 
activity. The most common one was purchasing inputs for crops (used by 65% of households). Less than 
19% of households used financial services. 
The baseline livestock focus included goats, poultry, and oxen. 23.5% of households raised goats, 19.6% 
raised poultry, and 6.4% raised oxen. On average, around 83% of the households raising these animals 
reported them in good or normal condition. In addition, most households raising these animals reported 
having used at least one improvement or NRM practice. 
Poverty Measurement 
Based on the daily per-adult equivalent expenditure of less than $1.90 per day (2011 purchasing power 
parity (PPP)), more than 92% of the surveyed households are considered poor. The depth of poverty of 
the poor is 54%, which means that the average poor person is 54% below the poverty line. In monetary 
terms, it would require an additional $1 per person per day to bring every poor person out of poverty. 
Gender Dynamics 
Around 23% of men and women in a union reported earning cash in the past 12 months. Among women 
earning cash, a majority (70%) reported that they participate in decisions about how to use self-earned 
cash, while 44% reported they participate in decisions about how to use their partner’s self-earned cash. 
Around 59% of people in a union are members of a community group, with more men (66%) than 
women (51%) belonging to a community group. In terms of access to credit, around 29% of men have 
access to credit, while the share of women with access to credit is slightly lower (26%). Of those with 
access to credit, around 30% make decisions about credit, with very similar rates for men and women. 
Resilience 
Researchers captured resilience indicators through several questions, including indices constructed to 
assess overall resilience capacities. Generally, households perceive their ability to meet their current 
needs as worse than the previous year and suspect that their future ability to meet these needs will 
deteriorate. The most common shock reported by households was drought or little rain (89%). The 
second most common shock was increased food prices (71%). Out of the average 3.6 shocks 
experienced across the sample, households perceived them to be severe, likely impacting the perceived 
ability to recover. Regarding the capacity to absorb shocks, very few households (less than 2%) have 
access to financial resources or crop insurance to absorb shocks. Another way to mitigate the impact of 
shocks is to draw on social networks. In this case, the social capacity index for the sample was 50 (out of 
100), which means that the average household could give and receive help from 1.5 out of three groups 
of people. 
This page is left intentionally blank.
Baseline Study of the Ifaa RFSA in Ethiopia (Vol. I)
Introduction 1 
1. INTRODUCTION
Since its inception in 2005, the Productive Safety Net Program (PSNP) has been a cornerstone of the 
Ethiopian government’s strategy for poverty alleviation, disaster risk management, and rural 
development. The PSNP provides food or cash transfers targeted to poor households in the form of 
payments for seasonal labor on public works (PW) or as direct support (DS) to households whose 
primary income earners are elderly or disabled. The PSNP has played an important role in improving the 
lives of poor Ethiopian households by reducing household food insecurity, increasing asset holdings, and 
improving agricultural productivity (Berhane et al. 2014; Hoddinott et al. 2017). PSNP has evolved 
through several phases, the fifth phase of the Productive Safety Net Program (PSNP5) began operating 
in 2021. 
Under the Implementer-Led Evaluation and Learning (IMPEL) Associate Award funded by the United 
States Agency for International Development (USAID) Bureau for Humanitarian Assistance (BHA), Causal 
Design is conducting an impact evaluation (IE) of the Ifaa resilience food security activity (RFSA) in the 
region of Oromia in Ethiopia. Catholic Relief Services (CRS) implements Ifaa with the main goal of Ifaa of 
improving the food security of vulnerable households in targeted PSNP communities to contribute to a 
sustained reduction in rural poverty. Ifaa aims to provide a range of support interventions to more than 
60,000 households. The IE uses a cluster randomized control trial (RCT) design that randomly selects the 
kebeles that will receive the Ifaa Enhanced Package of interventions and the kebeles that will receive 
the PSNP Basic Package of interventions. This design allows the IE to answer the following research 
question: 
What is the impact of the Ifaa Enhanced Package of interventions for PSNP5 
households compared to the PSNP Basic Package on reducing food insecurity, 
nutrition, and other related outcomes?1
This is an important research result since CRS considers the Enhanced Package’s interventions to lead to 
significant improvements for the PSNP households. The evaluation seeks to inform the larger 
conversation around the efficacy of RFSA interventions among vulnerable populations. 
This report summarizes the results of the baseline study conducted in May–July 2022. The baseline data 
were collected in 120 IE study kebeles across eight woredas in Oromia. Households included in the 
sample were PSNP5 households with a girl or woman2
aged 15–49 years. Given that the sample was 
selected from such a specific target population, neither the study nor its conclusions can be considered 
representative of all RFSA households. Therefore, estimates represent PSNP households with a girl or 
woman aged 15–49 years. However, as a Strengthen PSNP Institutions and Resilience (SPIR) baseline 
study found, more than 80% of PSNP households have a woman of reproductive age (WRA). Thus, it is 1
The reader can consult the pre-analysis plan of the impact evaluation (IMPEL 2022) for more details. 2 Because of the focus on households with a girl or WRA, throughout the text, we will use the term women to refer to girls and 
women aged 15–49 years. It is important to highlight that under Ethiopian law, any individual under 18 cannot be considered 
an adult or defined as a “woman” or “man.” 
IMPEL | Implementer-Led Evaluation and Learning
2 Introduction
likely that the study is largely representative of PSNP households. In addition, Causal Design has worked 
closely with BHA and relevant stakeholders to identify key learning objectives and ensure that the 
baseline survey and study can contribute to this learning where possible. 
The baseline report is organized as follows: Section 2 presents an overall description of the Ifaa
interventions and discusses the characteristics of the two packages that will be the subject of the IE 
study. Section 3 provides an overview of the evaluation design and the baseline report and discusses 
some of the challenges and limitations of the study. Section 4 summarizes the baseline survey data and 
the balance test comparing the RCT’s two arms. Section 5 includes a similar exercise comparing 
livelihood and non-livelihood kebeles. Finally, Section 6 presents the conclusions of the report. 
Baseline Study of the Ifaa RFSA in Ethiopia (Vol. I)
Ifaa Activities and Services 3 
2. ACTIVITIES AND SERVICES 
Per the goals and objectives of reducing food insecurity and promoting wellbeing and welfare among 
participant kebeles, the Ifaa RFSA aims to provide a range of support interventions to more than 60,000 
households. These interventions aim to strengthen and improve government services; agriculture and 
livelihood opportunities; health and nutrition; water, sanitation, and hygiene (WASH); gender and youth 
empowerment; and natural resource management (NRM) and the environment. 
To improve the food security of vulnerable households in targeted PSNP communities and contribute to 
a sustained reduction in rural poverty in Oromia, Ifaa interventions are organized around three main 
purposes: 
1. Vulnerable households and individuals have sufficient quantity, quality, and diversity of food at 
all times. 
2. Vulnerable community members’ livelihoods are transformed. 
3. PSNP Systems deliver accountable, effective, and shock-responsive services. 
Ifaa will target nine woredas in the region of Oromia. These woredas were selected based on poverty 
level, the potential for scaling up the PSNP, the potential for the continuum of the response with other 
programs, the presence of partners, and the opportunity for impact. The nine woredas include four that 
were part of the previous Development Food Security Activity 1 (DFSA1) program implemented by CRS 
and five newly added woredas. The Ifaa activities and services will prioritize pregnant and lactating 
women (PLW) to increase the likelihood that children under 5 receive appropriate nutrition, the activity 
reaches vulnerable community members, and behavioral changes lead to a lasting change. 
2.1 Activities Under the Impact Evaluation 
Of the nine woredas and 241 kebeles in which CRS has implemented Ifaa, the IE will take place in a 
subset of 120 kebeles in eight woredas.3
These kebeles were randomly allocated to one of two 
packages:4
 PSNP Basic (control) and Ifaa Enhanced (treatment). Table 1 shows a subset of the 
interventions and how they are mapped to these two evaluation packages.5
The Basic PSNP package is a 
set of interventions selected and implemented by the Government of Ethiopia (GoE) to support its most 
vulnerable populations. The Ifaa Enhanced Package is the most comprehensive and contains all the 
interventions in the PSNP Basic Package, plus additional interventions under each domain. Each kebele 
that is part of the evaluation study will receive the interventions associated with only one of the two 
evaluation packages. The interventions under livelihoods are special because only a subset of the 
kebeles is eligible6
 to receive those interventions. To account for this, the IE study will include two 
groups of kebeles: (i) those eligible to receive livelihood interventions and (ii) those that are ineligible. 3
Section 3.1.2.1 discusses in more detail the selection of the 120 kebeles, and Table 71 in Annex C contains a list of the 120 
kebeles with their treatment and livelihood status. 4
Section 3.1 provides more detail on the randomization and sampling strategy. 5
Table 60 in Annex B contains the complete list of interventions. 6
The criteria used by CRS to target livelihood kebeles included the following factors: (i) access to markets, (ii) road access, (iii) 
access to financial service providers, (iv) access to agricultural extension services (public and private), (v) access to functional 
small-scale irrigation, (vi) potential for crop and livestock production, (vii) access to other services (public work, university, and 
agricultural research demonstration plots), (viii) kebele potential for Ifaa sectoral integration and (ix) PSNP caseload. 
IMPEL | Implementer-Led Evaluation and Learning
4 Ifaa Activities and Services 
Causal Design will take advantage of the layering approach and compare kebeles receiving the Ifaa
Enhanced Package against kebeles receiving the PSNP Basic Package. Because of the differentiation 
between livelihood-eligible and livelihood-non-eligible kebeles, we are effectively making two 
comparisons. Comparison 1 is livelihood kebeles receiving PSNP Basic (including livelihood interventions) 
against livelihood kebeles receiving the Ifaa Enhanced Package. Comparison 2 is non-livelihood kebeles 
receiving PSNP Basic (excluding livelihood interventions) against non-livelihood kebeles receiving the 
Ifaa Enhanced Package without livelihood interventions. 
Table 1. List of interventions across the two evaluation packages7
Interventions Basic
PSNP
Ifaa
Enhanced
Livelihoods
Saving group X X
Financial literacy training X X
Support and training in business plan development X X 
Credit guarantee fund (conditional capacity building) X
Value chain financing co-investment X 
Youth fund ($250) X 
Gender youth and social dynamics
Implementation of Program Implementation Manual (PIM) gender provisions: 
Monitoring implementation of PIM gender provisions (e.g., exemption of PLW) 
until the child is 2 years) 
X X 
Monitoring implementation of gender-based violence (GBV) action plan included 
in the PSNP5 (e.g., the program grievance redress mechanism supporting to 
accept appeals related to GBV and refer to locally available GBV response 
services) 
X X 
Leadership training for women and youth in leadership positions X 
Community conversations for adults and youth X
PSNP systems
Provision of three food components (wheat, oil, and pulse) to Permanent Direct 
Support and Public Work (PW) clients to meet the daily food kilo/calorie 
requirement 
X X 
Food security task force (FSTF) capacity building (woreda, kebele, community 
levels) 
X X 
FSTF capacity building, specifically focusing on leadership & communication 
essential training, in addition to the basic FSTF capacity building 
X 7 Readers interested in the complete list of interventions, as well as a short description summary of each one of them, can 
consult Annex B. 
Baseline Study of the Ifaa RFSA in Ethiopia (Vol. I)
Ifaa Activities and Services 5 
Interventions Basic
PSNP
Ifaa
Enhanced
Private sector engagement: transportation of food from Primary Distribution 
Points to Food Distribution Points (FDP), construction of SEIs (same as PSNP 
Basic)
X
Health and nutrition 
Linkage to services (ensuring transfers for temporary direct support (TDS), 
supporting TDS PW clients to attend PW Social and Behavioral Change (SBC) 
sessions, and other health and nutrition services like antenatal care (ANC), 
growth monitoring, immunizations, etc.) 
X X
Capacity building training for government and partner staffs: adolescent 
nutrition, Community-based management of Acute Malnutrition (CMAM), 
Religious Leaders (RLs) materials.
X X
GoE basic health extension program X X 
Additional health extension programs (includes remote trainings, lead parents, 
motivation of health development armies) 
X
System strengthening through the capacity building and provision of materials 
(referral pads, formats, reg. books, SBC materials, etc.)
X
Environment and natural resource management
Training woreda GoE on equitable allocation and disbursement of resources for 
PSNP plan implementation 
X X
Watershed management planning X X 
Implementation of the Environment and Social Management Framework (ESMF) X X
Integrated Water Resources Management: water supply, risk management, and 
water benefits calculator 
X 
Farmer Managed Natural Regeneration (FMNR): land restoration technique—
introducing FMNR approach, organizing user groups, and leveraging with 
agroforestry practices 
X 
WASH
Water development, monitoring, and governance X X 
Community-led total sanitation and hygiene X 
Market-based sanitation and hygiene X 
IMPEL | Implementer-Led Evaluation and Learning
6 Methodology and Limitations 
3. METHODOLOGY AND LIMITATIONS 
The IE of the Ifaa activity uses a cluster RCT design at the kebele level, a small administrative unit in 
Ethiopia. Causal Design conducted a baseline survey of households in May–July 2022. The endline survey 
is planned for the second quarter of 2025. Below we present an overview of the evaluation design, a 
description of the baseline data collection activities, and the challenges and limitations of the study. A 
more detailed description of the methodology can be found in the Pre-Analysis Plan. 
3.1 Evaluation Design
The evaluation will use a clustered RCT design, which randomizes the selection of kebeles that receive 
the Ifaa Enhanced Package and the kebeles that receive the PSNP Basic Package in the Oromia region of 
Ethiopia. The group of kebeles receiving the Ifaa Enhanced Package will be referred to as the “treatment 
group,” while the set receiving the PSNP Basic Package will be referred to as the “control group.” The 
two evaluation packages were proposed by the implementing partner. 
3.1.1 Randomization Strategy 
The research team worked with CRS to randomize the Ifaa Enhanced and PSNP Basic interventions 
rollout across 120 kebeles. Based on power-level calculations and implementor-specified targets, the 
research team randomly allocated 50 kebeles to the control group and the remaining 70 to the 
treatment group. Half of the control kebeles (25) are eligible to receive livelihood interventions, while 
the other half are not eligible to receive those interventions. Something similar happens for the 
treatment kebeles: half (35) are eligible to receive livelihood interventions, while the other half are not. 
The research team used a stratified randomization approach to guarantee a better balance. This 
technique involves dividing the sample of kebeles into groups sharing similar characteristics. Based on 
the information provided by CRS, the evaluation team used two strata: woredas8
 and livelihood 
eligibility. Kebeles within a given woreda and livelihood eligibility status were randomly allocated to 
treatment and control. This approach ensures that treatment and control groups are balanced by the 
strata used. To conduct the randomization, pseudo-random numbers generated by Stata9
 were used. 
3.1.2 Sampling Strategy 
3.1.2.1 Kebele Level 
Ifaa will implement interventions in nine woredas (Ethiopian districts), made up of 241 kebeles. CRS 
identified Meta, one of the nine woredas selected, as a “big push plus” woreda, which will receive 
additional interventions. As these additional interventions could not be randomized, Meta was not 
selected to be a part of the IE study. Thus, only eight of the nine woredas are participating in the IE. 
Eleven kebeles within these eight woredas were also not selected to be a part of the (IE) study due to 8
The woreda was used as the stratum of randomization for two main reasons: (i) it will allow providing a balance of treatment 
assignment geographically, and (2) the woreda is the main administrative structure for local government which shapes local 
public expenditure and public service delivery. 9
The command used in Stata was runiform. 
Baseline Study of the Ifaa RFSA in Ethiopia (Vol. I)
Methodology and Limitations 7
persistent security issues, making it difficult for CRS to implement interventions effectively. Additionally, 
11 kebeles were purposefully selected to receive integrated watershed management+10 and thus were 
not included in the IE study. Out of the remaining 196 kebeles, 120 were randomly selected for the 
study.
Figure 1 presents a diagram illustrating the selection of the 120 kebeles for the IE study, while Figure 2
shows a map with the location of these kebeles and their respective woredas.11 As was discussed in 
previous sections, only half of the 120 kebeles in the IE study are eligible to receive livelihood 
interventions. Within each group, kebeles will be randomly assigned to control or treatment (resulting in 
four groups of kebeles). Twenty-five livelihood kebeles assigned to the control group will receive all the 
Basic PSNP interventions (including the livelihood ones), while the remaining 35 livelihood kebeles in the 
treatment group will receive the Ifaa Enhanced Package. For the kebeles not eligible to receive 
livelihood interventions, 25 control kebeles will receive the Basic PSNP interventions but without 
livelihood interventions, while the remaining 35 treatment kebeles will receive all the Ifaa Enhanced 
Package but without livelihood interventions.
Figure 1. Selection of 120 kebeles part of the IE
10 Since these kebeles were selected based on specific characteristics and not randomly selected, it is impossible to construct a 
valid control group. 11 Babile, Deder, Melka Belo, Gursum, Fedis, Midgega Tola, Chinaksen, and Jarso. All 120 kebeles can be found in Table 23 in 
Annex C.
Ifaa Kebeles
241
Livelihood IE
60
PSNP Basic
with Basic 
Livelihood
Intervention 
Packages:
Control
25
Ifaa Enhanced
with Enhanced 
Livelihood
Ifaa Enhanced
NO Livelihood
PSNP Basic
NO Livelihood
Random 
selection
Non-random 
selection
120 IE 
Kebeles
Livelihood
98
Not Selected
38
Non￾Livelihood IE
60
Control
25
Treated
35
Treated
35
Non￾Livelihood
98
Not Selected
38
Eligible for IE study
196
Excluded
45
Random 
selection
IMPEL | Implementer-Led Evaluation and Learning
8 Methodology and Limitations
Figure 2. Map of the Ifaa woredas and kebeles participating in the Impact Evaluation
3.1.2.2 Household Level
Based on power calculations,12 the research team decided to survey 39 households per kebele in each of 
the 120 kebeles, for a total of 4,683 households. Out of these, 1,950 households are located in control 
kebeles, while the remaining 2,730 are in treatment kebeles.
Household inclusion criteria. To select the sample of households for their inclusion in the survey, we 
used two inclusion criteria: households must be PSNP participants and must have at least one girl or 
woman of reproductive age (GWRA), which is defined as girls and women 15–49 years old.13 The first 
criterion was used because CRS targets their package of interventions to PSNP households. The second 
criterion was motivated by the fact that adolescent girls and GWRA are also targets of many Ifaa
activities, in particular, those related to nutrition and gender, and thus we expect to benefit the most
from these activities. Moreover, many BHA indicators are specific to GWRA and their young children, so 
by sampling GWRA, we increase the likelihood that the household will have a child under 5.
This strategy primarily excludes elderly households and is justified by the following reasons: (i) most 
PSNP households have at least one WRA,14 and (ii) because of the nature of the Ifaa Enhanced Package, 
elderly households are not likely to be substantially affected by the Enhanced Package. Most 
interventions, such as those related to nutrition, youth, and livelihoods, target households with children, 
youth, or working-age members. In addition, many nutrition and health outcomes are specific to young 12 The power calculation exercises can be found in Vol. II Section 2.3.1 “Power Calculations” of this baseline report. 13While the term GWRA is used to define girls and women aged 15–49, under Ethiopian law, any individual under the age of 18 
cannot be considered an adult nor defined as a “woman” or “man.” 14 Based on authors’ calculations from a PSNP4 dataset, 82.5% of PSNP households had a woman of reproductive age. 
Baseline Study of the Ifaa RFSA in Ethiopia (Vol. I)
Methodology and Limitations 9 
children or GWRA. Accordingly, the sample selected will represent the households predominantly 
targeted by the PSNP and RFSA interventions. 
Individual selection within the household. The research team selected randomly or purposively15 one 
target individual for each outcome16 instead of interviewing every eligible individual. The primary reason 
for this choice is that interviewing every eligible individual in the household is very time-consuming and 
costly for the household and for data collection. In addition, we do not need more than one individual 
per household for power reasons, and outcomes would generally be highly correlated within 
households. Thus, the additional information provided is limited. Typically, interviewing multiple 
individuals per household is useful if the objective is to compare outcomes across individuals in the 
same household: for example, in polygamous households comparing outcomes between first and 
second wives.17 However, this is not part of the IE design, and thus it is not worth the additional costs. 
Given the previous considerations, Table 2 provides more detail about the sampling strategy. 
Table 2. Sampling strategy 
Module Sampling choice
D: Children’s nutritional status and 
feeding practices18
Prevalence of exclusive breastfeeding (children 0–5 months): 
randomly sample one child in this age range. 
Children’s feeding practices and diets (children 6–23 
months): randomly sample one child in this age range. 
Children’s diarrhea (children 0–59 months): randomly 
sample one child in this age range. 
E: Women’s health, nutritional status, 
dietary diversity, and family planning
Randomly sample one woman 15–49 years old
G: Agriculture Select the person most informed about agriculture 
production in the household19
J: Gender (cash) Select adult most knowledgeable about household affairs 
and spouse20
K: Gender access to credit and group 
participation 
Select adult most knowledgeable about household affairs 
and spouse21
15 This selection was used when we wanted to target one person more knowledgeable about the set of questions. 16 The random household selection processes were embedded in the survey tool. When a random member of the household 
needs to be selected, the survey tool runs a randomization algorithm using the household roster. 17 For polygamous households, we randomly selected one wife from available wives. 18 Causal Design randomly selected children 0–5 months, 0–23 months and 0–59. This selection will be done independently, so 
that the same child could be selected twice (e.g., a child 8 months old could be selected for age bracket 6–23 months and age 
bracket 0–59 months). 19 This selection was done by asking the HH. To address issues around ownership and control, the survey contains follow-up 
questions on specific individuals involved in different activities. 20 In the case that the HH is not married or in a union, we did not ask the questions related to a couple. In the case of 
polygamous households, we randomly selected from available wives. 21 In the case that the HH is not married or in a union, we did not ask the questions related to a couple. In the case of 
polygamous households, we randomly selected from available wives. 
IMPEL | Implementer-Led Evaluation and Learning
10 Methodology and Limitations 
Sampling Frame
To sample the households, we used a sample frame of PSNP households provided by CRS. The limitation 
of this dataset is that no information on the age of the household members was provided. For each 
kebele, PSNP households were randomly ranked,22 and the first 39 households were selected to be 
surveyed. Enumerators went to these households, and if a household did not have a woman of 
reproductive age, the survey was immediately terminated. Then, the next ranked household was 
included in the list to be surveyed. This happened with 38 out of the 4,683 households. 
3.1.3 Questionnaire Development
The baseline survey was developed with previous baseline surveys used by BHA and refined in 
consultation with BHA and the implementing partners. 
The following survey modules were included: 
Module A: Household identification 
Module B: Roster 
Module C: Food access 
Module D: Child nutrition and health 
Module E: Women’s nutrition and health 
Module F: Water, sanitation, and hygiene 
Module G: Agriculture 
Module H: Household expenditure 
Module J: Gender and cash use 
Module K: Gender and credit 
Module R: Resilience 
3.2 Baseline Data Collection 
The data collection activities occurred during May and June 2022, coinciding with the lean season. 
Researchers conducted the training during the first 6 days of the data collection activities, and a 2-day 
practical piloting at the field level followed the training. Teams conducted the pilot testing in Woreda 
Kimbibit in the North Shewa zone of the Oromia region. 
For the data collection activities, 30 enumerators (eight women and 22 men) and six supervisors were 
deployed. The enumerators were divided into six teams. Each team was assigned a supervisor and was 
assigned to survey 1–2 woredas. To ensure quality data, the teams performed a series of activities 
during and after data collection. The data quality assurance undertaken during the data collection 
included: office-level phone call-backs and a 10% back-check on the interviewed households. 
