Publication Detail
Robots, Functional Limitations & Within-Disability Heterogeneity
Automation and Disability: How Functional Limitations Shape Vulnerability to Technological Change
Zones
IV Countries
1993–2014
Employment Loss
Abstract
Broad demographic categories used to identify populations at risk from technological change can aggregate workers with opposing exposures and obscure who is actually affected. We document this for disability and industrial robot adoption. Instrumenting U.S. robot exposure across 722 commuting zones with industry-level penetration in seven European economies, we find that robot adoption over 1993 to 2014 reduces employment by 3.4 percentage points for workers with disabilities and 3.7 for workers without — effects that are statistically indistinguishable. This aggregate similarity masks substantial heterogeneity by functional limitation. Workers with sensory impairments experience employment declines of 5.5 percentage points, roughly 50 percent above the aggregate, while workers with cognitive difficulties show no significant effect. Within-disability heterogeneity exceeds the between-group difference, suggesting that vulnerability to automation is determined by the interaction between specific functional limitations and the task content that robots displace, rather than by disability status per se.
Key Findings
- Aggregate effects look similar. Workers with and without disabilities experience statistically indistinguishable employment declines from robot exposure (3.4 vs 3.7 percentage points).
- Within-disability heterogeneity exceeds the between-group difference. The variation across functional limitations is larger than the variation between people with and without disabilities — the policy-relevant variation has been hidden by aggregation.
- Sensory impairments bear the largest cost. Workers with sensory impairments lose 5.5 percentage points of employment, about 50% above the aggregate.
- Cognitive difficulties show no significant effect. Suggesting these workers may occupy task niches less exposed to robotic displacement.
- Functional limitations matter more than disability status. Vulnerability is determined by the interaction between specific functional limitations and the task content robots displace — not by whether someone is classified as having a disability.
Heterogeneity by Functional Limitation
Methods
Research Design
Local labor market analysis across 722 U.S. commuting zones, tracking employment outcomes from 1993 to 2014 by disability status and functional limitation type.
Identification Strategy
Shift-share instrumental variables design. U.S. robot exposure is instrumented with industry-level robot adoption in seven European economies — isolating exogenous technology-side variation that does not respond to local U.S. labor market conditions.
Subgroup Decomposition
The analytic novelty is unbundling the aggregate "disability" category into specific functional limitations: sensory, cognitive, ambulatory, self-care, and independent-living. Each subgroup is allowed its own exposure-response coefficient.
Policy Implications
- Aggregate disability statistics mislead. Policy designed around the overall "people with disabilities" category risks under- and over-targeting at the same time.
- Functional limitation, not disability status, should structure targeting. Programs aimed at automation transition should differentiate by functional limitation and the task content workers occupy.
- Sensory impairments warrant priority attention. The empirical evidence points to this group as the most exposed to ongoing automation displacement.
- Workforce policy under task displacement is a within-group problem. Vocational rehabilitation, retraining, and assistive-technology programs should be designed against task content, not demographic labels.
Forthcoming policy brief
A RISEI Lab policy brief built on this paper is in development. It will translate the within-disability heterogeneity finding into actionable targeting guidance for VR agencies, workforce development boards, and policymakers planning automation transitions. Check back for the brief, or contact risei@northwestern.edu to be notified.
Citation
Vu, H., & Yin, M. (2026). Automation and disability: How functional limitations shape vulnerability to technological change. SSRN Working Paper. https://ssrn.com/abstract=6752423
Read the Working Paper
Available on SSRN. The paper is 56 pages, written April 1, 2026 and posted May 14, 2026.
Download from SSRN →