Annex_IIa-_Di_Lottery_Baseline_DQR.pdf

IMPAQ International, LLC Page i Non-PAPs Di Irrigation Investment Data Quality Review Report

Millennium Challenge Corporation

Final Data Quality Review Report Di Irrigation Investment on Persons Not Affected by the Project

Contract No:

Task Order No. MCC-10-0111-CON-20 TO02

September 2014

Submitted to:

Jack Molyneaux

Project Officer: Jennifer Witriol

Millennium Challenge Corporation 875 15th Street, NW

Washington, DC 20005

Submitted by:

Laurence Dessein

IMPAQ International, LLC 10420 Little Patuxent Parkway

Suite 300 Columbia, MD 21044 www.impaqint.com http://www.impaqint.com/

IMPAQ International, LLC Page i Non-PAPs Di Irrigation Investment

TABLE OF CONTENTS

TABLE OF CONTENTS .......................................................................................................................... I

TABLE OF EXHIBITS ............................................................................................................................ II

1. INTRODUCTION

2. SAMPLING DOCUMENTATION

2.1 Source of Sample

2.2 Data Collection

2.3 Data File

3. KEY VARIABLE DISTRIBUTION

3.1 Demographic information on the lottery candidates

3.2 Characteristics of lottery candidates’ households

3.3 Lottery Candidates’ Household and Land Assets

3.4 Experience in agricultural production

3.5 Land Ownership of the Lottery Candidates

3.6 Land Cultivated by the Lottery Candidates

3.7 Land rented by the lottery candidate from or to others

3.8 Income and debt of lottery candidates

4. DATA DOCUMENTATION

APPENDIX A: LAND ALLOCATION CRITERIA FOR NON-PAPS

APPENDIX B: ADDITIONAL VARIABLE DISTRIBUTIONS

IMPAQ International, LLC Page ii Non-PAPs Di Irrigation Investment

TABLE OF EXHIBITS

Exhibit 1: Di Irrigation Investment Activity Area Exhibit 2: Research Questions Exhibit 3: Boucle du Mouhoun Exhibit 4: Power Calculations Exhibit 5: Number of Households Selelected and Surveyed Exhibit 6: Distribution of Lottery Candidates by Province Exhibit 7: Distribution of Lottery Candidates by Gender Exhibit 8: Distribution of Lottery Candidates by Age Elligibility Exhibit 9 provides the breakdown of the age distribution of the lottery candidates. The majority of candidates (80.4%) are between the ages of 25 and 55. The distribution of candidates by age is approximately similar across treatment and control groups Exhibit 9: Distribution of Lottery Candidates by Age Exhibit 10: Distribution of Lottery Candidates by Education Level Exhibit 11: Distribution of Lottery Candidates’ Houses by Type of Wall Exhibit 12: Distribution of Lottery Candidates’ Houses by Type of Roof Exhibit 13: Distribution of Households with Land Rights (Formal or Informal) Exhibit 14: Distribution of Candidates’ Households with Right of Access to Community Land Exhibit 15: Distribution of Lottery Candidates’ Households by Agricultural Equipment Owned Exhibit 16: Distribution of Lottery Candidates’ Households by Farm Animals Owned Exhibit 17: Distribution of Lottery Candidates’ Households by Type of Agricultural Inputs Used Exhibit 18: Distribution of Lottery Candidates who are Farmers Exhibit 19: Distribution of Lottery Candidates by Profession, if not Farmer Exhibit 20: Level of Technical Training in Agricultural Production Exhibit 21: Household members 15 years of age or Older Supporting Agricultural Production Exhibit 22: Distribution of Lottery Candidates by Number of Cultivable Plots Owned Exhibit 23: Distribution of Lottery Candidates By the Number of Hectares Cultivated in the last 12 months Exhibit 24: Plots of Cultivable Land Rented from Others Exhibit 25: Distribution of Lottery Candidates by Plots of Cultivable Land Rented to Others Exhibit 25: Distribution of Income Sources of Lottery Candidates in Last 12 months Exhibit 26: Distribution of Debt of Lottery Candidates in Last 12 months Exhibit 27: Distribution of Lottery Candidates who were Previously Farmers* Exhibit 28: Number of Years of Farming Experience of Lottery Candidates* Exhibit 29: Years of Experience in Irrigation of Other HH Members Exhibit 30: Years of Experience in Rice Production of Other HH Members Exhibit 31: Years of Experience in Other Relevant Farming Experience of Other HH Members Exhibit 32: Number of Cultivable Plots Owned by Lottery Candidate that are Irrigated* Exhibit 33: Total Area (in ha) of Cultivable Plots Owned by the Lottery Candidate* Exhibit 32: Distance (in Km) Between Home of Lottery Candidate and his/her Closest Cultivable Plot* . 20

