US Secretary and Administrative Assistant Headcount Will Contract 12-18% by Q4 2027 Relative to the May 2026 BLS Baseline

Prediction

US Bureau of Labor Statistics Occupational Employment and Wage Statistics (OEWS) net employment for SOC code 43-6000 (Secretaries and Administrative Assistants) and the major sub-codes within it — 43-6011 (Executive Secretaries and Executive Administrative Assistants), 43-6014 (Secretaries and Administrative Assistants, Except Legal, Medical, and Executive), 43-6012 (Legal Secretaries and Administrative Assistants), and 43-6013 (Medical Secretaries and Administrative Assistants) — will, in aggregate, be down between 12% and 18% in the Q4 2027 release relative to the May 2026 baseline release.

The May 2026 OEWS release covers roughly 4.2 million employed workers in the combined 43-6000 codes. The Q4 2027 OEWS release should land between 3.44 million and 3.70 million workers, a net reduction of 500,000 to 760,000 positions, with the reduction concentrated in 43-6014 (the largest administrative-assistant sub-code) and 43-6013 (medical administrative).

Reasoning

The displacement curve worked out in the HAR-series analysis published alongside this prediction lands a roughly 60% net reduction in the US administrative-assistant cohort by 2030, with the displacement concentrated in 2027 and 2028 as the agentic-OS layer landed in 2025-2026 (Gemini Intelligence on Android, Microsoft Copilot on Windows 12, Apple Intelligence, Salesforce Agentforce) runs through the Microsoft Enterprise Agreement renewal cycle and the firm- side workforce-planning cycle.

The 12-18% by Q4 2027 anchor is the realistic first-eighteen-months section of that curve. The capability to drive the bulk of the displacement is already in production today. The IT renewal cycle and the workforce-planning cycle compound through 2026 and into 2027. The 12-18% range is the central case that emerges from triangulating Microsoft's leaked internal 2025 modeling (roughly 55% by 2030), Salesforce's public guidance (roughly 65% by 2030), and staffing-firm Q1 2026 placement-volume data showing a measurable deceleration in administrative-assistant placements through the second half of 2025 that has not yet reflected in the OEWS series.

The prediction is intentionally narrower than the full-curve analysis. It covers a single 18-month window, anchored to a specific BLS series, and is designed to be falsifiable from a public data source on a fixed date.

Confidence factors

Supporting (+):

• Microsoft Copilot generally available since late 2025; Windows 12 Copilot agent suite in production through Q1 2026
• Salesforce Agentforce vertical agents shipping through 2025 and 2026 with documented enterprise-customer deployments at scale
• Microsoft EA renewal cycle: 2024-class renewals running through 2027 with explicit Copilot-versus-headcount per-seat sales motion
• Q1 2026 staffing-firm placement-volume data showing decelerating administrative-assistant placements
• Historical labor-market response time from technology shock to OEWS- visible headcount change: typically 12-24 months, well inside the prediction window

Detracting (−):

• OEWS series lags real labor-market state by 6-12 months; the Q4 2027 release may not fully capture in-progress roll-offs
• Agentic-OS production error rates are not yet publicly disclosed at scale; if real-world error rates land worse than vendor demos suggest, firm-side rollout accelerates more slowly
• Regulatory or labor-law response could slow the curve at the federal or state level
• Macro-economic conditions: a recession would accelerate the curve; unexpected expansion could delay the headcount-reduction decisions in large firms

Validation criteria

The prediction is validated against the Q4 2027 BLS Occupational Employment and Wage Statistics release (typical publication: late March 2028 for Q4 2027 reference period). Specifically:

• Pass (12-18% reduction): OEWS net employment for the combined 43-6000 series lands between 3.44 million and 3.70 million workers in the Q4 2027 reference release
• Beat (over 18% reduction): Net reduction exceeds 18%; curve is running faster than central case
• Miss-slow (8-12% reduction): Net reduction lands between 8% and 12%; curve is running slower than central case but still consistent with the displacement thesis
• Miss-flat (under 8% reduction or growth): Net reduction is less than 8%, or the series shows net growth; the displacement thesis is substantively wrong on the timing

The 8% lower bound for the "miss-slow" zone is set against the rough rate of incidental administrative-headcount attrition that would occur from non-AI sources (firm-size shifts, normal turnover, demographic effects) over an 18-month window. A reduction below 8% would indicate that the agentic-OS displacement curve is meaningfully delayed, regardless of vendor adoption metrics.

Key indicators to watch

• Quarterly BLS Job Openings and Labor Turnover Survey (JOLTS) data for the Office and Administrative Support sector
• Robert Half, Kelly Services, and OfficeTeam quarterly staffing reports for administrative-assistant placement volume and starting compensation
• Microsoft and Salesforce earnings-call disclosures on Copilot and Agentforce seat-count growth
• Public Gartner and Forrester analyst reports on agentic-OS deployment maturity through 2026 and 2027
• The agentic-workforce-substitution boundary prediction that tracks the broader white-collar displacement thesis