Additionally, the research team undertook high-frequency checks regularly. 
Sample Surveyed 
The field data collection was expected to cover 4,683 households in 120 kebeles in eight woredas. Table 
3 shows the number of households surveyed in each woreda. 22 Within a kebele each household will be assigned a random number. Households will be ranked by this number. 
Baseline Study of the Ifaa RFSA in Ethiopia (Vol. I)
Methodology and Limitations 11 
Table 3. Number of households per woreda
Woreda Number of households
Babile 546 
Chinaksan 819 
Deder 706
Fedis 429
Gursum 740
Jarso 507 
Melka Belo 507 
Midega Tola 429
Total 4,683
During the survey activities, the enumerators faced issues surveying individual households and whole 
kebeles. In the first case, there were 126 households from the initial list that could not be surveyed. All 
126 households with issues were successfully replaced with valid households.23 Table 2 groups these 
households according to the reason for replacement. More than half of the replaced households had 
either migrated from the area or did not satisfy the inclusion criterion of having a woman of 
reproductive age. 
Table 4. Household replacement cases 
Reason for replacement Number of households replaced 
Migration 45
No eligible woman 15–49 39
Unknown in kebele 17
Not available after different visits 12
Duplicated household 9
Death of household member 4
Total 126
In the second case, the field team experienced issues accessing some kebeles because of the 
topographical characteristics of the area and poor road infrastructure. The field team used different 
strategies (e.g., traveling by foot or spending the night in the area of study) to overcome these problems 
with the result that only two kebeles (listed in Table 5) needed to be replaced. 
Both kebeles used as replacements were assigned the same treatment status as the initial kebeles. Ibbsa 
(Gursum) was replaced by a livelihood kebele in Deder. This kebele was selected because it was located 
in one of two woredas where data collection was still happening. In addition, we used the random 23 The replacement was done following the random rank assigned to the households in the sample frame. 
IMPEL | Implementer-Led Evaluation and Learning
12 Methodology and Limitations 
number generated initially to select the livelihood kebeles that would be part of the study. In the case of 
the second kebele, Hosale (Chinaksan), we chose one of the three kebeles that had been initially 
surveyed but randomly excluded when the design had to be adjusted. 
Table 5. Kebele replacement cases 
Initial kebele Replacement kebele Reason 
Ibbsa, Gursum Hake Bas, Deder No sample frame available
Hosale, Chinaksan Buyo Negeya, Gursum Security Issues
3.3 Challenges and Limitations
3.3.1 Survey Challenges 
The data collection activities took more time than initially expected because of the topographical 
conditions and the poor quality of the roads. In addition, some changes to the IE design were introduced 
at the beginning of the data collection activities,24 which required excluding some of the kebeles initially 
selected and including new ones. 
The characteristics of the sample frame also led to survey challenges. The sample frame provided by CRS 
did not have information about household members’ ages, household location information was 
imprecise in many cases, and there were duplicated households. We addressed these issues by working 
with local authorities and using replacement households. 
The final set of challenges related to the data collection tool. During data collection activities, we 
identified several issues related to the Computer-Assisted Personal Interviews (CAPI). These issues were 
promptly corrected but led to a loss of information in some modules.25 In future survey rounds, we will 
allocate additional time for the survey company and Causal Design staff to quality-check the CAPI and 
make plans for a more extensive field pilot. 
3.3.2 Limitations
The IE is characterized by several limitations, including: 
There are several large development organizations working in the region. Randomization of 
kebeles should ensure that kebeles are balanced in terms of other activities in the area. 
However, this could present challenges in identifying the effects of the Ifaa activity if 
interventions are concentrated in a certain part of the region. Therefore, during the study, CRS 
will collect information at the kebele level about the activities of other development 
organizations. 
An additional challenge is related to the permanence of the program beneficiaries. Because of 
the characteristics of the PSNP program, households might graduate from the program, and new 24 These changes were introduced after CRS specified that only some of the Ifaa kebeles would be receiving any type of 
livelihood activities. Therefore, to guarantee the validity of the IE study, the research team excluded some of the kebeles 
initially selected and included new ones to have the same number of kebeles receiving and not receiving livelihood activities. 25 Most of the issues were associated with coding errors that led to questions being skipped. Section 4 discusses these issues 
and their implications in terms of sample size. 
Baseline Study of the Ifaa RFSA in Ethiopia (Vol. I)
Methodology and Limitations 13 
households might enter the program. To account for this, the sample calculations accounted for 
the possibility of attrition. 
An additional challenge is tracking households over time for the impact evaluation. In general, 
household migration rates are not expected to be high, and therefore the evaluation team 
anticipates that the majority of baseline households can be surveyed during the endline. In 
addition, the evaluation design has accounted for attrition, a pattern in which households 
observed at baseline are not observed at the endline. 
The survey was conducted during the COVID-19 pandemic, and anthropometric measures were 
not collected at baseline to maintain distance between enumerators and respondents and to 
minimize contact. Pending the improvement of the COVID-19 pandemic, these measures will be 
collected at the endline. 
This section summarizes the characteristics of the sample and the main outcome variables. The section 
comprises nine subsections, each corresponding to a module in the baseline survey. Each subsection will 
present the main variables for the entire sample and discuss the main differences across woredas.26
Balance test comparing treatment and control, as well as livelihood and non-livelihood kebeles, will be 
presented in Sections 5 and 6. In each table, the mean is presented in the “All” column, with the 
standard deviation represented by the figure in parentheses found below. Column “N” shows the total 
number of each group. 
3.4 Characteristics of the Study Population 
This section provides basic demographic information for the baseline sample. In each household, a 
roster was administered to collect information on each household member. Across the 4,683 
households, there were 27,869 individuals. Table 6 presents some of the individual-level characteristics. 
Roughly 50% of the household members were female, and around 23% were GWRA. Figure 3 shows the 
sample population by age and gender. The average age is 19, and about 14% of the sample are children 
under 5. Among those 15 years old and older, 26% have at least some schooling, and 66% engage in 
some type of farming or livestock activity. For those over 10 years old, 31% worked for cash in the 
previous year. 
Table 6. Individual-level sample characteristics 
Outcome N All
Average age 
27,869 19.16
(16.13)
Percent of children under 5 years old 
27,869 14.03
(34.73)
Percent of children 5–14 years old 
27,869 38.96
(48.77)
Percent of adults (over 15 years old) 27,869 47.01
26 In the interest of space, we present the main indicators for the entire sample and focus on the most important differences 
across woredas. Annex B contains the complete list of indicators for the entire sample. 
IMPEL | Implementer-Led Evaluation and Learning
14 Methodology and Limitations
Outcome N All
(49.91)
Percent of females
27,869 49.96
(50.00)
Percent of GWRA
27,869 22.77
(41.94)
Percent of adults who are married or in a union
13,102 57.2
(49.48)
Percent of adults with any schooling
13,102 26.35
(44.05)
Percent of adults who are farmers
13,098 66.19
(47.31)
Percent of people older than 10 years who did any work and were 
paid in cash
13,478 31.45
(46.43)
Notes: Estimates from the RFSA Ifaa baseline survey sample. Standard deviations are in parentheses.
Figure 3. Age and gender composition of the sample
Table 7 provides basic household-level information about the sample. The average household size is 
close to six individuals, and the average number of children under 5 is 0.83. A large share (77%) of the 
household heads (HHs) is married or in a union, and close to 25% are female. Interestingly, the share of 
HHs with schooling (13.5%) is low compared to the overall sample of adults (26.35%). Furthermore, the 
difference in schooling between male and female HHs is substantial; while around 17% of male HHs 
have some schooling, only 2.7% of female HHs do. 
Baseline Study of the Ifaa RFSA in Ethiopia (Vol. I)
Methodology and Limitations 15 
Table 7. Household-level sample characteristics
Outcome N All
Average household size
4,683 5.95
(2.16)
Average number of children under the age of 5 in the household 
4,683 0.83
(0.94)
Average age of HH
4,683 41.77
(10.97)
Percent of HHs who are married or in a union
4,683 77.19
(41.96)
Percent of female headed households 
4,683 25.43
(43.55)
Percent of HH with some schooling 
4,683 13.5
(34.17)
Percent of female HHs with some schooling 
1,191 2.69
(16.18)
Percent of male HHs with some schooling 
3,492 17.18
(37.73)
Notes: Estimates from the RFSA Ifaa baseline survey sample. Standard deviations are in parentheses. 
Figure 3 presents some variables disaggregated at the woreda level to show the variability across 
woreda. There is some variability in the share of adults with any schooling. For example, in Chinaksen 
and Fedis, the share hovers around 19%, while in woredas like Jarso or Gursum, the share is over 30%. A 
similar variability can be observed when we look at the schooling of HHs, even though, in this case 
Gursum, stands out with over 20% of HHs with schooling. 
The graphs below examine work-related variables. There is a very large variability in the share of people 
older than 10 who worked and were paid in cash. The difference between the woreda with the lowest 
share (Gursum, 20%) and the highest (Melka Belo, 57%) is almost 37 percentage points, which is very 
close to the average share in the whole sample (31%). There is not a clear correlation between this 
variable and the share of workers that are farmers. There are woredas like Melka Belo, which has both a 
high share of people working for cash and a high share of farmers. There are also woredas like Fedis, 
which has one of the lowest shares of people working for cash and the highest share of farmers. 
IMPEL | Implementer-Led Evaluation and Learning
16 Methodology and Limitations 
Figure 4. Selected household- and individual-level sample characteristics, by woreda
Baseline Study of the Ifaa RFSA in Ethiopia (Vol. I)
Methodology and Limitations 17 
IMPEL | Implementer-Led Evaluation and Learning
18 Methodology and Limitations 
3.5 Food Security 
Researchers assessed household food insecurity using the Food Insecurity Experience Scale (FIES) 
created by the Food and Agriculture Organization of the United Nations (FAO). The scale consists of 
eight questions on a household’s difficulty accessing food due to a lack of money or other resources in 
the last 30 days27 (FAO, 2016). There are binary indicators for each question. If a respondent answered 
“yes,” the question would be allocated 1, and then questions would be summed up to create a raw 
score on a scale of 0–8. To assess the quality of the data collected, Causal Design used the FAO’s Rasch 
model.28 After applying the model, researchers obtained infit statistics for each question and a reliability 
score for the model. Infit statistics compare the misfit of each item with the extent of misfit expected if 
the assumptions of the Rasch model are valid (FAO, 2017). Acceptable infit statistics are in the range of 
0.7–1.3. Two questions29 had infit values below 0.7 (0.55 and 0.65), while one question30 had a value of 
1.8. The low infit of the first two questions suggests that these questions may be redundant with other 
items. The high infit of the last question suggests that the question performs poorly.31 The reliability 
score measures the degree to which observed results can be replicated by the Rasch model. The Rasch 
reliability score for this survey was 0.77, which suggests a good model fit. 
Researchers calculated the prevalence of food insecurity using the sum of the binary answers to the 
eight questions, which we will refer to as the raw FIES score. Two indicators are created for the 
proportion of households experiencing moderate to severe food insecurity, categorized as households 
with a raw FIES score of 4 or more, and the proportion experiencing severe food insecurity, which is 
considered households with a raw FIES score of 7 or 8. People experiencing moderate levels of food 
insecurity will typically eat low-quality diets and might have been forced, at times, to also reduce the 
quantity of food they would normally eat, while those experiencing severe levels would have gone for 
entire days without eating due to lack of money or other resources. 
Figure 4 reveals that households in the sample are very food insecure. Nearly all households show some 
indication of food insecurity, with 96.2% having experienced moderate-to-severe food insecurity and 
77.7% experiencing severe food insecurity. These estimates are higher than the average for East Africa, 
where the prevalence of moderate-to-severe food insecurity is 66.9% (FAO, 2022) and reflects that 
households are PSNP clients. Figure 5 displays the data by woreda and reveals that Melka Belo has the 
highest rates of severe food insecurity at almost 100%, while Jarso has the lowest at 60%. 27 As stated in FAO (2016), the reference period is flexible, and thus we used the last 30 days, which is easier for households to 
remember than the last 12 months. 28 The Rasch Model is a logit function that assumes that it gets progressively more difficult to answer “yes” to a subsequent 
question (e.g., it is more likely to answer yes to question 2 than to question 3). Furthermore, respondents answering “yes” to 
more questions will have a more severe trait of food insecurity. (FAO, 2022) 29 The two questions were: question 6, “During the past 30 days, was there a time when your household did not have food 
because of a lack of money or other resources?” and question 7, “During the past 30 days, was there a time when you or others 
in your household were hungry but did not eat because there was not enough money or other resources for food?”. 30 The question was number 2, “During the past 30 days, was there a time when you or others in your household were unable 
to eat healthy and nutritious food because of a lack of money or other resources?” 31 FAO suggests that questions with an infit above 1.3 could be dropped and that work should be done to improve this question 
for future surveys. Here we decided not to drop the questions and to improve them for the endline survey. The survey will 
include additional probes to ensure the respondent has a proper understanding of the question. 
Baseline Study of the Ifaa RFSA in Ethiopia (Vol. I)
Methodology and Limitations 19 
Table 8. Food insecurity experience scale
Outcome N All
Raw FIES score (scale 0–8) 
4,678 6.88
(1.39)
Percent of households with some indication of food insecurity
4,678 99.49
(7.15)
Percent of households that experienced approx. moderate-to-severe food 
insecurity
4,678 96.15
(19.24)
Percent of households that experienced approx. severe food insecurity
4,678 77.66
(41.66)
Percent of households that answered yes to all eight questions
4,678 37.77
(48.49)
Percent of households that answered no to all eight questions
4,678 0.51
(7.15)
Notes: Estimates from the RFSA Ifaa baseline survey sample. Standard deviations are in parentheses.
Figure 5. Severe food insecurity, by woreda
The second indicator under household food access is the Food Consumption Score (FCS). It is a measure 
that reflects food intake in frequency, dietary diversity, and nutritional importance. For each food item, 
enumerators asked households how many days out of the past week they consumed it. Then, 
enumerators group the food items into nine groups: staples, pulses, vegetables, fruit, meat/fish, 
IMPEL | Implementer-Led Evaluation and Learning
20 Methodology and Limitations 
milk/dairies, sugar/honey, oils/fats, and condiments. The number of days the household consumes each 
food group is weighted based on the nutritional value of the food group and then summed up. The FCS 
has a range of 0–112, and researchers separate households into those with poor food consumption 
(values of 0–21), borderline food consumption (values of 22–35), and acceptable food consumption 
(values of 35 or more) (WFP, 2008). The FCS has been found to correlate well with caloric availability at 
the household level (Wiesmann et al., 2009). 
Table 9 reveals that the mean FCS is low at 20.5. Only 7.8% of households have an acceptable score, 
while the majority (69.85%) have a poor score. This reflects that households mainly consume staples 
daily and rarely consume meat, fish, milk, and dairy (Table 10). Melka Belo is the woreda with the 
highest percentage of households with poor consumption, while Jarso has the lowest percentage, which 
is consistent with the FIES score (Figure 6). 
Table 9. Food consumption score 
Outcome N All
FCS (0–112) 
4,601 20.5 
(9.60)
Percent of households with poor consumption score (< 22) 
4,601 69.85 
(45.89) 
Percent of households with borderline consumption score (22–35) 
4,601 22.34 
(41.66) 
Percent of households with acceptable consumption score (> 35) 
4,601 7.8 
(26.82) 
Notes: Estimates from the RFSA Ifaa baseline survey sample. Standard deviations are in parentheses. 
Table 10. Number of days a household consumed food groups 
N All 
Staples
4,601 6.79
(0.92)
Pulses 
4,601 0.32
(1.21)
Vegetables 
4,601 1.62
(2.50)
Fruit 
4,601 0.17
(0.82)
Meat and fish 
4,601 0.15
(0.84)
Baseline Study of the Ifaa RFSA in Ethiopia (Vol. I)
Methodology and Limitations 21 
N All
Milk and dairy
4,601 0.26
(1.18)
Sugar
4,601 1.81
(2.70)
Oil
4,601 3.21
(3.21)
Condiments
4,601 6.25
(1.93)
Notes: Estimates from the RFSA Ifaa baseline survey sample. 
Figure 6. Poor food consumption score, by woreda
3.6 Child Nutrition and Feeding Practices
Infant and young child feeding (IYCF) practices are important determinants of children’s nutritional 
status under 2 years of age. We analyze four core IYCF indicators: (i) the percentage of infants 0–5 
months who are fed exclusively with breast milk the day preceding the survey; (ii) the percentage of 
children 6–23 months who receive a minimum dietary diversity (MDD), defined as consuming foods 
from five or more food groups; (iii) percentage of children 6–23 months who receive minimal meal 
frequency defined as receiving solid, semi-solid, or soft foods the minimum number of times; and (iv) 
the percentage of children 6–23 months who receive a minimum acceptable diet (MAD) which combines 
the MDD and minimum meal food frequency (WHO 2021).
IMPEL | Implementer-Led Evaluation and Learning
22 Methodology and Limitations 
Table 11 reveals low levels of IYCF practices among children less than 2 years old. Less than half 
(39.77%) of children 0–5 months old are exclusively breastfed, and only 1.92% of children 0–23 months 
consume a diet of minimum diversity. While 64.82% of children consumed solid, semi-solid, or soft 
foods the minimum number of times, only 1.65% of children 6–23 months have a MAD. The low rates of 
dietary diversity drive low rates of MAD. Table 12 reveals that most children consume breastmilk32 or 
grains, roots, and tubers; very few consume other food groups. For the MDD to improve, children need 
to increase their consumption of other food groups such as meats, eggs, legumes, nuts, vegetables, and 
fruits. These low rates of MDD and MAD mirror those found in other reports on PSNP clients in Ethiopia 
at 3.2% and 1.2%, respectively (Alderman et al., 2019). Figure 7 reveals that no children receive a MAD 
in four woredas (Fedis, Gursum, Melka Belo, and Midega Tola). 
Table 11. Infant and young children feeding practices 
Outcome N All
Percent of children under 6 months who are exclusively breastfed 8833 39.77
 (49.22)
Percent of children 6–23 months receiving a minimum acceptable diet 72534 1.54
 (12.31)
Percent of children 6–23 months receiving a minimum meal frequency 725 64.82
(47.78)
Percent of children 6–23 months consuming a diet of a minimum diversity 726 1.92
 (13.76)
Notes: Estimates from the RFSA Ifaa baseline survey sample. Standard deviations are in parentheses. 
Table 12. Food groups consumed by children 6–23 months in the last 24 hours 
Outcome N Breastfed N Non￾Breastfed
Percent consuming breastmilk
582 100 144 0
(0.00) (0.00)
Percent consuming grains, roots, tubers 
582 73.02 144 67.36
(44.42) (47.05)
Percent consuming legumes and nuts 
582 2.75 144 4.17
(16.37) (20.05)
Percent consuming dairy products 
582 14.26 144 11.81
(35.00) (32.38) 32 582 out of 726 children 6–23 months old consume breastmilk. 33 Due to a coding error in the Computer-Assisted Personal Interviews (CAPI), not every child under 6 months answered this 
question. Out of 241 children under 6 months, about 37% (88 children) answered the question. 34 There were 901 selected children 6–23 months of age. Due to a CAPI error, 175 of these children were not interviewed. In 
addition, one child in the age range didn’t answer the questions related to the consumption of solid, semi-solid, or soft foods. 
We excluded this child from the minimum acceptable diet and minimum meal frequency indicators. 
Baseline Study of the Ifaa RFSA in Ethiopia (Vol. I)
Methodology and Limitations 23 
Outcome N Breastfed N Non￾Breastfed
Percent consuming meats
582 0.52 144 0.69
(7.17) (8.33)
Percent consuming eggs
582 2.06 144 3.47
(14.22) (18.37)
Percent consuming vitamin-A-rich vegetables and fruits
582 8.76 144 10.42
(28.30) (30.65)
Percent consuming other vegetable and fruits
582 6.36 144 7.64
(24.42) (26.65)
Notes: Estimates from the RFSA Ifaa baseline survey sample. Standard deviations are in parentheses.
Figure 7. Minimum acceptable diet, by woreda
In addition to the IYCF indicators, the percentage of children 0–5 years with diarrhea and the percentage 
of children treated with oral rehydration therapy (ORT) are analyzed in the 2 weeks preceding the 
survey. Table 13 reveals that 11.26% of children reported having diarrhea in the prior 2 weeks, and 
nearly all these children (93.3%) were treated with ORT. Between woredas, there are large differences in 
the percentage of children with diarrhea, with about one in four children in the woredas of Fedis and 
Midega Tola having diarrhea compared to less than 5% in the woreda of Deder (Figure 8).
Table 13. Young children (0–59 months) diarrhea incidence and treatment
Outcome N All
Percent of children under 5 who had diarrhea in the prior 2 weeks 2,167 11.26
IMPEL | Implementer-Led Evaluation and Learning
24 Methodology and Limitations
Outcome N All
(31.62)
Percent of children under 5 who had diarrhea treated with ORT
244 93.03
(25.51)
Notes: Estimates from the RFSA Ifaa baseline survey sample. Standard deviations are in parentheses.
Figure 8. Percentage of children under 5 who had diarrhea in the prior 2 weeks, by woreda
3.7 Women’s Health, Maternal Nutrition, and Reproductive 
Health
This section focuses on the health and reproductive decisions of women of childbearing age. Table 14
includes the dietary diversity and the percentage of GWRA consuming a diet of MDD. The dietary 
diversity score is constructed by counting the number of food groups a woman consumes. The 10 food 
groups are grains (i.e., bread, rice, enjera), white roots, and tubers; pulses (i.e., beans, peas, lentils); nuts 
and seeds; dairy; meat, poultry, and fish; eggs; dark leafy greens (i.e., kale, pumpkin leaves, spinach); 
other vitamin-A-rich fruits and vegetables (i.e., mangos, red palm nut, pumpkin); other vegetables; and 
other fruits. A woman is said to consume a diet of MDD if she consumed during the previous day and 
night at least five of the 10 food groups specified.
Table 14. Indicators for women's nutrition
Outcome N All
Dietary diversity score 4,63535 1.69
35 Forty-five women didn’t answer at least one of the questions related with this indicator.
Baseline Study of the Ifaa RFSA in Ethiopia (Vol. I)
Methodology and Limitations 25 
Outcome N All
(1.01)
BL11: Percent of GWRA consuming a diet of MDD
4,635 0.86
(9.25)
Notes: Estimates from the RFSA Ifaa baseline survey sample. Standard deviations are in parentheses.
Table 15 shows that only about 0.86% of women consumed a diet that met the MDD requirements. 
Figure 9 illustrates the distribution of the MDD score.36 Over 70% of the women ate only one or two
food groups. The high level of food insecurity in the population of interest is additionally evidenced by 
the large number of women (420 or approximately 10% of the sample) who did not consume any of the 
food groups the day before the survey.37 Table 15 shows GWRA’s consumption per food group. The 
most common food group consumed was grains, white roots, and tubers (89%). The next most common 
food groups consumed are vitamin-A-rich fruits and vegetables (48%), other vegetables (17%), dark leafy 
greens (6%), dairy (4%), nuts and seeds (1%), and eggs (1%).