IMPAQ International, LLC Page iii Non-PAPs Di Irrigation Investment

Exhibit 33: Total Area (in ha) of Community Plots Used by Lottery Candidates*

IMPAQ International, LLC Page 1 Non-PAPs Di Irrigation Investment

1. INTRODUCTION

This report reviews the quality of the baseline data collected to assess the Millennium Challenge Corporation (MCC) Irrigation Investment on Persons Not Affected by the Project (Non-PAPs) in Burkina Faso. The data collection was contracted by the Millennium Challenge Account – Burkina Faso (MCA-BF) to The Centre d'Etudes, de Recherches et de Formation pour le Développement Economique et Social (CERFODES). As presented in Exhibit 1, the Di Irrigation Investment activity is located in the Di commune of Sourou province in Western Burkina Faso.

MCC collaborated with the government of Burkina Faso to build new irrigation infrastructures in a perimeter of 2,240 hectares (ha) of land in the Di commune.

The Commission for the Attribution of Land (CAT), composed of local stakeholders and officials, allocated plots of land in this perimeter to beneficiaries.

While most plots in the perimeter were distributed to households with pre-emptive rights (Persons Affected by the Project, PAPs) and to households living in disadvantaged villages near the perimeter, some of the plots were randomly allocated by a lottery system to Non-PAP households not located in disadvantaged villages. In addition to irrigated plots of land, the beneficiaries also received starter kits containing seeds, fertilizers, and tools to use for leveling the land and completing tertiary canals. Beneficiaries also received training on production related to the irrigated land.

IMPAQ has developed a randomized control trial (RCT) methodology to estimate the impact of the Di Irrigation Investment activity on Non-PAPs. IMPAQ will measure program impacts by comparing the outcomes for land recipients (treatment group) and non-recipients (control group). Since selection into the treatment group is random, the treatment status will be independent of observed and unobserved characteristics of the applicant. Thus, any changes observed between the two groups over time can be attributed to the effects of the intervention.

The impact evaluation will address research questions related to the following outcome categories:

1) Agricultural Production

2) Household Income

3) Land tenure

Exhibit 1: Di Irrigation Investment Activity Area

IMPAQ International, LLC Page 2 Non-PAPs Di Irrigation Investment

The specific questions to be addressed are listed in Exhibit 2.

Exhibit 2: Research Questions

Outcome Category Research Question

1) Agricultural Production

Has the land under production and the level of intensity of the land under production increased?

Has the volume of agricultural production and yield increased, including high-value crops (vegetables) or crops that require intensive water use (rice)?

Has there been a change in the adoption of new technologies/techniques?

2) Household Income Has there been a change in the components of household income?

Has there been a change in the total household income?

3) Land Tenure Has the perception of land security improved?

Has the enhanced land security affected land investments?

The purpose of this data quality review is to assess the quality of the baseline data for the impact analysis of the Di Irrigation Investment on Non-PAPs. This baseline data will be used in conjunction with the follow-up data (to be collected in the future) for an impact evaluation of the Di Irrigation Investment. In this report, we focus on the quality of the baseline data and discuss the sampling frame, sampling problems, data management issues, skip pattern problems, descriptive statistics of key variables and data documentation.

IMPAQ International, LLC Page 3 Non-PAPs Di Irrigation Investment

2. SAMPLING DOCUMENTATION

Source of Sample

The Non-PAPs Di Irrigation Investment activity had 600 ha of land available for distribution to Non-PAPs who did not receive land in Phase 1 of the project. To be eligible for the lottery and win a plot of land, Non-PAPs had to meet the following criteria:

Be 18 years of age or above.

Be a resident of one of the six provinces of the Boucle du Mouhoun region (Kossi, Banwa, Mouhoun, Balé, Nayala, Sourou), as shown in Exhibit 3.

Have at least two other people of age 15 or above who can cultivate the plot of land with the applicant.

Not be a PAPs.

The verification of eligibility of Non-PAPs was based on an application statement that included an oath about the truthfulness of the information provided in the application, including the home address. The eligibility criteria was known to applicants before they filled out the application.

To assess the required sample size for the impact evaluation, IMPAQ used the Burkina Faso household income data from 2011 for the Boucle du Mouhoun region (Deuxième Programme National de Gestion des Terroirs, PNGT2). IMPAQ conducted power calculations to estimate the minimum detectable effects (MDE) for different sample sizes. The MDE is expressed in terms of changes in agricultural income. Exhibit 4 provides the output of the power calculations.

