HOUSEHOLD INCOME SURVEY 2019

Household Income Survey (HIS) is one of surveys conducted by the Department of Statistics, Malaysia (DOSM). The survey was conducted for the first time in 1973 and subsequently carried out twice every five years, two surveys in each period of the Malaysian Development Plan. The main aims of HIS was to measure the economic well-being of the population, collecting information on household income distribution patterns according to various socio-economic features and providing basic data for calculating Poverty Line Income (PLI). Household income and poverty statistics are used for policy formulation and development plans especially for poverty eradication strategies and income distribution programs. This paper aimed to describe the sampling methodology used in designing HIS 2019 in Malaysia. In addition, this paper will emphasize on enhancement made in HIS 2019. A two stage stratified random sampling was adopted in the survey. The sampling frame was stratified by administrative district and urban/rural localities. The Primary Sampling Unit (PSU) is Enumeration Block (EBs) based on the information from 2010 Population Census. A total of 11,529 EBs were selected from the total EBs in Malaysia, where 7,557 and 3,962 EBs was selected from urban and rural areas respectively. The Secondary Sampling Unit (SSU) is Living Quarters (LQs) within the selected EBs. On the average, eight LQs were randomly selected from each selected EBs. All households within the selected LQs were included in the study. Sampling methodology is one of the critical success factors of a survey. Hence, a robust sampling methodology was used in executing this survey to ensure the representativeness of the study, thus will provide reliable findings for users particularly policy makers. Household Income Survey (HIS) is one of surveys conducted by Price, Income & Expenditure Statistics Division, Department of Statistics, Malaysia (DOSM) and is carried out twice in five years. The previous HIS was conducted in 2016 through 2017. HIS make available statistics concerning the household income and poverty to measure the economic well-being of the population. These statistics is used primarily by the government as inputs for the formulation of national development plans and monitoring the Malaysia Plan. In addition, these statistics is also widely used by the researchers as well as individuals for further analysis and research purposes.

Survey Instruments and Data Collection Techniques

Personal interview approach is used in collecting HIS data. Training was given to the officers in DOSM state’s office who were involved as enumerator in this survey. They will visit selected households to collect information on demography and income using a set of bi-lingual (Malay and English) structured questionnaires on the scopes of the survey. There were several modules included in the questionnaire. The household income questionnaire contains identification particulars (would be head of household as any members whether male or female which is an income recipient and age 15 years and over), household member particulars, individual and household income, annual household income payment during the last twelve (12) months and current transfer payment. To ensure the quality of the data, supervisor will perform certain checking procedures to detect and correct any error or omission during the survey.

Scope and Coverage

Target population for HIS 2019 is all households in Malaysia in both urban and rural areas. Individuals who live in residential institutions such as hostels, hotels, hospitals, old folk homes, prisons and welfare homes were excluded from the survey.

Sampling Frame

The sampling frame used for the HIS 2019 was based on the Household Sampling Frame i.e. a list of enumeration blocks (EBs) that were updated from time to time subsequent to the 2010 Population and Housing Census. Malaysia was geographically divided into 79,000 enumeration blocks (EBs). Furthermore, EBs were recognized as geographical contiguous areas of land with identified boundaries, contains about 80 to 120 living quarters with an average population of 500 to 600 people. Basically, all EBs were based on the population size of the gazette boundaries.

The EBs in the sampling frame was classified into either urban or rural areas. Urban areas were gazette area, with their adjoining built-up areas, which has a combined population of 10,000 or more as defined in the 2010 Population and Housing Census. Meanwhile, gazette area with population less than 10,000 could be classified as rural area. In year 2018, there were about 79,000 EBs in HIS sampling frame, with about 59,000 and 20,000 urban and rural EBs, respectively.

The sampling frame used to draw a HIS sample includes 78,976 enumeration blocks (EB) for urban and rural areas in strata 1 to strata 4 for Peninsular Malaysia as well as strata 1 to 0 in Sabah and Sarawak. The definition of strata is as below:

Sampling Design

The two-stage stratified random sampling was adopted in this survey. The strata were the primary stratum, which made up of the states in Malaysia, including Federal Territories. The second stratum, which was made up of the administrative districts by state and the tertiary stratum was made up of the urban and rural formed within the second stratum. The sampling involved two stages; the primary sampling unit (PSU) was the enumeration blocks (EBs) which the selections of the sample is done using Probability Proportionate to Size Sampling. Meanwhile, the secondary sampling unit (SSU) was living quarters (LQs) within each selected EB which the selections is done using Systematic Random Sampling. On an average, eight (8) LQs were randomly selected from each selected EBs. All households within the selected LQs were included in the study.

Sample Size

The optimum sample size estimation was based on several inputs i.e. total number of household 2018, average of household income, design effect and response rate from the previous HIS 2016/17. The estimation is calculated for each domain i.e. urban and rural at administrative district levels with several choices of Relative Standard Error (RSE) value (5.0%, 6.0%, 7.0% and 10.0%) to give an option of the sample size estimation. The final sample size for HIS 2019 is determined taking into consideration sample size and the RSE from previous survey (HIS 2016/17) as well as capability in terms of burden for data collection and cost involved. In general, sample size is increased for domain with considerably high RSE. In contrast, the previous sample size is maintained for domain with low RSE. This is to ensure the reliability of the estimates produced.

Table 1: Sample size (EB) by state and strata for HIS 2019 and HIS 2016/2017

Improvements in HIS 2019 Sampling Design

Representative samples are important as they ensure that all relevant types of people are included in the sample and that the right mix of people is interviewed. If the sample isn’t representative it will be subject to bias. Certain groups may be over-represented and their opinions magnified while others may be under-represented.

Generally, household's income has a correlation with the type of LQ and areas they are residing. Since the information of income for all household is not available in the sampling frame, this assumption is used as a proxy to ensure the sample of EBs and LQs consists of all income level in Malaysia. Hence, for HIS 2019, the number of sample size allocated for each urban and rural at administrative district level is proportionately distributed into type of LQ and details strata based on distribution of EB according to type of LQ as well as distribution of EB according to detail strata in the sampling frame. Thus, would increase the representativeness of the samples. Table 2 and Table 3 below shows the EBs distribution by strata detail and type of living quarters in Selangor respectively.

Table 2: Allocation of EBs by each strata detail for population and sample HIS 2019 in Selangor

Table 3: Allocation of EBs by type of living quarters for population and sample HIS 2019 in Selangor

Way Forward

Ideally, a good sampling design for HIS requires income information for each household in Malaysia particularly in the sampling frame. This is because based on this information strata according to income group can be formed to ensure that the selected sample covers all income groups. This will increase the representativeness which will ultimately produce the best estimate for the average household income in Malaysia. As an effort to realize this, starting from 2017, DOSM has begun collecting income information for every household. This information will also be collected during the 2020 Population Census. Hopefully with this effort, the sampling design for HIS will be improved in the near future.

1.