# Frontier Signal preview: data and methods

Snapshot checked: **17 November 2026**. This is a static editorial preview, not a live feed. Dates on figures are observation years; publication and retrieval dates are recorded separately. No displayed value is filled from memory.

## Workforce: one comparable release

Country and regional workforce figures now use **one World Bank World Development Indicators release, updated 11 November 2026**, retrieved through the public API on 17 November 2026. The previous separately published regional table has been replaced by an explicit sum of the country data. The two are similar but do not have identical rounded values.

Coverage: **11 Southeast Asian economies**, including Timor-Leste. Ages 15 and over, both sexes. Labour force means employed people plus unemployed people; people outside the labour force are excluded. Modelled estimates are designed for comparison and may differ from national headline survey rates. Low unemployment does not establish job quality, job security or formal employment.

Primary indicators:

- [Labour force, total — SL.TLF.TOTL.IN](https://data.worldbank.org/indicator/SL.TLF.TOTL.IN): ILO/UN/World Bank estimates.
- [Unemployment, total — SL.UEM.TOTL.ZS](https://data.worldbank.org/indicator/SL.UEM.TOTL.ZS): ILO modelled estimate, percent of total labour force.

The exact API inputs and response metadata are stored in `data/country-workforce.json`. All 11 annual observations from 2015 through 2025 are retained for each country. The chart displays **2015, 2020, 2024 and 2025**, positioned at their actual year intervals. Connecting lines join selected observations and are not additional measured values.

For each country and year:

- `unemployed = labour force × unemployment rate / 100`
- `employed = labour force − unemployed`

Derived counts are approximate and do not constitute an independent employment survey. Calculations retain exact source rates; rounding occurs only in display. The regional labour force, employed and unemployed values are sums across the same 11 countries and year. Regional unemployment is `sum(unemployed) / sum(labour force) × 100`, not a simple average of country rates.

| Year | Labour force (millions) | Derived employment (millions) | Derived unemployment (millions) | Unemployment rate |
| --- | ---: | ---: | ---: | ---: |
| 2015 | 319.6 | 310.4 | 9.2 | 2.9% |
| 2020 | 332.9 | 323.0 | 9.9 | 3.0% |
| 2024 | 353.7 | 345.0 | 8.7 | 2.5% |
| 2025 | 357.5 | 348.8 | 8.7 | 2.4% |

The 2015–2025 change is **+11.9%** in labour force (**38.0 million** people) and **+12.4%** in derived employment. Changes are computed within this release only. Independently rounded displayed values may not reproduce the exact calculation.

## Employment by sector: matched 2025 observations

All country sector figures are **2025 ILO modelled estimates** from the same World Bank release as the workforce data. This replaces the earlier 2023 UNData snapshot. The source response metadata and exact country values are stored in `data/country-sectors.json`.

- [Agriculture — SL.AGR.EMPL.ZS](https://data.worldbank.org/indicator/SL.AGR.EMPL.ZS)
- [Industry — SL.IND.EMPL.ZS](https://data.worldbank.org/indicator/SL.IND.EMPL.ZS)
- [Services — SL.SRV.EMPL.ZS](https://data.worldbank.org/indicator/SL.SRV.EMPL.ZS)

All three shares are retrieved directly. **Industry is not an inferred remainder.** In particular, the API independently confirms Lao PDR's low 7.4% modelled industry share, which the supplied research had left unverified.

Regional shares use same-year derived employment as weights: `sum(country share × country employment) / sum(country employment)`. They are not simple averages. The denominator is employed people, not companies or the total population.

| Country | Services | Agriculture | Industry |
| --- | ---: | ---: | ---: |
| Southeast Asia (weighted) | 49.5% | 26.6% | 23.9% |
| Brunei | 78.4% | 1.2% | 20.4% |
| Cambodia | 37.7% | 33.4% | 28.9% |
| Indonesia | 50.2% | 27.3% | 22.5% |
| Laos | 24.0% | 68.6% | 7.4% |
| Malaysia | 64.9% | 9.2% | 26.0% |
| Myanmar | 36.9% | 45.0% | 18.2% |
| Philippines | 59.5% | 20.5% | 20.0% |
| Singapore | 85.9% | 0.1% | 14.0% |
| Thailand | 49.2% | 28.6% | 22.2% |
| Timor-Leste | 50.2% | 46.0% | 3.8% |
| Vietnam | 40.1% | 25.0% | 34.8% |

The UI retains published precision for calculations and rounds display labels to one decimal. Any normalization is limited to donut geometry for negligible floating-point differences. Positive small sectors remain visible. These modelled estimates should not be merged into ranked comparisons with national survey figures from different dates or definitions.

## Small businesses: national snapshots, no regional total

The preview no longer presents the prior “70M+ / 85% / 44.8%” regional business claims as a comparable Southeast Asia snapshot. Those historical ASEAN statements did not establish one common reference year, national definition or 11-country scope. **Regional business count and contribution values are null.** The regional view explains why and directs readers to a country.