The results of the power calculations indicate that a total sample size of 488 individuals for the treatment group and 488 for the control group with plot sizes of 2 hectares for rice and 0.75 hectares for polyculture will be able to detect an impact on income of 18% or more. This level of impact is in line with findings from agriculture literature, which has generally shown positive effect sizes of irrigation projects on real income.1,2

1 Hussain, I. and Wijerathna, D., Irrigation and Income-Poverty Alleviation: A Comparative Analysis of Irrigation Systems in Developing Asia, International Water Management Institute (IWMI), February 2004.

2 Sawada, Y., Shoji, M., Sugawara, S. and Shinkai, N.,The Role of Infrastructure in Mitigating Poverty Dynamics: The Case of an Irrigation Project in Sri Lanka, World Bank, August 2008.

Exhibit 3: Boucle du Mouhoun

IMPAQ International, LLC Page 4 Non-PAPs Di Irrigation Investment

Exhibit 4: Power Calculations

Household Income, 2011 Standard deviation 719,111 719,111 719,111

MDE 107626.6 128901.7 152207

MDE/mean 14.91% 17.85% 21.08% Multiple for 80% power and 5% significance level (two-sided)

2.8 2.8 2.8 n1 700 488 350 n2 700 488 350 Total sample size (including control group) 1,400 976 700

1.5 rice 2 rice 3 rice

0.5 PC 0.75 PC 1 PC

decided The size of the plots was changed from two hectares for rice and 0.75 hectares for polyculture to two hectares of rice and one hectare for polyculture because in order to provide parcels large enough to provide an economical output sufficient for the new farmers/households and the need to maintain a particular sample size for the evaluation.

All applications for land were reviewed and scored by SHER-GRET, an MCA contractor also known as “AD7”. The criteria used in the scoring are listed in Appendix A. In the first round, 2,178 applicants met the elligibily criteria. MCA allowed for a restitution period during which an applicant could appeal the decision that resulted in an ineligible designation, and as a result, the total number of eligible applicants increased to 2,229.

To ensure that a sufficient number of individuals were classified as eligible and to help ensure that 20 percent of lottery winners were women, CAT in collaboration with MCC and MCA-BF set the threshold score for eligibility for the lottery at 60. AD7 reviewed and selected 1,528 applications with a score of 60 points or above. This group entered the lottery pool. Those below the threshold score were dropped from the pool of applicants.

Initially, only those who entered the lottery (1,528) were to be surveyed. However, MCC and MCA-BF decided to survey all applicants who met the eligibility criteria. As a result, CERFODES surveyed the 2,178 applicants who met the eligibility criteria in the first round. The 1,528 applicants who scored 60 points or above represented the sampling frame that was used to randomly assign applicants to the treatment group (land recipients) and to the control group (non-land recipients).

The randomization was organized into a two-tier process. First, CAT randomly selected 503 names from the pool of 1,528 lottery candidates. Then, based on the selected candidates’ stated preference for plots of rice or polyculture land, CAT randomly selected a plot for the candidates from either a rice pool or a polyculture pool. The 503 selected lottery candidates who received plots formed the treatment group. The rest of the 1,025 candidates who did not receive plots formed the control group. CAT selected an additional 150 names to form a waiting list in the event that any of the 503 selected winners could not be found.

IMPAQ International, LLC Page 5 Non-PAPs Di Irrigation Investment

Data Collection

As agreed by MCC and MCA-BF, CERFODES was contracted to survey all 2,178 applicants who met eligibility criteria. In total, CERFODES actually surveyed 2,130 (97.8%) applicants and could not find the remaining 48 applicants. The distribution of selected households and respondent households by province is presented in Exhibit 5.

Exhibit 5: Number of Households Selelected and Surveyed

Province HH Selected (Nbr.) HH Surveyed (Nbr.)

Balé 10 9

Banwa 43 43

Kossi 20 19

Mouhoun 43 40

Nayala 72 78

Sourou 1990 1,941

TOTAL 2178 2130

CERFODES surveyed five additional applicants in Nayala province. These were applicants that had submitted their application in Sourou Province but, after verification, CERFODES identified them as residents of Nayala Province.

Data File

On March 19, 2014, IMPAQ received a data file from MCC on the Non-PAPs Di Baseline Data (Base de données non PAP de DI.sav). In a memo dated April 23, 2014, IMPAQ submitted the results of the data quality assessment to MCC (DQA Memo_Non-PAP Di data_received 14 04 19.docx). IMPAQ checked the dataset for completeness and internal consistency. Overall, IMPAQ found that the data was of high quality and could be used to construct an analysis file for the impact evaluation. There were no significant issues related to missing individuals, individual subsititions, orphans or skip patterns.

