Labour Force Survey 2016-2017
Bangladesh, 2016 - 2017
BANGLADESH BUREAU OF STATISTICS
Identification
idno
BGD-BBS-LFS-2016-17-v01
Title
Labour Force Survey 2016-2017
Country
| Name | Country code |
|---|---|
| Bangladesh | BGD |
Abstract
Bangladesh continues to grow and integrate further with the global economy, and that’s why the access to decent and productive work remains one of the most viable means of poverty reduction. Despite of major achievements, there are lingering and emerging policy challenges confronting the country that will influence the achievement of its decent work goals. As the working-age population expands in the coming years, the pressure on the labour market to provide quality jobs will also rise. Addressing deficits in both the quantity and quality of jobs therefore remains a major policy challenge for Bangladesh. In general, several noteworthy trends emerged from the past decade in labour market - growth reduced poverty rates, even though unemployment rates are low and falling. This raises concerns about the quality of jobs, especially with such a large proportion of workers in vulnerable employment. The projected population trends indicate a rise in the adult working-age population, which is likely to add to the challenge of creating decent work opportunities for an expanding labour force.
Bangladesh Bureau of Statistics (BBS), the National Statistical Organization (NSO) of the country, has been conducting Labour Force Survey (LFS) since 1980 and continue it every three/four year until 2013. BBS has started implementation of quarterly labour force survey (LFS) to provide labour market indicators from July 2015 under a development project. The LFS 2016-17 report is the second annual report with quarterly breakdown of the estimates. Gender disaggregated data on labour force, employment, unemployment, underemployment, not in labour force, hours worked, earnings, informal employment. non-economic activities, volunteer activities are available in this report. Relevant sex and age-specific labour market information is provided in this report for informed decision-making and setting up an effective labour market information system. Additional efforts are taken to boost the information base and to achieve more robust and detailed labour and social trends analysis and monitoring, which will in turn provide a more credible basis for labour market policy formulation.
The survey report has provided a complete picture of labour statistics as well as the following key Indicators of labour market:
* Labour force participation rate
* Employment-to-population ratio
* Status in employment
* Employment by sector
* Employment by occupation
* Hours of work
* Employment in the informal economy
* Unemployment and youth unemployment
* Not in labour force
* Educational attainment
* Average monthly wages
Objectives of the survey
The primary objective of the survey was to collect comprehensive data on the labour force, employment and unemployment of the population aged 15 or older for use by the Government, international organizations, NGOs, researchers and others to efficiently provide targeted interventions. Specific objectives of the survey:
* Provide relevant information regarding the characteristics of the population and household that relate to housing, household size, female-headed households etc.
* Provide detailed information on education and training, such as literacy, educational attainment and vocational training.
* Provide relevant information on economic activities and the labour force regarding the working-age population, economic activity status and labour force participation.
* Provide detailed information on employment and informal employment by occupation and industry, education level and status in employment.
* Provide relevant information on unemployment, the youth labour force participation, youth employment, and youth unemployment.
* Provide other information on decent work regarding earnings from employment, working hours and time-related underemployment, quality and stability of employment, social security coverage, and safety at work, equal opportunities etc.
* Provide relevant information on non-economic activities, volunteer activities etc.
Bangladesh Bureau of Statistics (BBS), the National Statistical Organization (NSO) of the country, has been conducting Labour Force Survey (LFS) since 1980 and continue it every three/four year until 2013. BBS has started implementation of quarterly labour force survey (LFS) to provide labour market indicators from July 2015 under a development project. The LFS 2016-17 report is the second annual report with quarterly breakdown of the estimates. Gender disaggregated data on labour force, employment, unemployment, underemployment, not in labour force, hours worked, earnings, informal employment. non-economic activities, volunteer activities are available in this report. Relevant sex and age-specific labour market information is provided in this report for informed decision-making and setting up an effective labour market information system. Additional efforts are taken to boost the information base and to achieve more robust and detailed labour and social trends analysis and monitoring, which will in turn provide a more credible basis for labour market policy formulation.
The survey report has provided a complete picture of labour statistics as well as the following key Indicators of labour market:
* Labour force participation rate
* Employment-to-population ratio
* Status in employment
* Employment by sector
* Employment by occupation
* Hours of work
* Employment in the informal economy
* Unemployment and youth unemployment
* Not in labour force
* Educational attainment
* Average monthly wages
Objectives of the survey
The primary objective of the survey was to collect comprehensive data on the labour force, employment and unemployment of the population aged 15 or older for use by the Government, international organizations, NGOs, researchers and others to efficiently provide targeted interventions. Specific objectives of the survey:
* Provide relevant information regarding the characteristics of the population and household that relate to housing, household size, female-headed households etc.
