The main objective of this survey is to help improve the impact of migration and remittances on the economic and social situation in Kenya. At present, our knowledge base on migration and remittances in Kenya is quite limited. By providing rich and detailed information on the impact of migration and remittances at the household level, this survey will greatly increase our ability to maximize the socio-economic impact of migration and remittances in Kenya. To these ends, the survey will collect nationally-representative information in various African countries on three types of households: non-migrant households, internal migrant households and international migrant households. Comparisons between these three types of households will help policymakers identify the socio-economic impact of migration and remittances in Kenya.
Kind of Data
Sample survey data [ssd]
Unit of Analysis
The scope of the Migration Household Survey includes:
- Household Roster
- Housing Conditions
- Household Assets and Expenditure
- Household Use of Financial Services
- Internal and International Migration and Remittances from Former Household Members
- Internal and International Migration and Remittances from Non-Household Members
- Return Migrants
17 out of 69 districts in Kenya were selected using procedures described in the methodology report
Producers and sponsors
University of Nairobi
University of North Carolina
The World Bank
The study used the Kenya National Bureau of Statistics (KNBS) National Sample Survey and Evaluation Programme (NASSEP IV) sampling frame which has 69 districts as stratum comprising both urban and rural areas. The sample design for the study was multi-stage with the first stage covering the primary sampling units (PSUs) which was a sample of clusters developed during the 1999 census. The second stage was selection of households within the clusters. A re-listing of all households in sampled clusters was carried out to up-date the 1999 and also to be able to classify households into the three strata of interest in this study: international migrant households, internal migrant households, and non-migrant households. At the household level, interviews were held with the household head/spouse or other responsible adult with the requisite information about the household.
The study uses a purposive survey methodology that first selected districts with the largest concentration of international migrants, and then selected clusters also with the highest concentration of international migrants. This was done based on the information of previous household surveys and the knowledge of the administrative officers, statistical officers and cluster guides.
At the time of the study, the available National Census was conducted in 1999. This census did not contain questions on remittances but had questions on migration. The migration question asked then was where family members were living in the last one year. This means that the census captured either those who had come back or those who had come visiting and were to return to where they migrated to. It did not distinguish clearly the migration component. Further, the census was conducted 10 years ago which meant it does not provide the current status on aspects of migration. The Kenya Integrated Household Budget Survey (KIHBS) 2005/06 and the Financial Services Deepening survey (FSD) are two surveys that have recently been conducted with an element of migration and remittances. However, the information is not adequate for the current survey. For example, the KIHBS has a question that captures issues of remittance linking them to the transfers received from abroad. Although it has about 13,000 households, only about 125 households indicated they had received such transfers. This was a very small sample compared to what was envisaged by the current study. The Financial Services Deepening survey (FSD) (2006/07) also has a question on cash transfers from abroad but all this is related to issues of access to financial services and not to issues sought in the current study. Thus, it could not be used for the current study. The KIHBS and FSD surveys was based on the KNBS NASSEP IV and although one may have thought of revisiting the households that were covered for additional information, it is against the KNBS regulations to conduct such follow-ups and the households identities are not provided.
The Kenya National Bureau of Statistics household survey sampling frame, the National Sample Survey and Evaluation Programme (NASSEP IV), is based on the 1999 population and housing census. The objective of NASSEP IV frame was to construct a national master sampling frame of clusters of households in both rural and urban areas in Kenya using a sound sampling design. This sampling frame has a total of 1,800 clusters of which 1,260 are rural and 540 are urban as indicated in Appendix Table 1. Each cluster holds about 80 to 100 households. The framework is based on the old administrative units comprising of 69 districts in 8 Provinces. Currently, the districts have been subdivided and increased to 265 but this does not distort our sampling frame based on NASSEP IV as the new districts are curved out of the old districts.
This study utilized the NASSEP IV frame to select 102 clusters (5.6% of the total clusters) in 19 districts which yielded a total sample of 2,448 households assuming an average of 24 households in each cluster. The districts were selected first, then the clusters in each district and finally the households in each cluster. Households in each cluster were re-listed (updated) and grouped into three strata--international migrant, internal migrant and non-migrant households. In the selection of clusters in each district, at least one of the targeted five clusters was urban with exception of Nairobi and Mombasa which are purely urban.
The study however ended up covering 92 clusters (5.1% of the total clusters in NASSEP IV) from 17 districts. Two targeted districts-Kajiado and Baringo- were not covered due to logistical problems. First of all, the team was expected to finalize the field by 15th December so that the analysis could begin and be on time. When the fieldwork was winding up on 22nd December, the two districts were yet to be covered. Two, the two districts have more transport challenges and the team was therefore expected to use KNBS transport facilities and more research assistants to capture the households which are more widely spread on the ground. This required adequate funding and by the time the fieldwork was winding up no funds had been received from World Bank. Third, even when the funds were received in January, the team considered that the study would be capturing households in a different consumption cycle, having just gone through the festive season. Given all these factors, this saw a total of 2,123 household covered out of 2, 208 (96% of the total targeted). Of these, some households were later dropped due to a lot of missing data especially due to non response, and at the end a total of 1,942 households were cleaned up for analysis. This including 953 are urban and 989 rural drawn from 51 rural and 40 urban clusters.
