The Timor-Leste Household Income and Expenditure Survey 2010 is the first survey of this type ever conducted in the country. It is also special in the sense that it was undertaken entirely by the staff of the National Directorate of Statistics and funded from national resources. Only limited and occasional technical assistance was provided from abroad. The survey was undertaken primarily to address two important issues in the national statistical system. The first is the method of calculating consumer price indices (CPI) and the second issue is the system of national accounts.
While the HIES2011 is a new enquiry, some of the results can be compared to those obtained by the Living Standards Surveys of 2001 and 2007 and the Demographic and Health Surveys of 2003 and 2009, thus allowing trend analysis. The present survey also benefitted from the sampling frame provided by the Population Census of 2010.
The purpose of the survey is to obtain statistically useful information about the incomes and expenditures of private households in Timor-Leste. In this context ancillary information is collected about household composition, household members’ individual characteristics, housing situation, ownership of durable goods, access to facilities, and more.
This type of household enquiry provides direct insight in the economic situation of households. More importantly, by repeating the survey on a periodic basis, trends can be discerned. This informs statistics users about whether households have improved their economic situations over the intervening period, or possibly that they are now less well off than before. Such trends are caused by economic developments that are not necessarily under the control of policy makers. Nevertheless the effect of policy decisions can be studied through a HIES, for example the result of measures intended to support economically vulnerable segments of the population.
The 2011 Timor-Leste Household Income and Expenditure Survey covered the following topics:
- Household identification, geographic area location, and enumerator information
- Person general information
- Consumption and expenditure on food and non-food Items
- Consumption and expenditure on housing and household items
- Household income from work and trade
- Ownership of durable goods
- Dwelling, Crops, Livestock
- Person income information
- Other income and money transfers
Producers and sponsors
National Statistics Directorate (NSD)
Ministry of Finance
Government of Timor-Leste
Funding the survey
The sampling model developed by Sampling expert Abdul Wahab aims at statistical significance of the results at the national, national urban and national rural level. Besides that the sample has been chosen so as to also provide meaningful results at the level of the Oecusse District which is an outlying enclave of Timor-Leste in the Western part of Timor Island.
The sampling method is two-stage, in both stages systematic sampling is applied. The first stage is the EA-level. From the 1,828 EA's defined for the 2010 census, 188 were selected by listing and numbering the set, selecting a random starting number and appropriate interval, then drawing up the list of EA's to be covered. Inorder to obtain roughly equal sampling errors for urban and rural Timor-Leste, urban EA's were to be over-sampled. From the 397 urban EA's eventually 43 were selected, while from the 1431 rural EA's 145 became part of the HIES sample. Thus the sampling fraction at the first stage, the EA, in urban areas was 10.8% and in rural areas 10.1%.
A total of 24 private households are selected from each participating EA. Calculating on the basis of households, the sample thus contains 43 * 24 = 1,032 urban households, out of the 43,938 enumerated by the 2004 Census. This represents at the second stage, the household level, a sampling fraction of 2.35%. For the rural stratum the figures are that the sample contains 145 * 24 = 3,480 households out of a universe of 151,024 found by the 2004 Census. The rural sampling fraction for households thus becomes 2.30%.
(Refer Section 3 of the final survey report for detail sampling information)
Dates of Data Collection
Data Collection Mode
Data Collection Notes
A pilot survey was undertaken over the period September 1-8, 2010 in two Suco’s, one in Dili District, the other in rural Aileu. As was planned for the survey proper, 24 households were interviewed in each Suco, resulting in a data sample of 48 households. Beforehand six interviewers were trained in the details of administering the survey. Most trainees later held positions of responsibility in the HIES operation itself.
The results of the pilot were analyzed and lessons learned. Among other things, some questions in the questionnaire were reformulated. The pilot census also gave a better idea about the working time required per household and per Suco.
The training of the field staff: interviewers, data editors and team supervisors, took place from 10 to 21 January 2011. The training consisted of an initial theoretical review of the procedures and discussion of the questionnaire, followed by extensive fieldwork. Each participant completed at least four household interviews. The resulting questionnaires were then reviewed and any omissions, mistakes or lack of clarity discussed individually with its author.
While most of the practice interviews were conducted in Dili, the training also included a field exercise in Suco Tibar, Liquica District. Field work for the survey proper started immediately after this training had been completed.
The survey had been carried out over the course of 12 months.
The three field teams each consisted of a team supervisor, a travelling editor, five interviewers and a driver. There were seven women among the staff, one of whom worked as a supervisor, the others have been interviewers. The teams stayed in contact with NSD by mobile phone where coverage permits. The teams worked six days per week for an extended period of four to six weeks. They then had a short break and the team was reassigned to another geographical area. This assisted in obtaining geographic coverage that is more or less evenly distributed over the year.
