National Labor Force Survey 2021
Uganda
Get MicrodataCreated on
Feb 19, 2026
Last modified
Feb 19, 2026
Page views
6952
Downloads
50
Identification
Survey ID Number
DDI-UGA-UBOS-NLFS-2021-v01
Title
National Labor Force Survey 2021
Country
| Name | Country code |
|---|---|
| Uganda | UGA |
Abstract
he National Labour Force Survey (NLFS) 2021 was conducted by the Uganda Bureau of Statistics (UBOS) as the third in a series of stand-alone labour force surveys, following editions in 2011/12 and 2016/17. Its primary purpose was to generate updated labour market indicators to support evidence-based policymaking and monitor employment trends across Uganda. The survey covered a nationally representative random sample of 11,000 households across 15 sub-regions, using a multi-stage cluster sampling design aligned with the resolutions of the 20th International Conference of Labour Statisticians (ICLS). It achieved a 95% response rate.
Uganda's total population was estimated at approximately 43 million, with a working-age population (14–64 years) of 23.5 million. Of these, 20.5 million were working, predominantly in agriculture, forestry, and fishing. The Labour Force Participation Rate stood at 48% (58% male, 39% female), and the Employment-to-Population Ratio was 43%. The unemployment rate was 12%, with long-term unemployment accounting for 47% of all unemployed persons. The composite measure of labour underutilisation reached 42% (6.6 million people).
Informality was pervasive — nearly 88% of employed persons outside agriculture were in informal employment. The median monthly cash earnings for paid workers were UGX 200,000. Youth (18–30 years) faced particular challenges, with a 17% unemployment rate and 41% classified as NEET (Not in Education, Employment or Training). Child labour was a significant concern, with 40% of children aged 5–17 years engaged in child labour (excluding household chores). The survey also covered unpaid care work, subsistence agriculture, older persons' activities, and youth transitions to employment, providing a comprehensive picture of Uganda's labour market landscape.
Uganda's total population was estimated at approximately 43 million, with a working-age population (14–64 years) of 23.5 million. Of these, 20.5 million were working, predominantly in agriculture, forestry, and fishing. The Labour Force Participation Rate stood at 48% (58% male, 39% female), and the Employment-to-Population Ratio was 43%. The unemployment rate was 12%, with long-term unemployment accounting for 47% of all unemployed persons. The composite measure of labour underutilisation reached 42% (6.6 million people).
Informality was pervasive — nearly 88% of employed persons outside agriculture were in informal employment. The median monthly cash earnings for paid workers were UGX 200,000. Youth (18–30 years) faced particular challenges, with a 17% unemployment rate and 41% classified as NEET (Not in Education, Employment or Training). Child labour was a significant concern, with 40% of children aged 5–17 years engaged in child labour (excluding household chores). The survey also covered unpaid care work, subsistence agriculture, older persons' activities, and youth transitions to employment, providing a comprehensive picture of Uganda's labour market landscape.
Unit of Analysis
Primary Units of Analysis
Individuals/Persons — This was the main unit of analysis for labour market indicators. Specifically, the survey focused on persons aged 14–64 years (Uganda's working age population) for employment-related measures, though some modules extended to different age groups.
Other Units of Analysis by Age Group
Children aged 5–17 years — used for analysis of child labour and children's activities
Youth aged 18–30 years — used for youth employment, NEET status, and transition to employment analysis
Older persons (65+ years) — used for analysis of older persons' activities and labour participation
Persons aged 10 years and above — used for education attainment, marital status, and access to financial services
Persons aged 5 years and above — used for unpaid work analysis
Secondary Unit of Analysis
Households — used for socio-demographic background characteristics such as household size, headship, main source of income, and perceived financial situation. The sample itself was drawn from 11,000 households across the country.
Individuals/Persons — This was the main unit of analysis for labour market indicators. Specifically, the survey focused on persons aged 14–64 years (Uganda's working age population) for employment-related measures, though some modules extended to different age groups.
Other Units of Analysis by Age Group
Children aged 5–17 years — used for analysis of child labour and children's activities
Youth aged 18–30 years — used for youth employment, NEET status, and transition to employment analysis
Older persons (65+ years) — used for analysis of older persons' activities and labour participation
Persons aged 10 years and above — used for education attainment, marital status, and access to financial services
Persons aged 5 years and above — used for unpaid work analysis
Secondary Unit of Analysis
Households — used for socio-demographic background characteristics such as household size, headship, main source of income, and perceived financial situation. The sample itself was drawn from 11,000 households across the country.