Selection of Districts
There was a particular interest in investigating households that had international migrants and which may have received transfers from abroad. A random sample of the population would not produce adequate number of households that had received transfers or had international migration, as we learnt from the KIHBS data set. As indicated earlier, out of 13,000 households surveyed under KIHBS only 125 households receiving remittances from abroad. With this experience and information, this study selected the top nineteen districts from KIHBS (2005/07) that showed households with migration characteristics. The key factor used was that the households indicated they received cash transfers from abroad. Districts with more than one household fulfilling this criterion of having received transfers from abroad were considered. In addition, Financial Services Deepening survey (FSD) survey results were used to confirm that the selected districts had reported having received money from abroad. In addition, since this is a relatively rare phenomenon in Kenya, the selection of districts is designed such that households with the relevant characteristics have a high probability of being selected. As such those districts with a presence of cash transfers mechanisms such as M-PESA, Western Union, or Money Gram services were considered. All these information was used to update the information from KIHBS.
Selection of Clusters
In each district, 5 clusters were selected of which at least one cluster was an urban cluster as defined by KNBS, except for Nairobi and Mombasa which are purely urban. Some other district had more than one urban cluster selected based on their number of clusters and accessibility to rural clusters for example Garissa. The study covered 10 clusters in Nairobi and 6 in Mombasa with an attempt made to capture this across various income group levels.
In selection of the clusters, the supervisors sat down with the KNBS statistics officers, cluster guides, village elders, administrative officers (Chiefs and sub-chiefs) to map out clusters where the probability of getting an international migrant was high. Of this probabilities were very subjective as it was based on how well these people understood the composition of the households in the areas they represent. This helped to identify the five clusters targeted for study.
Selection of Households
The selection process involved re-listing of the households in each cluster so as to update the list of occupied households and identify the three groups of households. Each group or stratum was treated as an independent sub-frame and random sampling was used to select households in each group. The listing exercise was facilitated by the respective District Statistical Officer (DSO). An instrument was developed to capture the basic characteristics of households in terms of household headship, number of members, and whether contain international or internal migrants and number.
A total of 24 households per cluster were targeted for selection for interview making a minimum of 120 households in each district except Mombasa and Nairobi. Experience on the ground was different and this saw variations on the household selected in the sample. The factors considered included availability of international migrants and a rule that in each cluster a maximum of 10 non-migrant households were surveyed (the expected rule was 10 non-migrants, 7 local migrants and 7 international migrant). However due to the rarity of international migrants, in the sampling process, international migrants were given a priority so that in each cluster almost all were included in the sample.
In the re-listing exercise, households were assigned identification numbers. When the number of households listed in a stratum exceeds the number to be sampled, households in that stratum were selected randomly. However, the procedure results in unequal selection probabilities of households in the three strata, which needs to be kept track of through the supervisor sampling sheets.
For the identification of the respondent household and clusters, the NASSEP IV sampling frame had adequately documented facilitating identification of selected clusters and households on the ground. Each cluster had a map and a listing of the households within it. The maps indicate the location of the structures and the households. This enabled the interviewers reach the selected household.
Deviations from the Sample Design
At the planning stage of the survey, the aim was to select the districts purposively and the clusters and households randomly. However, from the pre-testing we realised that it was possible to end up with clusters without international migrants, which was the key target, if and the design did not allow substituting the clusters once the selection has been made. We also realised that the DSOs, village elders and administrative officers had an idea of households in their locality and could identify those with international migrants. Thus we changed the methodology of selecting the clusters.
Dates of Data Collection
Data Collection Mode
The research team headed by a team leader provided the overall intellectual coordination for the project and in consultation with the World Bank Migration team, took final decisions on all matters relating to the project. The team leader worked with the principal researchers for day to day management of logistics and technical aspects the project. The supervisors oversaw the data gathering at the district level.
Data Collection Notes
Two levels of training were conducted for the instruments and sampling procedure. At the initial point the supervisors were trained on the questionnaire, re-listing process and selection of households. First of all, the research team in Nairobi held a one day workshop with Prof Bilsborrow and Ms. Plaza to finalise on the instruments and sampling procedures. Then the research team trained the supervisors for two days going through the re-listing instrument, the listing sheet, the selection process of households across the three groups and the field questionnaire. The training on the questionnaire reviewed the background, objectives and rationale of the household survey for the project, overview of the proposed scientific methodology for the survey, and an item-by-item familiarization with the instrument for fieldwork, contextualization of issues raised and potential responses and case studies. The training also outlined the project management structure, including reporting lines, roles and conduct of each team member, potential challenges and implications, documentation and field reports as well as relationships with respondents. Reports from the pilot study also featured.
