Ethiopia Households (ERHS)

The Ethiopian Rural Household Survey

Availability

The data and information on this website is made available by Stefan Dercon and the Centre for the Study of African Economics at the University of Oxford. Permission for use of the data for academic research is given by them and its collaborating institutions on the different rounds of the survey. We would ask that the use of the data and its source be acknowledged.

In particular, the 1989 survey was collected by the International Food Policy Research Institute (IFPRI). The 1994-1995 survey rounds (round 1 to 3) were collected by the Economics Department, Addis Ababa University and the Centre for the Study of African Economies. The 1997 survey round (round 4) was collected by the Economics Department, Addis Ababa University and the International Food Policy Research Institute, with limited input by the Centre for the Study of African Economies. The 1999 survey round was collected by the Economics Department, Addis Ababa University with input from the Centre for the Study of African Economies. The 2004 round was collected by the Economics Department, Addis Ababa University, the International Food Policy Research Institute and the Centre for the Study of African Economies.

Users should understand that, from 1994, the ERHS data collection occurred as part of a capacity-building project financed by SIDA. Users are therefore encouraged to conduct their research on the ERHS in collaboration with researchers of the Economics Department of Addis Ababa University. For further information on this, please contact Stefan Dercon (This email address is being protected from spambots. You need JavaScript enabled to view it.) for details.

We would also ask that the financing for the surveys be acknowledged from the Economic and Social Research Council (ESRC), the Swedish International Development Agency (SIDA) and the United States Agency for International Development (USAID).

For any queries, please contact me This email address is being protected from spambots. You need JavaScript enabled to view it.) but please understand that I cannot provide technical support in using these data. Access to the subsequent rounds will materialise in the future pending further discussions with our collaborating partners.

Stefan Dercon's home page

Abstract

The Ethiopia Rural Household Survey is a unique panel data set covering households in a number of villages in rural Ethiopia. Data collection started in 1989, when a team visited 6 farming villages in Central and Southern Ethiopia to conduct a household survey with a focus on the crisis and recovery in the 1980s. In 1994, the survey was expanded to cover 15 villages across the country. From the sample in 1989, about 360 households were successfully traced – implying an attrition rate of less than 5 percent. From 1994, the sample was expanded to about 1470 households. Data were subsequently collected in during further rounds in 1994 (later in the year), 1995, 1997 and 1999. A sixth round of data collection is taking place in 2004, in collaboration with IFPRI. Data are currently available on the web for 1989, 1994 (two rounds) and 1995. Other rounds will become available. A strong commitment to first use by Ethiopian researchers at the University of Addis Ababa and a serious backlog in manpower and documentation has meant that further data cannot be released yet, but this will happen in the future.

The data are provided in a clean format, using file, variable and label names corresponding closely to the questionnaire using SPSS. Note that transferring the data into other software, such as STATA, is straightforward but labels may be shortened in this procedure, so access to SPSS for basic data management is recommended.

The data contain numerous variables that are expressed in local measurement units (land and quantities) and are not yet valued in monetary terms. To facilitate conversion of the data, you can use the following files.

In 1994, the Centre for the Study of African Economies and the Economics Department of Addis Ababa University started a panel survey incorporating the villages surveyed earlier by IFPRI, who assisted. Six of the seven villages earlier surveyed were complemented with nine additional villages allowing for a total of 15 village studies, covering 1477 households. The nine additional communities were selected to account for the diversity in the farming systems in the country. Within each village, random sampling was used.

The selected households were interviewed thrice: in the first part of 1994, again later in the same year and in the first part of 1995. For 1994, the data collection was a complicated process and the data was put in per survey site.

A detailed description of the sampling method and the reorganised data is available for downloading as datadoc94a.pdf (PDF file, 47KB).

More advice and explanation on the rules followed for cleaning the data is available in the notes for each questionnaire contained in the three files: 94a-notes.pdf, 94b-notes.pdf, and 95-notes.pdf.

You can also download the questionnaires themselves: 1994a, 1994b, and 1995. The questionnaires have straight forward structure. When printed on A4, blank pages should be added where appropriate and the questionnaire should be copied recto verso. This enables a straightforward interpretation of the codes used.
 