Coverage
Geographic Coverage
The NLFS 2021 had the following geographic coverage:
National Scope
The survey covered the entire country of Uganda, designed to be nationally representative of the population living in Uganda at the time.
Sub-national Breakdown
The country was divided into 15 sub-regions (strata), which served as the primary domains for generating estimates. Results were also disaggregated at 4 regional levels and for urban and rural areas separately. The major regional cities were also specifically catered for in the sample design.
Sampling Frame
The enumeration areas (EAs) used in the survey were drawn from the 2014 National Population and Housing Census frame. In the first stage of the multi-stage cluster sampling design, EAs were selected across all sub-regions; in the second stage, ten households were selected within each EA.
National Scope
The survey covered the entire country of Uganda, designed to be nationally representative of the population living in Uganda at the time.
Sub-national Breakdown
The country was divided into 15 sub-regions (strata), which served as the primary domains for generating estimates. Results were also disaggregated at 4 regional levels and for urban and rural areas separately. The major regional cities were also specifically catered for in the sample design.
Sampling Frame
The enumeration areas (EAs) used in the survey were drawn from the 2014 National Population and Housing Census frame. In the first stage of the multi-stage cluster sampling design, EAs were selected across all sub-regions; in the second stage, ten households were selected within each EA.
Universe
The universe of the NLFS 2021 was the civilian, non-institutional household population of Uganda residing within the country's 135 districts and 11 cities as of July 2020. The survey explicitly excluded persons living in institutions (such as military barracks, prisons, hospitals, and boarding schools) and other special areas.
Geographic Boundary
The population was restricted to persons residing within the sovereign territory of Uganda, covering both rural and urban areas across all 15 statistical sub-regions.
Nationality
While the survey covered all household residents regardless of nationality, the population was overwhelmingly Ugandan - 99.4% were Ugandan nationals, with less than 1% being non-Ugandans (such as refugees and migrants).
Age Boundaries
The universe was defined differently depending on the module of analysis:
5 years and above - the broadest eligible age group for individual interviews; used for unpaid work, schooling status, ICT access, and children's activities
10 years and above - used for education attainment, marital status, financial inclusion, and access to information
14-64 years - Uganda's defined working age population, the primary universe for all core labour market indicators (employment, unemployment, labour underutilisation)
15 years and above - used for internationally comparable labour market indicators aligned with ILO standards
18-30 years - Uganda's defined youth population, used for youth employment and NEET analysis
5-17 years - used for child labour analysis
65 years and above - used for older persons' labour force participation analysis
Sex
Both male and female persons were included, with results disaggregated by sex throughout the report.
Residence
Both rural and urban residents were included, with residence being a key dimension of disaggregation.
Geographic Boundary
The population was restricted to persons residing within the sovereign territory of Uganda, covering both rural and urban areas across all 15 statistical sub-regions.
Nationality
While the survey covered all household residents regardless of nationality, the population was overwhelmingly Ugandan - 99.4% were Ugandan nationals, with less than 1% being non-Ugandans (such as refugees and migrants).
Age Boundaries
The universe was defined differently depending on the module of analysis:
5 years and above - the broadest eligible age group for individual interviews; used for unpaid work, schooling status, ICT access, and children's activities
10 years and above - used for education attainment, marital status, financial inclusion, and access to information
14-64 years - Uganda's defined working age population, the primary universe for all core labour market indicators (employment, unemployment, labour underutilisation)
15 years and above - used for internationally comparable labour market indicators aligned with ILO standards
18-30 years - Uganda's defined youth population, used for youth employment and NEET analysis
5-17 years - used for child labour analysis
65 years and above - used for older persons' labour force participation analysis
Sex
Both male and female persons were included, with results disaggregated by sex throughout the report.
Residence
Both rural and urban residents were included, with residence being a key dimension of disaggregation.