The second level of training involved the interviewers. The first group was trained in Nairobi, by the supervisors and the research team. For the supervisors this was again a training ground for them as they were expected to train interviewers in each district they were allocated. The interviewers were trained on the questionnaire and re-listing process. In each selected district, the supervisors spent the two days training the interviewers recruited on the ground, sampling the clusters, re-listing households and sampling households. They also took the interviewers through some sort of pre-testing to ensure that they understood the questionnaire and were comfortable with the questions.
· Questionnaire was designed by the World Bank; updated and contextualized by the research team School of Economics University of Nairobi
· Tested in a pilot survey conducted in Nairobi
· Language of design was English. Enumerators however came from the local areas covered in the survey and were trained to elicit information as contained in the questionnaire in local languages
The survey questionnaire has 7 sections. Every question is important, and should be completed as fully and accurately as possible. The goal is to collect the BEST possible information from household respondents on each and every question. All answers are to be recorded on the actual survey questionnaire. Each household member is assigned an ID and this ID remains the same throughout the questionnaire. In other words, the ID of a household member does NOT change from one section of the questionnaire to another.
The cover sheet asks for basic information on the household, such as location of household, name of household head, village/town and enumeration area. It is very important to fill this cover sheet out correctly, so that – if some information is incorrect - the household can be located again.
SECTION 1: HOUSEHOLD ROSTER
This section collects basic demographic data on each and every member of the household. It is VERY important that this section be completed accurately and completely.
SECTION 2: HOUSEHOLD CONDITIONS
Section 2 collects information on household conditions and characteristics. Please interview the main person responsible for each dwelling. In this section there is only one response for each household. The Q2.1 asks for the tenure status of the dwelling whereas Q2.2 and Q2.3 ask about its construction and the material of its exterior walls. Information on cooking room and the total number of rooms are in Q2.4 and Q2.5. The last questions in this section ask about the presence of electricity and the source of drinking water for the household.
SECTION 3: HOUSEHOLD ASSETS AND EXPENDITURE
Section 3 is VERY important because it collects information on the assets and expenditures of the household. The goal of this section is to understand the current welfare status of the household, that is, is the household rich or poor.
SECTION 4: USE OF FINANCIAL SERVICES
Section 4 collects information on the household use of financial services. The purpose of this section is to see if households with migrants and remittances tend to use banks more often than households without migrants.
SECTION 5: MIGRATION AND REMITTANCES FROM FORMER HOUSEHOLD MEMBERS
Section 5 collects information on internal and international migration and remittances from former household members who are currently living outside of the household. “Former household members” here means any person who used to live in the household, but is CURRENTLY living away from the household in another place within the country OR in another country. The person who is CURRENTLY living outside the household may have moved away for the purposes of work, marriage, education, or other reason.
SECTION 6: MIGRATION AND REMITTANCES FROM NON-HOUSEHOLD HOUSEHOLD MEMBERS
Section 6 collects information on internal and international remittances received from non-household members. In other words, households without migrants may well receive remittances from friends, relatives and other people who are not members of their households. Households without migrants can also receive remittances from people for the repayment of loans or debts.
SECTION 7: RETURN MIGRANTS
Section 7 collects information on household members who used to live in another part of the country OR in another country, and have since returned to the household to live. For the purposes of this survey, these people are defined as “return migrants.” “Return migrants” here include all members of the household who used to live outside of the household for at least 3 months during the last 5 years, and have since returned to the household to live.
Data editing was conducted through a period of nearly three months. After data entry, the data analyst together with the team leader worked together to clean up the data and check for inconsistencies. The team also corresponded with Prof Mario Navarrete in identifying and correcting all inconsistencies in the original data.
Data was entered in SPSS. This was done at a central point in Nairobi with a group of data analysts. Descriptive statistics were conducts at different stages in the process of data cleaning. Questionnaires were entered as they came from the field. The supervisors spent some time confirming the questionnaires against the re-listing sheet and also checking for quality of information. With the help of data analyst who selected a sample of questionnaires from each district assisted in checking for consistency in the responses and addressing any missing information.
Initially the team had one data analyst who prepared the code sheet and managed the data entry for the first bunch of questionnaires. However as the data collection picked up, more data analysts were employed to assist with data entry. With the first one hundred questionnaires entered some statistics were generated to check on the quality of information and this helped with the cleaning exercise. Similar exercise was repeated as half of the questionnaires were entered.
Use of the dataset must be acknowledged using a citation which would include:
- the Identification of the Primary Investigator
- the title of the survey (including acronym and year of implementation)
- the survey reference number
- the source and date of download
Disclaimer and copyrights
The user of the data acknowledges that the original collector of the data, the authorized distributor of the data, and the relevant funding agency bear no responsibility for use of the data or for interpretations or inferences based upon such uses.