Development of a Cost-Effective Poverty Assessment Tool
Contractor’s Final Report
Lima, Peru. October 2004
Acknowledgements
We wish to thank Ms. Julia Johannsen and Mr. Manfred Zeller for their help and valuable guidance in every stage of the survey, especially in relation to quality assurance procedures for data gathering and processing.
We are also very thankful to all the staff who was directly involved in the project, including interviewers who had to work long hours under harsh weather and even unsafe public conditions. This survey would not have been possible without their valuable efforts.
Contents
Key Staff in Peru 4
Definitions 4
Territorial / administrative divisions 4
Geographical Regions 4
Other Terms 5
Survey Organization and Execution 5
Functions 5
Selection and Training of Survey Personnel 6
Fieldwork 6
Fieldwork Planning and Programming 6
Fieldwork Execution 7
Fieldwork Control and Supervision 7
Data Entry and Validation 8
Sampling 8
Sample of MFI Clients 8
Sample of Non-Clients 8
Routes 10
Description of routes and assigned personnel 10
Supervision 14
Routes selected for supervision. 14
Supervision Procedure 14
Supervisors’ Reports 15
Major Findings 15
Learning Curve 15
Improved Wording Needed 15
Estimated workload based on the actual sample 15
Errors in the lists of entering clients in areas of recent expansion 16
Most common errors when filling questionnaires out 16
Difficulties found in MFI selection 16
Curious happenings during fieldwork 18
The Case of the CRAC Cruz de Chalpón. 18
Final Report
Key Staff in Peru
The Peruvian staff involved in the project included the following professionals:
M.Sc. Luis Castillo Quintana. Project Manager and Director
Econ. Pedro Llontop Ledesma. Fieldwork Manager,
Est. Mario Reyna Farje Espinoza. Statistician.
Definitions
Many terms are peculiar to a given country, i.e. the meaning they are given there may not be the same –or as common– as elsewhere.
Territorial / administrative divisions
The Peruvian territory is administratively divided first into departments (departamentos) and therefore a department is the largest territorial division. Departments are in turn broken down into provinces (provincias), and finally provinces are subdivided into districts (distritos.)
As the decentralizing regionalization process consolidates, a further governance tier is currently in progress –that of regions. For the time being, however, regions and departments encompass the same territories and consequently we will no longer consider regions in the discussion that follows.
Geographical Regions
Traditionally, the Peruvian territory has been divided into three geographical regions as a result of the influence of the Andes, the large mountain range that runs North-South throughout the territory. Accordingly, such regions take the shape of three North-South longitudinal strips. The strip between the Sea and the Western slopes of the Andes is called the Coast. The highlands crowned by the Andes, with its fertile and habitable valleys are called the Sierra. The land East of the Andes where tropical forests and rainforests thrive is called the Jungle, and is sometimes referred to as the Amazonian region/area. Climatic and cultural differences are believably assumed to influence and/or determine their inhabitants’ distinct patterns of living.
While such names refer directly to dominant geographical features, it should not be understood as restricted to them, but only as labels of such extensive tracts of land. It is within such a context that we should understand such apparent oxymora as “Urban Jungle” (neither should it be thought of as equivalent to “asphalt jungle”, but rather as the urban areas within the geographical region called “the Jungle”) and the like.
Other Terms
Route. An established or selected course of travel and the assigned territory to be systematically covered
CRAC. Initials of Caja Rural de Ahorro y Crédito, Spanish for Rural Savings and Credit Union
CMAC. Initials of Caja Municipal de Ahorro y Crédito, Spanish for Municipal Savings and Credit Union
CAC. Initials of Cooperativa de Ahorro y Crédito, Spanish for Savings and Credit Union
EDPYME. Initials of Entidad de Desarrollo de la Pequeña y Micro Empresa, Spanish for Micro and Small Business Development Institution.
Survey Organization and Execution
Functions
Instituto Cuánto was responsible for the overall implementation of the survey in compliance with the instructions by, and in coordination with, the IRIS Center. The survey direction and management was structured as follows:
100 Project Manager. Responsible for both technical and managerial direction of the survey in all stages. He also comprehensively assesses, controls, and supervises the progress of the survey.
200 Translator. Responsible for translating and back-translating documents in Spanish into English and vice versa, as well as facilitating communication between IRIS researches visiting Lima and local staff.
300 Statistician. Responsible for preparing the sampling frame, as part of the survey design, as well as for determining the procedures leading to the selection of sampling units.
400 Fieldwork Manager. Responsible for implementing the survey fieldwork, from staff training and planning to actual fieldwork to the submission of completed questionnaires for data entry.
500 Regional Supervisors: Responsible for coordinating the fieldwork in each area surveyed.
600 Logistics Manager: Responsible for delivering any materials on a timely manner, and managing current expenses.
700 Data Processing Officer. Responsible for processing data, designing the data entry shell, and validating the data entered.