Producers and sponsors
Primary investigators
| Name | Affiliation |
|---|---|
| Uganda Bureau of Statistics | Ministry of Finance Planning and Economic Developement |
Producers
| Name | Affiliation | Role |
|---|---|---|
| Uganda Bureau of Statistics | MoFPED | Funding the survey |
| International Labour Organisation |
Sampling
Sampling Procedure
Type of Sample
The NLFS 2021 used a probability-based random sample, meaning every household in Uganda had a known, non-zero chance of being selected. The sample was designed to be nationally representative of the civilian, non-institutional household population of Uganda.
Sample Design The survey followed a stratified two-stage cluster sampling design, structured as follows:
Stage 1 — Stratification (Explicit and Implicit) The country was first divided into 15 strata (sub-regions), referred to as explicit stratification. These sub-regions served as the primary domains for generating estimates. Within each explicit stratum, the Primary Sampling Units (PSUs) were further sorted geographically, which constituted implicit stratification. The 15 sub-regions used were:
Kampala
Buganda South
Buganda North
Busoga
Bukedi
Elgon
Teso
Lango
Acholi
Karamoja
West Nile
Bunyoro
Tooro
Ankole
Kigezi
Each sub-region was further disaggregated into rural and urban domains, and regional cities were specifically catered for in the design.
Stage 2 — First Stage Selection (PSU/EA Level) Within each stratum, Enumeration Areas (EAs) were selected as the Primary Sampling Units (PSUs). A total of 1,100 EAs were targeted across the country, allocated across the 15 sub-regions as follows:
Sub-region Rural EAs Urban EAs Total EAs
Acholi 43 19 62
Ankole 54 36 90
Bukedi 50 10 60
Bunyoro 51 19 70
Busoga 67 24 91
Elgon 59 14 73
Kampala — 60 60
Karamoja 43 7 50
Kigezi 39 11 50
Lango 53 7 60
North Buganda 62 33 95
South Buganda 48 56 104
Teso 56 11 67
Tooro 61 19 80
West Nile 70 18 88
National 756 344 1,100
The EAs were drawn from a sampling frame based on the 2014 National Population and Housing Census (NPHC), updated through a listing exercise conducted prior to the main survey to capture new households formed after the census.
Stage 3 — Second Stage Selection (Household Level) Within each selected EA, 10 households were systematically selected, giving a total target sample of 11,000 households nationally (1,100 EAs × 10 households per EA).
Individual Respondents Within each selected household, all household members aged 5 years and above were eligible for individual interview, making the individual selection essentially a complete enumeration within selected households rather than a sub-sample.
Target Sample Size
EAs targeted: 1,100
Households targeted: 11,000
Households successfully completed: 8,668 (95% response rate)
Individual interviews completed: 39,522 (99% response rate)
Sampling Fraction and Precision The sample size was determined by balancing several factors including:
The accuracy required for estimates at each domain level
Available resources and operational constraints
The need to control both sampling error (measurable through variance estimation) and non-sampling error (from response errors, coding errors, data entry errors, etc.)
The sample was specifically designed to produce reliable estimates at four levels of disaggregation:
National level
Urban/rural residence
15 sub-regions
Regional cities
Limitation A noted limitation of the design was that the sample size was insufficient to generate district-level estimates — disaggregation was only reliable up to the 15 sub-regional groupings. Additionally, the 2014 NPHC sampling frame was considered inadequate to fully capture statistics on labour migration (inflows and outflows).
In summary, the NLFS 2021 employed a stratified two-stage cluster sampling design with 15 explicit strata, 1,100 EAs as PSUs, and 10 households per EA, targeting a total sample of 11,000 households — large enough to generate reliable estimates down to the sub-regional and city level across Uganda.
The NLFS 2021 used a probability-based random sample, meaning every household in Uganda had a known, non-zero chance of being selected. The sample was designed to be nationally representative of the civilian, non-institutional household population of Uganda.
Sample Design The survey followed a stratified two-stage cluster sampling design, structured as follows:
Stage 1 — Stratification (Explicit and Implicit) The country was first divided into 15 strata (sub-regions), referred to as explicit stratification. These sub-regions served as the primary domains for generating estimates. Within each explicit stratum, the Primary Sampling Units (PSUs) were further sorted geographically, which constituted implicit stratification. The 15 sub-regions used were:
Kampala
Buganda South
Buganda North
Busoga
Bukedi
Elgon
Teso
Lango
Acholi
Karamoja
West Nile
Bunyoro
Tooro
Ankole
Kigezi
Each sub-region was further disaggregated into rural and urban domains, and regional cities were specifically catered for in the design.