Upon receiving additional information and documentation from MCC on July 3, 2014, IMPAQ found that 51 individuals on the list of individuals admitted to the lottery had not been surveyed by CERFODES. IMPAQ initially understood that CERFODES did not survey these individuals because they were added to the list of admitted individuals to the lottery after the restitution period. And, since CERFODES surveyed individuals using the pre-restitution list, these 51 individuals were never surveyed. However, after receipt of additional project documentation from MCC, IMPAQ found that 21 of the 51 individuals were in fact in the pre-restitution list. Only 4 of the 21 individuals had not been surveyed because CERFODES did not have the right number or because the indivudal was not present in the area. IMPAQ is still

IMPAQ International, LLC Page 6 Non-PAPs Di Irrigation Investment waiting for additional documentation from MCC to clarify why the 17 remaining candidates were not surveyed.

Additionally, the baseline data contains 6 identifiers (IDs) (3 sets of duplicates) in which the individual identifier is not unique. The IDs are:

N0037/N0037

D00334/ D00334

N00378/ N00378

The values for the variables for these duplicates are different. Therefore, these IDs are likely different cases. IMPAQ raised this issue in the DQA memo. However, we are still waiting for additional information to clarify why the three listed cases do not have unique household IDs.

We removed the duplicates in the baseline dataset to generate the descriptive statistics below.

The above issues do not present major challenges for the impact analysis, however addressing them will improve the quality of the eventual public use data.

IMPAQ International, LLC Page 7 Non-PAPs Di Irrigation Investment

3. KEY VARIABLE DISTRIBUTION

Demographic Information on the Lottery Candidates

Exhibit 6 presents the distribution of lottery candidates by province. The lottery assigned 1,025 individuals to the control group and 503 to the treatment group. Consistent with the eligibility criteria for lottery admission, all of the lottery candidates in the sample are residents of one of the six provinces of the Boucle du Mouhoun region (Kossi, Banwa, Mouhoun, Balé, Nayala and Sourou). The majority of the candidates (90.4%) come from Sourou province. The distribution of candidates by province is similar across the treatment and control groups.

Exhibit 6: Distribution of Lottery Candidates by Province

Province Control Treatment Total

No. % No. % No. % Sourou 931 90.8 451 89.7 1,382 90.4 Kossi 2 0.2 2 0.4 4 0.3 Mouhoun 8 0.8 7 1.4 15 1 Balé 2 0.2 1 0.2 3 0.2 Banwa 11 1.1 7 1.4 18 1.2 Nayala 36 3.5 19 3.8 55 3.6 Missing 35 3.4 16 3.2 51 3.3 Total 1,025 100 503 100 1,528 100

An examination of the lottery candidates by age (Exhibit 7) indicates that 77.2 percent of the candidates are males and 19.4 percent are females. The distribution of candidates by gender is approximately similar across treatment and control groups 22.7 percent of the candidates in the treatment group are females.

Exhibit 7: Distribution of Lottery Candidates by Gender

Gender Control Treatment Total

Male 78.7 74.2 77.2 Female 17.9 22.7 19.4

Missing 3.4 3.2 3.3 Total 100 100 100

Exhibit 8 indicates that all of the candidates, except one, are 18 years old or older, which is consistent with the eligibility criteria for lottery admission.

IMPAQ International, LLC Page 8 Non-PAPs Di Irrigation Investment

Exhibit 8: Distribution of Lottery Candidates by Age Elligibility

Age Eligibility No. % Non Eligible (Under 18 years of age) 1 0.1

Eligible (18 years old or above) 1,474 96.5 Missing 53 3.5 Total 1,475 97.6

Exhibit 9 provides the breakdown of the age distribution of the lottery candidates. The majority of candidates (80.4%) are between the ages of 25 and 55. The distribution of candidates by age is approximately similar across treatment and control groups.

Exhibit 9: Distribution of Lottery Candidates by Age

Age Group (In Years) Control Treatment Total

Younger than 25 11.1 10.3 10.9 25–55 80.3 80.5 80.4 Older than 56 5.2 5.6 5.3 Missing 3.4 3.6 3.5

Exhibit 10 presents the distribution of lottery candidates by education level. Approximately one-quarter of the candidates (23.4%) completed primary school. A higher percentage of lottery candidates (40.2%) did not complete a primary level of education. Only 3.5 percent of the candidates completed post-primary schooling. Another 13 percent completed secondary school, and 4 percent completed tertiary level schooling. The distribution of candidates by education is approximately similar across treatment and control groups.

