+691 over the quarter
+1,677 over the year
SLPA evidence explainer · 9 September 2026
Unemployment at 10.1%.
What improved, who benefited, and how we know.
More people are reported to be working. That deserves recognition. Understanding the full improvement also means asking who is counted, what changed in the labour force, and whether better work is reaching the people who need it.
Officially reported unemployment
Q2 202513.4%↘Q2 202610.1%
−3.3 percentage pointsover one year
April–June 2026 · Government release citing the CSO Labour Force Survey
Read the result and its limits01 / The result
A welcome improvement. An incomplete explanation.
The 10.1% figure is an official announcement of a survey estimate. Its arithmetic is consistent with the reported employment gain. A complete independent check still needs the underlying Q2 tables and methodology.
From 12.1% in Q1 2026
and 13.4% in Q2 2025
From 24.1% a year earlier
−8.6 percentage points
Government / Office of the Prime Minister · Q2 announcement ↗. Employment changes count the net difference in people employed; they are not a count of new positions, gross hires, permanent contracts or full-time-equivalent jobs.
Established public record
Government reports more employment and lower unemployment.
Both are favourable movements in the published point estimates. “Down 3.3 percentage points” is the correct annual comparison; it is different from a 3.3% relative decrease.
Still to be established
The full distribution, precision and explanation.
We could not locate the full Q2 report on the CSO pages checked by 9 September. This limits verification; it does not show the headline is false. Sampling uncertainty and comparability need their own evidence.
The Briefing Room · 7 September 2026
Pierre on the unemployment result.
Prime Minister Philip J. Pierre discusses the unemployment rate. Watch the supplied recording, then follow the survey evidence, the labour-force arithmetic and the policy questions.
02 / Follow the numbers
More work explains part of the movement. The labour force matters too.
The unemployment rate divides unemployed people by the labour force: everyone employed plus everyone meeting the unemployment definition. It does not divide by the whole population.
Unemployment rate=
Unemployed peopleEmployed + unemployed people
× 100Using 99,528 employed and a rounded 10.1% rate implies a labour force of approximately 110,710 and approximately 11,182 unemployed people. Compared with the Q1 published counts, employment rose by 691 while the implied labour force fell by about 1,703. These are changes in estimated totals, not a record of individual transitions.
Q1 2026CSO published counts
112,413labour force
Q2 2026Unemployment reconstructed
≈110,710labour force
Employed peopleUnemployed peopleSame scale · 0–120,000 people
Employed+691net increase in people
Labour force≈−1,703implied net change
Unemployed≈−2,394implied net change
Holding the Q1 2026 labour force fixed, 99,528 employed would produce unemployment of 11.46%. The announced result is 10.1%. This shows why the denominator matters; it does not identify who left, why they left, or which policy caused a change.
Starting counts: CSO Q1 2026, Table 2 ↗ and CSO Q2 2025, Table 1 ↗. Q2 2026 employment and rate: Government announcement ↗. Q2 labour-force and unemployment counts are SLPA calculations.
Calculation, rounding and interpretation
Labour force = 99,528 ÷ (1 − 0.101) = 110,709.68. Unemployed = labour force − 99,528 = 11,181.68. Published estimates rounded to people would differ slightly. If 10.1% was rounded to the nearest tenth, its underlying rate is approximately 10.05%–10.15%, implying roughly 110,648–110,771 in the labour force. That range reflects rounding alone; it is not a confidence interval.
The net labour-force difference could reflect participation, migration, population estimation, sample variation, classification changes or a combination. The headline cannot identify these contributions. Nor can it reveal how many individuals moved directly from unemployment into work.
The announcement’s 89.9% “employment rate” is the complement of 10.1% within the labour force. It does not mean 89.9% of all adults are employed. A Q2 employment-to-population ratio needs a compatible working-age population denominator, which the release does not provide.
Download the labelled comparison data (CSV) · Read the calculation notes
Try the denominator yourself.
Each scenario starts from the same hypothetical population. Watch how employment and the unemployment rate can move together—or in different directions.
Illustration · 1,000 adults · each square represents 10 people
11.1%unemployment rate
Of 1,000 adults, 900 are in the labour force: 800 employed and 100 unemployed. The unemployment rate is 100 ÷ 900 = 11.1%.
- Employed
- 800
- Unemployed
- 100
- Outside labour force
- 100
03 / How we know
The CSO measures. The survey classifies. Government communicates.
A household survey estimates the labour market, including work that payroll and benefit registers miss. Its credibility comes from a transparent method and reproducible checks.