A team required an average of some four working days to cover an EA. One to one-and-a-half days were needed to produce the household listing, which should be exhaustive. The remaining time was used to conduct the 24 interviews. An average interview did take 2 to 3 hours.
In a data collection operation like this much depends on the ability of the respondent(s) to recall food consumed over the past week and non-food items procured over the last month or year. It is easy to forget an item. In order to jog the memory of respondents, interviewers asked specifically about every item in the questionnaire, to inquire whether or not it had been consumed and/or purchased. This amounts to 190 different food stuffs and 89 categories of non-food items. While this is a tedious procedure for both interviewer and respondents (very often the answer has to be “No”), it is essential for obtaining the best possible information.
Preparing a household listing immediately before the interviewing avoids problems where older listings might turn out to have become obsolete. Therefore interviewers usually encountered no problem identifying the dwellings and the households to be interviewed. Despite the considerable response load placed on participating households, refusal to cooperate has been minimal, as is usual in Timor-Leste. When sometimes respondents where not available during working hours, interviewers have revisited at another more convenient time.
National Statistics Directorate
Ministry of Finance
For questionnaire design, much benefit was derived from the United Nations methodological document “Household Sample Surveys in Developing and Transition Countries” [UN05]. Available questionnaires from other countries were studied: Australia, Brazil, Eritrea, Ethiopia, Ghana, Iraq, Kenya, Republic of Korea, Liberia, Maldives, Mexico, Micronesia, Mongolia, Mozambique, Pakistan, Philippines, Samoa, South Africa and Sri Lanka. Not all of this material was equally useful, since methodologies, subjects of inquiry and even the very idea of what constitutes a household income and expenditure survey differ widely.
In the event, the design of the present questionnaire was mostly inspired by two sources: the National Economic Survey of Indonesia (SUSENAS) and the 2007 Timor-Leste Survey of Living Standards, which contains modules on household income and expenditure. This is not to say that the HIES is in any way a successor to the TL-SLS, which was much more ambitious in its non-economical chapters. As compared to the TL-SLS the HIES is a modest operation using a smaller questionnaire and requiring fewer resources in terms of manpower, respondent time, foreign expertise and funding. It is also more focused on the micro-economy of the household.
The draft questionnaire was circulated for comments among interested parties within and outside the NSD, which resulted in a limited number of reactions. Useful responses were obtained from the NSD National Accounts adviser and from various staff members of the Asian Development Bank. These responses resulted in several minor and a few major improvements.
The data had been editied in the field, after field editing completed, batches of completed EA’s (24 households each) were sent to Dili where data entry and computer batch editing took place. Computer edit did normally produce a number of warnings and error messages. Questionnaires and computer records were then visually inspected to resolve the issues. Often the mistakes resulted from typing errors or multi-interpretable writings on the questionnaire. There are also some outliers in the form of households with unusually high income and expenditure levels. In exceptional cases the team supervisor was consulted, which occasionally led to a revisit of the household to clear up questions, omissions or inconsistencies.
Data entry was done using CSPro.
The efforts resulted in a full-fletched data dictionary, two stand-alone programs and a number of table generating routines. The data-entry program contains a considerable number of range checks and consistency controls that will alert operators to mistakes. These mistakes can be erroneous entries in the questionnaires or keying mistakes. In most cases the operators were able to correct what appeared to be wrong, but sometimes they needed to ask the assistance of one of the supervisors to solve a particular issue.
Not all mistakes can be caught by a data-entry program, if only because it would lead to unacceptable delays. Therefore there is also a batch-editing program that generates a list of mistaken or suspect fields in a large number of individual questionnaires. This list was then used for guidance by the data-entry operator and/or supervisor to manually correct entries. On rerunning the batch-edit program the error messages should have disappeared.
There can be entries that seem unlikely but might be correct. This could concern unusually high wages or yearly expenditures that hardly surpass monthly expenditures for the same item category. The supervisor reviewed such error messages in the context of what was otherwise known about the household. In some exceptional cases the household needed to be revisited. In this respect it was useful that the questionnaire contains geo-coordinates (latitude, longitude) for each household in the sample.
Automatic corrections – imputations – were not used in this survey. The sample size was considered too small for this relatively unsophisticated correction method to be applicable.
The use of the datasets must be acknowledged using a citation which would include:
- the identification of the Primary Investigator (including country name)
- the full title of the survey and its acronym (when available), and the year(s) of implementation
- the survey reference number
- the source and date of download (for datasets disseminated online)
Timor-Leste National Statistics Directorate. Household Income and Expenditure Survey (HIES) 2011. Ref. TLS_2011_HIES_v01_M_v01_A_PUF. Dataset downloaded from [URL] on [date].
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.