Exhibit 10: Distribution of Lottery Candidates by Education Level

Education Level Control Treatment Total

None 39 42.5 40.2 Primary 23.1 24.1 23.4 Rural School 3 2.4 2.8 Madrassa 10.6 7.8 9.7 Post-Primary 2.9 4.8 3.5 Secondary 13.8 11.5 13 Tertiary 4.1 3.8 4

IMPAQ International, LLC Page 9 Non-PAPs Di Irrigation Investment

Characteristics of Lottery Candidates’ Households

Exhibits 11 and 12 present the characteristics of the houses of the lottery candidates. The majority of candidates’ houses have walls made of mudbrick (81.8%) and roofs made of concrete screeds (80.1%). For both the wall and roof types, the distributions are approximately similar across treatment and control groups.

Exhibit 11: Distribution of Lottery Candidates’ Houses by Type of Wall

Type of Walls Control Treatment Total

Straw or Other Plant 2.1 1.8 2

Mudbrick 81.7 82.1 81.8 Cinderblock 10.7 11.9 11.1

Exhibit 12: Distribution of Lottery Candidates’ Houses by Type of Roof

Type of Roof Control Treatment Total

Concrete screed 79.4 81.5 80.1

Mudbrick 15.1 11.9 14.1 Straw or Other Plant 2 3.4 2.5

Lottery Candidates’ Household and Land Assets

Exhibit 13 and 14 shows the percentage of households that have formal or informal rights over the plots of land they cultivate or over community land.

As presented in Exhibit 13, more than half (60.2%) have rights over the land they cultivate. The distribution is approximately similar across treatment and control groups.

Exhibit 13: Distribution of Households with Land Rights (Formal or Informal)

Land Rights Control Treatment Total

No 39.1 30.8 36.4

Yes 57.4 66 60.2

Missing 3.5 3.2 3.4

IMPAQ International, LLC Page 10 Non-PAPs Di Irrigation Investment

Out of the 36.4 percent of households without land rights over their own plots (see Exhibit 13),

61.9 percent have rights to access community land, as shown in Exhibit 14. The distribution is approximately similar across treatment and control groups.

Exhibit 14: Distribution of Candidates’ Households with Right of Access to Community Land

Exhibit 15 presents the agricultural equipment owned by the lottery candidates. The majority of candidates have animal plows (73.9%) and traditional plows (71.1%), followed by fertilizer application device (43.5%) and wheelbarrows (29.6%). A smaller percentage have pumps (8.5%) and tractors (1.8%). The distribution of candidates by the type of agricultural equipment owned is approximately similar across treatment and control groups.

Exhibit 15: Distribution of Lottery Candidates’ Households by Agricultural Equipment Owned

Agricultural Equipment Control Treatment Total

Animal plow 73.4 75 73.9 Traditional Plows 71.2 70.8 71.1 Pumps 8.7 8.2 8.5 Tractor 1.7 2 1.8 Fertilizer application device 43.2 44.1 43.5

Wheelbarrow 29.9 29 29.6

Exhibit 16 presents the type of farm animals owned by the lottery candidates. The five most common animals owned by lottery candidates are chickens (86.3%), oxen (68.3%), donkeys (59.5%) and sheep (50.5%). The distribution of candidates by the type of farm animal owned is

Right of Access to Community Land

Control Treatment Total

No 38.7 36.8 38.1 Yes 61.3 63.2 61.9

IMPAQ International, LLC Page 11 Non-PAPs Di Irrigation Investment

Exhibit 16: Distribution of Lottery Candidates’ Households by Farm Animals Owned

Farm Animals Control Treatment Total

Oxen 68.1 68.6 68.3 Other Cattle 31.3 30.6 31.1 Donkeys 59 60.4 59.5 Other Asses 22 20.3 21.4 Horses 1.4 2.2 1.6 Other Equidae 1.2 1.4 1.2 Sheep 51 49.5 50.5 Goats 43.6 46.7 44.6 Pigs 8.3 11.1 9.2 Chickens 85.6 87.9 86.3 Turkeys 18.1 18.5 18.3

Exhibit 17 presents the type of agricultural inputs used by the households of lottery applicants during the last harversting season. The most common agricultural input is traditional seed (79.1%), followed by organic manure (72.2%) and mineral fertilizer (70.2 %). The distribution of is approximately similar across treatment and control groups.

