AI Adoption and Employment Expectations: Evidence from a Survey of Small Business Owners
A · Strong evidence New York Fed economists analysing the Federal Reserve's 2025 Small Business Credit Survey (about 5,000 weighted employer firms) with regression controls; strong data, but a blog post on expectations rather than outcomes.
Summary based on the full text·Original: about 7 minutes to read
In short
A Liberty Street Economics post (8 October 2026) by Will Aarons and Asani Sarkar of the Federal Reserve Bank of New York. Using the 2025 Small Business Credit Survey, it compares 12-month employment and revenue expectations of US small businesses that use AI with those that do not. AI users are significantly more likely to expect both more employees and more revenue, and the gap survives controls for firm and owner characteristics. The revenue link is strongest for firms in weaker financial condition and those struggling with technology.
Key numbers
- 33 vs 15 ppnet share expecting higher employment next 12 months, AI users versus non-users
- 10 ppemployment-expectation gap that remains after firm, owner, location and past-employment controls
- 46%employer small businesses using AI tools in the 2025 survey, with another 15% planning to
- 77%respondents reporting no change in current labour costs from AI use
As reported by the source.
English·Web page·opens libertystreeteconomics.newyorkfed.org
What it reports
- 46% of firms with at least one employee use AI tools; another 15% plan to within 12 months. Among users, 63% call AI somewhat or very important to production and 51% have integrated it partially or fully.
- 77% report no change in current labour costs from AI; only 31% report increased sales from AI.
- Net expectation of higher employment: 33 percentage points for AI users, 15 for non-users.
- With firm, owner and location controls the employment gap is 13 pp; adding past-12-month employment change leaves 10 pp.
- Among owners in fair or poor financial condition the employment gap is 21 pp; among those in good or better condition, 10 pp.
- Net expected revenue growth: 48 pp for users, 21 for non-users; 14 pp gap after controls and past revenue. Gap is 24 pp for firms in fair or poor condition, 15 pp for those in good condition.
- AI users facing technology challenges (website, social media, e-commerce, cybersecurity) have 16 pp higher net revenue expectations than users without such challenges.
- Depth of integration did not change employment expectations compared with firms still experimenting.
How they got it
The survey was fielded September-November 2025; the AI module covers 5,248 employer firms, weighted to match the national population of employer firms. Regression samples are 4,769 (employment) and 4,698 (revenue). Outcomes are diffusion indexes: share expecting an increase minus share expecting a decrease. Financial condition is used as a proxy test for owner optimism.
Limits
- Measures expectations, not realised hiring or revenue; the authors leave actual outcomes to future research.
- Cross-sectional association; controls reduce but cannot rule out selection into AI use.
- US small employer firms only; large firms are not covered.
Why it matters for companies
The data cut against the simple story that AI adoption means planned headcount cuts, at least among small firms: users expect to hire more, and most see no change in labour costs yet. The strongest effect appears where firms lack technical capacity, which suggests AI is filling skill gaps more than replacing staff. For managers in larger organisations, the useful signal is that suppliers and small partners are adopting fast and expect to grow. Whether those expectations turn into jobs is still open.
Summary based on the full text. Read the original for the full argument and data.Read the original on libertystreeteconomics.newyorkfed.org