* Provide detailed information on education and training, such as literacy, educational attainment and vocational training.
* Provide relevant information on economic activities and the labour force regarding the working-age population, economic activity status and labour force participation.
* Provide detailed information on employment and informal employment by occupation and industry, education level and status in employment.
* Provide relevant information on unemployment, the youth labour force participation, youth employment, and youth unemployment.
* Provide other information on decent work regarding earnings from employment, working hours and time-related underemployment, quality and stability of employment, social security coverage, and safety at work, equal opportunities etc.
* Provide relevant information on non-economic activities, volunteer activities etc.
Kind of Data
Sample survey data [ssd]
Version
Version Date
2018-01
Producers and sponsors
Authoring entity/Primary investigators
| Agency Name | Affiliation |
|---|---|
| BANGLADESH BUREAU OF STATISTICS | Statistics and Informatics Division, Ministry of Planning |
Producers
| Name |
|---|
| Improving of Labour Statistics and Labour Market Information System through Panel Survey |
Funding Agency/Sponsor
| Name | Abbreviation |
|---|---|
| Statistics and Informatics Division | SID |
Sampling
Sampling Procedure
The frame used for the selection of sample for the survey was based on the Population and Housing Census 2011. Sampling Frame which was made up of preparing of PSUs that is consists of collapsing one or more Enumeration Area (EAs) that was created for the Population and Housing Census 2011. EAs is geographical contiguous areas of land with identifiable boundaries. On average, each PSUs has 225 households. All the Enumeration areas of the country was identified into three segments viz. Strong, Semi-strong and not-strong based on the housing materials. The frame has 1284 PSUs/EAs spread all over the country, and covers all socio-economic classes and hence able to get a suitable and representative sample of the population. The survey was distributed into twenty-one domains viz. Rural, Urban and City corporations of seven administrative divisions.
From each selected PSUs/EAs, an equal number of 24 households were selected systematically, with a random start. The systematic sampling method was adopted as it enables the distribution of the sample across the cluster evenly and yields good estimates for the population parameters. Selection of the households was done at the HQ and assigned to the Enumerators, with strictly no allowance for replacement of non-responding households.
The Bangladesh Quarterly Labor Force Survey (QLFS) sample will be selected in two stages, with small area units called Primary Sampling Units (PSUs) in the first stage and a cluster of 24 households per PSU in the second stage. Both stages are random selections. The survey will implement a rotational panel strategy, in which some of the households in each cluster will be replaced by new households each quarter.
The survey administered with a total sample about 123 thousand households, intended to deliver reliable quarterly estimates of unemployment and other relevant labor force indicators for of the country’s seven divisions and locality viz. national level estimates with disaggregation by City Corporations, Rural and Urban.
From each selected PSUs/EAs, an equal number of 24 households were selected systematically, with a random start. The systematic sampling method was adopted as it enables the distribution of the sample across the cluster evenly and yields good estimates for the population parameters. Selection of the households was done at the HQ and assigned to the Enumerators, with strictly no allowance for replacement of non-responding households.
The Bangladesh Quarterly Labor Force Survey (QLFS) sample will be selected in two stages, with small area units called Primary Sampling Units (PSUs) in the first stage and a cluster of 24 households per PSU in the second stage. Both stages are random selections. The survey will implement a rotational panel strategy, in which some of the households in each cluster will be replaced by new households each quarter.
The survey administered with a total sample about 123 thousand households, intended to deliver reliable quarterly estimates of unemployment and other relevant labor force indicators for of the country’s seven divisions and locality viz. national level estimates with disaggregation by City Corporations, Rural and Urban.
Response Rate
The number of households’ n needed to estimate an individual-level prevalence P with a margin of error E at the confidence level a is given by
where Deff is the design effect, due to stratification and clustering, c is the average number of relevant individuals per household, and ta is the normal deviate corresponding to the confidence level a.
