Philippine outsourcing employment falls 25%

Current status

• 2025 baseline. IBPAP (the Philippine outsourcing industry association) reports 1.9 million workers and $40.3B in revenue for 2025. Its homepage shows "1.9 Million Talent Workforce". Some 2026 briefings use 1.89M.
• July 2026 roadmap review. IBPAP cut its 2028 goal from 2.5M to a range of 1.85M (downside) to 2.14M (best case) (GMA News, Jul 14, 2026; BusinessWorld, Jul 15, 2026). At that briefing CEO Jack Madrid said AI "has not really scaled yet".
• September 2026 summit (International IT-BPM Summit). IBPAP projected 1.94M full-time-equivalent workers for 2026 (+2.6%) and 1.99M for 2027 (+2.6%). It said revenue will grow faster than headcount (Manila Times, Sep 25; GMA, ~Sep 24; Philstar, Sep 27, 2026).
• Even the downside case is far from the threshold. In the slower-growth scenario employment only eases from 1.89M to 1.85M (Daily Tribune, Sep 24, 2026). Nothing published since the question opened is anywhere near 1,425,000.
• Independent cross-check. I pulled government labour-survey data from the Philippine Statistics Authority (OpenSTAT table 0101B3FEMP2, openstat.psa.gov.ph). Employment in "Administrative and support service activities", the sector that includes call centres, averaged 3.07M in January–August 2026. That is +10.8% on January–August 2025. "Information and communication" was +2.6%. So far the data show no contraction.
• Global pressure is real but modest. Concentrix's global headcount fell 4.9% year on year by March 2026 (Revelio). US customer-service-representative employment fell from about 2.7M to 2.6M between May 2024 and May 2025. Microsoft, Uber and Hyatt have cut support staff (LA Times/Bloomberg, Jul 28, 2026). Still, ING (Sep 2026), HSBC and Apollo's Torsten Slok find Philippine outsourcing employment still growing.

Publication timing

IBPAP issues a year-end "on track" report in late December and a results briefing in late January. It adds mid-year roadmap updates and the September summit report, whose figures are projections and do not count. So the qualifying figures inside the window are the annual numbers for 2026–2030. The 2031 figure counts only if the late-December 2031 report states it as the year's level. In practice, YES needs the reported 2029 or 2030 figure to be about 27% below the expected ~1.94M for 2026, roughly −7 to −10% a year from 2027–28.

Reference class and base rate

IBPAP has never reported a year-on-year fall. Even in 2020, during COVID, it reported growth (1.30M to 1.32M). A 25% drop within about five years in a large national services-export industry is historically rare outside bans or crashes. Technology-driven occupational declines, such as travel agents or telephone operators, have usually taken a decade or more. The historical base rate is therefore close to zero. AI is the one plausible structural break.

Paths and estimate

Path to YES. AI voice and agent systems would have to become cheap and reliable enough by 2027–28 to roughly halve demand for customer-service workers. Those make up about half of the Philippine workforce. The other segments (back office, health-records processing) would also have to shrink, and growth in in-house global capability centres would have to fail to offset the losses.

• I put about 6–8% on actual employment falling at least 25% by 2030.
• I put 0.65–0.75 on IBPAP then publishing a comparable figure promptly. It may lag, understate the decline, or shift to measures such as "AI-enabled workers" or revenue per worker.

Path to NO. Growth simply slows: IBPAP's own range, plus a shift towards higher-value work.

My check. A Monte Carlo with regime switching gave about 3% when I allowed only severe declines of about 10% a year. Adding a small chance of a 20%-a-year collapse brings it to about 5%. I spread quarterly hazards around the December and January releases, close to zero until 2028 and rising through 2029–2031. The cumulative probability is about 1.3% by end-2029, 3% by end-2030 and 5.2% by end-2031.

Strongest case against my estimate

AI capability is moving very fast. Gartner expects AI to resolve 80% of common service issues by 2029. A play-money Manifold market gives about 81% that AI "replaces most call-centre workers by 2030". Kalshi gives about 22% to a severe US macro shock by July 2028, though that market measures economic conditions, not AI as such.

If call-centre jobs halve by 2030, YES becomes likely. I discount these signals for four reasons:

• vague resolution criteria and long-shot bias in those markets;
• current Philippine data showing growth;
• slow enterprise adoption (NY Fed: AI use is wide but shallow);
• IBPAP's smoothed, upbeat reporting, which still has to cross a high bar inside the window.

They do keep me from going below about 3%.

Question. By the last day of each quarter, has IBPAP reported Philippine IT-BPM sector headcount at least 25% below its 2025 level?

Definitions.

• IBPAP: the IT and Business Process Association of the Philippines, including a renamed institutional successor.
• Sector headcount: IBPAP's reported total direct employment in the Philippine information technology and business process management industry, including its full-time-equivalent employment measure where that is the headline industry employment measure. This is the nationwide industry total, not only IBPAP members, contact centers, a region, or an individual employer. Indirect and induced employment are excluded.
• 2025 level: 1,900,000, the reported 2025 industry employment total. The threshold is therefore 1,425,000 or fewer. This baseline is fixed; subsequent revisions to the 2025 figure do not change it.
• Reported headcount: a figure described as achieved or estimated actual employment for a completed reporting period or specified as-of date. Annual and interim observations count; targets, forecasts and hypothetical scenarios do not. A point-in-time observation must be after the opening date; for period observations, the Observation period convention applies.

Resolves YES if an eligible IBPAP publication reports qualifying sector headcount of 1,425,000 or fewer. A preliminary estimate or a revision of an eligible post-opening observation counts. Use the published numerical precision without inferring unreported digits. If only a range is reported, its upper bound must be at or below the threshold. An explicit IBPAP statement that actual sector employment has fallen at least 25% relative to the fixed 1.9 million baseline also counts.

The reported total must cover the same nationwide industry scope as the baseline. A smaller total produced solely by excluding subsectors or changing coverage does not qualify unless IBPAP also supplies a comparable total meeting the threshold. No attribution to AI is required.

Resolves NO if no qualifying report has appeared by the quarter-end. Lowering a future employment target, reporting layoffs or vacancies rather than total employment, reporting a decline in employment growth, or repeating figures for pre-opening observations does not qualify. In particular, the July 2026 revised 2028 roadmap scenarios are forecasts, not qualifying employment observations.

Resolving source. IBPAP-authored publications only: its official website at ibpap.org, its Knowledge Hub publications (including The Philippine IT-BPM Industry Overview and The IT-BPM Navigator), and official industry performance releases, reports or presentations. A publicly accessible complete reproduction of an IBPAP-authored document counts; a media paraphrase, executive interview or unattributed estimate does not. This is a narrower source requirement than Public confirmation (A2). The fixed baseline is corroborated by IBPAP's website and GMA News's July 14, 2026 account of IBPAP's roadmap briefing. Apply the Resolution window and Resolution date (A3) conventions; the publication date, not the employment observation date, determines the qualifying quarter.