--- title: "Artificial Intelligence, Productivity, and the Workforce: Evidence from Corporate Executives" aliases: - "Atlanta Fed CFO Survey 2026" - "Baslandze et al 2026" tags: - us - survey - productivity - workforce-composition - firm-size - working-paper - central-bank source: "/research/2026-03-fedres-atlanta-ai-productivity-workforce-cfos.pdf" extract: "/research/extracts/2026-03-fedres-atlanta-ai-productivity-workforce-cfos.md" authors: "Salomé Baslandze; Zachary Edwards; John R. Graham; Ty McClure; Michael Sparks; Brent Meyer; Sonya Ravindranath Waddell; Daniel Weitz" publisher: "Federal Reserve Bank of Atlanta" date: "2026-03" --- # Artificial Intelligence, Productivity, and the Workforce: Evidence from Corporate Executives > [!info] Source > PDF (human): [2026-03-fedres-atlanta-ai-productivity-workforce-cfos.pdf](/research/2026-03-fedres-atlanta-ai-productivity-workforce-cfos.pdf) · Raw extract (machine): [2026-03-fedres-atlanta-ai-productivity-workforce-cfos.md](/research/extracts/2026-03-fedres-atlanta-ai-productivity-workforce-cfos.md) > Working Paper 2026-4 — Federal Reserve Bank of Atlanta, March 2026 — Survey-based empirical study (US corporate CFOs), 56 pages > **JEL:** O33, D22, J24 · **DOI:** https://doi.org/10.29338/wp2026-04 ## TL;DR A survey of ~750 US CFOs finds that more than half of firms have already invested in AI, with productivity gains concentrated in high-skill services and finance (≈0.8% labor productivity growth in 2025, expected to exceed 2% in 2026). Aggregate employment is expected to fall by less than 0.4% in 2026 — but the composition shifts: routine clerical roles decline >2pp over three years, with offsetting growth in skilled-technical roles. Large firms expect AI-driven workforce reductions; small firms expect modest gains. ## Key findings - **Adoption is widespread but uneven.** >50% of firms have invested in AI; many smaller firms only beginning in 2026. Small firms invest more intensively per employee than large firms; their AI spend is mostly operating expense (subscriptions). Large firms also have hardware + internal-development spend. - **Perceived productivity gains exceed measured ones** — a "productivity paradox" wedge consistent with delayed revenue realisation. - **Sectoral heterogeneity in productivity gains (2025):** high-skill services / finance ≈ 0.8% labor productivity growth; low-skill services, manufacturing, construction ≈ 0.4%. **2026 expectations roughly double**, with finance/high-skill services exceeding 2%. - **Capital deepening explains only a small share of gains** — bulk is residual revenue-based TFP, linked to innovation- and demand-oriented motivations (new products, better customer reach), not pure cost cutting. - **Aggregate employment effect is small but heterogeneous.** Firm-size-and-sector-weighted aggregate employment expected to decline <0.4% due to AI in 2026. Large firms expect to shed workers; small firms anticipate modest growth. - **Compositional shift:** share of routine clerical employment expected to decline >2 percentage points over three years (mostly large firms), offset by skilled-technical roles (engineers, data analysts, scientists) and other positions (mostly small firms). - **New "Negative Exposure Index (NEI)"** maps CFO open-ended responses to BLS SOC groups. Office and administrative support roles show the most negative exposure (consistent with automation of routine clerical activities such as data entry). Professional, technical, and sales-related occupations are more frequently described as **enhanced** by AI. ## Methodology in brief Survey of ~750 financial executives (Atlanta Fed CFO Survey + collaboration with Duke / Richmond Fed). Asks directly about AI-attributed changes to productivity and workforce, sidestepping the inference problem in standard outcome data. Constructs implied revenue-based TFP from revenue/employment changes; decomposes productivity into capital-deepening vs residual. ## Implications for AdaptAI ### Calculator (`/calculator`) - The NEI is a credible alternative ranking of role exposure that should inform `jobTitlesWeighted[*].weight` for **office & administrative support** roles (push weights upward), and **professional / technical / sales** roles (push weights downward — augmentation, not replacement). - Reinforces current `industryModifier` direction: Finance & Insurance (+9), IT/Digital (+10), Professional & Business Services (+8). The 2-percentage-point clerical decline is consistent with the existing high weight for `Customer Service / Retail / Call Centres` (+6) — could justify a modest increase. - Time-horizon deltas (currently +15/+30/+45) appear conservative against the doubling of productivity gains 2025→2026 — review if 2-year delta should rise. ### Big Picture (`/big-picture`) - `bigPictureData.industryExposure` — confirms admin & customer support as highest-exposure tier; finance and high-skill services as high-exposure-with-augmentation. - `bigPictureData.skillsDemand.declining` — "Routine Analysis" and "Standard Reporting" trajectories supported. - `bigPictureData.skillsDemand.growing` — "AI/ML Engineering", "Complex Problem Solving" supported. - `aiInvestment.investment` — Finance ranks #1 in this paper's productivity gains, consistent with the £1.8B estimate. ### Industry Insights - `/industries/finance` — supports current narrative of largest near-term productivity gains; refresh prose to cite 2025 ≈ 0.8% / 2026 > 2% gains. - `/industries/admin-support` and `/industries/admin-customer-support` — supports the >2pp clerical decline framing; useful citation. - `/industries/professional-services`, `/industries/tech` — supports augmentation, not replacement, framing. ### How to Adapt - `/adapt/upskilling` — emphasise transition into skilled-technical roles (engineers, data analysts, scientists) which are the offsetting demand sink. - `/adapt/repetitive-tasks` — reinforces case that routine clerical work is the front line. - `/adapt/human-advantage` — innovation/demand orientation (new products, customer reach) is where productivity gains concentrate, supporting the "be the curator/customer-facing layer" advice. ## Related notes - [[BIS European Firm-Level Study]] — independent causal confirmation that AI augments rather than displaces in the short run, in EU + US firms. - [[CBP UK Evidence Review]] — UK lens on the same productivity and composition questions; finds programmers and finance analysts continuing to grow, admin/clerical contracting. - [[Same Storm, Different Boats]] — direct Swedish-register evidence of the within-employer compositional shift this paper predicts. ## Caveats & limitations - Self-reported survey data; productivity numbers are CFO perceptions until validated against revenue. - US sample — UK applicability needs care; finance-heavy UK economy means gains may be similar but adoption lag is real. - "Less than 0.4%" aggregate employment decline is a 2026 figure only — does not address multi-year cumulative impact. - Survey was fielded in 2025; results are forward-looking expectations for 2026.