The earlier LFS 2013 reported an unemployment rate of 4.2%, with a design effect of 1.77 and an average of 1.92 persons in the labor force per household. Using these figures as referential parameters, the number of households needed to estimate this indicator with a margin of error of 1% at the 95% confidence level is which implies that a total sample of around 29,400 households would be needed to achieve the required precision in all 21 estimation domains. Since these domains have very unequal populations – ranging from less than half a million in the smaller city corporations to nearly 30 million in rural Dhaka – the distribution of the sample into such domains should arbitrate between doing it equitably (which would deliver estimates of similar precision for all of them) and doing it proportionally (which would deliver nearly optimal estimates for Bangladesh as a whole). Consistently with the criterion used by the BBS for the 2013 QLFS, the new survey will do it on the basis of Kish’s allocation, which is generally considered the best compromise between these two extremes: the sample will be thus distributed in proportion to the factors ?1 ? / ?? ? 1 / ??, where H is the number of strata (24 in this case) and Nh (1 = h = H) are the number of households reported by the 2011 census in each domain.
In addition to the above theoretic considerations, the QLFS sampling design needs to account for two practical constrains imposed by fieldwork management:
* First, the total sample size obviously needs to be a multiple of the cluster size (24 households).
* Second, since the survey will be fielded by dedicated interviewers, each responsible for visiting 12 PSUs per quarter (one per week), to make an efficient use of human resources, and to keep all interviewers working within zila boundaries, the number of PSUs per zila should be a multiple of 12.
This in turn implies that, in addition to the ten city corporations, the rural and urban portions of all 64 zilas should become de facto sampling strata.
where Deff is the design effect, due to stratification and clustering, c is the average number of relevant individuals per household, and ta is the normal deviate corresponding to the confidence level a.
The earlier LFS 2013 reported an unemployment rate of 4.2%, with a design effect of 1.77 and an average of 1.92 persons in the labor force per household. Using these figures as referential parameters, the number of households needed to estimate this indicator with a margin of error of 1% at the 95% confidence level is which implies that a total sample of around 29,400 households would be needed to achieve the required precision in all 21 estimation domains. Since these domains have very unequal populations – ranging from less than half a million in the smaller city corporations to nearly 30 million in rural Dhaka – the distribution of the sample into such domains should arbitrate between doing it equitably (which would deliver estimates of similar precision for all of them) and doing it proportionally (which would deliver nearly optimal estimates for Bangladesh as a whole). Consistently with the criterion used by the BBS for the 2013 QLFS, the new survey will do it on the basis of Kish’s allocation, which is generally considered the best compromise between these two extremes: the sample will be thus distributed in proportion to the factors ?1 ? / ?? ? 1 / ??, where H is the number of strata (24 in this case) and Nh (1 = h = H) are the number of households reported by the 2011 census in each domain.
In addition to the above theoretic considerations, the QLFS sampling design needs to account for two practical constrains imposed by fieldwork management:
* First, the total sample size obviously needs to be a multiple of the cluster size (24 households).
* Second, since the survey will be fielded by dedicated interviewers, each responsible for visiting 12 PSUs per quarter (one per week), to make an efficient use of human resources, and to keep all interviewers working within zila boundaries, the number of PSUs per zila should be a multiple of 12.
This in turn implies that, in addition to the ten city corporations, the rural and urban portions of all 64 zilas should become de facto sampling strata.
Weighting
With the sampling strategy described here, the probability phij of selecting household hij in PSU hi
of stratum h in any given quarter is given by7
is the number of PSUs selected in stratum h,
is the total number of households in PSU hi, as reported by the 2011 Census, is the total number of households in PSU hi, as reported by the QLFS household listing operation, an mhi is the number of households visited in PSU hi (normatively always 24).To obtain unbiased estimators from the sample, the data reported for the household should be affected by a sampling weight (or raising factor) whij, equal to the inverse of its selection probability (whij=1/phij). If nhi and n’hi were equal in all PSUs, the formula would simplify to a constant and the sample would be self-weighted within each stratum. In practice, nhi and n’hi will rarely by equal but often similar, so the sample will not be exactly self-weighted, but quite approximately so. As the quarterly survey started from July 2015, survey base weights were post-adjusted to estimate total population of July 2015 for the first quarter and and kept same for the successive three quarters of the QLFS 2015-16. Similarly, survey base weights were post-adjusted to estimate total population of July 2016 for the QLFS 2016-17. However, post-adjusted survey base weights to estimate total population of January 2017 for the QLFS 2016-17 is also available in the microdata.