Figure 9. Dietary diversity score for GWRA
Table 15. GWRA consumption
Outcome N All
Percent consuming grains and roots 4,63538 89.06
36 The figure shows the distribution of the MDD score across the full sample of WRA. 37 This striking result is corroborated by information provided in Module C. About 76% of the households with WRA who didn’t 
eat any of the food groups also stated that, during the past 30 days, there was at least one time when one or more members of 
the households went without eating for a whole day. Based on information from the field team, because of the existing drought 
and shortage of food, women give priority to children, and they might consume less or consume food items not included in the 
list (e.g., coffee, tea). Moreover, some households also have a perception of aid dependency where they consider the interview 
will bring some kind of support for them. 38 Forty-five women didn’t answer at least one of the questions related with this indicator.
IMPEL | Implementer-Led Evaluation and Learning
26 Methodology and Limitations
Outcome N All
(31.22)
Percent consuming pulses 4,635 0.78
(8.78)
Percent consuming nuts and seeds 4,635 1.36
(11.58)
Percent consuming dairy 4,635 4.38
(20.47)
Percent consuming meat, poultry, and fish 4,635 0.54
(7.33)
Percent consuming eggs 4,635 1.34
(11.49)
Percent consuming dark leafy greens 4,635 5.89
(23.55)
Percent consuming other vitamin-A-rich fruits and vegetables 4,635 47.81
(49.96)
Percent consuming other vegetables 4,635 17.32
(37.85)
Percent consuming other fruits 4,635 0.95
(9.70)
Notes: Estimates from the RFSA Ifaa baseline survey sample. Standard deviations are in parentheses.
While the dietary diversity score by woreda remained relatively consistent within a +/- 0.5 margin, the 
share of women with an MDD score by woreda had greater variability (Figure 10). In most woredas, only 
between 0.25% and 0.75% of women ate a diet of minimum diversity. Babille and Jarso were exceptions, 
with shares of 1.39 and 2.22. 
Figure 10. Percentage of women with a minimum dietary diversity score, by woreda
Baseline Study of the Ifaa RFSA in Ethiopia (Vol. I)
Methodology and Limitations 27 
The contraceptive prevalence rate (CPR) measures non-pregnant GWRA (15–49) in a union who are 
currently using or whose partners are currently using at least one contraceptive method. Table 16 
displays this rate. Nearly 11% of non-pregnant women between the ages of 15–49 in a union are using 
birth control, with nearly all of them (99%) using a form of modern birth control. Modern birth control 
includes female and male sterilization, injectables (Depo-Provera), intrauterine devices (IUDs), 
contraceptive pills, implants, female and male condoms, diaphragm with spermicide (cream, foam, or 
gel), the standard days method, the lactation amenorrhea method, and emergency contraception. 
Traditional birth control methods include rhythm and withdrawal. Injectables are the most popular form 
of birth control among women using a birth control method (used by 64% of this group of women). 
Table 16. Indicators for reproductive health 
Outcome N All
BL20: CPR, percent of non-pregnant women 15–49 in a union 
2,568 11.37
 (31.75)
BL26: Percent of births receiving at least four ANC visits during pregnancy 
1,622 15.72
 (36.41)
BL36: Percent of women in a union who have knowledge of modern family 
planning methods that can be used to delay or avoid pregnancy39
2,936 29.9
 (45.79)
Number of contraceptive methods woman in a union know (0–12) 
2,936 1.72
 (2.01)
BL37: Percent of women in a union who made decisions about modern family 
planning methods in the past 12 months 
26240 50.38
 (50.09)
Notes: Estimates from the RFSA Ifaa baseline survey sample. Standard deviations are in parentheses. 
Table 16 shows the percentage of GWRA with a live birth in the past 5 years who received at least four 
ANC visits from skilled health personnel during their most recent pregnancies. “Skilled health personnel” 
refers to doctors, nurses, midwives, skilled birth attendants, or clinical officers. Women qualify as having 
received ANC if they have had at least four ANC visits. About 16% of these women met these criteria 
during their most recent pregnancy. 98% of women received ANC from skilled health personnel, with 
more than 64% of these visits with nurses or midwives. 
Table 16 also depicts the percentage of women in a union who know of modern family planning 
methods that can delay or avoid pregnancy. Of the women in a union between ages 15 and 49, 30% 
reported having knowledge of modern family planning methods. The rate at which women in a union 
reported having knowledge of modern family planning methods is similar across the age groups 
between 20–29 (31%) and between 30–49 (30%). The rate of women from the age group of 15–19 is 
lower at 24%. The average number of contraceptive methods married women or in a union know is 1.68. 39 "A woman is counted as having knowledge of modern family planning methods if she can identify at least three modern 
methods of family planning" (BHA, 2021). 40 Following BHA (2021), this question was only asked to GWRA in a union using birth control methods. The reduced sample size 
is explained by the low number of women in a union who used any modern contraceptive method in the past 12 months. 
IMPEL | Implementer-Led Evaluation and Learning
28 Methodology and Limitations 
Of those who reported knowledge of modern family planning methods, 68% knew three methods or 
fewer. 
Table 16 presents findings on decision-making about family planning. The questions used to calculate 
this indicator are the following: “Did you or your partner use any of these modern contraceptive 
methods in the past 12 months?” and “Who usually makes the decision on whether or not you should 
use contraception: you, your husband/partner, you and your husband/partner jointly, or someone 
else?” Of the women who have used modern family planning methods in the past 12 months, almost 
51% report making that decision either by themselves or with their partner jointly. 
Figure 11 presents the variability in family planning knowledge and usage across woredas. Women in a 
union in Gurum, closely followed by Jarso, are far more familiar with the largest number of 
contraceptive methods, while women in Babile know the least number of methods out of all of the 
woredas. Notably, the data suggests that knowing the most contraceptive methods has little correlation 
with CPR. While the women surveyed in Gursum know the most contraceptive methods, Gursum has 
the third lowest CPR. Conversely, Babile, where women know the least contraceptive methods, has the 
second highest CPR. Chinaksan has the third highest number of methods known but significantly 
outranks all other woredas in terms of the CPR. 
Figure 11. Contraceptive prevalence rate and number of contraceptive methods women in a union 
know, by woreda 
 
Baseline Study of the Ifaa RFSA in Ethiopia (Vol. I)
Methodology and Limitations 29 
The data across woredas for women in a union making decisions about modern family planning fall 
within similar levels (hovering around 50%), with the exception of Jarso, where less than 20% of women 
self-reported as decision makers. A similar pattern is evident for ANC. The data was relatively uniform 
across woredas (about 10%) except for Babile, which surpassed all others significantly by 20%. 
3.8 Water, Sanitation, and Hygiene Practices
Water, sanitation, and hygiene (WASH) practices are captured through five indicators in this section. 
This section targeted households rather than individuals, which is reflected in the table below. 
Table 17. Water indicators
Outcome N All
BL16: Percent of households using basic drinking water services41
353 4.25
(20.20)
Percent of households using basic drinking water services based on four of 
five42 criteria
4,670 11.2
(31.54)
Percent of households with water available year-round
4,681 67.49
(46.85)
Percent of households with water available every day in the past 2 weeks 4,681 67.49
41 Due to a skip error, data was only collected for households that used additional filtration to make water potable (353 out of 
4681). 42 The sample size for this indicator is all the households for which four of the five criteria for basic drinking water services can 
be constructed (excludes per person per day production criterion).
IMPEL | Implementer-Led Evaluation and Learning
30 Methodology and Limitations 
Outcome N All 
 (46.85)
Percent of households using an improved drinking water source 
4,681 82.08
(38.36)
Percent of households able to fetch water in 30 minutes or less 
4,681 47.87
 (49.96)
Per capita volume of water a household draws per day
4,670 29.74
 (45.72)
Table 18 Sanitation and hygiene indicators
Outcome N All
BL17: Percent of households with soap and water at a handwashing station 
on the premises 
3,81543 2.88 
(16.74) 
BL18: Percent of households practicing correct use of recommended 
household water treatment technologies 
4,67644 7.36 
(26.11) 
BL19: Percent of households practicing open defecation 
4,681 58.73 
(49.24) 
BL27: Percent of households using improved sanitation facilities (not shared) 
4,681 15.87 
(36.55) 
Notes: Estimates from the RFSA Ifaa baseline survey sample. Standard deviations are in parentheses. 
Indicator 16 measures the percentage of households with access to basic drinking water services. BL16 
uses five criteria to define access to basic drinking water services for households: (i) the water source is 
an improved source, (ii) the total collection time must be 30 minutes or less for a round trip (including 
wait time), (iii) the source consistently produces basic drinking water, (iv) water is available every day in 
the past 2 weeks and (v) the service produces 20 liters per day for each person counted. Following 
USAID (2021b), questions related to the fifth criterion were only answered by those who use additional 
filtration to make the water potable. 
Table 17 shows that of 353 households that answered all five questions, around 6% reported having 
access to basic water services. When we look at households satisfying only four out of five criteria,45 the 
data shows that around 11% have access to basic water services. Furthermore, around 82% used an 43 One of the inputs for this indicator is question F14: “Please show me where members of your household most often wash 
their hands.” We excluded households that either didn’t answer the question, answered “I don’t know,” or refused to answer. 44 One of the inputs for this indicator is question F09: “Do you do anything to the water to make it safer to drink?”. We excluded 
households that either didn’t answer the question, answered “I don’t know,” or refused to answer. 45 Due to a skip error, only a reduced set of households answered the questions associated with producing at least 20 liters per 
person. Therefore, researchers present an alternative indicator to have an idea of the access to basic drinking water for the 
whole sample. 
Baseline Study of the Ifaa RFSA in Ethiopia (Vol. I)
Methodology and Limitations 31 
improved drinking water source,46 67% had access to water all year round, and 48% could fetch water in 
30 minutes or less.
Table 18 shows sanitation and hygiene indicators. Indicator 17 measures the percentage of households 
with soap and water at a handwashing station on the premises. Only about 3% of households were 
observed to have these resources. Indicator 18 illustrates that only about 7% of households practiced 
the correct use of recommended household water treatment technologies,47 while Indicator 19 shows 
that 59% of households practiced open defecation.48 As seen in Indicator 27, less than 16% have 
household-level improved sanitation facilities that are not shared with other families. 
Figure 12 presents the percentage of households with access to water services disaggregated by 
woreda, when the indicator is constructed using four out of five criteria. The figure shows that Gursum, 
followed by Melka Belo and Babile, have the most access to water services, with over 15% of 
participants reporting access. Participants in Midega Tola have the least access, with roughly 1% of 
participants noting access, followed by Fedis at roughly 4% and Deder at 7%. 
Figure 12. Percentage of households with access to water services, by woreda 
The variation in the share of households practicing open defecation across woredas is very large (Figure 
13); the share of Fedis, the woreda with the highest share (approximately 90%), is four times the share 
in Jarso (22%). The data suggests a negative correlation between the use of improved sanitation facilities 
and open defecation (Figure 13). Across WASH indicators, Fedis is the woreda with the least access to 46 The main source of drinking water is public tap/standpipe (used by 32.3% of households), followed by surface water (used by 
25.1% of households). The first is considered an improved water source, while the second is not. 47 Water treatment technologies included in the study are chlorination, flocculant/disinfection, filtration, solar disinfection, and 
boiling. The most common technology to treat water is by flocculation before drinking (used by 6.3% of households) 48 Open defecation is the most common type of sanitation facility, followed by pit latrines without slab/open pit (used by 24% 
households) and pit latrine with slab/ventilated improved pit latrine with slab (used by 16% of households). 
IMPEL | Implementer-Led Evaluation and Learning
32 Methodology and Limitations
WASH services, including sanitation facilities, and consequently has the highest rate of open defecation 
at nearly 90%. Accordingly, Jarso, which has the highest reported use of improved sanitation facilities, 
has the lowest rate of open defecation.
Figure 13. Use of improved sanitation facilities and open defecation by woreda
Baseline Study of the Ifaa RFSA in Ethiopia (Vol. I)
Methodology and Limitations 33
In most woredas, only a small share of the population uses recommended water treatment technologies 
(Figure 14). Interestingly, Gursum has a much larger share of households using recommended water 
treatment technologies (about 26%) than the other woredas.
Figure 14. Percentage of households practicing correct use of recommended household water 
treatment technologies, by woreda
3.9 Agriculture
Access to financial services allows farmers to diversify their livelihood strategies, invest, manage risks,
and protect assets during shocks. 49, 50, 51 While a handful of farming households (15.59%) have taken out
an agriculture credit in the last 12 months, very few have used crop insurance (1.26%) or saved (2.74%) 
(Table 19). Consequently, fewer than one in five farming households have accessed financial services52
in the last 12 months. Babile has the highest percentage of farming households accessing financial 
services at nearly 30%, while Midega Tola has the lowest at less than 10% (Figure 15). Only 23% of 
households reported cultivating crops or raising livestock with the specific intent to sell or resell to earn 
income. For this subset of households, 76% report practicing at least one value chain activity promoted 49 Enumerators only ask the questions in this module to households that have at least one farmer (4561 out of 4683). In 
addition to this, 23 households didn’t give their consent to participate or were not available. Besides this, most of the indicators 
in this section look at smaller subsamples (e.g., enumerators only ask agricultural insurance questions to households with land). 
We will mention this when relevant. 50 Only one farmer per household (the lead farmer) answered the questions in Module G. The standard structure of Module G 
questions is to first ask the lead farmer if someone in the household did the specified activity and subsequently ask who in the 
household had done so. In this section, we will present the answers at the household level to reduce measurement errors 
related to any recall mistakes of the lead farmer. 51 Following BHA’s recommendations, we did not collect yield information for crops. In the case of livestock, we only collected
information about the number of animals and their health condition. 52 A household has access to financial services if it satisfies at least one of the following conditions: (i) took any agricultural 
credit in the past, (ii) saved any cash, or (iii) bought agricultural insurance.
IMPEL | Implementer-Led Evaluation and Learning
34 Methodology and Limitations 
by Ifaa. Figure 16 shows the share of farming households practicing the different value chain activities. 
The most common practice is purchasing crop inputs (65.23%), followed by processing the product 
(16.65%). Most farming households (89.42%) adhere to at least one natural resource management 
(NRM) practice, most commonly hedge row planting (63.38%) followed by natural regeneration 
(42.72%). 
Table 19. Financial services, value chain activities, and natural resource management practices 
Outcome N All
Percent of farming households using financial services53
4,450 18.36
(38.72)
Percent of farming households using agricultural credit54
4,450 15.69
(36.37)
Percent of farming households who saved 
4,447 2.74
(16.34)
Percent of farming households using crop insurance 
4,449 1.26
(11.15)
Percent of farming households who cultivate any crop or raise/buy livestock 
with the specific intention to sell or resell to earn income 
4,448 22.95
(42.06)
Percent of farming households reporting at least one value chain activity 
promoted by CRS 
1,021 76
(42.73)
Percent of farming households using at least one NRM practice 
4,450 89.42
(30.77)
Notes: Estimates from the RFSA Ifaa baseline survey sample. Standard deviations are in parentheses. Enumerators asked 
financial services and NRM practices questions to the 4,450 that reported having access to a plot of land or animals. In addition, 
only households that cultivate crops or raise/buy livestock with the specific intent to sell or resell to earn income (1021) 
answered the value chain activity question. 
53 A farming household uses financial services if at least one of the following is satisfied: (i) the household uses agricultural 
credit, (ii) the household saves, or (iii) the household uses crop insurance. 54 Any type of credit (i.e., both formal and informal) is included in this question. 
Baseline Study of the Ifaa RFSA in Ethiopia (Vol. I)
Methodology and Limitations 35 
Figure 15. Percent of farming households who used financial services in the past 12 months, by 
woreda 
Figure 16. Share of farming households using different value chain activities 
3.9.1 Crops
Most farming households (90.89%) in the sample have planted at least one crop (Table 20). Of the 
households that have planted at least one crop, the most common crop is maize (73.10%), followed by 
sorghum (56.33%) and then wheat (14.61%). Few farming households grow the crops promoted by Ifaa, which are haricot beans, groundnuts, and potatoes (Table 21). These crops are promoted by Ifaa as 
value-chain crops and not for consumption. CRS expects very few farmers to plant them and only targets 
these crops for 10% of the farmers by the end of the RFSA. 
IMPEL | Implementer-Led Evaluation and Learning
36 Methodology and Limitations 
Table 20. Main crops grown55
N All 
Percent of farming households with a plot of land that planted at least one crop
4,404 90.89
(28.77)
Percent of farming households growing maize 
4,003 73.10
(44.35)
Percent of farming households growing sorghum 
4,003 56.33
(49.59)
Percent of farming households growing wheat 
4,003 14.61
(35.32)
Percent of farming households growing ground nut56
 4,003 7.77
(27.77)
Percent of farming households growing sweet potatoes 
4,003 3.17
(17.52)
Percent of farming households growing barley 
4,003 3.15
(17.46)
Percent of farming households growing millet 
4,003 1.80
(13.29)
Percent of farming households growing common bean 
4,003 1.30
(11.32)
Percent of farming households growing cabbage 
4,003 1.05
(10.18)
Notes: Estimates from the Ifaa baseline survey sample. Standard deviations are in parentheses. 
Table 21. Improved management practices or technologies57 in target crops 
Outcome N All
Percent of farming households growing haricot beans 4,003 0.05 55 For each crop in the table, the percentages were computed with the number of households that grow the specific crop 
divided by the number of households that grow at least one crop in the plots over which it makes decisions. 56 The CAPI included both ground nuts and peanuts as possible crops. In the area of study, these crops are confusing, and 
people use them interchangeably. Due to this, we grouped both answers under the crop ground nuts. 57 Following BHA (2021), a household is said to have applied improvement practices or technologies if the household uses either 
at least one of the promoted improvement practices for the target crop or one of the NRM practices. This indicator will depend 
to a great extent on the NRM practices (not specific to a particular crop), because almost 90% of households use at least one 
NRM practice. 
Information related only to the use of improvement management practices (i.e. without considering the use of NRM practices) 
is presented using the variable “Percent of farming households using at least one of the promoted improvement management 
practices”. 
Baseline Study of the Ifaa RFSA in Ethiopia (Vol. I)
Methodology and Limitations 37 
Outcome N All
(2.23) 
Percent of farming households growing groundnuts58
4,003 1.45
(11.95)
Percent of farming households using at least one of the promoted improvement 
management practices for ground nut59
58 70.69 
(45.92) 
Percent of farming households who have applied improvement practices or 
technologies for groundnuts 
58 81.03
 (39.55) 
Percent of farming households growing potatoes 
4,003 0.72 
(8.48)
Percent of farming households using at least one of the promoted improvement 
management practices for potatoes60
29 93.1 
(25.79) 
Percent of farming households who have applied improvement practices or 
technologies for potatoes 
29 100 
 (0.00) 
Notes: Estimates from the RFSA Ifaa Baseline Survey Sample. Standard deviations are in parentheses. The sample on improved 
management practices for haricot beans was too small for analysis (N=2). A household is said to have applied improvement 
practices or technologies if the household uses either at least one of the promoted improvement practices for the target crop or 
one of the NRM practices. 
3.9.2 Livestock 
The target animals for Ifaa are goats, poultry, and oxen. 23% of farming households raise goats, with an 
average of 3.77 goats (Table 22). The percentage of farming households raising goats varies substantially 
by woreda, with about 45% of households in Midega Tola reporting raising goats and less than 10% in 
Deder (Figure 17). In terms of improved management practices promoted by Ifaa, 48% of households 
use at least one practice for their goats.61 However, since most households report one NRM practice, 
most households are classified as having applied improvement practices or technologies for goats. 
Similarly, one in five households raises poultry, with an average of 3.38 poultry. In terms of improved 
management practices promoted by Ifaa, 41% reported using at least one practice for their poultry.62
However, given that most households report one NRM practice, most households are classified as 
having applied improvement practices or technologies for poultry. Few households raise oxen (6.35%), 58 Only farming households that selected ground nuts as one of the crops they grow were considered here. As discussed in 
footnote 56, people use peanuts and ground nuts interchangeably in the areas of study. The results related with ground nuts 
presented in this table are therefore incomplete, because households that selected they grow peanuts were not considered. 59 The most important practices are organic manure, compost and crop association, used by 54%, 7% and 7% of households 
growing groundnuts. 60 The most important practices are organic manure and performing weedings, both used by 34% of households growing 
potatoes. 61 Improved shelter/housing and vaccinations are the most common activities used by 22% and 19% of households. 62 Improved shelter and improved poultry variety/breed are the most common activities used by 21% and 16% of households. 
IMPEL | Implementer-Led Evaluation and Learning
38 Methodology and Limitations 
and of these households, 41% reported using at least one improvement management practice.63
However, given that most households report one NRM practice, most households are classified as 
having applied improvement practice or technologies for oxen. 
Table 22. Improved management practices or technologies64 in target animals
Outcome N All 
Percent of farming households raising goats
4,537 23.45
 (42.37)
Percent of farming households who have applied improvement practices or 
technologies for goats 
1,064 47.93
(49.98)
Percent of farming households using improved management practices for goats 
1,064 93.52
 (24.64)
Percent of farming households raising poultry 
4,537 19.57
(39.68)
Percent of farming households who have applied improvement practices or 
technologies for poultry 
888 41.33
(49.27)
Percent of farming households using improved management practices for poultry 
888 94.93
 (21.95)
Percent of farming households raising oxen 
4,537 6.35
 (24.38)
Percent of farming households who have applied improvement practices or 
technologies for oxen 
288 47.92
(50.04)
Percent of farming households using improved management practices for oxen 
288 85.07
 (35.70)
Notes: Estimates from the RFSA Ifaa baseline survey sample. Standard deviations are in parentheses. A household is said to have 
applied improvement practices or technologies if the household uses either at least one of the promoted improvement practices 
for the target crop or one of the NRM practices. 63 Vaccinations and Improved shelter/housing are the most common activities used by 27% and 19% of households. 64 Following BHA (2021), a household is said to have applied improvement practices or technologies if the household uses either 
at least one of the promoted improvement practices for the target animal or one of the NRM practices. This indicator will 
depend to a great extent on the NRM practices (not specific to a particular crop) because almost 90% of households use at least 
one NRM practice. 
Information related only to the use of improved management practices (i.e., without considering the use of NRM practices) is 
presented using the variable “Percent of farming households using at least one of the promoted improvement management 
practices.” 
Baseline Study of the Ifaa RFSA in Ethiopia (Vol. I)
Methodology and Limitations 39 
Figure 17. Percent of farming households raising goats, by woreda
3.10 Poverty Measurement 
This section presents different measures of poverty, all based on the household consumption module of 
the survey. The baseline survey did not include a section for non-food items, so enumerators did not 
collect this information. To construct poverty measures, the poverty line was multiplied by the share of 
food consumption in total expenditure (0.85) in the Strengthen PSNP4 Institutions and Resilience I 
baseline survey (Alderman et Al., 2019). We will present the information for food consumption 
expenditure. The percentage of the poor and depth of poverty measures will be based on the adjusted 
poverty line and thus we will refer to these two variables as food poverty and depth of food poverty. We 
computed the percentage of food-poor people as those living on less than US$1.61 per day65 (2011 
purchasing power parity). The equivalent of US$1.61 in current Ethiopian Birr was determined to be 
44.26.66 67
Table 23 summarizes the food poverty measures. The mean daily per-capita food expenditure is 17.64 
Birr or about US$0.64. The mean per-adult equivalent68 measure is slightly higher, 21.36 Birr or about 
US$0.78. The food poverty rate is approximately 98.61% when using per-capita consumption and 65 1.9 * 0.85 = 1.61. 66 The 2011 PPP used was 5.439 Birr/dollar. To adjust the price by inflation, we used the food and non-alcoholic beverages 
consumer price index (food CPI). The food CPI used for 2011 was 65.25, and 329 for June 2022. This gives us the food poverty 
line of (1.9 * 0.85) * 5.439 * 329/65.28 = 44.26. The food CPI information from 2011 until March 2022 comes from IMF (2022), 
while the information for April-June 2022 comes from CSA (2022). Neither source had food CPI series for the complete period of 
interest. 67 The expenditure module of the survey followed standard practices for expenditure calculation. Frequent items, primarily 
foods, used 7-day recall. 68 The per capita measures are computed by dividing the household consumption measure by the number of members of the 
household. The per adult equivalent measures divide household consumption by a weighted measure of the number of 
members of the households. The weights differ by age and gender and are used to account for the difference between the cost 
of children and adults. We used the weights specified in Ethiopia’s Household Consumption Expenditure Survey 2010/11. 