Selection and Training of Survey Personnel
This involved two stages: First, personnel recruiting and technical training for fieldwork supervisors and interviewers, and second, a course for critics and codification operators.
Instituto Cuánto has available a large directory of experienced interviewers, which take part of social and economic studies on a regular basis.
Pretest Survey
This survey will allow us to administer the questionnaires proposed for the study so as to make it possible to correct any errors, measure the time required to complete a questionnaire per household, as well as to improve the instructions given on the manner both conglomerates and households were to be located.
Fieldwork
This part of the survey refers to the execution of a number of tasks, sub-tasks, and operations intended to gather data from selected households to meet survey objectives.
Fieldwork involved the following tasks:
Fieldwork Planning and Programming
This task involved providing information on fieldwork routes, time required to complete the fieldwork, detailed determination of goods and services required for the successful completion of each task, fieldwork budgeting and allocation of funds to each route. Planning and programming required taking into account sample size and distribution. One of the basic elements of this process is determining the time required to get to the selected sampling unit. In this regard, the following movements should be taken into account:
· Within the route
· Between routes
· Between districts
· Between provinces
· Between departments
To that end, district-, province-, and department-level maps, roadmaps and street maps of the districts selected, as well as auxiliary information indicating distances, types of roads, and the usual means of transportation were provided. It was very important the broad expertise and knowledge of the staff involved in this task.
Fieldwork Execution
This task involves gathering information from selected households by administering the questionnaires designed for this purpose. The main guidelines to execute the survey are listed below:
The information from each selected household is gathered through direct interview.
Fieldwork schedule is flexible, depending on circumstances and the appointment arranged by the interviewer or Team Leader with the household head.
The Team Leader allocates the workload to each interviewer
Work is done Monday to Saturday. Only previously appointed interviews are carried out on Sundays, as required, to complete the weekly workload or to retrieve information.
As the information from individual households is completed, each interviewer submits all completed questionnaires to his/her Team Leader for data cleaning and validation
The Team Leader supervises interviews to correct interviewers’ initial mistakes and check for compliance with methods and procedures.
As Team Leaders are responsible for the technical and administrative performance of survey fieldwork, they had to be continually in contact with the Project Manager, who consequently controls and supervises the overall progress of the survey.
Team Leaders are required to submit fieldwork progress reports every other day, either via email o telephone.
Fieldwork Control and Supervision
It involves checking the survey fieldwork for adequate technical performance and compliance with methods, budget, and scheduled deadlines as planned for the fieldwork. Control and supervision is performed by the Team Leader.
Data Entry and Validation
This stage involves gathering, coding, and entering the data from completed questionnaires. Data entry is carried out under the Data Processing Officer in a roomy and adequately equipped computer room by experienced data entry operators. Double entry of data is performed to prevent potential data entry errors.
Sampling
Factors to be taken into consideration for sample selection
Sample of MFI Clients
Step 1.
Once we had available the lists provided by the selected MFI, entering clients were sorted by district. Then two districts were selected by applying PPS sampling.
Step 2.
Where selected districts had over 100 clients each, 100 were chosen from each to get a sample size of 200.
Where districts had fewer than 100 clients, two districts were randomly selected and the sampling units were allocated to both districts according to their sizes.
There was only one exception to this rule. The CRAC Cruz de Chalpón had too few new clients. As a result, we had to select 5 districts to cover the total sample of 200 clients.
Step 3.
Sample units were allocated as shown in Table 1.
Sample of Non-Clients
Step 1. (Using PPS method)
At the beginning we decided to select 6 departments from which the sample will be taken. The following departments were selected:
- Arequipa
- Cusco
- La Libertad
- Lima (twice)
- Piura
Table 1. Allocation of the MFI-Client sample among selected MFIs.
CAC San Isidro / Lima / Huaral / Chancay / 57 / 32 / 3
Lima / Huaral / Huaral / 299 / 168 / 17
Total / 356 / 200 / 20
CMAC de Chincha / Ica / Chincha / Chincha Alta / 349 / 100 / 10
Ica / Chincha / Pueblo Nuevo / 244 / 100 / 10
Total / 593 / 200 / 20
NGO Caritas del Perú / Huánuco / Huánuco / Amarilis / 412 / 111 / 11
Junín / Chanchamayo / Chanchamayo / 331 / 89 / 9
Total / 743 / 200 / 20
Edpyme Edyficar / Lima / Lima / Chorrillos / 252 / 100 / 10
Lima / Lima / San Juan de Lurigancho / 783 / 100 / 10
Total / 1035 / 200 / 20
CRAC Cruz de Chalpón / Lambayeque / Chiclayo / Chiclayo / 136 / 119 / 12
Cajamarca / Jaén / Jaén / 52 / 46 / 5
Lambayeque / Ferreñafe / Ferreñafe / 24 / 21 / 2
Lambayeque / Lambayeque / Lambayeque / 16 / 14 / 1
Total / 228 / 200 / 20
CAC San Pedro / Apurímac / Chincheros / Ranracancha / 98 / 25 / 3
Apurímac / Chincheros / Anco – Huallo / 700 / 175 / 17
Total / 798 / 200 / 20
Step 2
Since Lima was taken twice, we decided to draw one more department out of the remaining ones. The selected department was Cajamarca
Step 3.