Exhibit 17: Distribution of Lottery Candidates’ Households by Type of Agricultural Inputs Used

Type of Agricultural Inputs

Control Treatment Total

Traditional Seed 78.7 79.9 79.1 Improved Seed 48.1 50.7 48.9 Mineral Fertilizer 69.6 71.6 70.2 Herbicide 62.6 67.2 64.1 Pesticide 61.4 63.6 62.1 Compost 60.2 60.8 60.4 Organic Manure 72 72.6 72.2 Other Input 2.1 2.2 2.2

Experience in Agricultural Production

Exhibit 18 shows the percentage of lottery candidates who are farmers at the time of the lottery. The majority of them (83.1%) are farmers. The distribution is approximately similar across treatment and control groups.

IMPAQ International, LLC Page 12 Non-PAPs Di Irrigation Investment

Exhibit 18: Distribution of Lottery Candidates who are Farmers

Farmers Control Treatment Total

No 13.7 13.3 13.5 Yes 82.9 83.5 83.1

Exhibit 19 shows the distribution of lottery candidates by profession for those who are not farmers. Out of the 13.5 percent of candidates who are not farmers, the majority (6.9%) hold salaried positions. The distribution is approximately similar across treatment and control groups.

Exhibit 19: Distribution of Lottery Candidates by Profession, if not Farmer

Profession Control Treatment Total

Trader 3.1 2.6 2.9 Craftsman (mason, carpenter, mechanic) 1.2 1.4 1.2

Salaried 7.5 5.8 6.9 Worker / day laborer 0.5 1.2 0.7 Unemployed 0.8 1.4 1 Missing 0.6 0.9 0.8 Total 13.7 13.3 13.5

Exhibit 20 presents the level of technical training of lottery candidates in agricultural production. Over half of the candidates (54.8%) never received technical training in agricultural production. 18.2 percent received one training and 22.4 percent received more than one training in agricultural production. The distribution of candidates by level of technical training in agricultural production is approximately similar across treatment and control groups.

Exhibit 20: Level of Technical Training in Agricultural Production

Number of Technical Training in Agriculture

Production

Control Treatment Total

None 54.5 55.3 54.8 One Training 18.2 18.1 18.2 More Than One Training 22.2 22.9 22.4

Missing 5 3.8 4.6

IMPAQ International, LLC Page 13 Non-PAPs Di Irrigation Investment

As indicated in Exhibit 21, a vast majority of households (83%) have 2 to 10 household members (15 years and older) supporting the household’s agricultural production. The distribution is

Exhibit 21: Household members 15 years of age or Older Supporting Agricultural Production

Number of HH Members 15 or Older

Control Treatment Total

Less than 2 5.2 4.4 4.9 2–10 82.8 83.3 83 11–15 6.7 7.0 6.8 16 or More 1.9 2.2 2.0

Land Ownership of the Lottery Candidates

Exhibit 22 presents the number of cultivable plots owned by the lottery candidates. The majority of candidates (53.9%) do not own any plots, 19.7 percent own one plot, 12.2 percent own two plots and the remaining 10.9 percent own more than two plots. The distribution is

Exhibit 22: Distribution of Lottery Candidates by Number of Cultivable Plots Owned

Number of Plots of Cultivable Land

Control Treatment Total

0 53.5 54.9 53.9 1 20.7 17.7 19.7 2 11.8 13.1 12.2 3 6.8 6.2 6.6 4 1.7 3.0 2.1 5 1.0 1.0 1.0

6 or more 1.2 1.0 1.2

Missing 3.4 3.1 3.3

Land Cultivated by the Lottery Candidates

Exhibit 23 shows the number of hectares that lottery candidates cultivated in the last 12 months for those who owned cultivable plots. The distribution is approximately similar across treatment and control groups.

IMPAQ International, LLC Page 14 Non-PAPs Di Irrigation Investment

Exhibit 23: Distribution of Lottery Candidates By the Number of Hectares Cultivated in the last 12 months

Number of Hectares Cultivated in the last 12 months

Control Treatment Total

Less than 0.5 11.6 9.7 11 0.5–1 17.1 17.9 17.4

1–1.99 16.7 14.5 16 2–2.99 15.5 16.4 15.8 3–3.99 12.3 10.1 11.6

4 or more 26.9 30.4 28 Does not know 0 1 0.3

Land Rented by the Lottery Candidate from or to Others

Exhibit 24 shows the number of cultivable plots rented from others. Approximately two-thirds (68.4%) of candidates do not rent land from others. The distribution is approximately similar across treatment and control groups.