IMPEL | Implementer-Led Evaluation and Learning
40 Methodology and Limitations 
96.17% when using adult equivalent measures. Even the lower per-adult equivalent measure is very 
high, which signals that the target population is among the poorest groups in Ethiopia.69
Table 23 includes measures of the depth of poverty of the food poor. This measure is used to 
understand how far below the food poverty line the target population is. The depth of poverty of the 
food poor is 64.47% by per capita consumption and 58.65% by adult equivalent measures. This means 
the average food-poor person is 64% below the food poverty line. Figure 18 shows the distribution of 
food expenditure values across the full sample, where the food poverty line is shown in green. 
Table 23. Poverty measurements
Outcome N All
Daily per capita food expenditure (Birr) 
4,677 17.64
(11.01)
Daily per adult equivalent food expenditure (Birr) 
4,677 21.36
(13.26)
Household food expenditure per day (Birr) 
4,677 96.73
(54.42)
Percent food poor (per-capita consumption expenditure) 
4,677 98.61
(11.69)
Percent food poor (per adult equivalent consumption expenditure) 
4,677 96.17
(19.20)
Depth of poverty of food poor (using per-adult equivalent consumption 
expenditure) 
4,558 64.47
(19.05)
Depth of poverty of food poor (using per-capita consumption expenditure)
4,399 58.65
(20.91)
Notes: Estimates from the RFSA Ifaa baseline survey sample. Standard deviations are in parentheses.
69 Ethiopia’s national poverty rate is 30.8% (World Bank, n.d.). Even though the measure of food poverty in this report is not 
directly comparable to the national poverty rate, the large difference between both measures is evidence that we are looking at 
some of the poorest households in Ethiopia. 
Baseline Study of the Ifaa RFSA in Ethiopia (Vol. I)
Methodology and Limitations 41
Figure 18. Distribution of log food consumption expenditure per adult equivalent per day in 
purchasing power parity USD
Figure 19 provides a picture of the food poverty measures at the woreda level. Even though all of the 
woredas are very poor, there are some sizable differences: while woredas like Fedis or Melka Belo have 
close to 100% food poverty rates, Jarso has a much lower food poverty rate of 82%. The graph for the 
depth of food poverty portrays a similar picture: woredas with higher food poverty rates tend to have a 
higher depth of food poverty. 
Figure 19. Poverty rate and depth of poverty per adult equivalent measure, by woreda
IMPEL | Implementer-Led Evaluation and Learning
42 Methodology and Limitations 
3.11 Gender Dynamics 
Gender dynamics are captured through six indicators in this section. As people in a union were the focus 
of this survey’s section, they were the only ones included.70 These indicators explore men’s and 
women’s financial resources and access to credit within the households. 
3.11.1 Use of Financial Resources
This section presents findings on participation in cash-earning activities focusing on men and women in 
a union as well as women’s participation in decisions around their and their husbands’ incomes.71
Table 24. Gender (cash) indicators 
Outcome N All
BL32: Percent of women and men in a union who earned cash in the past 12
months 
3,444 23.05
 (42.12)
BL32: Cash-earning men in a union 
1,830 29.07
 (45.42)
BL32: Cash-earning women in a union 
1,614 16.23
 (16.74) 70 The questions in this section were only asked to the HH and their partner if the HH was married or in a union. Even though 
the questions were intended to be answered by both, there are many cases where only one of them was available. Additionally, 
some households didn’t give consent to answer the module (5 out of 3,689 households), or the HH was not present (1,009 out 
of 3,689). 71 The sample size for this question was men and women in a union.
Baseline Study of the Ifaa RFSA in Ethiopia (Vol. I)
Methodology and Limitations 43 
Outcome N All 
BL33: Percent of women in a union and earning cash who report participation 
in decisions about the use of self-earned cash 
262 70.61
(45.64)
BL34: Percent of women in a union and earning cash who report participation 
in decisions about the use of spouse/partner’s self-earned cash 
262 43.89
(49.72)
BL 35: Percent of men in a union and earning cash who report spouse/partner 
participation in decisions about the use of self-earned cash 
532 40.98
 (49.23)
Notes: Indicators BL33 and BL34 are constructed on the population of women in a union that earned money in the last 12 
months. BL35 is constructed on the population of men in a union that earned money in the last 12 months. Estimates from the 
RFSA Ifaa baseline survey sample. Standard deviations are in parentheses. 
As Table 24 illustrates, just 23% of men and women in a union earned cash in the past year overall, with 
29% of men in a union and 16% of women in a union participating in cash-earning activities. Indicator 33 
shows that most women in a union (71%) report having decision-making power over the use of their 
self-earned cash. About 44% of cash-earning women also reported having decision-making power over 
their spouse/partner’s self-earned cash. Similarly, 41% of men in a union reported that their 
spouse/partner also made decisions about using self-earned cash.
Figure 20 presents the share of men and women in a union earning cash disaggregated at the woreda 
level. Even though in all woredas less than half of the women and men earned cash, there is a high 
variability: while in Midega Tola more than 55% of men and 40% of women in a union earned cash, the 
share is 4.21% in Jarso. Across the board, the data shows that a lower proportion of women earn cash 
than men.
Figure 20. Men/women in a union who reported earning cash, by woreda
IMPEL | Implementer-Led Evaluation and Learning
44 Methodology and Limitations 
3.11.2 Credit 
This section presents findings on credit access and decision-making among people in a union, as well as 
their participation in community groups. Indicator 41 shows the percentage of men and women in a 
union who took any type of credit and are members of a community group.72 The list of community 
groups that were included in the survey were: agriculture/livestock/fisheries producer’s groups; water 
users’ groups; forest users’ groups; credit or microfinance groups; mutual help or insurance groups; 
trade and business association groups; civic groups; local governments; informal/formal organizations in 
the community; and religious groups. 
Table 25. Access to credit and group participation indicators, by gender 
Outcome N All
BL41: Percent of women/men in a union who are members of a community 
group73
999 59.36
(49.14)
BL41: Men in a union who are members of a community group 
563 65.9
(47.45)
BL41: Women in a union who are members of a community group 
436 50.92
(50.05)
BL42: Percent of women/men in a union with access to credit 
3,444 27.73
(44.77)
BL42: Men in a union who have access to credit 
1,830 29.45
(45.60)
BL42: Women in a union who have access to credit 
1,614 25.77
(43.75)
BL43: Percent of women/men in a union who make decisions about credit
955 29.53
(45.64)
BL43: Men in a union who report making decisions about credit 
539 28.94
(45.39)
BL43: Women in a union who report making decisions about credit 
416 30.29
(46.01)
Notes: BL43 was only asked to people that have taken credit. Due to a skip pattern, BL41 (membership to a community group) 
was also only asked to women/men that have taken credit. Estimates from the RFSA Ifaa baseline survey sample. Standard 
deviations are in parentheses. 72 Due to a skip pattern, questions related to community membership were only asked to people who took credit (999 out of 
3,444 individuals). 73 The most common community groups are religious groups and mutual help or insurance groups, attended by 54% and 42% of 
people in a union. 
Baseline Study of the Ifaa RFSA in Ethiopia (Vol. I)
Methodology and Limitations 45 
Table 25 shows that 59% of people in a union are members of a community group, with 66% of men in a 
union and 51% of women in a union reporting membership. The other two indicators in the table look at 
credit-related variables for people in a union. 28% of people in a union reported having access to credit￾with less than one-third (30%) of men and about one-quarter (26%) of women in a union reporting 
credit access. Of those who had access to credit, approximately 30% reported making decisions about 
that credit. Men and women both made decisions about credit at similar rates, with 29% of men and 
30% of women in a union stating that they made credit decisions. 
Across all woredas surveyed, there was high variability in access to credit by men and women (Figure 
21). Gursum had the highest access rate (45% for men and 39% for women), while Melka Belo had the 
lowest (3.5% for men and 2.8% for women), followed by Chinaksan (11.3% for men and 1% for women). 
The shares disaggregated by gender follow a similar pattern. Generally, a larger proportion of men have 
access to credit, except for Fedis and Deder. In Fedis, roughly 3% more women report access to credit 
than men, while in Deder, the difference is 9%. 
Figure 21. Share of men and women in a union with access to credit, by woreda 
3.12 Resilience 
This section presents five different indexes that provide a picture of households’ resilience. These 
indexes are based on subindexes. In each subsection, we will present the main index and the 
IMPEL | Implementer-Led Evaluation and Learning
46 Methodology and Limitations 
subindexes. Some of the main indexes are constructed using principal components analysis,74 and, in 
most cases, it is difficult to interpret the final number. Relying on the subindexes in addition to the main 
index helps provide a better understanding of households’ resilience capacities. 
3.12.1 Ability to Recover from Shocks and Stresses Index 
The ability to recover from shocks and stresses index reflects the ability to recover from negative events 
that have impacted the household. On average, households score 3 on this index. This index is 
composed of indices that reflect how households perceived their recovery and the total number and 
severity of shocks the household experienced over the past year. 
Table 26. Ability to recover from shocks and stresses index 
Outcome N All 
BL23: Mean ability to recover from shocks and stresses index
4,545 3
(1.36)
Ability to recover subindex (2–6) 
4,649 2.98
(1.36)
Shock exposure subindex (0–168) 
4,575 24.4
(14.41)
Total shocks experienced (0–21) 
4,683 3.6
(2.19)
Notes: Estimates from the RFSA Ifaa baseline survey sample. Standard deviations are in parentheses. 
Households experienced an average of 3.6 out of 21 possible shocks in the previous year. The most 
reported shocks are illustrated below in Figure 22, which shows the shocks experienced by at least 20% 
of the households. The two most important shocks experienced by more than 70% of the households 
are increases in food prices (72%) and drought conditions. Other common shocks listed included a delay 
in PSNP food assistance, increased prices of inputs, and crop pests. 
In terms of the number and intensity of shocks experienced, households scored an average of 24.4 out 
of 168, which accounts for 21 shocks and four different levels of severity (regarding both the impacts on 
the household economic situation and household consumption). This suggests that out of the average of 
3.6 shocks experienced, households perceived those shocks to be severe. 74 The indexes relying on principal component analysis are constructed as follows: first, we calculated the first principal 
components of all the available subindexes associated with the main index; we then predicted the score for each household 
based on the weighted combination of the first component and the subindexes. This score was rescaled to be between 0 and 
100. 
Baseline Study of the Ifaa RFSA in Ethiopia (Vol. I)
Methodology and Limitations 47
Figure 22. Most common shocks reported by households
Figure 23 shows the variability across woredas in the ability to recover from shocks and stresses. Fedis 
has the highest score (4.08), and Melka Belo (2.25) and Jarso (2.93) have the lowest scores. The shock 
exposure and total shock experiences follow a similar pattern (Figure 24). The households in all woredas 
perceived the shock to have had a severe impact on their income and food consumption.
Figure 23. Mean ability to recover from shocks and stresses, by woreda
IMPEL | Implementer-Led Evaluation and Learning
48 Methodology and Limitations 
Figure 24. Shock indicators
Baseline Study of the Ifaa RFSA in Ethiopia (Vol. I)
Methodology and Limitations 49 
3.12.2 Absorptive Capacity Index
The absorptive capacity index reflects the ability of households to prepare for, deal with, and mitigate 
the impact of shocks and stressors on well-being outcomes through preventive measures and positive 
coping strategies. Overall, households scored around 29 out of 100 on this index, reflecting a low ability 
to absorb shocks.75 There are few resources that households have access to that would enable them to 
mitigate shocks better. 
A key aspect of the absorptive capacity index is financial resources (e.g., cash savings, remittances) that 
households can access to absorb shocks. Overall, very few households have access to financial resources 
for absorbing shocks, lowering their ability to mitigate the impact of shocks on well-being outcomes. 
Specifically, only 78 households (1.6%) have cash savings, 112 (2.3%) reported receiving remittances, 
and 56 (1.2%) have access to insurance. Households, on average, own 8.27 different types of household 
and productive assets (out of 45), although this does not necessarily mean that households have a large 
asset stock.
Table 27. Absorptive capacity index and related indexes 
Outcome N All
BL09: Absorptive capacity index (0–100) 
4,442 28.68
(15.49)
BL09: Absorptive capacity index (0–100), excluding access to insurance 
4,662 29.96
(15.12)
Bonding social capital index (0–6) 
4,681 2.55
(1.88)
Access to cash savings index (0–1) 
4,680 0.02
(0.13)
Remittances index (0–1) 
4,683 0.02
(0.15)
Asset ownership index—total type (0–45) 
4,665 8.27
(3.65)
Shock preparedness and responsiveness index (0–3) 
4,681 0.39
(0.56)
Access to insurance index (0–1) 
4,450 0.01
(0.11)
Access to humanitarian assistance index (0–1) 4,679 0.5
75 For the absorptive capacity index, it was impossible to compute the subindex “availability of informal safety nets,” which was 
excluded from the calculations. In Table 25, we present two versions of the absorptive capacity index: one with the subindex 
“access to insurance” and another one without. This was done because we couldn’t calculate that subindex for more than 200 
households. 
IMPEL | Implementer-Led Evaluation and Learning
50 Methodology and Limitations 
Outcome N All
(0.50)
Notes: Estimates from the RFSA Ifaa baseline survey sample. Standard deviations are in parentheses. 
The absorptive capacity index also captures the level of social capital that households can access to help 
them absorb shocks. Overall, findings suggest that most households have low to medium social capital. 
The bonding social capital index reflects the number of types of individuals that households could get 
and give help to inside of their communities (out of three groups). On average, households feel able to 
get and give help from 1.25 of these types of individuals. 
Another element of absorptive capacity is how well a household is prepared to mitigate shocks76
through the availability of disaster preparedness groups in the community and other household shock 
mitigation strategies. On average, households score 0.39 out of 3 on this index, suggesting a low ability 
to mitigate shocks. 
The last dimension of absorptive capacity is the availability of humanitarian assistance in the 
community. Around half of the households reported either having used humanitarian emergency 
assistance77 or its availability in the village. 
Most woredas have an absorptive capacity index between 20 and 33. The exception is Jarso, with a 
score of 46.86 (Figure 25). Looking at the subindexes, we observe a high variability in the bonding social 
capital index, shock preparedness and responsiveness index, and access to humanitarian assistance 
index (Figure 26). The first two indexes follow a similar pattern as the absorptive capacity index: Jarso 
has the highest value, while Melka Belo has the lowest. The last index behaves differently, whereby 
woredas with low absorptive capacity indexes tend to have higher scores. 
Figure 25. Absorptive capacity index, by woreda 
76 This index does not include whether the household reports participating in any of the following activities: soil conservation 
activities, flood diversion structures (i.e., protection of land/infrastructure from flooding), planting trees on communal land, or 
improving access to health services given available data. 77 The survey question associated with this index is “Has your household received any kind of formal support from the 
government or non-governmental organizations over the past 12 months?” 
Baseline Study of the Ifaa RFSA in Ethiopia (Vol. I)
Methodology and Limitations 51 
Figure 26. Bonding social capital index, shock preparedness and responsiveness index and access to 
humanitarian assistance index, by woreda 
IMPEL | Implementer-Led Evaluation and Learning
52 Methodology and Limitations 
3.12.3 Social Capital Index 
These indices convey the ability of households to draw on social networks78 to get support to reduce the 
impact of shocks and stresses on their households. They measure both the degree of bonding (social 
capital among households within their communities) and the degree of bridging (social capital between 
households in the area and households outside their community). Findings suggest that households can 
draw on other households within their community slightly more (average score of 54.2) than they can 
draw on other households outside of their community (average score of 46.04). 
Table 28. Social capital index and related indexes 
Outcome N All
Index of social capital at household level (0–100)
4,681 50.12
(33.89) 
Bonding index (0–100) 
4,681 54.2 
(35.77) 
Bridging index (0–100) 
4,681 46.04 
(36.46) 
Notes: Estimates from the RFSA Ifaa baseline survey sample. Standard deviations are in parentheses. 78 Three different groups of individuals are considered: (i) relatives, (ii) non-relatives in the same ethnic groups/clans, and (iii) 
non-relatives in other ethnic groups/clans. Both indexes (bridging and bonding) are based on these three groups. 
Baseline Study of the Ifaa RFSA in Ethiopia (Vol. I)
Methodology and Limitations 53 
When we disaggregate the indexes at the woreda level, we observe that both the bonding and the 
bridging index behave similarly (Figure 27). Jarso has by far the highest score in both indexes while 
Melka Belo has the lowest. 
Figure 27. Bonding and bridging indexes, by woreda 
IMPEL | Implementer-Led Evaluation and Learning
54 Methodology and Limitations 
3.12.4 Adaptive Capacity Index
The adaptive capacity index measures the ability of households to manage resources and make 
proactive and informed choices to better prepare for and adapt to future shocks. The index is composed 
of several components that reflect different resources or adaptive abilities.79 The adaptive capacity 
index, excluding two problematic indexes (“access to financial resources” and “aspirations/confidence to 
adapt index”), is, on average, 35.93 out of 100, which suggests that households have a limited ability to 
manage resources and adapt to future shocks.
Table 29. Adaptive capacity index and related indexes 
Outcome N All 
BL08: Adaptive capacity index (0–100) 
516 43.45
(22.57)
BL08: Adaptive capacity index (0–100), excluding aspirations/confidence to 
adapt 
1,964 40.88
(19.81)
BL08: Adaptive capacity index (0–100), excluding access to finance s. practices 
848 37.65
(21.42)
BL08: Adaptive capacity index (0–100), excluding access to finance s. practices 
and aspirations/confidence to adapt 
3,960 35.93
(18.93)
Aspirations/confidence to adapt index (0–16) 
4,290 9.73
(2.27)
Bridging social capital index (0–6) 
4,681 2.17
(1.86)
Linking social capital index (0–4) 
4,678 0.47
(0.98)
Social network index (adjusted range 0–3) 
4,368 1.08
(1.32)
Education/training index (0–8) 
4,669 0.98
(1.22)
Livelihood diversification index (0–20) 
4,681 1.81
(0.76) 79 In Table 27, we present four versions of the adaptive capacity index: (i) includes all of the available subindexes, (ii) excludes 
“access to financial resources,” (iii) excludes “aspirations/confidence to adapt index,” and (iv) excluding the two subindexes 
previously mentioned. This was done because the inclusion of the subindexes mentioned led to a substantial reduction in the 
sample size. 
Baseline Study of the Ifaa RFSA in Ethiopia (Vol. I)
Methodology and Limitations 55 
Outcome N All 
Adoption of improved practices index (0–1)80
1,021 0.35
(0.48)
Access to financial resources index (0–1) 
2,242 0.34
(0.47)
Exposure to information index (0–19)
4,681 1.75
(2.38)
Asset ownership index—total type (0–45) 
4,665 8.27
(3.65)
Notes: Estimates from the RFSA Ifaa baseline survey sample. Standard deviations are in parentheses. 
One aspect of the adaptive capacity index captures households’ social capital and networks, considering 
that if households can leverage these networks more effectively, they may better prepare for and adapt 
to future shocks. Across these indicators, households score poorly, suggesting that households are 
unable to effectively leverage social capital and networks to adapt to shocks. In particular, the bridging 
social capital index reflects the number of types of individuals whom households could draw on outside 
of their communities (out of three groups). On average, households feel able to draw on 1.08 of these 
types of individuals. The linking social capital index reflects how well-connected households are to 
government or non-governmental organization leaders and whether they can draw on them for help. 
Households score very low (0.47 out of 4) on this, suggesting that most households neither know leaders 
nor can they ask leaders for help. Finally, the social network index81 captures household access to and 
participation in various support groups. Households score a 1 out of 3, suggesting that less than half of 
households have access to and/or participate in these groups. 
Another aspect of the adaptive capacity index captures the human resources, assets, and financial 
resources available to households to mitigate shocks. Overall, households have low levels of human 
capital and asset resources, suggesting constraints on the overall resource pool they can draw on in the 
face of shocks. The education/training index reflects the level of human capital in the household, 
specifically adult literacy; whether any adult has surpassed primary school; and the number of trainings 
in which household adults have participated. Households score low (0.98 out of 8) on this indicator, 
reflecting that overall household human capital is low. The asset ownership index illustrates the number 
of different types of assets a household owns (out of 45 types). On average, a household owns 8.27 
different types of assets. This could mean that overall household asset stock is low, although this does 
not reflect the value of each asset. Finally, the access to financial resources index reflects the financial 80 This subindex is slightly different than indicator BL30 in section 4.6. While indicator BL30 looks at farming households using at 
least one of the value chain activities promoted by CRS, the subindex “Adoption of improved practices index” considers 
households adopting three or more value chain activities promoted by CRS. 81 This subindex was adapted to account for the lack of some village-/community-level variables. The original subindex has 
three individual-level variables and three village-level variables, while the one presented here only includes the individual ones. 
IMPEL | Implementer-Led Evaluation and Learning
56 Methodology and Limitations 
resources available in the village through credit and savings institutions.82 34% of the households 
reported the presence of a credit or microfinance group in their village. 
A third aspect of the adaptive capacity index reflects how diversified and improved household livelihood 
activities are. The livelihood diversification index reflects the number of different livelihood activities 
households were engaged in over the past year. Overall, households were engaged in an average of 1.81 
out of 20 activities, indicating that activities are not well diversified. The adoption of improved practices 
index83 reflects whether households adopted improved crop or livestock practices, NRM practices, or 
improved storage practices. Overall, households score 0.34 out of 1, indicating that about 34% of 
households have adopted improved practices. The exposure to information index captures the number 
of topics that households have received information on in the past year, which relates directly to a 
household’s ability to make informed choices to better prepare for shocks. On average, households have 
received information on 1.75 out of 19 available topics, highlighting that households have had very 
limited exposure to information to help inform shock mitigation strategies. 
Finally, the aspirations/confidence to adapt index reflects a household adult’s aspirations, confidence to 
adapt, and a sense of control over one’s life. On average, adults score 9.73 out of 16 on this index, 
reflecting a moderate sense of confidence to adapt.
The adjusted adaptive capacity index fluctuates at about 25% and 35% for most woredas (Figure 28). 
Jarso and Chinaksan are exceptions. As we have observed with other indexes, Jarso has by far the 
highest score (65), almost three times the score of Fedis (23.21)—the woreda with the lowest score. 
Most individual subindexes behave similarly, except for the education/training subindex. In Figure 29, it 
is evident that Melka Belo has one of the highest scores, while Jarso has the second-highest score, 
behind Gursum. 
82 This index differs from the one in BHA’s manual because of the lack of community-level variables. Instead of using two 
separate questions (one for savings and another for credit institutions), we relied on the following question in Module K: “Is 
there a credit or microfinance group including Savings and Credit Cooperative Organization/merry-go-rounds/Village Savings 
and Loan Association in your community?” 83 This subindex does not include the adoption of storage methods, which was not collected. Additionally, the subindex does 
not include a measure of quality for each of the service types. Instead, it only captures whether the service exists. Health 
services reflect whether non-governmental organizations are currently conducting health activities and not whether local 
health institutions are available. 