The base data files are available as Zip archives:

94a-base.zip (775KB)
94b-base.zip (kB)
95-base.zip (kB)

These files are more user friendly reorganised data, following the structure of the questionnaire. (The original data was put in per survey site)

The files are in SPSS-format (extension .sav) and contain a data dictionary. If you have SPSS, it is easy to know the content of the files. To find out which variables each file contains, 1) start SPSS, 2) choose 'file' from the main menu, 3) from the drop down menu, choose 'Display Data info', 4) in the interactive window choose the filename and file type, and choose 'Open'. A list of the variables and the questions they refer to appears in the output window.

To translate the files into a different format is very straightforward using a transfer programme like Stat/Transfer or DBMS-copy.

You can also download the questionnaires themselves:

When printed, blank pages should be added where appropriate and the questionnaire should be copied recto verso. This enables a straightforward interpretation of the codes used. More advice and explanation on the codes used and the rules followed for cleaning the data is available in the notes for each questionnaire contained in the three files:

A detailed description of the reorganised data is available for downloading as datadoc-ifpri.pdf (40KB).

In 1989, the International Food Policy Research Institute (IFPRI) conducted a survey in seven communities (Peasant Associations) located in the regions Amhara, Oromiya and the Southern Ethiopian People's Association.

The study collected mainly consumption, asset and income data. The roughly 450 households were randomly selected within the Peasant Asssociations. The selected Peasant Associations were in areas which had suffered from famine. The sample does not pretend to be statistically representative.

The data collection was a complicated process, using different questionnaires - which contained common parts - for the different Peasant Associations. The data was put in per Peasant Association. Because of the complicated structure of the original questionnaire, the data was reorganised per theme. This more user friendly version of the data is available in the base data files.

Six of the seven Peasant Associations surveyed in 1989 were later incorporated in the 1994 Ethiopian Rural Household Survey (ERHS) to form a panel of roughly 400 households for 1989-1994.

The files are available for downloading as a Zip archive of the 1989 base data files (580KB).

The original questionnaire consisted of a tailor made questionnaire per Peasant Association. The raw data files are organised as such, that is, per survey side. These files can be found in 1989 survey raw data files. To save a long process of data restructuring, you can also download a more user friendly reorganised version of the data. This reorganised data is described on this page. It also contains some derived variables and is organised in 11 themes:

1. Household income files form all sources
2. Area planted, harvested and values of production by year
3. Household members, age sex, birth, death etc.
4. Number of adults, females, males, etc. at household level
5. Livestock current holdings and recall
6. Livestock mortality and types of food consumed during the worst year
7. Values of farm and non-farm assets sold by year
8. Present values of farm and non-farm assets
9. Assets
10. Harvest use
11. Farm history

Within each theme, you can choose for the pooled files across all sites or for the data per Peasant Association. Only for the last three themes (assets, harvest use and farm history) there are no site-specific files available. For the themes 4,5, 8 and 9 on the other hand, a pooled file is not available for the moment. The table underneath lists the files.

The files are in SPSS/PC+ format (extension .sys) and contain a data dictionary. If you have SPSS, it is easy to know the content of the files. To find out which variables each file contains, 1) start SPSS, 2) choose 'file' from the main menu, 3) from the drop down menu, choose 'Display Data info', 4) in the interactive window choose the filename and file type, and choose 'Open'. A list of the variables and the questions they refer to appears in the output window.

To transfer the data into another format, first open the .sys file in SPSS and save it as a .sav file. To translate the files into a different format is very straightforward using a transfer programme like Stat/Transfer or DBMS-copy.

If you have access to a transfer programme like Stat/transfer, it is very straightforward to get the data into another format (like SAS, Stata, …)

More detailed information on the files and their content can be found in the PDF file of the 1989 Survey Data Documentation. The exact phrasing of the questions can be found using the 1989 Questionnaire.