Work for pay or profit.
The Q1 report includes as little as one hour in the reference week, alongside qualifying temporary absence. Informal or self-employed work can count. The threshold measures participation in employment; it does not establish adequate income or decent working conditions.
Without employment, searching and available.
The Q1 definitions use a four-week job-search window and availability within two weeks, with a specified future-job-start exception. Being without a job alone does not automatically place someone in this category.
Neither employed nor unemployed under the definitions.
This can include students, retirees, carers and people who want work but do not satisfy the search or availability criteria. They have different circumstances and should not be treated as one policy problem.
CSO Q1 definitions ↗ · UN/ILO definitions and interpretation ↗. Q2 operational definitions still require confirmation.
- Select a sample
CSO’s published historical method uses geographically stratified household sampling. A survey interviews a sample and estimates the population; it is not a count of everybody.
- Ask about actual activity
Interviewers collect work, search and availability answers for defined reference periods. Questions, translations and interview practice affect classification.
- Check and weight
Records are edited, coded and weighted to represent the population. Response patterns and population controls matter as much as the raw sample size.
- Estimate uncertainty
Survey design, clustering, weights and any overlap between quarterly samples should inform standard errors and uncertainty around changes.
- Release and reconcile
Publish tables, methods, revisions and dates together. The OPM communicates results; the CSO supplies statistical accountability. ILO standards and assistance are not independent certification of every quarterly number.
Process context: CSO 2016 survey documentation ↗ and UN/ILO statistical metadata ↗. Historical sample sizes and field arrangements are not presented as the 2026 design.
Three checks before calling the comparison fully verified.
1. Confirm when the updated survey began
Government announced labour-survey modernisation in April 2026, with technical training beginning on 13 April. The timing makes comparability an immediate question, but it does not establish that a new method was used for Q2.
Ask for the first affected reference quarter, old and new questionnaires, classification and weighting changes, and a bridge or backcast where needed. The announcement’s reference to the 19th and 21st international standards alone cannot tell us how the actual measured series changed.
Government / NCPC · Survey modernisation announcement ↗2. Reconcile the youth comparison and precision table
The September announcement gives Q1 youth unemployment as 18.5%; the Q1 CSO report gives 18.4%. The annual comparison, 24.1% to 15.5%, is unaffected by this Q1 discrepancy. The quarterly difference needs a revision or correction note.
On PDF page 6 of the Q1 report, Table 1’s precision figures describe an 18.20% estimate and a 16.80%–19.60% confidence interval. Table 2 on that page gives national unemployment of 12.1%. Those precision figures cannot establish uncertainty around the national result without clarification.
These are specific publication-quality questions. They do not establish manipulation or invalidate every estimate in the report.
CSO · Labour Force Report, Q1 2026 ↗3. Test the change, not just the headline
Q1 reports 1,305 completed responses against an expected 1,456, or 89.6%. A response rate alone does not establish representativeness. Q2 needs its own sample, response, weighting and uncertainty information.
“Significant improvement” in a press release need not mean statistically significant. Testing the change requires the variance of the difference, including covariance where quarterly samples overlap. Comparing two individual confidence intervals is not a substitute.
Seasonality also matters: Q2-to-Q2 is a more useful seasonal comparison than adjacent quarters, but it does not remove changes in methods, composition or economic conditions.
04 / Who benefited?
Youth show a lower rate. The full map of gains is still missing.
The available release gives a direction for youth unemployment and selected industries. It does not show which households, women and men, districts or income groups gained secure work.
National · Q2 2025
13.4%
National · Q2 2026
10.1%
Youth · Q2 2025
24.1%
Youth · Q2 2026
15.5%
Youth are defined as ages 15–29 in the CSO Q1 report. Confirm the same definition in Q2 before joining series. A 15.5% youth unemployment rate is not the share of all young people without work.
Government’s annual comparison ↗ · CSO youth definition ↗Industry signals
The release reports construction employment up 14.0% and administrative and support services up 35.1%, without an explicit comparison period for those changes. Get the base headcounts and comparable tables before ranking contributions. A large percentage gain can come from a small base.
Women and place
Q1 unemployment was 16.2% for women and 8.0% for men. That is dated context, not proof of who benefited in Q2. Publish sex and district results with sample precision before making claims about inclusion or using small-area rates to allocate funding.
Quality and access
Ask about hours wanted and worked, real earnings, informal work, job tenure and transport or care constraints. For youth, add the share not in employment, education or training (NEET), alongside placements and retention.
Industry headlines ↗ · Q1 sex comparison ↗. The measures proposed here are SLPA recommendations, not additional Q2 findings.