Stage 2 — First Stage Selection (PSU/EA Level) Within each stratum, Enumeration Areas (EAs) were selected as the Primary Sampling Units (PSUs). A total of 1,100 EAs were targeted across the country, allocated across the 15 sub-regions as follows:
Sub-region Rural EAs Urban EAs Total EAs
Acholi 43 19 62
Ankole 54 36 90
Bukedi 50 10 60
Bunyoro 51 19 70
Busoga 67 24 91
Elgon 59 14 73
Kampala — 60 60
Karamoja 43 7 50
Kigezi 39 11 50
Lango 53 7 60
North Buganda 62 33 95
South Buganda 48 56 104
Teso 56 11 67
Tooro 61 19 80
West Nile 70 18 88
National 756 344 1,100
The EAs were drawn from a sampling frame based on the 2014 National Population and Housing Census (NPHC), updated through a listing exercise conducted prior to the main survey to capture new households formed after the census.
Stage 3 — Second Stage Selection (Household Level) Within each selected EA, 10 households were systematically selected, giving a total target sample of 11,000 households nationally (1,100 EAs × 10 households per EA).
Individual Respondents Within each selected household, all household members aged 5 years and above were eligible for individual interview, making the individual selection essentially a complete enumeration within selected households rather than a sub-sample.
Target Sample Size
EAs targeted: 1,100
Households targeted: 11,000
Households successfully completed: 8,668 (95% response rate)
Individual interviews completed: 39,522 (99% response rate)
Sampling Fraction and Precision The sample size was determined by balancing several factors including:
The accuracy required for estimates at each domain level
Available resources and operational constraints
The need to control both sampling error (measurable through variance estimation) and non-sampling error (from response errors, coding errors, data entry errors, etc.)
The sample was specifically designed to produce reliable estimates at four levels of disaggregation:
National level
Urban/rural residence
15 sub-regions
Regional cities
Limitation A noted limitation of the design was that the sample size was insufficient to generate district-level estimates — disaggregation was only reliable up to the 15 sub-regional groupings. Additionally, the 2014 NPHC sampling frame was considered inadequate to fully capture statistics on labour migration (inflows and outflows).
In summary, the NLFS 2021 employed a stratified two-stage cluster sampling design with 15 explicit strata, 1,100 EAs as PSUs, and 10 households per EA, targeting a total sample of 11,000 households — large enough to generate reliable estimates down to the sub-regional and city level across Uganda.
Weighting
The NLFS 2021 applied sampling weights to all statistical estimates to account for the complex stratified cluster sampling design used. The report makes consistent reference to weighted and unweighted figures throughout, confirming that weights were applied to produce population-representative estimates.
Why Weights Were Necessary
Weights were necessary for three main reasons arising from the survey design:
Unequal probability of selection — the stratified design with 15 sub-regions and differential EA allocations (e.g., Kampala had 60 EAs versus Karamoja's 50) meant different households had different probabilities of being selected
Non-response adjustment — the household response rate of 95% and individual response rate of 99% meant some sampled units did not respond, requiring adjustment to avoid bias
Population representation — to enable the sample of 8,668 completed households and 39,522 individual interviews to produce estimates that correctly represent Uganda's total population of approximately 43 million people
Evidence of Weighting Applied
The report explicitly distinguishes between weighted and unweighted figures. The response rate tables (Tables 1.2 and 1.3) are explicitly labelled (un-weighted), while all substantive statistical tables throughout the report present weighted population estimates. For example, the survey's weighted estimate of the national population was 42,885,900 compared to the unweighted sample of 47,690 individuals — confirming the application of expansion/inflation factors to project sample figures to population-level estimates.