Files contained in the Zip archive:

1. Food consumption and expenditure files

Survey site File name
Debre Berhan debxcly.sys
Dinki dinxcly.sys
Arssi arsxcly.sys
Harerghe harclxpy.sys
Wolayta wolxcly.sys
Gamogofa gamxcly.sys
Sidamo sidxcly.sys
Pooled file across sites ethxly.sys

2. Household income files from all sources

Survey site File name
Debre Berhan debinc5.sys
Dinki dininc5.sys
Arssi arsinc5.sys
Harerghe haryrev4.sys
Wolayta wolinc5.sys
Gamogofa gaminc6.sys
Sidamo sidinc5.sys
Pooled file across sites ethyrev.sys

 

Survey site File name
Debre Berhan debprodv.sys
Dinki dinprodv.sys
Arssi arsprodv.sys
Harerghe harprodv.sys
Wolayta wolprodv.sys
Gamogofa gamprodv.sys
Sidamo sidprodv.sys
Pooled file across sites ethprodv.sys

4. Household members age, sex, birth, death, etc.

Survey site File name
Debre Berhan debdemo4.sys
Dinki dindemo4.sys
Arssi no file
Harerghe hardemo4.sys
Wolayta woldemo4.sys
Gamogofa no file
Sidamo siddemo4.sys
Pooled file across sites ethdemo.sys - not available yet

 

Survey site File name
Debre Berhan debfmly.sys
Dinki dinfmly.sys
Arssi no file
Harerghe harfmly.sys
Wolayta wolfmly.sys
Gamogofa no file
Sidamo sidfmly.sys
Pooled file across sites ethfamly.sys - not available yet

6. Livestock current holdings and recall

Survey site File name
Debre Berhan deblvs5.sys
Dinki dinlvs5.sys
Arssi arslvs5.sys
Harerghe harlvs5.sys
Wolayta wollvs5.sys
Gamogofa gamlvs5.sys
Sidamo sidlvs5.sys
Pooled files across sites ethlvs2.sys
Pooled files across sites tlsu80.sys

7. Livestock mortality and types of food consumed during the worst year

Survey site File name
Pooled files across sites wyrlvs.sys
Pooled files across sites crisfud.sys

8. Values of farm and non-farm assets sold by year

Survey site File name
Debre Berhan solddeb.sys
Dinki solddin.sys
Arssi soldars.sys
Harerghe soldhar.sys
Wolayta soldwol.sys
Gamogofa soldgam.sys
Sidamo soldsid.sys - not available yet

9. Present values of farm and non-farm assets

Survey site File name
Debre Berhan assetdeb.sys
Dinki assetdin.sys
Arssi assetars.sys
Harerghe assethar.sys
Wolayta assetwol.sys
Gamogofa assetgam.sys
Sidamo assetsid.sys - not available yet

10. Assets

Survey site File name
Pooled file across sites ethasets.sys

11. Harvest use

Survey Site File name
Pooled file across sites  harvuse.sys

12. Farm history

Survey Site File name
Pooled file across sites pre74.sys

The questionnaire can be downloaded as quest89.pdf.

The original questionnaire for the 1989 household survey existed in different forms for the different survey sites. The questionnaire available below is a compilation of the common parts of the site-specific questionnaire. It contains 7 parts:

A. Household demographics
G. Household drought experience
B. Household Assets
I . Ox Seed Project
C . Production an Inputs
D. Non-farm income
E. Mother's questionnaire

The questionnaire should be seen as a reference to how the question was phrased, and not as a guide to how the data is organised. When you read the data in SPSS, all variables have the question to which they refer attached as label. Linking the variables to the questions is straightforward.

Note that some parts of the questionnaire come from an older version. If there is a conflict between the question in the questionnaire and the one attached to the variable in the data file (e.g. between dates as in 'pre-74'), consider the latter as the correct one.
 

The files described here are available for downloading as a Zip archive, link84n94a.zip (1KB)

The files in the 1989 survey all contain the variable HHID89, which identifies the household. This variable is also the key to link with 1994a ERHS data, which included the variable q2, which is the household number in the survey in 89. Note that only a small subset of households in the 1994a ERHS data were interviewed in 1989.

Because the 1989 survey used other price information, conversion codes and other principles for the data cleaning that could not be replicated in 1994, most of the variables are not directly comparable between 1989 and 1994a. To solve this problem, a series of variables were recalculated using the initial 1989 raw data and questionnaire. This information is available in the two SPSS files: the file cons89.sav contains aggregate information on consumption data in 89; the file demo89.sav contains aggregate information on household characteristics. Note that linking information from 1989 to 1994a may lead to spurious or unreliable results for the reasons mentioned above.