Exhibit 24: Plots of Cultivable Land Rented from Others

Number of Plots of Cultivable Land

Control Treatment Total

0 67.1 71 68.4 1 20.3 19.1 19.9 2 6.3 5 5.9 3 1.6 0.8 1.3 4 0.9 0.8 0.9 6 0.3 0 0.2

10 0 0.2 0.1

12 0.1 0 0.1

Exhibit 25 shows the number of cultivable plots lottery candidates rent to others. Nearly all candidates (96.4%) do not rent any plots of land to others. The distribution is approximately

IMPAQ International, LLC Page 15 Non-PAPs Di Irrigation Investment

Exhibit 25: Distribution of Lottery Candidates by Plots of Cultivable Land Rented to Others

Number of Plots of Cultivable Land

Control Treatment Total

0 96.4 96.4 96.4 1 0.1 0.4 0.2 3 0.1 0 0.1

Income and Debt of Lottery Candidates

Exhibit 26 presents the lottery candidates’ income from four different sources, including crop yield (wet season), crop yield (dry season), trade/commerce, animal sale, income generating activity and other sources of income. As indicated by the large median value, the most common source of income for candidates is crop yields (dry season) and it is followed by crop yields (wet season). The distribution is approximately similar across treatment and control groups.

Exhibit 25: Distribution of Income Sources of Lottery Candidates in Last 12 months

Income Source Control Treatment Total

Mean

(FCFA)

Median

(FCFA)

Mean

(FCFA)

Median

(FCFA)

Mean

(FCFA)

Median

(FCFA)

Crop Yields (Wet Season) 173,221 25,000 194,699 42,000 180,303 30,000

Crop Yields (Dry Season) 468,910 100,000 354,224 75,000 431,252 100,000

Trade/Commerce 189,723 0 157,671 0 179,191 0

Animal Sale 96,478 4,500 105,250 3,500 99,362 4,000

Income Generating Activity 241,444 0 176,567 0 220,097 0

Other Income 60,997 0 67,736 0 63,214.4 0

Exhibit 27 presents the amount of debt for the 269 lottery candidates who declared they had debt. Out of the 269 candidates, a large number (46.5%) of candidates have debts higher than FCFA 100,000. The treatment group has slightly less debt on average than the control group.

Exhibit 26: Distribution of Debt of Lottery Candidates in Last 12 months

Household Debt Control Treatment Total

No debt 29.2 36.1 31.4 Debt less than or equal to FCFA

100.000 20.5 25.6 22.1

Debt higher than FCFA 100.000 50.3 38.4 46.5

IMPAQ International, LLC Page 16 Non-PAPs Di Irrigation Investment

4. DATA DOCUMENTATION

Based on our review of the value label against the questions in the survey, we have not found any significant issues. The data seems to be well-documented.

Below we list of files and datasets reviewed:

Files (received between May and July 2014):

• 130917_Brochure informations - non PAP phase aleatoire VF revAD7.docx

• 130917_dossier demande Non-PAP concurrentielle VF revAD7.docx

• 130917_GUIDE DU DEMANDEUR VF revAD7.docx

• 140109_Méthodologie de tirage au sort_VF.docx

• Allocation des terres à Di MCA version finale.doc

• Arrêté Cahier de charges specifique_Di.pdf

• Cahier Général des charges amenagements hydro-agricoles de type familial.doc

• Cahiers de charges sp PI Di vf 100912.docx

• Circulaire n2011-153-MAH-CAB_Adoption land allocation Di.pdf

• Communiqué de presse.doc

• COMMUNIQUES presse écrite rev MCABF.DOCX

• COMMUNIQUES radio-télé_ rev MCABF.DOCX

• GOBF adoption Cahier de charge general 9.6.12.pdf

• Processus d'élaboration du cahier de charges spécifiques de Di MCA version finale rev

06 10 11.docx

• RAPPORT DE LA COLLECTE _NON_PAP DEPOSE LIVRABLE 3.doc

• Rapport final _NONPAP.pdf

• Rapport final _NONPAP2r2.doc

• Processus enquête ME7Bbis.docx

• Grille de ponderation pour l'évaluation des candidats MCA version finale.xlsx

• ~$180214_bdd_nonPAPConcurrentielle - eligibles.xlsx

• 141212_bdd_nonPAPaleatoire_VF.xlsx

• 180214_bdd_nonPAPConcurrentielle - eligibles.xlsx

• Liste des admissibles loterie Non PAP Di.xlsx

• LISTE DES CANDIDATS AD7 180114 Situation des NON-PAP non enquetes AD7FINAL

FINAL110214.xls

• Liste des gagnants et liste d'attente Non PAP_2.xls

Datasets

• Base de données non PAP de Di.sav. (Received April 3, 2014)