Baseline Study of the Ifaa RFSA in Ethiopia (Vol. I)
Methodology and Limitations 57
Figure 28. Adjusted84 adaptive capacity index, by woreda
Figure 29. Education/training index by woreda
Like the previous section, for subindexes associated with social capital, Jarso has by far the highest score
(Figure 30). The differences between Jarso and the rest of the woredas are quite large for both 
subindexes; the score for Jarso is two and three times higher than the score for the woreda, with the 
second-largest score. 84 This version of the index excludes the subindexes Access to Finance Service practices and Aspirations/Confidence to adapt.
IMPEL | Implementer-Led Evaluation and Learning
58 Methodology and Limitations 
Figure 30. Social capital related indexes, by woreda
3.12.5 Transformative Capacity Index 
The transformative capacity index85 captures system-level resources, governance, and institutions that 
make up the enabling environment that promotes or limits a household’s capacity to respond to shocks 85 For the transformative capacity index, it was not possible to compute the subindexes “access to markets,” “access to 
infrastructure,” and “collective action.” In Table 28, we present four versions of the adaptive capacity index: (i) includes all the 
available subindexes, (ii) excludes “access to agricultural services,” (iii) excludes “gender equity subindex,” and (iv) excludes the 
two subindexes previously mentioned. This was done because the inclusion of the subindexes mentioned leads to a substantial 
reduction of the sample size. 
Baseline Study of the Ifaa RFSA in Ethiopia (Vol. I)
Methodology and Limitations 59 
and stressors. The transformative capacity index, excluding two problematic indexes, is, on average, 
41.53 (out of 100), indicating that there are not very strong institutions available to enhance household 
capacity to respond to shocks. 
Table 30. Transformative capacity index and indexes 
Outcome N All
BL25: Transformative capacity index (0–100)
200 43.86
(22.10)
BL25: Transformative capacity index (0–100), excluding agricultural services 
and gender equity 
4,290 41.53
(20.33)
BL25: Transformative capacity index (0–100), excluding gender equity index
4,290 36.25
(21.37)
BL25: Transformative capacity index (0–100), excluding agricultural services 
200 44.76
(22.13)
Access to formal safety nets index (0–11) 
4,462 2.05
(0.79)
Access to communal natural resources index (0–4) 
4,681 1.29
(1.64)
Basic services index (0–1), including only the police variable 
4,682 0.98
(0.13)
Access to agricultural services index (0–1) 
4,683 0.12
(0.33)
Bridging social capital index (0–6) 
4,681 2.17
(1.86)
Linking social capital index (0–4) 
4,678 0.47
(0.98)
Social cohesion index (0–3) 
4,573 1.1
(1.10)
Local decision-making index (0–1) 
4,619 0.91
(0.28)
Local government responsiveness index (0–2) 
4,683 1.84
(0.38)
Gender index (0–3) 4,682 2.68
IMPEL | Implementer-Led Evaluation and Learning
60 Methodology and Limitations 
Outcome N All
(0.56)
Gender equitable decision-making index (0–1)
259 0.71
(0.44)
Notes: Estimates from the RFSA Ifaa baseline survey sample. Standard deviations are in parentheses.
One dimension of the transformative capacity index is access to basic services. Because of a lack of 
information, the basic services index only considers the availability of police/security/force.86 This index 
was close to 1, which means that almost all households have access to government security forces (local 
or national) that can reach a village within 1 hour. 
Another dimension of the transformative capacity index is the availability of economic institutions to 
support livelihoods. Access to these economic institutions is varied. Only 12% of households report 
having access to agricultural extension services.87 A slightly larger number of households have access to 
natural communal resources; on average, households have access to only 1.29 out of four natural 
communal resources (communal grazing land, water source, firewood, and irrigation source). 
Another aspect of the transformative capacity index reflects the strength of households to support 
themselves through their networks. Overall, households have a low ability to draw on their networks. As 
mentioned in the Adaptive Capacity Index section, households feel able to draw on 1.08 out of three 
types of individuals (bridging social capital index), and the low linking social capital score suggests most 
households neither know leaders nor can they ask leaders for help. A related index is the social cohesion 
index88 which illustrates how active households have been in various support groups in the community. 
On average, households report engaging in 1.1 out of three support groups, reflecting low participation 
in support groups. Finally, the local decision-making index reflects how actively households participate 
in groups in their communities. 91% of households report active participation. 
The gender index reflects constraints to gender-neutral behavior at the community level. On average, 
communities report that 2.68 out of three gender-neutral behaviors are norms.89 The gender equitable 
decision-making index90 reflects how equitable decision-making is across male and female adults within 
the same household. On average, households score 0.71 out of 1 on this index, reflecting that out of two 
key household decisions,91 on average, more than one involves both male and female household 
members. 86 Due to a CAPI error, questions on the other basic services (primary school, health services, and financial services) were not 
collected. 87 This index was calculated based on the percentage using agricultural extension services versus those with access, given 
available data. 88 This index ranges from 0–3 instead of 0–4 as we do not have data on whether community members came together for social 
events. 89 The three behaviors considered were: (i) men and women sit and eat together within their households, (ii) men and women 
regularly sit together at public meetings, and (iii) men in the village help with childcare. 90 This index does not include measures of equitable decision-making around nutrition and child health, as well as savings. The 
index was constructed using information from households where both the HH and his/her spouse were available. The results of 
this index should be considered cautiously due to the low sample size. 91 The two major decisions are: (i) how the cash earned will be used and (ii) decisions about making major household purchases.
Baseline Study of the Ifaa RFSA in Ethiopia (Vol. I)
Methodology and Limitations 61
The final dimension of the transformative capacity index measures how available and reliable external 
sources of support are. Overall, households have access to a moderate number of these external 
resources. The formal safety nets index reflects the number of external safety nets (e.g., emergency 
food or cash assistance, agricultural inputs) available in the community. Overall, households have access 
to two out of 11 formal safety nets. The government responsiveness index reflects whether households 
have access to a reliable police force and a peace committee. On average, households have access to 
1.84 of two of these resources.
At the woreda level, Jarso has the highest score (close to 80), while the rest of the woredas have values 
in the range of 23 to 43 (Figure 31). Even though there is high variability in all the subindexes that 
comprise the transformative capacity index, it is worth highlighting the access to agricultural services 
index.92 Figure 32 shows that almost all woredas have scores below 0.1. Jarso is the exception, with a 
score above 0.9. 
Figure 31. Adjusted93 transformative capacity index, by woreda
92 A household gets an index of 1 if it uses training and extension services. 93 Does not include agricultural services and gender equity indexes.
IMPEL | Implementer-Led Evaluation and Learning
62 Methodology and Limitations
Figure 32. Access to agricultural services index, by woreda
IMPEL | Implementer-Led Evaluation and Learning
63 Comparison of Treatment and Control Groups
4. COMPARISON OF TREATMENT AND CONTROL GROUPS
This section presents the results of balance tests run across treatment arms. The different subsections 
display tables containing similar outcome variables to those discussed in Section 4. For each variable we 
run: 
Equation 1, where denotes the household or individual, stands for the outcome variable, and
is equal to 1 if household lives in a treated kebele.
Equation 1. Balance test regression across treatment arms = + +
The variable of interest in this section is , which represents the difference between the treatment and 
control groups. For each outcome variable, we show the mean value for the control and treatment 
groups, the coefficient, ` its associated T-statistic and the p-value. Standard errors are clustered at the 
kebele level. 
Causal Design ran more than 140 balance tests. In an experimental study using random assignment, the 
difference between treatment arms is expected to be equal to zero. This is more likely to happen in 
situations where the number of clusters is very large. In samples of practical size, it is possible to obtain 
a significant difference in means across treatment arms for some variables by chance. In the case of this 
impact evaluation, we worked with 120 clusters, which were allocated to the treatment or control 
groups. Even though the allocation was done randomly, it is possible that by chance some variables have 
a statistically significant difference across treatment arms. Even for statistically identical groups, at a 5% 
significance level, one out of every 20 tests is expected to indicate a significant difference due to 
random chance. 
From the 141 balance tests presented in the tables of this section, there were 15 variables with a 
statistically significant difference across treatment arms at a 10% significance level, four at a 5% 
significance level, and none at a 1% significance level. Even though the number of variables with a 
statistically significant difference at a 10% significance level is slightly higher than what would be 
expected (15 versus 14), the opposite happens when we look at a significance level of 5% (four versus 
seven) or 1% (zero versus 1.4). These results show that the randomization led to a sampling error within 
the expected ranges. 
The sections with the highest share of significant differences were characteristics of the study 
population; women’s health, maternal nutrition, and reproductive health; agriculture; and gender 
dynamics. 
4.1 Characteristics of the Study Population 
Table 31. Individual-level sample characteristics 
Outcome N Control N 
Treat￾ment 
Difference T-Stat 
Diff
P-value 
Diff
Average age 11,739 19.075 16,130 19.217 0.143 0.496 0.621
IMPEL | Implementer-Led Evaluation and Learning
64 Comparison of Treatment and Control Groups
Outcome N Control N 
Treat￾ment 
Difference T-Stat 
Diff
P-value 
Diff
(16.16) (16.10) (0.29)
Percent of children under 5 
years old
11,739 14.413 16,130 13.751 -0.663 -0.915 0.362
(35.12) (34.44) (0.72)
Percent of children 5–14 
years old
11,739 38.853 16,130 39.033 0.179 0.229 0.819
(48.74) (48.78) (0.78)
Percent of adults (older than 
15 years old) 
11,739 46.733 16,130 47.216 0.483 0.522 0.603
(49.90) (49.92) (0.93)
Percent of females 11,739 49.689 16,130 50.155 0.466 0.835 0.406
(50.00) (50.00) (0.56)
Percent of GWRA 11,739 22.523 16,130 22.951 0.428 0.858 0.392
(41.78) (42.05) (0.50)
Percent of adults who are in 
a union
5,486 58.476 7,616 56.276 -2.200 -1.57 0.119
(49.28) (49.61) (1.40)
Percent of adults with at 
least some schooling
5,486 26.85 7,616 25.985 -0.865 -0.496 0.621
(44.32) (43.86) (1.74)
Percent of adults who are 
farmers 
5,485 66.582 7,613 65.914 -0.668 -0.262 0.794
(47.17) (47.40) (2.55)
Percent of people older than 
10years who did work and 
were paid in cash 
5,498 34.503 7,980 29.348 -5.155* -1.745 0.084 
(47.54) (45.54) (2.95) 
Notes: Standard deviations are in parentheses. * denotes significance at 10 pct., ** at 5 pct., and *** at 1 pct. Standard errors 
are clustered at the kebele level. 
Table 32. Household-level sample characteristics 
Outcome N Control N Treat￾ment
Difference T-Stat 
Diff
P-value 
Diff
Average household size 
1,950 6.02 2,733 5.902 -0.118 -1.024 0.308
(2.11) (2.20) (0.12)
Average number of children 
under the age of 5 in the 
household
1,950 0.868 2,733 0.812 -0.056 -1.081 0.282
 (0.94) (0.94) (0.05) 
Average age of HH
1,950 41.583 2,733 41.911 0.328 0.709 0.48
(10.76) (11.11) (0.46)
Percent of HH who are in a 
union 
1,950 79.744 2,733 75.375 -4.369** -2.538 0.012
(40.20) (43.09) (1.72)
Percent of HH who are female
1,950 23.487 2,733 26.82 3.333* 1.789 0.076
(42.40) (44.31) (1.86)
1,950 13.692 2,733 13.355 -0.337 -0.203 0.84
Baseline Study of the Ifaa RFSA in Ethiopia (Vol. I)
Comparison of Treatment and Control Groups 65 
Outcome N Control N Treat￾ment
Difference T-Stat 
Diff
P-value 
Diff
Percent of HH with some 
schooling (34.39) (34.02) (1.66)
Percent of female HH with some 
schooling
458 2.183 733 3.001 0.818 0.755 0.452
(14.63) (17.07) (1.08)
Percent of male HH with some 
schooling
1,492 17.225 2,000 17.15 -0.075 -0.037 0.97
(37.77) (37.70) (2.02)
Notes: Standard deviations are in parentheses. * denotes significance at 10 pct., ** at 5 pct., and *** at 1 pct. Standard errors 
are clustered at the kebele level. 
4.2 Food Security 
Table 33. Food Insecurity Experience Scale
Outcome N Control N Treat￾ment 
Difference T-Stat 
Diff
P￾value 
Diff
Raw FIES score (scale 0–8) 
1,949 6.884 2,729 6.87 -0.014 -0.15 0.881
(1.35) (1.43) (0.09)
Percent of households with some 
indication of food insecurity
1,949 99.538 2,729 99.45 -0.088 -0.368 0.714
(6.78) (7.39) (0.24)
Percent of households that 
experienced approx. moderate-to￾severe food insecurity 
1,949 96.973 2,729 95.566 -1.407 -1.536 0.127 
(17.14) (20.59) (0.92) 
Percent of households that 
experienced approx. severe food 
insecurity
1,949 77.014 2,729 78.124 1.110 0.371 0.711 
(42.09) (41.35) (2.99) 
Percent of households that 
answered yes to all eight 
questions
1,949 37.045 2,729 38.292 1.248 0.305 0.761
(48.30) (48.62) (4.09) 
Percent of households that 
answered no to all eight questions
1,949 0.462 2,729 0.55 0.088 0.368 0.714
(6.78) (7.39) (0.24)
Raw FIES score (scale 0–8) 1,949 6.884 2,729 6.87 -0.014 -0.15 0.881
Notes: Standard deviations are in parentheses. * denotes significance at 10 pct., ** at 5 pct., and *** at 1 pct. Standard errors 
are clustered at the kebele level. 
IMPEL | Implementer-Led Evaluation and Learning
66 Comparison of Treatment and Control Groups
Table 34. Food consumption score
Outcome N Control N Treat￾ment 
Difference T-Stat 
Diff
P￾value 
Diff
Food consumption score (FCS) (0–
112)
1,916 20.897 2,685 20.212 -0.684 -0.824 0.411
(10.03) (9.28) (0.83)
Percent of households with poor 
FCS (<22)
1,916 69.259 2,685 70.279 1.020 0.248 0.805
(46.15) (45.71) (4.12)
Percent of households with 
borderline FCS (22–35)
1,916 21.555 2,685 22.905 1.350 0.438 0.662
(41.13) (42.03) (3.08)
Percent of households with 
acceptable FCS (>35)
1,916 9.186 2,685 6.816 -2.370 -1.483 0.141
(28.89) (25.21) (1.60)
Notes: Standard deviations are in parentheses. * denotes significance at 10 pct., ** at 5 pct., and *** at 1 pct. Standard errors 
are clustered at the kebele level. 
4.3 Child Nutrition and Feeding Practices 
Table 35. Infant and young children feeding practices 
Outcome N Control N Treat￾ment 
Difference T-Stat 
Diff
P￾value 
Diff
Percent of children under 6 
months exclusively breastfed
40 30 48 47.917 17.917 1.59 0.117
(46.41) (50.49) (11.27)
Percent of children 6–23 months 
receiving a minimum acceptable 
diet
302 2.318 423 1.182 -1.136 -0.98 0.329 
(15.07) (10.82) (1.16) 
Percent of children 6–23 months 
receiving a minimum meal 
frequency
302 64.57 423 65.012 0.442 0.085 0.933 
(47.91) (47.75) (5.22) 
Percent of children 6–23 months 
consuming a diet of a minimum 
diversity
303 2.64 423 1.418 -1.222 -0.863 0.39 
(16.06) (11.84) (1.42)
Notes: Standard deviations are in parentheses. * denotes significance at 10 pct., ** at 5 pct., and *** at 1 pct. 
Standard errors are clustered at the kebele level. 
Table 36. Food groups consumed by children (6–23 months) in the last 24 hours
Outcome N Control N Treat￾ment 
Difference T-Stat 
Diff
P￾value 
Diff
Percent consuming breastmilk 303 81.518 423 79.196 -2.322 -0.634 0.528
Baseline Study of the Ifaa RFSA in Ethiopia (Vol. I)
Comparison of Treatment and Control Groups 67 
Outcome N Control N Treat￾ment
Difference T-Stat 
Diff
P￾value 
Diff
(38.88) (40.64) (3.66)
Percent consuming grains, roots,
tubers 
303 67.987 423 74.704 6.718 1.248 0.214
(46.73) (43.52) (5.38)
Percent consuming legumes and 
nuts 
303 3.3 423 2.837 -0.463 -0.349 0.728
(17.89) (16.62) (1.33)
Percent consuming dairy products
303 15.842 423 12.293 -3.548 -1.027 0.307
(36.57) (32.87) (3.46)
Percent consuming meats
303 0.33 423 0.709 0.379 0.734 0.465
(5.74) (8.40) (0.52)
Percent consuming eggs 
303 2.97 423 1.891 -1.079 -0.82 0.414
(17.00) (13.64) (1.32)
Percent of consuming vitamin-A￾rich vegetables and fruits
303 12.211 423 6.856 -5.355** -2.122 0.036
(32.80) (25.30) (2.52)
Percent consuming other 
vegetable and fruits
303 6.271 423 6.856 0.585 0.297 0.767
(24.28) (25.30) (1.97)
Notes: Standard deviations are in parentheses. * denotes significance at 10 pct., ** at 5 pct., and *** at 1 pct. Standard errors 
are clustered at the kebele level. 
Table 37. Young children (0–59 months) diarrhea incidence and treatment 
Outcome N Control N Treat￾ment 
Difference T-Stat 
Diff
P- value 
Diff
Percent of children under 5 (0–59 
months) who had diarrhea in the 
prior 2 weeks
946 9.619 1221 12.531 2.911 1.359 0.177 
(29.50) (33.12) (2.14) 
Percent of children under 5 (0–59 
months) with diarrhea treated 
with oral rehydration therapy 
(ORT) 
91 94.505 153 92.157 -2.349 -0.677 0.5 
(22.91) (26.97) (3.47) 
Notes: Standard deviations are in parentheses. * denotes significance at 10 pct., ** at 5 pct., and *** at 1 pct. Standard errors 
are clustered at the kebele level. 
IMPEL | Implementer-Led Evaluation and Learning
68 Comparison of Treatment and Control Groups
4.4 Women’s Health, Maternal Nutrition, and Reproductive 
Health 
Table 38. Indicators for women's health, maternal nutrition, and reproductive health
Outcome N Control N Treat￾ment 
Difference T-Stat 
Diff
P￾value 
Diff
Percent of GWRA with MDD
1,928 0.985 2,709 0.775 -0.210 -0.759 0.449
(9.88) (8.77) (0.28)
CPR of non-pregnant GWRA in a 
union
1,110 12.432 1,458 10.562 -1.870 -0.958 0.34
(33.01) (30.75) (1.95)
Percent of births receiving at least 
four ANC visits during pregnancy
693 14.863 929 16.362 1.499 0.494 0.622
(35.60) (37.01) (3.03)
Percent of GWRA in a union who 
have knowledge of modern family 
planning methods
1,270 33.465 1,666 27.191 -6.274* -1.764 0.08
(47.21) (44.51) (3.56) 
Number of contraceptive methods
married women or in a union 
know (0–12)
1,270 1.849 1,666 1.626 -0.223 -1.283 0.202 
(1.97) (2.05) (0.17) 
Percent of women in a union who 
made decisions about modern 
family planning methods in the 
past 12 months
119 47.899 143 52.448 4.548 0.554 0.581 
(50.17) (50.12) (8.21) 
Notes: Standard deviations are in parentheses. * denotes significance at 10 pct., ** at 5 pct., and *** at 1 pct. Standard errors 
are clustered at the kebele level. 
Table 39. Woman of reproductive age’s consumption 
Outcome N Control N Treat￾ment 
Difference T-Stat 
Diff
P- value 
Diff
Grains and roots 
1,928 88.071 2,707 89.767 1.697 0.701 0.484
(32.42) (30.31) (2.42)
Pulses (beans, peas, and lentils) 
1,928 0.83 2,707 0.739 -0.091 -0.275 0.784
(9.07) (8.57) (0.33)
Nuts and seeds (including 
groundnuts) 
1,928 1.089 2,707 1.552 0.462 1.253 0.213
(10.38) (12.36) (0.37)
Dairy 
1,928 4.512 2,707 4.285 -0.227 -0.239 0.811
(20.76) (20.26) (0.95)
Baseline Study of the Ifaa RFSA in Ethiopia (Vol. I)
Comparison of Treatment and Control Groups 69 
Outcome N Control N Treat￾ment
Difference T-Stat 
Diff
P￾value 
Diff
Meat, Poultry, and fish
1,928 0.674 2,707 0.443 -0.231 -1.026 0.307
(8.19) (6.64) (0.23)
Eggs 
1,928 1.66 2,707 1.108 -0.552 -1.509 0.134
(12.78) (10.47) (0.37)
Dark green leafy vegetables 
1,928 7.573 2,707 4.692 -2.881* -1.697 0.092
(26.46) (21.15) (1.70)
Other vitamin-A-rich fruits and 
vegetables 
1,928 48.963 2,707 46.989 -1.973 -0.594 0.554
(50.00) (49.92) (3.32)
Other veg
1,928 18.361 2,707 16.587 -1.774 -0.778 0.438
(38.73) (37.20) (2.28)
Other fruit 
1,928 1.452 2,707 0.591 -0.861* -1.7 0.092
(11.97) (7.67) (0.51)
Notes: Standard deviations are in parentheses. * denotes significance at 10 pct., ** at 5 pct., and *** at 1 pct. Standard errors 
are clustered at the kebele level. 
4.5 Water, Sanitation, and Hygiene Practices 
Table 40. Water, sanitation, and hygiene indicators 
Outcome N Control N Treat￾ment 
Difference T-Stat 
Diff
P￾value 
Diff
BL16: Percent with access to basic 
drinking water services 
133 4.511 220 4.091 -0.420 -0.181 0.857
(20.83) (19.85) (2.33)
Percent with access to basic drinking 
water services based on four of five of 
the criteria
1,949 10.57 2,721 11.65 1.081 0.55 0.583
(30.75) (32.09) (1.96) Percent of households with soap and
water at a handwashing station on the
premises
1,949 3.079 2,732 1.83 -1.248 -1.563 0.121
(17.28) (13.41) (0.80) Percent of households practicing use 
of recommended household water 
treatment technologies
1,947 6.574 2,729 7.915 1.341 0.713 0.477 
 (24.79) (27.00) (1.88) 
Percent of households practicing open 
defecation 
1,949 58.851 2,732 58.638 -0.212 -0.04 0.968
(49.22) (49.26) (5.33)
IMPEL | Implementer-Led Evaluation and Learning
70 Comparison of Treatment and Control Groups
Outcome N Control N Treat￾ment
Difference T-Stat 
Diff
P￾value 
Diff
Percent using improved sanitation 
facilities (not shared)
1,949 16.06 2,732 15.739 -0.320 -0.113 0.91
(36.73) (36.42) (2.83)
Notes: Standard deviations are in parentheses. * denotes significance at 10 pct., ** at 5 pct., and *** at 1 pct. Standard errors 
are clustered at the kebele level. 