of stratum h in any given quarter is given by7
is the number of PSUs selected in stratum h,
is the total number of households in PSU hi, as reported by the 2011 Census, is the total number of households in PSU hi, as reported by the QLFS household listing operation, an mhi is the number of households visited in PSU hi (normatively always 24).To obtain unbiased estimators from the sample, the data reported for the household should be affected by a sampling weight (or raising factor) whij, equal to the inverse of its selection probability (whij=1/phij). If nhi and n’hi were equal in all PSUs, the formula would simplify to a constant and the sample would be self-weighted within each stratum. In practice, nhi and n’hi will rarely by equal but often similar, so the sample will not be exactly self-weighted, but quite approximately so. As the quarterly survey started from July 2015, survey base weights were post-adjusted to estimate total population of July 2015 for the first quarter and and kept same for the successive three quarters of the QLFS 2015-16. Similarly, survey base weights were post-adjusted to estimate total population of July 2016 for the QLFS 2016-17. However, post-adjusted survey base weights to estimate total population of January 2017 for the QLFS 2016-17 is also available in the microdata.
Data Collection
Dates of Data Collection (YYYY/MM/DD)
| Start date | End date |
|---|---|
| 2016-07 | 2017-06 |
Mode of data collection
Face-to-face [f2f]
Type of Research Instrument
The Quarterly Labour Force Survey 2015-16 questionnaire comprised 14 sections, as follows:
Section 1. Household basic information
Section 2. Household roster (members’ basic information)
Section 3. General education (for persons aged 5 years or older) & vocational training (for persons aged 15 years or older)
Section 4. Working status (for persons aged 15 years or older)
Section 5. Main activities (for persons aged 15 years or older)
Section 6. Secondary activities (for employed persons aged 15 years or older)
Section 7. Occupational safety and health within the previous 12 months (for persons aged 15 years or older)
Section 8. Underemployment (for employed persons aged 15 years or older)
Section 9. Unemployment (for not employed persons aged 15 years or older)
Section 10. Own use production of goods (for persons aged 15 years or older)
Section 11. Own use provision of services (for persons aged 15 years or older)
Section 12. Unpaid trainee/apprentice work (for persons aged 15 years or older)
Section 13. Volunteer work (for persons aged 15 years or older)
Section 14. Migration (for persons aged 15 years or older)
Section 1. Household basic information
Section 2. Household roster (members’ basic information)
Section 3. General education (for persons aged 5 years or older) & vocational training (for persons aged 15 years or older)
Section 4. Working status (for persons aged 15 years or older)
Section 5. Main activities (for persons aged 15 years or older)
Section 6. Secondary activities (for employed persons aged 15 years or older)
Section 7. Occupational safety and health within the previous 12 months (for persons aged 15 years or older)
Section 8. Underemployment (for employed persons aged 15 years or older)
Section 9. Unemployment (for not employed persons aged 15 years or older)
Section 10. Own use production of goods (for persons aged 15 years or older)
Section 11. Own use provision of services (for persons aged 15 years or older)
Section 12. Unpaid trainee/apprentice work (for persons aged 15 years or older)
Section 13. Volunteer work (for persons aged 15 years or older)
Section 14. Migration (for persons aged 15 years or older)
Data Collectors
| Name | Abbreviation | Affiliation |
|---|---|---|
| Statistics and Informatics Division | SID | Ministry of Planning |
Data Processing
Cleaning Operations
Initial manual editing and coding of industry and occupation classification was done in the BBS headquarters by the selected editors and coders. The supervising officers further checked the questionnaires and validated the data randomly sampled edited questionnaires. Data was captured using Census and Survey Processing System (CSPro) through a data entry screen specially created and incorporated with checks to ensure accuracy during data entry. Erroneous entries and potential outliers were then verified and corrected appropriately. A total of 12 data entry personnel were engaged during the exercise. Weights were developed to account for the selection probabilities. The weights were developed using the design weights of the PSUs. The non-response adjustment and urban-rural calibration was also used. The captured data were exported to STATA format for cleaning and analysis. The cleaned data was weighted before final analysis.
Metadata production
Document ID
DDI-BGD-BBS-LFS-2016-17-v01
Producers
| Name | Abbreviation | Affiliation | Role |
|---|---|---|---|
| BANGLADESH BUREAU OF STATISTICS | BBS | Statistics and Informatics Division, Ministry of Planning | Documentation of the study |