Precision of Weighted Estimates
The survey errors appendix also reports Standard Errors (SE), Relative Errors (Coefficient of Variation/CV), and 95% Confidence Intervals for all key weighted estimates, for example:
National population: 42,885,900 ± 2,279,700 (CV = 5.3%; 95% CI: 38.4M–47.4M)
Working age population: 23,493,896 ± 1,239,120 (CV = 5.3%; 95% CI: 21.1M–25.9M)
Unemployment rate: 0.119 ± 0.005 (CV = 3.8%; 95% CI: 11.1%–12.8%)
LFPR: 0.483 ± 0.009 (CV = 1.9%; 95% CI: 46.5%–50.2%)
Limitation on Weighting Formula
While the report confirms that weights were applied and reports their effect through design effects and standard errors, it does not explicitly publish the full weighting formula or the step-by-step calculation of the sampling weights in the main report. The specific mathematical formula for the weight coefficient — including the base weight (inverse of selection probability), non-response adjustment factor, and any post-stratification/calibration factors — is not detailed in this main report. Such technical details would typically be contained in a separate technical report or methodology annex not included in this main report document.
In summary, the NLFS 2021 used probability-based sampling weights to produce population-representative estimates, necessitated by the stratified cluster design and differential sampling fractions across sub-regions. The weights effectively inflated the sample of ~39,500 interviewed individuals to represent Uganda's population of ~43 million. Design effects ranging from approximately 1.0 to 21.0 across different indicators confirm the significant role of weighting and design-based estimation in producing accurate, unbiased national and sub-national statistics.
Why Weights Were Necessary
Weights were necessary for three main reasons arising from the survey design:
Unequal probability of selection — the stratified design with 15 sub-regions and differential EA allocations (e.g., Kampala had 60 EAs versus Karamoja's 50) meant different households had different probabilities of being selected
Non-response adjustment — the household response rate of 95% and individual response rate of 99% meant some sampled units did not respond, requiring adjustment to avoid bias
Population representation — to enable the sample of 8,668 completed households and 39,522 individual interviews to produce estimates that correctly represent Uganda's total population of approximately 43 million people
Evidence of Weighting Applied
The report explicitly distinguishes between weighted and unweighted figures. The response rate tables (Tables 1.2 and 1.3) are explicitly labelled (un-weighted), while all substantive statistical tables throughout the report present weighted population estimates. For example, the survey's weighted estimate of the national population was 42,885,900 compared to the unweighted sample of 47,690 individuals — confirming the application of expansion/inflation factors to project sample figures to population-level estimates.
Precision of Weighted Estimates
The survey errors appendix also reports Standard Errors (SE), Relative Errors (Coefficient of Variation/CV), and 95% Confidence Intervals for all key weighted estimates, for example:
National population: 42,885,900 ± 2,279,700 (CV = 5.3%; 95% CI: 38.4M–47.4M)
Working age population: 23,493,896 ± 1,239,120 (CV = 5.3%; 95% CI: 21.1M–25.9M)
Unemployment rate: 0.119 ± 0.005 (CV = 3.8%; 95% CI: 11.1%–12.8%)
LFPR: 0.483 ± 0.009 (CV = 1.9%; 95% CI: 46.5%–50.2%)
Limitation on Weighting Formula
While the report confirms that weights were applied and reports their effect through design effects and standard errors, it does not explicitly publish the full weighting formula or the step-by-step calculation of the sampling weights in the main report. The specific mathematical formula for the weight coefficient — including the base weight (inverse of selection probability), non-response adjustment factor, and any post-stratification/calibration factors — is not detailed in this main report. Such technical details would typically be contained in a separate technical report or methodology annex not included in this main report document.
In summary, the NLFS 2021 used probability-based sampling weights to produce population-representative estimates, necessitated by the stratified cluster design and differential sampling fractions across sub-regions. The weights effectively inflated the sample of ~39,500 interviewed individuals to represent Uganda's population of ~43 million. Design effects ranging from approximately 1.0 to 21.0 across different indicators confirm the significant role of weighting and design-based estimation in producing accurate, unbiased national and sub-national statistics.
Data Collection
Data Collection Mode
Mode of Data Collection
The primary method of data collection was face-to-face interviews conducted by trained field enumerators visiting sampled households in person. The interviews were administered using Computer Assisted Personal Interviewing (CAPI) technology, replacing the traditional paper-based data collection approach.
The primary method of data collection was face-to-face interviews conducted by trained field enumerators visiting sampled households in person. The interviews were administered using Computer Assisted Personal Interviewing (CAPI) technology, replacing the traditional paper-based data collection approach.
Metadata production
DDI Document ID
DDI-UGA-UBOS-NLFS-2021-v01