National counting units, observation years and employment/output denominators remain distinct. Missing values mean no comparable value was verified for this preview, not zero and not proof that a statistic has never been published. Never divide household-survey workforce totals by enterprise counts, sum these counts across countries, or apply enterprise employment shares to regional workforce totals.

`data/country-enterprises.json` contains each metric's year, original sources and scope. Existing country count snapshots are retained; the Malaysia contribution measures were updated from the primary DOSM release. Country figures are not harmonised or presented as a ranked comparison.

| Country | Count | Count year | Counting unit / coverage | Employment share (year) | Output share (year) |
| --- | ---: | --- | --- | --- | --- |
| Brunei | 6,134 | 2022 | MSME enterprises | 52.3% (2022) | 5.9% (2022) |
| Cambodia | 512,870 | 2014 | MSME establishments | 71.8% (2014) | — |
| Indonesia | 56,142,687 | 2024 | MSME business units | — | 59.64% (2024) |
| Laos | 133,721 | 2020 | MSME enterprises | — | — |
| Malaysia | 1,086,386 | 2024 | MSME establishments | 48.7% (2025) | 39.7% (2025) |
| Myanmar | 46,014 | 2023 | registered small & medium enterprises | — | — |
| Philippines | 1,236,908 | 2024 | MSME establishments | 66.6% (2024) | — |
| Singapore | 354,600 | 2024 | small & medium enterprises | 69.6% (2024) | 47.3% (2024) |
| Thailand | 3,255,957 | 2024 | MSME enterprises | 68.8% (2024) | 34.9% (2024) |
| Timor-Leste | — | — | micro, small & medium businesses | — | — |
| Vietnam | 757,788 | 2023 | MSME enterprises | 37.1% (2023) | — |

Malaysia's **2025** MSME employment share is **48.7%** of national employment and GDP share is **39.7%**, per [DOSM, MSME Performance 2025](https://www.dosm.gov.my/portal-main/release-content/micro-small--medium-enterprises-msmes-performance-2025), published 30 July 2026. Its count remains a separately dated **2024 establishment** count. The 2025 release's employment count (8.09 million) is not a business count.

Other employment shares generally cover workers in enterprises/establishments. Singapore's output share is **enterprise value added**, not GDP. Cambodia's 2014 observation is explicitly historical. Myanmar covers registered SMEs and is approximate. Vietnam's count excludes household/informal businesses. Indonesia is a survey-based business-unit estimate, not a register of incorporated companies.

Country sources:

- [ADB Asia SME Monitor 2025](https://www.adb.org/publications/asia-sme-monitor-2025), [full report](https://www.adb.org/sites/default/files/publication/1096981/asia-sme-monitor-2025.pdf): landscape tables 14.2, 15.2, 17.2, 19.2, 20.2, 21.2 and 22.2. These values retain the report's vintage; revisions can differ.
- [OECD Malaysia country chapter](https://www.oecd.org/en/publications/financing-smes-and-entrepreneurs-2026_bb7fc5c4-en/malaysia_ebf7a4e9-en.html): 2024 count.
- [OECD Indonesia country chapter](https://www.oecd.org/en/publications/financing-smes-and-entrepreneurs-2026_075d8058-en/full-report/indonesia_6fd1f81b.html): 2024 count and GDP share.
- [Myanmar Ministry of Information](https://www.moi.gov.mm/moi%3Aeng/article/16079): 2023 SME Development Centre registration list, reported November 2024.

## Editorial preview signals

The curated signal data and primary references are in `data/preview-signals.json`, bundled by the same build script when present. These editorial signals can use different observation periods from the explorer, but their dates and definitions must remain visible. They do not alter the harmonised explorer data.

The ILO generative-AI analysis was published in July 2026 but its employment exposure estimates refer to **2025**. Exposure describes occupational tasks that could interact with generative AI; it does not predict job losses or measure layoffs. National unemployment, youth and MSME figures must retain their survey reference periods and denominators.

## Map

Geometry comes from [Natural Earth 1:50m administrative boundaries](https://raw.githubusercontent.com/nvkelso/natural-earth-vector/master/geojson/ne_50m_admin_0_countries.geojson), downloaded 17 November 2026. Natural Earth is [public domain](https://www.naturalearthdata.com/about/terms-of-use/).

The SVG has an equirectangular projection, approximately 90°E–142°E and 12°S–30°N, with a 680 × 460 viewBox. Country anchors are layout positions, not capital coordinates. City markers use approximate city-centre coordinates. Decorative routes and motion do not encode measured activity or a live market signal.

## Rebuilding and validation

Run `python3 scripts/build-intelligence-data.py` after intentional changes to the source JSON. It performs no network requests. It checks country coverage, common release dates, complete observation years, valid rates and sector totals, then rebuilds `assets/intelligence-data.js`.

The browser uses the bundled local file without external API requests. Source calculations retain full precision; display rounding, animation and chart geometry must not change the underlying figures. One shared country selector updates all five explorer views, including dates, units, source links and accessible descriptions. The editorial AI comparison remains separate.