4.6 Agriculture
Table 39. Financial services, value chain activities, and natural resource management practices
Outcome N Control N Treat- ment
Difference T-Stat 
Diff
P-value 
Diff
Percent of households using 
financial services
1,860 16.774 2,590 19.498 2.724 1.254 0.212
(37.37) (39.63) (2.17)
Percent of households using 
agricultural credit
1,860 13.817 2,590 17.027 3.210 1.564 0.121
(34.52) (37.59) (2.05)
Percent of households who saved 
1,858 2.799 2,589 2.704 -0.095 -0.139 0.889
(16.50) (16.22) (0.68)
Percent of households using crop
insurance
1,860 1.667 2,589 0.966 -0.701 -1.532 0.128
(12.81) (9.78) (0.46)
Percent of households reporting 
at least one value chain activity
453 76.38 568 75.704 -0.675 -0.089 0.929
(42.52) (42.92) (7.56)
Percent of households using at 
least one NRM practice
1,860 88.763 2,590 89.884 1.121 0.389 0.698
(31.59) (30.16) (2.88)
Notes: Standard deviations are in parentheses. * denotes significance at 10 pct., ** at 5 pct., and *** at 1 pct. Standard errors 
are clustered at the kebele level.
Table 40. Improved management practices or technologies in target crops 
Outcome N Control N Treat￾ment
Difference T-Stat 
Diff
P-value 
Diff
Percent of households growing 
haricot beans
1,693 0.059 2,310 0.043 -0.016 -0.217 0.829
(2.43) (2.08) (0.07)
Percent using at least one 
practice for haricot beans
1 100 1 100 0.000
(.) (.) (.)
Percent of households growing 
groundnuts
1,693 1.299 2,310 1.558 0.259 0.33 0.742
(11.33) (12.39) (0.79)
22 81.818 36 80.556 -1.263 -0.098 0.923
Baseline Study of the Ifaa RFSA in Ethiopia (Vol. I)
Comparison of Treatment and Control Groups 71 
Outcome N Control N Treat￾ment
Difference T-Stat 
Diff
P-value 
Diff
Percent using at least one 
practice for groundnut (39.48) (40.14) (12.88)
Percent of households growing 
potato
1,693 0.295 2,310 1.039 0.744* 1.907 0.059
(5.43) (10.14) (0.39)
Percent using at least one 
practice for potatoes
5 100 24 100 0.000
(0.00) (0.00) (.)
Notes: Standard deviations are in parentheses. * denotes significance at 10 pct., ** at 5 pct., and *** at 1 pct. 
Standard errors are clustered at the kebele level. 
Table 41. Improved management practices or technologies in target animals
Outcome N Control N Treat￾ment 
Difference T-Stat 
Diff
P￾value 
Diff
Percent of households raising 
goats
1,895 26.174 2,642 21.499 -4.675* -1.662 0.099
(43.97) (41.09) (2.81)
Percent of households using at 
least one practice for goats
496 92.54 568 94.366 1.826 0.785 0.434
(26.30) (23.08) (2.33)
Percent of households raising 
poultry
1,895 20.158 2,642 19.152 -1.006 -0.395 0.694
(40.13) (39.36) (2.55)
Percent of households using at 
least one practice for poultry
382 92.67 506 96.64 3.970 1.5 0.136
(26.10) (18.04) (2.65)
Percent of households raising 
oxen 
1,895 7.599 2,642 5.45 -2.149* -1.9 0.06
(26.51) (22.71) (1.13)
Percent of households using at 
least one practice for oxen
144 82.639 144 87.5 4.861 0.628 0.532
(38.01) (33.19) (7.75)
Notes: Standard deviations are in parentheses. * denotes significance at 10 pct., ** at 5 pct., and *** at 1 pct. 
Standard errors are clustered at the kebele level. 
4.7 Poverty Measurement 
Table 42. Poverty measurement indicators 
Outcome N Control N Treat- ment
Difference T-Stat 
Diff
P-value 
Diff
Daily per capita food expenditure 
(Birr) 
1,947 17.742 2,730 17.564 -0.178 -0.141 0.888
(11.04) (10.99) (1.26)
1,947 21.566 2,730 21.218 -0.348 -0.231 0.818
IMPEL | Implementer-Led Evaluation and Learning
72 Comparison of Treatment and Control Groups
Outcome N Control N Treat￾ment
Difference T-Stat 
Diff
P-value 
Diff
Daily per adult equivalent food 
expenditure (Birr) (13.38) (13.18) (1.51)
Household food expenditure per 
day (Birr)
1,947 99.109 2,730 95.035 -4.074 -0.616 0.539
(56.04) (53.17) (6.61)
Percent poor (per adult 
equivalent consumption 
expenditure)
1,947 95.591 2,730 96.587 0.996 0.888 0.376
 (20.54) (18.16) (1.12) 
Depth of poverty of poor (per 
adult equivalent consumption
expenditure)
1,820 58.448 2,579 58.791 0.343 0.14 0.889 
(21.46) (20.52) (2.45)
Notes: Standard deviations are in parentheses. * denotes significance at 10 pct., ** at 5 pct., and *** at 1 pct. 
Standard errors are clustered at the kebele level. 
4.8 Gender Dynamics 
4.8.1 Use of Financial Resources
Table 43. Gender (cash) indicators 
Outcome N Control N Treat￾ment 
Difference T-Stat 
Diff
P￾value 
Diff
Percent of cash earners in a union 
1,479 27.045 1,965 20.051 -6.994* -1.732 0.086
(44.43) (40.05) (4.04)
Percent of cash-earning men in a 
union
820 32.683 1,010 26.139 -6.544 -1.268 0.207
(46.93) (43.96) (5.16)
Percent of cash-earning women in 
a union
659 20.03 955 13.613 -6.418 -1.456 0.148
(40.05) (34.31) (4.41)
Percent of women in a union and 
earning cash who report 
participation in decisions about 
the use of self-earned cash
132 71.97 130 69.231 -2.739 -0.212 0.833
(45.09) (46.33) (12.92) Percent of women in a union and 
earning cash who report 
participation in decisions about 
the use of spouse/partner’s self￾earned cash
132 46.212 130 41.538 -4.674 -0.369 0.713 
(50.05) (49.47) (12.65) 
Baseline Study of the Ifaa RFSA in Ethiopia (Vol. I)
Comparison of Treatment and Control Groups 73 
Outcome N Control N Treat￾ment
Difference T-Stat 
Diff
P￾value 
Diff
Percent of men in a union and 
earning cash who report 
spouse/partner’s participation in 
decisions about the sue of self￾earned cash 
268 36.94 264 45.076 8.135 1.11 0.271
(48.35) (49.85) (7.33) 
Notes: Standard deviations are in parentheses. * denotes significance at 10 pct., ** at 5 pct., and *** at 1 pct. Standard errors 
are clustered at the kebele level. 
4.8.2 Credit 
Table 44. Gender access to credit and group participation indicators
Outcome N Control N Treatm
ent 
Difference T-Stat 
Diff
P- value 
Diff
Percent of people in a union who 
are members of a community
group
340 67.353 659 55.235 -12.118 -1.178 0.242
(46.96) (49.76) (10.29) 
Percent of men in a union who are 
members of a community group
209 72.249 354 62.147 -10.102 -0.967 0.336
(44.88) (48.57) (10.45)
Percent of women in a union who 
are members of a community
group
131 59.542 305 47.213 -12.329 -1.152 0.253
(49.27) (50.00) (10.70) 
Percent of people in a union with 
access to credit
1,479 22.583 1,965 31.603 9.020* 1.724 0.087
(41.83) (46.50) (5.23)
Percent of men in a union with 
access to credit
820 25.122 1,010 32.97 7.848 1.423 0.157
(43.40) (47.03) (5.52)
Percent of women in a union with 
access to credit
659 19.423 955 30.157 10.734* 1.876 0.063
(39.59) (45.92) (5.72)
Percent of men and women in a 
union who make decisions about
credit
334 29.042 621 29.791 0.749 0.119 0.906 
(45.46) (45.77) (6.31) 
Percent of men in a union who 
report making decisions about 
credit
206 24.757 333 31.532 6.774 0.919 0.361 
(43.27) (46.53) (7.37) 
Percent of women in a union who 
report making decisions about
credit
128 35.938 288 27.778 -8.160 -1.162 0.249 
(48.17) (44.87) (7.02)
IMPEL | Implementer-Led Evaluation and Learning
74 Comparison of Treatment and Control Groups
Notes: Standard deviations are in parentheses. * denotes significance at 10 pct., ** at 5 pct., and *** at 1 pct. Standard errors 
are clustered at the kebele level. 
4.9 Resilience 
4.9.1 Ability to Recover from Shocks and Stresses Index 
Table 45. Ability to recover from shocks and stresses index 
Outcome N Control N Treat￾ment
Difference T-Stat 
Diff
P-value 
Diff
Mean ability to recover from 
shocks and stresses index 
1,884 2.922 2,661 3.054 0.132 1.05 0.296
(1.34) (1.37) (0.13)
Ability to recover index (2–6) 
1,934 2.907 2,715 3.034 0.127 0.974 0.332
(1.33) (1.38) (0.13)
Shock exposure index (0–168) 
1,899 23.638 2,676 24.943 1.305 0.778 0.438
(13.48) (15.01) (1.68)
Total shocks experienced (0–21) 
1,950 3.511 2,733 3.665 0.154 0.632 0.528
(2.07) (2.27) (0.24)
Notes: Standard deviations are in parentheses. * denotes significance at 10 pct., ** at 5 pct., and *** at 1 pct. Standard errors 
are clustered at the kebele level. 
4.9.2 Absorptive Capacity Index 
Table 44. Absorptive capacity index and subindexes 
Outcome N Control N Treat￾ment 
Difference T-Stat 
Diff
P￾value 
Diff
Absorptive capacity index (0–100) 
1,857 29.449 2,585 28.132 -1.317 -0.726 0.469
(15.71) (15.30) (1.81)
Absorptive capacity index (0–100): 
No access to insurance 
1,944 30.604 2,718 29.493 -1.110 -0.629 0.53
(15.22) (15.04) (1.76)
Bonding social capital index (0–6) 
1,949 2.555 2,732 2.553 -0.002 -0.01 0.992
(1.89) (1.86) (0.21)
Access to cash savings index (0–1) 
1,949 0.018 2,731 0.015 -0.003 -0.554 0.581
(0.13) (0.12) (0.01)
Remittances index (0–1) 
1,950 0.024 2,733 0.024 -0.000 -0.044 0.965
(0.15) (0.15) (0.01)
1,946 8.6 2,719 8.036 -0.563* -1.83 0.07
Baseline Study of the Ifaa RFSA in Ethiopia (Vol. I)
Comparison of Treatment and Control Groups 75 
Outcome N Control N Treat￾ment
Difference T-Stat 
Diff
P￾value 
Diff
Asset ownership index—total type 
(0–45) (3.77) (3.54) (0.31)
Shock preparedness and 
responsiveness index (0–3)
1,949 0.394 2,732 0.389 -0.004 -0.088 0.93
(0.57) (0.56) (0.05)
Access to insurance index (0–1) 
1,860 0.017 2,590 0.01 -0.007 -1.533 0.128
(0.13) (0.10) (0.00)
Access to humanitarian assistance 
index (0–1)
1,947 0.503 2,732 0.492 -0.012 -0.184 0.854
(0.50) (0.50) (0.06)
Notes: Standard deviations are in parentheses. * denotes significance at 10 pct., ** at 5 pct., and *** at 1 pct. 
Standard errors are clustered at the kebele level.
4.9.3 Social Capital Index
Table 45. Social capital index and subindexes 
Outcome N Control N Treat￾ment 
Difference T-Stat 
Diff
P- value 
Diff
Index of social capital at 
household level (0–100) 
1,949 50.462 2,732 49.876 -0.585 -0.162 0.871
(34.10) (33.74) (3.61)
Bonding subindex (0–100) 
1,949 54.336 2,732 54.1 -0.236 -0.064 0.949
(35.70) (35.83) (3.69)
Bridging subindex (0–100) 
1,949 46.588 2,732 45.653 -0.935 -0.243 0.808
(36.58) (36.36) (3.84)
Notes: Standard deviations are in parentheses. * denotes significance at 10 pct., ** at 5 pct., and *** at 1 pct. Standard errors 
are clustered at the kebele level. 
IMPEL | Implementer-Led Evaluation and Learning
76 Comparison of Treatment and Control Groups
4.9.4 Adaptive Capacity index
Table 46. Adaptive capacity index and indexes 
Outcome N Control N Treat￾ment
Difference T-Stat 
Diff
P￾value 
Diff
Adaptive capacity index (0–100) 
216 47.065 300 40.839 -6.226 -0.948 0.347
(20.30) (23.77) (6.57)
Adaptive capacity index (0–100): 
No adopt improved practices 
785 43.692 1,179 39.003 -4.689 -1.099 0.274
(19.96) (19.50) (4.27)
Adaptive capacity index (0–100): 
No access to finance s. practices 
389 39.406 459 36.154 -3.252 -0.657 0.513
(21.12) (21.58) (4.95)
Adaptive capacity index (0–100): 
No access to finance s. practices 
and improved
1,690 36.92 2,270 35.199 -1.721 -0.625 0.533 
(19.27) (18.64) (2.75) 
Aspirations/confidence to adapt 
index (0–16) 
1,812 9.792 2,478 9.69 -0.103 -0.637 0.525
(2.31) (2.24) (0.16)
Bridging social capital index (0–6) 
1,949 2.202 2,732 2.149 -0.052 -0.247 0.805
(1.88) (1.84) (0.21)
Linking social capital (0–4) 
1,949 0.501 2,729 0.451 -0.050 -0.399 0.691
(1.01) (0.96) (0.13)
Social network index (adjusted 
range 0–3) 
1,830 1.139 2,538 1.041 -0.098 -0.677 0.5
(1.32) (1.32) (0.14)
Education/training index (0–8)
1,943 1.016 2,726 0.955 -0.061 -0.617 0.538
(1.21) (1.23) (0.10)
Livelihood diversification index (0–
20) 
1,949 1.848 2,732 1.776 -0.072 -1.184 0.239
(0.77) (0.76) (0.06)
Adoption of improved practices 
index (0–1) 
453 0.375 568 0.335 -0.041 -0.551 0.583
(0.48) (0.47) (0.07)
Access to financial resources (0–1) 
868 0.3 1372 0.365 0.066 0.629 0.531
(0.46) (0.48) (0.10)
Exposure to information index (0–
19) 
1,949 1.745 2,732 1.751 0.006 0.026 0.979
(2.26) (2.47) (0.23)
1,946 8.6 2,719 8.036 -0.563* -1.83 0.07
Baseline Study of the Ifaa RFSA in Ethiopia (Vol. I)
Comparison of Treatment and Control Groups 77 
Outcome N Control N Treat￾ment
Difference T-Stat 
Diff
P￾value 
Diff
Asset ownership index— total 
type (0–45) (3.77) (3.54) (0.31)
Notes: Standard deviations are in parentheses. * denotes significance at 10 pct., ** at 5 pct., and *** at 1 pct. Standard errors 
are clustered at the kebele level. 
4.9.5 Transformative Capacity Index 
Table 47. Transformative capacity index and indexes
Outcome N Control N Treat￾ment
Difference T-Stat 
Diff
P-value 
Diff
Transformative capacity index (0–
100) 
82 42.016 118 45.141 3.125 0.302 0.766
(25.51) (19.39) (10.34)
Transformative capacity index (0–
100): Excluding agricultural 
services and gender equity 
1,784 42.375 2,506 40.927 -1.448 -0.404 0.687 
(21.63) (19.34) (3.58) 
Transformative capacity index (0–
100): Excluding gender equity 
index 
1,784 37.484 2,506 35.365 -2.119 -0.545 0.587 
(23.05) (20.05) (3.89) 
Transformative capacity index (0– 100): Excluding agricultural 
Services 
82 41.949 118 46.717 4.768 0.497 0.625 
(22.42) (21.81) (9.59) 
Access to formal safety nets index 
(0–11) 
1,842 2.071 2,618 2.043 -0.028 -0.218 0.828
(0.84) (0.76) (0.13)
Access to communal natural 
resources index (0–4) 
1,950 1.372 2,732 1.23 -0.142 -0.52 0.604
(1.70) (1.59) (0.27)
Basic services index (0–1): Only 
police variable 
1,950 0.981 2,733 0.985 0.004 0.493 0.623
(0.14) (0.12) (0.01)
Access to agricultural services 
index (0–1) 
1,950 0.154 2,733 0.103 -0.051 -0.85 0.397
(0.36) (0.30) (0.06)
Bridging social capital index (0–6) 
1,949 2.202 2,732 2.149 -0.052 -0.247 0.805
(1.88) (1.84) (0.21)
Linking social capital (0–4) 
1,949 0.501 2,729 0.451 -0.050 -0.399 0.691
(1.01) (0.96) (0.13)
Social cohesion index (0–3) 
1,908 1.191 2,663 1.043 -0.148 -0.948 0.345
(1.14) (1.07) (0.16)
IMPEL | Implementer-Led Evaluation and Learning
78 Comparison of Treatment and Control Groups
Outcome N Control N Treat￾ment
Difference T-Stat 
Diff
P-value 
Diff 
Local decision making index (0–1)
1,930 0.929 2,689 0.899 -0.030 -1.205 0.23
(0.26) (0.30) (0.03)
Local government responsiveness 
index (0–2) 
1,950 1.829 2,733 1.843 0.014 0.299 0.765
(0.39) (0.36) (0.05)
Gender index (0–3) 
1,950 2.653 2,733 2.699 0.046 0.641 0.523
(0.57) (0.54) (0.07)
Gender equitable decision making 
index (0–1) 
105 0.762 154 0.682 -0.080 -0.387 0.702
(0.42) (0.46) (0.21)
Notes: Standard deviations are in parentheses. * denotes significance at 10 pct., ** at 5 pct., and *** at 1 pct. Standard errors 
are clustered at the kebele level. 
IMPEL | Implementer-Led Evaluation and Learning
79 Comparison of Livelihood and Non-Livelihood Kebeles
5. COMPARISON OF LIVELIHOOD AND NON-LIVELIHOOD 
KEBELES 
As was discussed in Section 3, half of the 120 kebeles in the impact evaluation study are eligible to 
receive livelihood interventions, while the other half are not. Researchers randomly selected the 60 
livelihood kebeles in the impact evaluation study from a list of 100 livelihood-eligible kebeles. Because 
the eligibility criteria to receive livelihood interventions was based on a specific set of criteria (e.g., 
proximity to market)- it is to be expected that the two groups of kebeles differ along different 
dimensions. To better understand these differences, in this section, we present the balance tests run 
between livelihood and non-livelihood kebeles. It is important to notice that differences across 
livelihood statuses don’t pose any threat to the IE. The information presented here is complementary to 
the IE and allows a better understanding of the main differences across livelihood status. 
The different subsections below display tables containing similar outcome variables as the ones 
discussed in Section 4. We run a similar regression as in the previous section. In Equation 2. Balance test 
regression for livelihood and non-livelihood kebeles, I denotes the household or individual, denotes 
the household or individual, stands for the outcome variable and is equal to 1 if 
household lives in a livelihood kebele. 
Equation 2. Balance test regression for livelihood and non-livelihood kebeles = + +
The variable of interest in this section is , which represents the difference between livelihood and non￾livelihood kebeles. For each outcome variable, we show the mean value for the livelihood and non￾livelihood groups, the coefficient, as well as its associated T-statistic and the p-value. Standard errors 
are clustered at the kebele level. 
We ran 141 balance tests. There were 32 variables with a statistically significant difference across 
treatment arms at a 10% significance level, 23 at a 5% significance level, and five at a 1% significance 
level. These values are higher than the ones found in the previous section, which reflects the fact that 
livelihood kebeles were purposefully selected. In the share of significant differences across sections, four 
sections account for most of the variables (26 out of 32) with a significant difference. These sections are 
characteristics of the study population; child nutrition and feeding practices; women’s health, nutrition, 
and reproductive health; and gender dynamics. Below we briefly discuss some of the differences. It is 
important to notice that even though the differences are statistically significant, most of them are small. 
Overall, we don’t find substantial differences between livelihood and non-livelihood kebeles. 
Table 48 presents the demographic variables with a statistically significant difference. Eleven (out of 18) 
demographic variables were statistically different, and these were distributed almost equally between 
individual-level and household-level characteristics. Livelihood kebeles have fewer children 5–14 years 
and more adults and women. Additionally, the average adult and HH living in a livelihood kebele is less 
likely to be married or in a union and, for female HHs, more likely to have some schooling. 
IMPEL | Implementer-Led Evaluation and Learning
80 Comparison of Livelihood and Non-Livelihood Kebeles
Even though the differences are statistically significant, it is important to notice that, for most variables, 
the differences are not very large. For example, the average household size in a livelihood kebele is 
5.795, compared to 6.107 in non-livelihood kebeles. 
Table 48. Characteristics of the study population, variables with a significant difference across 
livelihood status 
Outcome N 
Non￾liveli￾hood
N Liveli￾hood Difference T-Stat 
Diff
P-value 
Diff
Individual-level characteristics
Average age
14,309 18.817 13,560 19.516 0.699** 2.525 0.013
(16.05) (16.20) (0.28)
Percent of children 5–14 years 
old 
14,309 39.681 13,560 38.193 -1.488* -1.952 0.053
(48.93) (48.59) (0.76)
Percent of adults (older than 
15 years old) 
14,309 45.74 13,560 48.355 2.615*** 2.95 0.004
(49.82) (49.97) (0.89)
Percent of females 
14,309 49.235 13,560 50.723 1.488*** 2.753 0.007
(50.00) (50.00) (0.54)
Percent of GWRA 
14,309 22.084 13,560 23.496 1.412*** 3.007 0.003
(41.48) (42.40) (0.47)
Percent of adults who are in a 
union
6,545 58.824 6,557 55.574 -3.249** -2.392 0.018
(49.22) (49.69) (1.36)
Household-level characteristics
Average household size 
2,343 6.107 2,340 5.795 -0.312*** -2.706 0.008
(2.18) (2.14) (0.12)
Average number of children 
under the age of 5 in the 
household
2,343 0.89 2,340 0.779 -0.111** -2.223 0.028 
 (0.97) (0.91) (0.05) 
Percent of HH who are in a 
union 
2,343 78.873 2,340 75.513 -3.360* -1.904 0.059
(40.83) (43.01) (1.77)
Percent of HH that are female
2,343 23.517 2,340 27.35 3.834** 2.032 0.044
(42.42) (44.59) (1.89)
Percent of female HH with 
some schooling
551 1.452 640 3.75 2.298** 2.237 0.027
(11.97) (19.01) (1.03)
Notes: Standard deviations are in parentheses. * denotes significance at 10 pct., ** at 5 pct., and *** at 1 pct. Standard errors 
are clustered at the kebele level. 
Baseline Study of the Ifaa RFSA in Ethiopia (Vol. I)
Comparison of Livelihood and Non-Livelihood Kebeles 81 
Table 49 shows that in livelihood kebeles, fewer children 6–23 months of age tend to consume a 
minimum acceptable diet and a diet of minimum diversity. Keeping in mind that in the whole sample, 
the percentage of children is very low, children 6–23 months in non-livelihood kebeles are three and 
four times more likely to consume a minimum acceptable diet and a diet of minimum diversity. The 
differences in these two indicators are mostly determined by differences in children’s consumption 
patterns. Children 6–23 months in livelihood kebeles are less likely to consume breastmilk, meat, and 
vitamin-A-rich vegetables but are more likely to consume legumes and nuts. 