05 / Policy implications
Recognise the progress. Test the explanation. Make the gains last.
Investment and stronger demand can support employment. The unemployment rate alone cannot tell us how much a named project, tax concession or training programme contributed.
Demand, investment and recovery
Construction, tourism activity and business expansion can raise hiring. Workers and firms also respond to wages, skills, credit and infrastructure. Check sector output, vacancies, hours and payroll evidence against the employment estimates.
Participation, composition and measurement
Changes in who searches, who migrates, who is available and how survey answers are classified can change the rate. The reconstructed labour-force decline is a reason to investigate these channels; it is not evidence that discouraged workers explain it.
What would have happened otherwise?
Match implementation dates and exposure to outcomes. Compare credible untreated groups or staggered rollouts where feasible; disclose assumptions and pre-existing trends. Project payrolls establish direct employment, while a net national impact also needs displacement, duration and spillovers assessed.
The Government release links improvement to public and private investment and names construction projects. That is a plausible account to test. It is not a measured causal contribution. A sound evaluation would distinguish resident workers from positions, job-years from permanent jobs, and gross project hiring from net additional employment.
| Action and responsible bodies | Publish or measure | Decision it improves |
|---|---|---|
| CSO · complete the release | Q2 tables and reusable data; definitions; response and weights; uncertainty; revisions; first quarter of any method change. | Whether the size and comparability of the improvement are well supported. |
| CSO and Labour · widen the dashboard | Employment-to-population ratio, participation, hours, earnings, underemployment and potential labour force, split by sex and age. | Whether more people have adequate work, including those missed by the headline. |
| Training bodies and employers · follow entrants | Completion, paid placement, 3/6/12-month retention and real earnings; barriers involving childcare, transport and disability access. | Which pathways deliver durable first incomes and who remains excluded. |
| Finance and implementing agencies · evaluate programmes | Costs, timelines, direct resident employment, job duration and a credible comparison for net additional outcomes. | Which investments and incentives merit continuation, redesign or expansion. |
Aggregate National Insurance Corporation contributor trends, employer payrolls, vacancies and tax records could provide useful cross-checks, subject to coverage and privacy. Formal contributors do not cover the whole labour market, so these sources should complement the household survey rather than be treated as an exact substitute.
RIPPLE-4Follow the household consequences.
Track whether work produces steadier income, care and education continuity, local spending and stronger future capability.
Sovereign Option TheoryAsk which choices become possible.Reliable work, transferable skills and savings can widen a person’s options and improve resilience when one sector slows.
PITONS StandardCommit to an outcome and a review.Set the target, evidence, owner and review trigger before judging a programme by its headline achievements.
06 / Open evidence
Check the record. Reuse the calculations.
Reviewed 9 September 2026. Official estimates, SLPA accounting and policy proposals are identified separately. This page should be updated when the CSO’s complete Q2 tables, revision explanation or methods bridge become available.
Government / Office of the Prime Minister · Q2 announcement ↗7 September 2026
Source of the 10.1% estimate, 99,528 employed, youth and industry headlines. A communication of CSO results, not the full survey report.
CSO · Labour Force Report, Q1 2026 ↗2026 · Tables 1–2 and definitions
Published employment, unemployment and labour-force counts. The precision table and some narrative figures conflict with the main national table; see the evidence audit.
CSO · Statistical Snapshot, Q2 2025 ↗2025 reference quarter · Table 1
Year-earlier employment 97,851; labour force 113,041; unemployment 15,190; national rate 13.4%; youth rate 24.1%.
Government / NCPC · Survey modernisation announcement ↗20 April 2026
Announces technical training and updated work statistics. Does not identify the first quarter using a revised questionnaire or publish a comparable bridge series.
UN / ILO · SDG 8.5.2 unemployment metadata ↗Updated 27 March 2026
International definitions, data sources, quality assurance and interpretation limits. National survey estimates must be distinguished from harmonised or modelled international series.
CSO · Labour Force Survey documentation ↗2016 methodology · historical context
Documents household sampling, interviews, coding, editing and weighting. The 2016 sample design and response rate are not assumed to describe Q2 2026.
CSO · Publications catalogue ↗Checked 9 September 2026
Q1 2026 was the latest full Labour Force Survey report located in this catalogue at review. Absence here does not establish that no other release exists.
Video: recording supplied to SLPA, identified with the submission as Prime Minister Philip J. Pierre on The Briefing Room, 7 September 2026. It records the speaker’s account; it is not a substitute for the statistical tables.