• Base de données non PAP de Di.sav. (Received May 23, 2014)

IMPAQ International, LLC Page 17 Non-PAPs Di Irrigation Investment

APPENDIX A: Land Allocation Criteria for NON-PAPS

IMPAQ International, LLC Page 18 Non-PAPs Di Irrigation Investment

APPENDIX B: Additional Variable Distributions

Exhibit 27: Distribution of Lottery Candidates who were Previously Farmers*

Previous Farming Experience

Control Treatment Total

No 1.6 0.6 1.2 Yes 13.7 12.5 13.3 Missing 84.8 86.9 85.5 Total 100 100 100

* of those not currently farmers

Exhibit 28: Number of Years of Farming Experience of Lottery Candidates*

Farming Experience Control Treatment Total

Less Than One 0.3 0.4 0.3 1 to 5 12 13.1 12.4 6 to 10 27.3 23.7 26.1 More than 10 55.4 59 56.6 Missing 5 3.8 4.6 Total 100 100 100

* Of those who currently working as farmers or who previously worked as famers

Exhibit 29: Years of Experience in Irrigation of Other HH Members

Irrigation Experience of Other HH Members (n Years)

Control Treatment Total

None 34.1 34.6 34.3 Less Than 2 years 7.9 10.1 8.6 More Than 2 Years 53.8 51.7 53.1 Missing 4.2 3.6 4

Exhibit 30: Years of Experience in Rice Production of Other HH Members

Rice Production Experience of Other HH Members (In Years)

Control Treatment Total

None 40.6 41 40.7 Less Than 2 years 7.5 8 7.7 More Than 2 Years 47.5 47.3 47.4 Missing 4.4 3.8 4.2

IMPAQ International, LLC Page 19 Non-PAPs Di Irrigation Investment

Exhibit 31: Years of Experience in Other Relevant Farming Experience of Other HH Members

Other Relevant Farming Experience of Other HH Members (In Years)

Control Treatment Total

None 30.6 32.8 31.3 Less Than 2 years 7.9 8.9 8.2 More Than 2 Years 56.5 54.5 55.8 Missing 5 3.8 4.6

Exhibit 32: Number of Cultivable Plots Owned by Lottery Candidate that are Irrigated*

Number of Cultivable Plots Irrigated

Control Treatment Total

0 28.7 26.6 28 1 10.6 11.7 11 2 1.9 1.8 1.8 3 0.9 1 0.9 4 0.6 0.6 0.6 5 0.1 0 0.1 6 0.3 0 0.2 Missing 57 58.3 57.4

*Of those who own plots of land

Exhibit 33: Total Area (in ha) of Cultivable Plots Owned by the Lottery Candidate*

Total Area (in ha) of Cultivable Plots

Control Treatment Total

Less than 0,5 ha 5.8 4 5.2 0,5 - 1 ha 7.1 7.8 7.3 1 - 1.99 ha 7.3 5.8 6.8 2 - 2.99 ha 6.5 6.6 6.5 3 - 3.99 ha 4.7 4.2 4.5 4 ha or more 11.7 13.5 12.3 Don’t know 0 0.2 0.1 Missing 56.9 58.1 57.3

*Of those who currently own plots of land

IMPAQ International, LLC Page 20 Non-PAPs Di Irrigation Investment

Exhibit 34: Distance (in Km) Between Home of Lottery Candidate and his/her Closest Cultivable Plot*

Distance (in Km) Between Home and Closest Cultivable Plot

Control Treatment Total

Less than 1 km 9.5 11.7 10.2 1 - 1.99 km 9.8 9.7 9.8 2 - 2.99 km 8 8.7 8.2 3 - 3.99 km 6 4.2 5.4 4 km or more 9.8 7.6 9 Missing 57.1 58.1 57.4 Total 100 100 100

*Of those who currently own plots of land

Exhibit 35: Total Area (in ha) of Community Plots Used by Lottery Candidates*

Total Area (in ha) of Community Plots Control Treatment Total

Less than 0,5 ha 4.3 6 4.8 0,5 - 1 ha 6.9 6 6.6 1 - 1.99 ha 4.1 4.2 4.1 2 - 2.99 ha 3.7 4 3.8 3 - 3.99 ha 2.6 1.4 2.2 4 ha or more 3.1 3.6 3.3 Don’t know 0.1 0 0.1 Missing 75.1 75 75.1

*Of those who farm community plots