Table 49. Child nutrition and feeding practices, variables with a significant difference across livelihood 
status 
Outcome N 
Non
Liveli￾hood
N Liveli￾hood Difference T-Stat 
Diff
P￾value 
Diff
Infant and young children feeding practices
Percent of children 6–23 months 
receiving a minimum acceptable 
diet
373 2.681 353 0.567 -2.114** -2.074 0.04 
(16.17) (7.52) (1.02)
Percent of children 6–23 months 
consuming a diet of a minimum 
diversity
373 3.217 353 0.567 -2.651** -2.128 0.035 
(17.67) (7.52) (1.25) 
Food consumption groups 
Percent consuming breastmilk 
373 83.914 353 76.204 -7.710** -2.2 0.03
(36.79) (42.64) (3.51)
Percent consuming legumes and 
nuts 
373 1.609 353 4.533 2.924** 2.281 0.024
(12.60) (20.83) (1.28)
Percent consuming meats
373 1.072 353 0 -1.072** -2.082 0.04
(10.31) (0.00) (0.52)
Percent of consuming vitamin-A￾rich vegetables and fruits
373 11.528 353 6.516 -5.013** -2.023 0.045
(31.98) (24.72) (2.48)
Notes: Standard deviations are in parentheses. * denotes significance at 10 pct., ** at 5 pct., and *** at 1 pct. Standard errors 
are clustered at the kebele level. 
As with children, there is a significant difference in GWRA’s consumption patterns (Table 50). Women in 
livelihood kebeles tend to consume fewer nuts and seeds and less fruit but more vitamin-A-rich fruits 
and vegetables and other vegetables. Despite the differences in consumption patterns, the share of 
women with MDD is not statistically different across livelihood status. The share of women with MDD in 
the whole sample is very low (0.86%), and more than 90% of women consume less than three food 
groups. Variations in a few items consumed will not lead to sizable differences in the share of women 
consuming a diet of minimum diversity. 
 
IMPEL | Implementer-Led Evaluation and Learning
82 Comparison of Livelihood and Non-Livelihood Kebeles
Table 50. GWRA age consumption, variables with a significant difference across livelihood status
Outcome N 
Non
Liveli￾hood
N Liveli￾hood Difference T-Stat 
Diff
P￾value 
Diff
Nuts and seeds (including 
groundnuts) 
2,317 1.683 2,318 1.035 -0.648* -1.711 0.09
(12.87) (10.12) (0.38)
Other vitamin-A-rich fruits and 
vegetables 
2,317 44.281 2,318 51.337 7.056** 2.207 0.029
(49.68) (49.99) (3.20)
Other vegetables 
2,317 14.243 2,318 20.406 6.163*** 2.814 0.006
(34.96) (40.31) (2.19)
Other fruit
2,317 1.338 2,318 0.561 -0.777* -1.765 0.08
(11.49) (7.47) (0.44)
Notes: Standard deviations are in parentheses. * denotes significance at 10 pct., ** at 5 pct., and *** at 1 pct. Standard errors 
are clustered at the kebele level. 
Table 51 presents the variables with statistically significant differences in the gender dynamics section. 
Credit-related variables account for most of the differences. In general, more people in a union tend to 
belong to a community group and to a union with access to credit. In the case of financial variables, 
women in a union living in livelihood kebeles are more likely to participate in decisions related to using a 
spouse or partner’s self-earned cash. The magnitude of the differences in this section is not negligible. 
For example, in livelihood kebeles, people in a union are 11% more likely to belong to a union with 
access to credit. There are even higher differences in the financial resources variable, where the 
differences are more than 23%. 
Table 51. Gender dynamics, variables with a significant difference across livelihood status 
Outcome N 
Non
Liveli￾hood
N Liveli￾hood Difference T-Stat 
Diff
P￾value 
Diff
Use of financial resources
Percent of women in a union and 
earning cash who report 
participation in decisions about 
the use of spouse/partner’s self￾earned cash
105 29.524 157 53.503 23.979** 2.222 0.03 
(45.83) (50.04) (10.79) 
Credit
Percent of women in a union who 
are members of a community
group
153 54.248 283 49.117 -5.132 -0.507 0.613 
(49.98) (50.08) (10.12) 
Percent of people in a union with 
access to credit
1621 21.592 1,823 33.187 11.595** 2.272 0.025
(41.16) (47.10) (5.10)
Baseline Study of the Ifaa RFSA in Ethiopia (Vol. I)
Comparison of Livelihood and Non-Livelihood Kebeles 83 
Outcome N 
Non
Liveli￾hood
N Liveli￾hood Difference T-Stat 
Diff
P￾value 
Diff
Percent of men in a union with 
access to credit
865 23.699 965 34.611 10.912** 2.043 0.043
(42.55) (47.60) (5.34)
Percent of women in a union with 
access to credit
756 19.18 858 31.585 12.405** 2.151 0.033
(39.40) (46.51) (5.77)
Percent of people in a union who
make decisions about credit
350 28 605 30.413 2.413 0.403 0.688
(44.96) (46.04) (5.99)
Percent of men in a union who 
report making decisions about 
credit
205 24.878 334 31.437 6.559 0.932 0.354 
(43.34) (46.50) (7.03) 
Percent of women in a union who 
report making decisions about
credit
145 32.414 271 29.151 -3.263 -0.513 0.609 
(46.97) (45.53) (6.36) 
Notes: Standard deviations are in parentheses. * denotes significance at 10 pct., ** at 5 pct., and *** at 1 pct. Standard errors 
are clustered at the kebele level. 
5.1 Characteristics of the Study Population 
Table 52. Individual-level sample characteristics across livelihood status 
Outcome N 
Non￾liveli￾hood
N Liveli￾hood Difference T-Stat 
Diff
P￾value 
Diff
Average age 
14,309 18.817 13,560 19.516 0.699** 2.525 0.013
(16.05) (16.20) (0.28)
Percent of children under 5 
years old
14,309 14.578 13,560 13.451 -1.127 -1.624 0.107
(35.29) (34.12) (0.69)
Percent of children 5–14 years 
old 
14,309 39.681 13,560 38.193 -1.488* -1.952 0.053
(48.93) (48.59) (0.76)
Percent of adults (older than 
15 years old) 
14,309 45.74 13,560 48.355 2.615*** 2.95 0.004
(49.82) (49.97) (0.89)
Percent of females 
14,309 49.235 13,560 50.723 1.488*** 2.753 0.007
(50.00) (50.00) (0.54)
Percent of GWRA 
14,309 22.084 13,560 23.496 1.412*** 3.007 0.003
(41.48) (42.40) (0.47)
6,545 58.824 6,557 55.574 -3.249** -2.392 0.018
IMPEL | Implementer-Led Evaluation and Learning
84 Comparison of Livelihood and Non-Livelihood Kebeles
Outcome N 
Non￾liveli￾hood
N Liveli￾hood Difference T-Stat 
Diff
P￾value 
Diff
Percent of adults who are in a 
union (49.22) (49.69) (1.36)
Percent of adults with at least 
some schooling
6,545 25.561 6,557 27.131 1.570 0.896 0.372
(43.62) (44.47) (1.75)
Percent of adults who are 
farmers 
6,543 65.78 6,555 66.606 0.825 0.328 0.743
(47.45) (47.17) (2.51)
Percent of people older than 
10 years who did work and 
were paid in cash
6,763 30.977 6,715 31.929 0.951 0.332 0.741
(46.24) (46.62) (2.87)
Notes: Standard deviations are in parentheses. * denotes significance at 10 pct., ** at 5 pct., and *** at 1 pct. Standard errors 
are clustered at the kebele level. 
Table 53. Household-level sample characteristics across livelihood status 
Outcome N 
Non￾liveli￾hood 
N Liveli￾hood Difference T-Stat 
Diff
P-value 
Diff
Average household size 
2,343 6.107 2,340 5.795 -0.312*** -2.706 0.008
(2.18) (2.14) (0.12)
Average number of children 
under the age of 5 in the 
household
2,343 0.89 2,340 0.779 -0.111** -2.223 0.028 
 (0.97) (0.91) (0.05) 
Average age of HH
2,343 41.897 2,340 41.651 -0.246 -0.534 0.594
(11.20) (10.73) (0.46)
Percent of HH who are in a 
union 
2,343 78.873 2,340 75.513 -3.360* -1.904 0.059
(40.83) (43.01) (1.77)
Percent of HH that are female
2,343 23.517 2,340 27.35 3.834** 2.032 0.044
(42.42) (44.59) (1.89)
Percent of HH with some 
schooling 
2,343 13.487 2,340 13.504 0.017 0.011 0.992
(34.17) (34.18) (1.63)
Percent of female HH with some 
schooling
551 1.452 640 3.75 2.298** 2.237 0.027
(11.97) (19.01) (1.03)
Percent of male HH with some 
schooling
1,792 17.188 1,700 17.176 -0.011 -0.006 0.996
(37.74) (37.73) (2.00)
Notes: Standard deviations are in parentheses. * denotes significance at 10 pct., ** at 5 pct., and *** at 1 pct. Standard errors 
are clustered at the kebele level. 
Baseline Study of the Ifaa RFSA in Ethiopia (Vol. I)
Comparison of Livelihood and Non-Livelihood Kebeles 85 
5.2 Food Security 
Table 54. Food insecurity experience scale across livelihood status 
Outcome N 
Non￾liveli￾hood 
N Liveli￾hood Difference T-Stat 
Diff
P-value 
Diff
Raw FIES score (scale 0–8) 
2,343 6.921 2,338 6.827 -0.093 -1.002 0.318
(1.39) (1.41) (0.09)
Percent of households that
experienced severe food 
insecurity 
2,343 78.19 2,338 77.032 -1.159 -0.398 0.692
(41.30) (42.07) (2.91)
Percent of households that
experienced moderate-to-severe
food insecurity
2,343 96.714 2,338 95.552 -1.162 -1.214 0.227
(17.83) (20.62) (0.96) 
Percent of households with some 
indication of food insecurity
2,340 99.444 2,338 99.53 0.085 0.355 0.723
(7.43) (6.84) (0.24)
Percent of households that
answered yes to all eight
questions
2,343 40.546 2,338 34.944 -5.602 -1.404 0.163 
(49.11) (47.69) (3.99) 
Percent of households that 
answered no to all eight questions
2,343 0.555 2,338 0.47 -0.084 -0.352 0.725
(7.43) (6.84) (0.24)
Notes: Standard deviations are in parentheses. * denotes significance at 10 pct., ** at 5 pct., and *** at 1 pct. Standard errors 
are clustered at the kebele level. 
Table 55. Food consumption score across livelihood status 
Outcome N 
Non￾liveli￾hood
N Liveli￾hood Difference T-Stat 
Diff
P￾value 
Diff
FCS (0-112) 
2,301 20.264 2,300 20.731 0.467 0.578 0.564
(9.97) (9.22) (0.81)
Percent of households with poor 
FCS (<22)
2,301 72.056 2,300 67.652 -4.403 -1.113 0.268
(44.88) (46.79) (3.96)
Percent of households with 
borderline FCS (22–35)
2,301 19.731 2,300 24.957 5.226* 1.781 0.077
(39.81) (43.29) (2.93)
Percent of households with 
acceptable FCS (>35) 
2,301 8.214 2,300 7.391 -0.823 -0.529 0.598
(27.46) (26.17) (1.55)
Notes: Standard deviations are in parentheses. * denotes significance at 10 pct., ** at 5 pct., and *** at 1 pct. Standard errors 
are clustered at the kebele level. 
IMPEL | Implementer-Led Evaluation and Learning
86 Comparison of Livelihood and Non-Livelihood Kebeles
5.3 Child Nutrition and Feeding Practices 
Table 56. Infant and young children feeding practices across livelihood status 
Outcome N 
Non￾liveli￾hood 
N Liveli￾hood Difference T-Stat 
Diff
P-value 
Diff
Percent of children under 6 
months exclusively breastfed
46 34.783 42 45.238 10.455 0.859 0.394
(48.15) (50.38) (12.18)
Percent of children 6–23 months 
receiving a minimum acceptable 
diet
373 2.681 353 0.567 -2.114** -2.074 0.04
(16.17) (7.52) (1.02)
Percent of children 6–23 months 
receiving a minimum meal 
frequency
372 66.935 353 62.606 -4.329 -0.863 0.39
 (47.11) (48.45) (5.02) 
Percent of children 6–23 months 
consuming a diet of a minimum 
diversity
373 3.217 353 0.567 -2.651** -2.128 0.035 
 (17.67) (7.52) (1.25) 
Notes: Standard deviations are in parentheses. * denotes significance at 10 pct., ** at 5 pct., and *** at 1 pct. Standard errors 
are clustered at the kebele level. 
Table 57. Food groups consumed by children (6–23 months) in the last 24 hours across livelihood 
status 
Outcome N 
Non￾liveli￾hood 
N Liveli￾hood Difference T-Stat 
Diff
P-value 
Diff
Percent consuming breastmilk 
373 83.914 353 76.204 -7.710** -2.2 0.03
(36.79) (42.64) (3.51)
Percent consuming grains, roots, 
tubers 
373 72.118 353 71.671 -0.447 -0.086 0.932
(44.90) (45.12) (5.21)
Percent consuming legumes and 
nuts 
373 1.609 353 4.533 2.924** 2.281 0.024
(12.60) (20.83) (1.28)
Percent consuming dairy products
373 16.354 353 11.048 -5.306 -1.631 0.105
(37.04) (31.39) (3.25)
Percent consuming meats
373 1.072 353 0 -1.072** -2.082 0.04
(10.31) (0.00) (0.52)
Percent consuming eggs 
373 2.949 353 1.7 -1.249 -1.015 0.312
(16.94) (12.94) (1.23)
373 11.528 353 6.516 -5.013** -2.023 0.045
Baseline Study of the Ifaa RFSA in Ethiopia (Vol. I)
Comparison of Livelihood and Non-Livelihood Kebeles 87 
Outcome N 
Non￾liveli￾hood 
N Liveli￾hood Difference T-Stat 
Diff
P-value 
Diff
Percent of consuming vitamin-A- rich vegetables and fruits (31.98) (24.72) (2.48)
Percent consuming other 
vegetable and fruits
373 7.507 353 5.666 -1.841 -0.99 0.324
(26.39) (23.15) (1.86)
Notes: Standard deviations are in parentheses. * denotes significance at 10 pct., ** at 5 pct., and *** at 1 pct. Standard errors 
are clustered at the kebele level. 
Table 58. Young children diarrhea incidence and treatment (children 0–59 months) across livelihood 
status 
Outcome N 
Non￾liveli￾hood 
N Liveli￾hood Difference T-Stat 
Diff
P-value 
Diff
Percent of children under 5 (–-59 
months) who had diarrhea in the 
prior 2 weeks
1,161 10.767 1,006 11.829 1.062 0.488 0.626
 (31.01) (32.31) (2.18) 
Percent of children under 5 (0–
59 months) with diarrhea treated 
with ORT
125 95.2 119 90.756 -4.444 -1.205 0.232
 (21.46) (29.09) (3.69) 
Notes: Standard deviations are in parentheses. * denotes significance at 10 pct., ** at 5 pct., and *** at 1 pct. Standard errors 
are clustered at the kebele level. 
5.4 Women’s Health, Maternal Nutrition and Reproductive 
Health 
Table 59. Indicators for women's health, maternal nutrition, and reproductive health across livelihood 
status 
Outcome N 
Non￾liveli￾hood 
N Liveli￾hood Difference T-Stat 
Diff
P-value 
Diff
Percent of GWRA with MDD 
2,319 0.69 2,318 1.035 0.345 1.29 0.2
(8.28) (10.12) (0.27)
Contraceptive prevalence rate of 
non-pregnant GWRA in a union
1,325 11.925 1,243 10.78 -1.144 -0.608 0.544
(32.42) (31.03) (1.88)
Percent of births receiving at 
least four ANC visits during 
pregnancy
833 15.126 789 16.35 1.224 0.411 0.682 
 (35.85) (37.01) (2.98) 
Percent of GWRA in a union who 1,516 29.881 1,420 29.93 0.048 0.014 0.989 
IMPEL | Implementer-Led Evaluation and Learning
88 Comparison of Livelihood and Non-Livelihood Kebeles
Outcome N 
Non￾liveli￾hood 
N Liveli￾hood Difference T-Stat 
Diff
P-value 
Diff
have knowledge of modern 
family planning methods (45.79) (45.81) (3.58)
Number of contraceptive 
methods married women or in a 
union know (0–12)
1,516 1.699 1,420 1.747 0.048 0.274 0.785
 (1.95) (2.08) (0.18) 
Percent of women in a union 
who made decision about 
modern family planning methods 
in the past 12 months
143 53.147 119 47.059 -6.088 -0.739 0.462 
 (50.08) (50.12) (8.24) 
Notes: Standard deviations are in parentheses. * denotes significance at 10 pct., ** at 5 pct., and *** at 1 pct. Standard errors 
are clustered at the kebele level. 
Table 60. Woman of reproductive age’s consumption across livelihood status 
Outcome N 
Non￾liveli￾hood 
N Liveli￾hood Difference T-Stat 
Diff
P-value 
Diff
Grains and roots 
2,317 87.872 2,318 90.25 2.378 1.006 0.317
(32.65) (29.67) (2.36)
Pulses (beans, peas, and lentils) 
2317 0.518 2,318 1.035 0.517 1.542 0.126
(7.18) (10.12) (0.34)
Nuts and seeds (including 
groundnuts) 
2317 1.683 2,318 1.035 -0.648* -1.711 0.09
(12.87) (10.12) (0.38)
Dairy 
2317 4.92 2,318 3.84 -1.081 -1.169 0.245
(21.63) (19.22) (0.92)
Meat, poultry, and fish
2317 0.561 2,318 0.518 -0.043 -0.197 0.844
(7.47) (7.18) (0.22)
Eggs 
2317 1.424 2,318 1.251 -0.173 -0.486 0.628
(11.85) (11.12) (0.36)
Dark green leafy vegetables 
2317 4.704 2,318 7.075 2.371 1.477 0.142
(21.18) (25.65) (1.60)
Other vitamin-A-rich fruits and 
vegetables 
2,317 44.281 2,318 51.337 7.056** 2.207 0.029
(49.68) (49.99) (3.20)
Other vegetables 
2,317 14.243 2,318 20.406 6.163*** 2.814 0.006
(34.96) (40.31) (2.19)
Baseline Study of the Ifaa RFSA in Ethiopia (Vol. I)
Comparison of Livelihood and Non-Livelihood Kebeles 89 
Outcome N 
Non￾liveli￾hood 
N Liveli￾hood Difference T-Stat 
Diff
P-value 
Diff
Other fruit
2,317 1.338 2,318 0.561 -0.777* -1.765 0.08
(11.49) (7.47) (0.44)
Notes: Standard deviations are in parentheses. * denotes significance at 10 pct., ** at 5 pct., and *** at 1 pct. 
Standard errors are clustered at the kebele level. 
5.5 Water, Sanitation and Hygiene Practices 
Table 61. Water, Sanitation, and Hygiene Indicators across livelihood status 
Outcome N 
Non￾liveli￾hood 
N Liveli￾hood Difference T-Stat 
Diff
P-value 
Diff
BL16: Percent with access to 
basic drinking water services
177 3.39 176 5.114 1.724 0.745 0.459
(18.15) (22.09) (2.31)
Percent of households using 
basic drinking water services 
based on four of five of the 
criteria
2,340 10.556 2,330 11.845 1.290 0.659 0.511
(30.73) (32.32) (1.96) Percent of households with 
soap and water at a 
handwashing station on the 
premises
2,342 1.964 2,339 2.736 0.772 1.008 0.315
(13.88) (16.32) (0.77) Percent of households 
practicing correct use of 
recommended household water 
treatment technologies
2,339 7.183 2,337 7.531 0.348 0.186 0.853 
(25.83) (26.39) (1.88)
Percent of households 
practicing open defecation 
2,342 59.949 2,339 57.503 -2.446 -0.455 0.65
(49.01) (49.44) (5.38)
Percent using improved 
sanitation facilities (not shared)
2,342 14.603 2,339 17.144 2.541 0.926 0.356
(35.32) (37.70) (2.74)
Notes: Standard deviations are in parentheses. * denotes significance at 10 pct., ** at 5 pct., and *** at 1 pct. Standard errors 
are clustered at the kebele level. 
IMPEL | Implementer-Led Evaluation and Learning
90 Comparison of Livelihood and Non-Livelihood Kebeles
5.6 Agriculture 
Table 62. Financial services, value chain activities and natural resource management practices across 
livelihood status 
Outcome N 
Non￾liveli￾hood
N Liveli￾hood Difference T-Stat 
Diff
P￾value 
Diff
Percent of households who using 
financial services
2,218 18.35 2,232 18.369 0.019 0.009 0.993
(38.72) (38.73) (2.13)
Percent of households using 
agricultural credit
2,218 15.78 2,232 15.591 -0.189 -0.092 0.926
(36.46) (36.29) (2.04)
Percent of households who saved 
2,217 2.21 2,230 3.274 1.063 1.596 0.113
(14.70) (17.80) (0.67)
Percent of households using 
insurance
2,217 1.353 2,232 1.165 -0.188 -0.429 0.669
(11.56) (10.73) (0.44)
Percent of households reporting 
at least one value chain activity
459 72.113 562 79.181 7.068 0.942 0.349
(44.89) (40.64) (7.51)
Percent of households using at 
least one natural resource 
management practice
2,218 88.278 2,232 90.547 2.269 0.849 0.398 
(32.18) (29.26) (2.67) 
Notes: Standard deviations are in parentheses. * denotes significance at 10 pct., ** at 5 pct., and *** at 1 pct. 
Standard errors are clustered at the kebele level. 
Table 63. Improved management practices or technologies in target crops across livelihood status 
Outcome N 
Non￾liveli￾hood
N Liveli￾hood Difference T-Stat 
Diff
P-value 
Diff
Percent of households growing 
haricot beans
1,941 0 2,062 0.097 0.097 1.432 0.155
(0.00) (3.11) (0.07)
Percent using at least one 
practice for haricot beans
2 100
(0.00)
Percent of households growing 
groundnuts
1,941 1.391 2,062 1.503 0.112 0.141 0.888
(11.71) (12.17) (0.80)
Percent using at least one 
practice for groundnut
27 74.074 31 87.097 13.023 1.036 0.312
(44.66) (34.08) (12.57)
Percent of households growing 
potato
1,941 0.824 2,062 0.63 -0.194 -0.438 0.662
(9.04) (7.92) (0.44)
Baseline Study of the Ifaa RFSA in Ethiopia (Vol. I)
Comparison of Livelihood and Non-Livelihood Kebeles 91 
Outcome N 
Non￾liveli￾hood 
N Liveli￾hood Difference T-Stat 
Diff
P-value 
Diff
Percent using at least one 
practice for potatoes
16 100 13 100 0.000
(0.00) (0.00) (.)
Notes: Standard deviations are in parentheses. * denotes significance at 10 pct., ** at 5 pct., and *** at 1 pct. Standard errors 
are clustered at the kebele level. 
Table 64. Improved management practices or technologies in target animals across livelihood status 
Outcome N 
Non￾liveli￾hood
N Liveli￾hood Difference T-Stat 
Diff 
P￾value 
Diff
Percent of households raising 
goats
2,265 23.709 2,272 23.195 -0.513 -0.188 0.851
(42.54) (42.22) (2.73)
Percent of households using at 
least one practice for goats
537 94.786 527 92.22 -2.566 -1.128 0.262
(22.25) (26.81) (2.28)
Percent of households raising 
poultry
2,265 19.029 2,272 20.114 1.086 0.444 0.658
(39.26) (40.09) (2.44)
Percent of households using at 
least one practice for poultry
431 95.824 457 94.092 -1.732 -0.734 0.464
(20.03) (23.60) (2.36)
Percent of households raising 
oxen 
2,265 6.755 2,272 5.942 -0.813 -0.733 0.465
(25.10) (23.65) (1.11)
Percent of households using at 
least one practice for oxen
153 90.85 135 78.519 -12.331 -1.56 0.122
(28.93) (41.22) (7.90)
Notes: Standard deviations are in parentheses. * denotes significance at 10 pct., ** at 5 pct., and *** at 1 pct. Standard errors 
are clustered at the kebele level. 