When we analyzed the sample, we observed that no department in the Amazonian region had been selected. To solve this, we applied PPS sampling to the departments in the Amazonian region alone. The selected department was Loreto.
Step 4
With a sample of 7 departments, we allocated the 800 non-client sample among the selected domains as shown in Table 2.
Table 2. Non-Client sample allocation among selected domains.
Domain / Total / % Population / % SampleCoast and Lima / 400 / 0.518 / 0.500
Metropolitan Lima / 200 / 0.289 / 0.250
Urban Coast / 132 / 0.178 / 0.165
Rural Coast / 68 / 0.052 / 0.085
Sierra / 266 / 0.350 / 0.333
Urban Sierra / 99 / 0.126 / 0.124
Rural Sierra / 167 / 0.224 / 0.209
Jungle / 134 / 0.131 / 0.168
Urban Jungle / 66 / 0.060 / 0.083
Rural Jungle / 68 / 0.071 / 0.085
TOTAL / 800 / 1.000 / 1.000
Table 3. Districts selected
Cerro Colorado / Chao / Ate
Mariano Melgar / La Esperanza / El Agustino
Tiabaya / Trujillo / Lima
Rímac
San Juan de Miraflores
Cajamarca / Loreto / Santiago de Surco
Cajamarca / Iquitos
Encañada / Punchana
Querocoto / Yurimaguas
Cusco / Piura
Echarate / Chulucanas
Quiquijana / Pariñas
Wanchaq / Sullana
Step 5. (Using PPS Sampling)
We selected 6 districts from Lima and 3 from all other departments. Table 3 shows the selected districts.
Step 6
Since there is no recent information about which districts have rural areas, we selected randomly, according to our needs, which districts would be rural and which urban. The districts that were chosen as rural were verified is they indeed have rural areas from were to take the rural sample.
Step 7
The sample was finally allocated as shown in Table 4:
Routes
Fourteen routes were required to cover all selected areas from which the sample was to be taken. Table 5 shows the routes covering the MFI-Client sample areas and Table 6 shows the routes covering the Non-Client sample areas..
Description of routes and assigned personnel
Only two out of the fourteen routes (routes 13 and 14) required the use of Quechua-speaking interviewers. Most interviews were therefore held in Spanish.
Except for the team who covered route 6 only, all teams covered two routes, one after the other. Table 7 details the sequences followed.
Route 1 with 2. Route 3 with 4 and 5. Route 6 with 7. Route 8 with 9 and 10. Route 11 alone. Route 12 alone. Route 13 with 14.
Table 4. Detailed allocation of Non-Client Sample among selected districts
Urban / 497 / 332 / 99 / 66
Rural / 303 / 68 / 167 / 68
TOTAL / 800 / 400 / 266 / 134
Arequipa / 100 / - / 100 / -
Urban / 66 / - / 66 / -
Mariano Melgar / 33 / - / 33 / -
Tiabaya / 33 / - / 33 / -
Rural / 34 / - / 34 / -
Cerro Colorado / 34 / - / 34 / -
Cajamarca / 100 / - / 100 / -
Rural / 100 / - / 100 / -
Cajamarca / 34 / - / 34 / -
Encañada / 33 / - / 33 / -
Querocoto / 33 / - / 33 / -
Cusco / 100 / - / 66 / 34
Urban / 33 / - / 33 / -
Wanchaq / 33 / - / 33 / -
Rural / 67 / - / 33 / 34
Echarate / 34 / - / - / 34
Quiquijana / 33 / - / 33 / -
La Libertad / 100 / 100 / - / -
Urban / 66 / 66 / - / -
La Esperanza / 33 / 33 / - / -
Trujillo / 33 / 33 / - / -
Rural / 34 / 34 / - / -
Chao / 34 / 34 / - / -
Lima / 200 / 200 / - / -
Urban / 200 / 200 / - / -
Ate / 33 / 33 / - / -
El Agustino / 33 / 33 / - / -
Lima / 33 / 33 / - / -
Rímac / 34 / 34 / - / -
San Juan de Miraflores / 34 / 34 / - / -
Santiago de Surco / 33 / 33 / - / -
Loreto / 100 / - / - / 100
Urban / 66 / - / - / 66
Iquitos / 33 / - / - / 33
Punchana / 33 / - / - / 33
Rural / 34 / - / - / 34
Yurimaguas / 34 / - / - / 34
Piura / 100 / 100 / - / -
Urban / 66 / 66 / - / -
Pariñas / 33 / 33 / - / -
Sullana / 33 / 33 / - / -
Rural / 34 / 34 / - / -
Chulucanas / 34 / 34 / - / -
Table 5. Routes for MFI-Client areas