5.7 Poverty Measurement
Table 65. Poverty measurement indicators across livelihood status 
Outcome N 
Non￾liveli￾hood
N Liveli￾hood Difference T-Stat 
Diff
P- value 
Diff
Daily per capita food expenditure 
(Birr)
2,340 16.626 2,337 18.652 2.027* 1.71 0.09
(10.25) (11.63) (1.19)
Daily per adult equivalent food 
expenditure (Birr)
2,340 20.234 2,337 22.493 2.259 1.594 0.114
(12.51) (13.89) (1.42)
Notes: Standard deviations are in parentheses. * denotes significance at 10 pct., ** at 5 pct., and *** at 1 pct. Standard errors 
are clustered at the kebele level. 
IMPEL | Implementer-Led Evaluation and Learning
92 Comparison of Livelihood and Non-Livelihood Kebeles
5.8 Gender Dynamics 
5.8.1 Use of Financial Resources
Table 66. Gender cash indicators across livelihood status
Outcome N 
Non￾liveli￾hood
N Liveli￾hood Difference T-Stat 
Diff
P- value 
Diff
Percent of cash earners in a union 
1,621 21.407 1,823 24.52 3.113 0.785 0.434
(41.03) (43.03) (3.97)
Percent of cash-earning men in a 
union
865 27.977 965 30.052 2.075 0.41 0.682
(44.91) (45.87) (5.06)
Percent of cash-earning women in 
a union
756 13.889 858 18.298 4.409 1.077 0.284
(34.61) (38.69) (4.09)
Percent of women in a union and 
earning cash who report 
participation in decisions about 
the use of self-earned cash
105 68.571 157 71.975 3.403 0.257 0.798
(46.65) (45.06) (13.25) Percent of women in a union and 
earning cash who report 
participation in decisions about 
the use of spouse/partner’s self￾earned cash 
105 29.524 157 53.503 23.979** 2.222 0.03 
(45.83) (50.04) (10.79) 
Percent of men in a union and 
earning cash who report 
spouse/partner’ participation in 
decisions about the use of self￾earned cash 
242 35.537 290 45.517 9.980 1.393 0.168 
(47.96) (49.88) (7.17) 
Notes: Standard deviations are in parentheses. * denotes significance at 10 pct., ** at 5 pct., and *** at 1 pct. 
Standard errors are clustered at the kebele level. 
5.8.2 Credit 
Table 67. Gender access to credit and group participation indicators across livelihood status 
Outcome N 
Non￾liveli￾hood 
N Liveli￾hood Difference T-Stat 
Diff
P-value 
Diff
Percent of people in a union who 
are members of a community 
group
363 65.84 636 55.66 -10.180 -0.976 0.331
 (47.49) (49.72) (10.42) 
Baseline Study of the Ifaa RFSA in Ethiopia (Vol. I)
Comparison of Livelihood and Non-Livelihood Kebeles 93 
Outcome N 
Non￾liveli￾hood 
N Liveli￾hood Difference T-Stat 
Diff
P-value 
Diff
Percent of men in a union who 
are members of a community 
group
210 74.286 353 60.907 -13.379 -1.256 0.212
(43.81) (48.87) (10.65)
Percent of women in a union 
who are members of a 
community group 
153 54.248 283 49.117 -5.132 -0.507 0.613
(49.98) (50.08) (10.12)
Percent of people in a union with 
access to credit
1,621 21.592 1,823 33.187 11.595** 2.272 0.025
(41.16) (47.10) (5.10)
Percent of men in a union with 
access to credit
865 23.699 965 34.611 10.912** 2.043 0.043
(42.55) (47.60) (5.34)
Percent of women in a union 
with access to credit
756 19.18 858 31.585 12.405** 2.151 0.033
(39.40) (46.51) (5.77)
Percent of men and women in a 
union who make decisions about
credit
350 28 605 30.413 2.413 0.403 0.688
 (44.96) (46.04) (5.99) 
Percent of men in a union who 
report making decisions about 
credit
205 24.878 334 31.437 6.559 0.932 0.354
 (43.34) (46.50) (7.03) 
Percent of women in a union 
who report making decisions 
about credit
145 32.414 271 29.151 -3.263 -0.513 0.609
 (46.97) (45.53) (6.36) 
Notes: Standard deviations are in parentheses. * denotes significance at 10 pct., ** at 5 pct., and *** at 1 pct. 
Standard errors are clustered at the kebele level. 
5.9 Resilience
5.9.1 Ability to Recover from Shocks and Stresses Index 
Table 68. Ability to recover from shocks and stresses index across livelihood status 
Outcome N 
Non￾liveli￾hood
N Liveli￾hood Difference T-Stat 
Diff
P￾value 
Diff
Mean ability to recover from 
shocks and stresses index 
2,285 2.953 2,260 3.046 0.093 0.749 0.455
(1.32) (1.39) (0.12)
Ability to recover index (2–6) 
2,328 2.922 2,321 3.04 0.118 0.916 0.361
(1.33) (1.39) (0.13)
IMPEL | Implementer-Led Evaluation and Learning
94 Comparison of Livelihood and Non-Livelihood Kebeles
Outcome N 
Non￾liveli￾hood
N Liveli￾hood Difference T-Stat 
Diff
P￾value 
Diff
Shock exposure index (0–168)
2,297 26.073 2,278 22.716 -3.357** -2.032 0.044
(15.56) (12.93) (1.65)
Total shocks experienced (0–21) 
2,343 3.817 2,340 3.385 -0.432* -1.809 0.073
(2.34) (2.00) (0.24)
Notes: Standard deviations are in parentheses. * denotes significance at 10 pct., ** at 5 pct., and *** at 1 pct. Standard errors 
are clustered at the kebele level. 
5.9.2 Absorptive Capacity Index 
Table 69. Absorptive capacity index and subindexes across livelihood status 
Outcome N 
Non￾liveli￾hood 
N Liveli￾hood Difference T-Stat 
Diff
P-value 
Diff
Absorptive capacity index (0–
100) 
2,213 28.498 2,229 28.865 0.367 0.207 0.837
(15.82) (15.15) (1.77)
Absorptive capacity index (0–
100): No access to insurance 
2,332 29.914 2,330 29.998 0.084 0.049 0.961
(15.45) (14.79) (1.73)
Bonding social capital index (0–-
6) 
2,341 2.494 2,340 2.614 0.119 0.592 0.555
(1.88) (1.87) (0.20)
Access to cash savings index (0–
1) 
2,340 0.015 2,340 0.019 0.004 0.79 0.431
(0.12) (0.14) (0.01)
Remittances index (0–1)
2,343 0.026 2,340 0.021 -0.005 -0.707 0.481
(0.16) (0.14) (0.01)
Asset ownership index—total 
type (0–45) 
2,334 8.349 2,331 8.193 -0.156 -0.513 0.609
(3.78) (3.51) (0.30)
Shock preparednesss and 
responsiveness index (0–3) 
2,341 0.381 2,340 0.401 0.020 0.438 0.662
(0.57) (0.55) (0.05)
Access to insurance index (0–1) 
2,218 0.014 2,232 0.012 -0.002 -0.428 0.669
(0.12) (0.11) (0.00)
Access to humanitarian 
assistance index (0–1) 
2,340 0.477 2,339 0.516 0.039 0.626 0.532
(0.50) (0.50) (0.06)
Notes: Standard deviations are in parentheses. * denotes significance at 10 pct., ** at 5 pct., and *** at 1 pct. Standard errors 
are clustered at the kebele level. 
Baseline Study of the Ifaa RFSA in Ethiopia (Vol. I)
Comparison of Livelihood and Non-Livelihood Kebeles 95 
5.9.3 Social Capital Index
Table 70. Social capital index and subindexes across livelihood status 
Outcome N 
Non￾liveli￾hood 
N Liveli￾hood Difference T-Stat 
Diff
P-value 
Diff
Index of social capital at 
household level (0–100) 
2,341 49.076 2,340 51.165 2.088 0.594 0.554
(34.15) (33.60) (3.51)
Bonding subindex (0–100) 
2,341 53.481 2,340 54.915 1.433 0.396 0.692
(35.99) (35.54) (3.61)
Bridging subindex (0–100) 
2,341 44.671 2,340 47.415 2.743 0.731 0.466
(36.84) (36.02) (3.75)
Notes: Standard deviations are in parentheses. * denotes significance at 10 pct., ** at 5 pct., and *** at 1 pct. 
Standard errors are clustered at the kebele level.
5.9.4 Adaptive Capacity Index 
Table 71. Adaptive capacity index and indexes across livelihood status 
Outcome N 
Non￾liveli￾hood 
N Liveli￾hood Difference T-Stat 
Diff
P-value 
Diff
Adaptive capacity index (0–100) 
203 43.586 313 43.354 -0.233 -0.036 0.972
(21.70) (23.15) (6.49)
Adaptive capacity index (0–100): 
No adopt improved practices
851 41.822 1,113 40.154 -1.667 -0.396 0.693
(19.65) (19.92) (4.21)
Adaptive capacity index (0–100): 
No access to finance s. practices 
382 35.967 466 39.022 3.055 0.636 0.527
(20.11) (22.36) (4.81)
Adaptive capacity index (0–100): 
No access to finance s. practices 
and improved
1,942 35.507 2,018 36.344 0.837 0.311 0.756
 (18.40) (19.42) (2.69) 
Aspirations/confidence to adapt 
index (0–16) 
2,129 9.624 2,161 9.841 0.217 1.389 0.168
(2.31) (2.22) (0.16)
Bridging social capital index (0–6) 
2,341 2.094 2,340 2.248 0.154 0.747 0.457
(1.87) (1.85) (0.21)
Linking social capital (0–4) 
2,339 0.442 2,339 0.502 0.061 0.496 0.621
(0.97) (1.00) (0.12)
2,162 1.06 2,206 1.103 0.043 0.305 0.761
IMPEL | Implementer-Led Evaluation and Learning
96 Comparison of Livelihood and Non-Livelihood Kebeles
Outcome N 
Non￾liveli￾hood 
N Liveli￾hood Difference T-Stat 
Diff
P-value 
Diff
Social network index (adjusted 
range 0–3) (1.31) (1.33) (0.14)
Education/training index (0–8) 
2,334 0.947 2,335 1.014 0.066 0.678 0.499
(1.19) (1.25) (0.10)
Livelihood diversification index 
(0–20) 
2,341 1.771 2,340 1.84 0.069 1.133 0.26
(0.72) (0.80) (0.06)
Adoption of improved practices 
index (0–1) 
459 0.37 562 0.338 -0.032 -0.447 0.656
(0.48) (0.47) (0.07)
Access to financial resources (0–
1) 
996 0.423 1,244 0.273 -0.149 -1.46 0.148
(0.49) (0.45) (0.10)
Exposure to information index 
(0–19) 
2,341 1.712 2,340 1.785 0.073 0.319 0.75
(2.33) (2.43) (0.23)
Asset ownership index—total 
type (0–45) 
2,334 8.349 2,331 8.193 -0.156 -0.513 0.609
(3.78) (3.51) (0.30)
Notes: Standard deviations are in parentheses. * denotes significance at 10 pct., ** at 5 pct., and *** at 1 pct. Standard errors 
are clustered at the kebele level.Transformative Capacity Index 
5.9.5 Transformative Capacity Index 
Table 72. Transformative capacity index and indexes across livelihood status 
Outcome N 
Non￾liveli￾hood 
N Liveli￾hood Difference T-Stat 
Diff
P-value 
Diff
Transformative capacity index 
(0–100) 
83 35.075 117 50.092 15.017* 1.78 0.091
(15.78) (23.82) (8.44)
Transformative capacity index 
(0–100): Excluding agricultural 
services and gender equity
2,163 40.635 2,127 42.439 1.804 0.522 0.603 
 (20.16) (20.47) (3.46) 
Transformative capacity index 
(0–100): Excluding gender equity 
index
2,163 35.237 2,127 37.273 2.036 0.545 0.587 
 (20.91) (21.78) (3.74) 
Transformative capacity index 
(0–100): Excluding agricultural 
services 
83 36.84 117 50.382 13.542 1.587 0.129 
 (16.17) (24.05) (8.53) 
2,245 2.018 2,215 2.092 0.074 0.602 0.548
Baseline Study of the Ifaa RFSA in Ethiopia (Vol. I)
Comparison of Livelihood and Non-Livelihood Kebeles 97 
Outcome N 
Non￾liveli￾hood 
N Liveli￾hood Difference T-Stat 
Diff
P-value 
Diff
Access to formal safety nets 
index (0–11) (0.81) (0.78) (0.12)
Access to communal natural 
resources index (0–4) 
2,343 1.356 2,339 1.222 -0.134 -0.501 0.617
(1.65) (1.62) (0.27)
Basic services index (0–1): Only 
police variable 
2,343 0.984 2,340 0.983 -0.001 -0.111 0.911
(0.13) (0.13) (0.01)
Access to agricultural services 
index (0-1) 
2,343 0.108 2,340 0.141 0.032 0.557 0.579
(0.31) (0.35) (0.06)
Bridging social capital index (0–6) 
2,341 2.094 2,340 2.248 0.154 0.747 0.457
(1.87) (1.85) (0.21)
Linking social capital (0–4) 
2,339 0.442 2,339 0.502 0.061 0.496 0.621
(0.97) (1.00) (0.12)
Social cohesion index (0–3) 
2,283 1.075 2,288 1.134 0.059 0.388 0.699
(1.04) (1.16) (0.15)
Local decision making index (0–1) 
2,318 0.911 2,301 0.912 0.001 0.026 0.98
(0.28) (0.28) (0.03)
Local government 
responsiveness index (0–2) 
2,343 1.804 2,340 1.87 0.066 1.464 0.146
(0.41) (0.34) (0.04)
Gender index (0–3) 
2,343 2.622 2,340 2.738 0.117* 1.698 0.092
(0.58) (0.52) (0.07)
Gender equitable decision￾making index (0–1) 
106 0.67 153 0.745 0.075 0.336 0.74
(0.47) (0.42) (0.22)
Notes: Standard deviations are in parentheses. * denotes significance at 10 pct., ** at 5 pct., and *** at 1 pct. 
Standard errors are clustered at the kebele level. 
Baseline Study of the Ifaa RFSA in Ethiopia (Vol. I)
Conclusion 98 
6. CONCLUSION
The Ifaa baseline survey was conducted in eight woredas in the region of Oromia, Ethiopia. This baseline 
survey of 4,683 households shows an area with low levels of education, high poverty rates, and high 
levels of food insecurity. Around 66% of adults in the area are farmers, and 31.5% of people older than 
10 worked and were paid in cash in the past 12 months. Over 94% of households have access to a plot of 
land, and more than 90% of these households plant at least one crop; maize and sorghum are the most 
common crops. It is important to highlight that very few households grow haricot beans, groundnuts, or 
potatoes, which are Ifaa’s target crops. 
The baseline survey was administered to PSNP beneficiaries, some of the poorest households in 
Ethiopia, during the lean season in May–June 2022. Many shocks have affected these households in the 
last couple of years, including droughts, fall armyworms, desert locusts, and political conflict (Alderman 
et al., 2021). This is reflected in the extremely high poverty rates and levels of food insecurity in the 
region. More than 92% of people spend less than $1.59 per day94 (2011 PPP) on food. Using the FIES to 
measure food insecurity, we can conclude that nearly all households (96%) experience moderate to 
severe food insecurity, while more than 77% face severe food insecurity. The survey modules looking at 
children’s nutrition and women’s nutrition provide a similar picture. Less than 2% of all children 6 to 23 
months met MDD criteria or MAD standards, while less than 1% of women consumed a diet that met 
the MDD criteria. These low percentages are associated with a low-quality diet: people consume few 
food groups and, in cases of severe food insecurity, are forced to reduce the quantity of food they 
consume or spend days without eating. 
The survey’s resilience module demonstrated that households have low capacities to face negative 
shocks. The two most common shocks reported by households were too little rain/drought and 
increases in food prices. On average, people perceived shocks as severe, and households report not 
having many resources to cope with a shock’s impact. Access to financial resources, insurance, or formal 
safety nets is very low, and households have few assets available to mitigate a shock’s impact. 
Furthermore, institutions in the area that could improve households’ response capacity are weak. 
Causal Design performed two balance exercises. First, we compared treatment and control kebeles. The 
second balance exercise examined livelihood kebeles compared to non-livelihood kebeles. The baseline 
data shows a good balance between treatment and control kebeles, with only two (out of 141) variables 
showing statistically significant95 differences across treatment arms. In the case of livelihood and non￾livelihood kebeles, a larger number of statistically significant differences (23 variables) were observed, 
which reflects the fact that livelihood kebeles were purposefully selected. In addition to the balance 
exercises, this study examined the differences across woredas.96 Even though all woredas selected are 
among the poorest regions in Ethiopia, we saw sizable differences across many variables. These 94 This value reflects an adjustment to the poverty line of $1.90 per day (2011 PPP). See Section 4.7 for more details. 95 This number corresponds to a 5% significance level. 96 The survey was not designed to produce estimates at the woreda level. The results at this level are only informative and 
should be interpreted with care. 
Baseline Study of the Ifaa RFSA in Ethiopia (Vol. I)
Comparison of Livelihood and Non-Livelihood Kebeles 99 
differences tend to be correlated; woredas with higher levels of poverty tend to have higher levels of 
food insecurity and lower capacities to cope with negative shocks. 
Looking ahead to the endline survey in 2025, the most pressing challenge is the possibility of attrition. 
The gap of 3 years between the baseline and endline means households might migrate to different 
communities, making it difficult to interview the same households at the endline. Additionally, the 
provision of Ifaa interventions follows a graduation model in which households that graduate stop 
receiving interventions. Households interviewed for the baseline might graduate in the next years and 
thus would stop receiving Ifaa interventions. The main strategy to address the possibility of attrition is 
to include an inflation factor of 10% in the number of households surveyed. Going forward, constant 
communication with CRS will be very helpful in understanding possible changes in the endline survey 
population. 
IMPEL | Implementer-Led Evaluation and Learning
References 100 
7. REFERENCES 
Alderman H., Bachewe F., Gilligan D., Hidrobo M., Ledlie N., Ramani G., &Taffesse A. (2019). Impact 
Evaluation of the Strengthen PSNP4 Institutions and Resilience (SPIR) Development Food Security 
Activity (DFSA) Baseline Report. Washington, DC: The Implementer-Led Evaluation & Learning Associate 
Award. 
Alderman H., Bachewe F., Gilligan D., Hidrobo M., Ledlie N., Ramani G., & Taffesse A. (2021). Impact 
Evaluation of the Strengthen PSNP4 Institutions and Resilience (SPIR) Development Food Security 
Activity (DFSA) Endline Report. Washington, DC: The Implementer-Led Evaluation & Learning Associate 
Award. 
Banerjee A., Duflo E., Goldberg N., Karlan D., Osei R., Parienté W., Shapiro J., Thuysbaert B., and Udry C. 
(2015). A multifaceted program causes lasting progress for the very poor: Evidence from six 
countries. Science, (348)6236. https://www.science.org/doi/10.1126/science.1260799
Berhane, G., Gilligan, D. O., Hoddinott, J., Kumar, N., & Taffesse, A. S. (2014). Can social protection work 
in Africa? The impact of Ethiopia’s productive safety net programme. Economic Development and 
Cultural Change, 63(1), 1–26.
Central Statistical Agency. (2011). Household Consumption and Expenditure (HCE) Survey 2010/2011, 
Analytical Report. The Federal Democratic Republic of Ethiopia. http://www.csa.gov.et/survey￾report/category/75-hice-2011 
Central Statistical Agency. (2022). Country and regional level consumer price indices (CPI). The Federal 
Democratic Republic of Ethiopia.. http://www.csa.gov.et/price-indices/consumer-price-index 
FAO. (2016). Methods for estimating comparable rates of food insecurity experienced by adults 
throughout the world. Rome, FAO. 
FAO, IFAD, UNICEF, WFP, & WHO. (2022). The State of Food Security and Nutrition in the World 2022. 
FAO. (2017). The Food Insecurity Experience Scale: Measuring food insecurity through people’s 
experiences. United Nations. https://www.fao.org/3/i7835e/i7835e.pdf
FAO. (n.d.). Voices of the Hungry: Modeling Food Insecurity on Bivariate and Regression Analyses. http://www.fao.org/3/a-bp091e.pdf. Hirvonen, K., Hoddinott, J., Minten, B., & Stifel, D. (2017). Children’s diets, nutrition knowledge, and 
access to markets. World Development, (95) 303–315.
IMF. (2022). Country Indexes and Weight. Produced by: International Monetary Fund (IMF). Retrieved 
from: https://data.imf.org/regular.aspx?key=61015892
IMPEL. (2021). Baseline Study of the FIOVANA Resilience Food Security Activity in Madagascar (Vol. 1). 
Washington, DC: The Implementer-Led Evaluation & Learning Associate Award. 
IMPEL. (2022). RFSA Ifaa IE Pre-Analysis Plan. Washington, DC: The Implementer-Led Evaluation & 
Learning Associate Award. 
Baseline Study of the Ifaa RFSA in Ethiopia (Vol. I)
Annex B: List of Kebeles with Treatment and Livelihood Status 101 
Nord, M. (2014). Introduction to Item Response Theory applied to Food Security Measurement: Basic 
Concepts, Parameters, and Statistics. Technical Paper, FAO, Rome. 
http://www.fao.org/economic/ess/ess-fs/voices/en
Upton, J., Tennant, E., Florella, K. J., & Barrett, C. B. (2021). Caveat utilitor: A comparative assessment of 
resilience measurement approaches. Journal of Development Economics, (157). 
https://doi.org/10.1016/j.jdeveco.2022.102873 
USAID. (2021a). Bureau for Humanitarian Assistance Indicator Handbook Part I: Indicators for Baseline 
and Endline Surveys for Resilience Food Security Activities. United States Department of State. 
https://www.usaid.gov/document/indicator-handbook-part-i-indicators-baseline-and-endline-surveys
USAID. (2021b). Bureau for Humanitarian Assistance Indicator Handbook Supplement to Part I: BHA 
Baseline/Endline Questionnaire and Indicator Tabulations for Resilience Food Security Activities. United 
States Department of State. https://www.usaid.gov/document/indicator-handbook-supplemental-part￾i-baselineendline-questionnaire-and-indicator-tabulations WFP. (2008). Food consumption analysis: Calculation and use of the Food Consumption Score in food 
security analysis. United Nations. 
WHO and UNICEF. (2021). Indicators for assessing infant and young child feeding practices: definitions 
and measurement methods. United Nations. https://www.who.int/publications/i/item/9789240018389 Wiesmann, D., L. Bassett, T. Benson, & Hoddinott, J. (2009). Validation of the World Food Programme’s 
Food Consumption Score and Alternative Indicators of Household Food Security. IFPRI Discussion Paper 
00870. Washington, DC: International Food Policy Research Institute. 
https://www.researchgate.net/publication/46442105_Validation_of_the_World_Food_Programme's_fo
od_consumption_score_and_alternative_indicators_of_household_food_security World Bank. (n.d.). Poverty headcount ratio at $1.90 a day (2011 PPP) (% of population) – Ethiopia. 
https://data.worldbank.org/indicator/SI.POV.DDAY?locations=ET