TL;DR
A new NBER working paper by Joshua S. Gans and Avi Goldfarb models automation when tasks are quality complements in an O-ring production framework. The authors find that automating one task changes incentives for others, can produce discrete or bundled adoption choices, and may raise labor income under partial automation, undermining simple linear exposure measures.
What happened
Joshua S. Gans and Avi Goldfarb present a theoretical treatment of automation in settings where production requires many interdependent tasks whose qualities multiply — an "O-ring" technology — in NBER Working Paper 34639 (January 2026). In their model each worker has a fixed time endowment that must be allocated across the tasks they perform; machines can substitute for particular tasks by providing a specified quality, and workers then reallocate time across remaining manual tasks. The authors describe a "focus" mechanism: when some tasks are automated, worker time concentrates on remaining bottleneck tasks, changing the returns to further automation. From this setup they derive three main results: task-by-task substitution logic is incomplete because automating one task alters incentives for others; automation choices can be discontinuous and require bundled adoption even as machine quality improves smoothly; and partial automation can raise labor income by amplifying the value of remaining bottleneck tasks. They note that common exposure indices that linearly aggregate task-level automation risk will overstate displacement in such complementary-task environments.
Why it matters
- Challenges common linear exposure indices used to predict worker displacement when tasks are complements.
- Suggests automation can produce non-linear, threshold-driven adoption patterns rather than smooth gradual substitution.
- Indicates partial automation can increase labor income by magnifying the importance of remaining human tasks.
- Points to the need to analyze bottleneck task structure and time reallocation, not just average task exposure, when forecasting impacts.
Key facts
- Paper title: "O-Ring Automation" (NBER Working Paper 34639).
- Authors: Joshua S. Gans and Avi Goldfarb.
- Issue date: January 2026; DOI 10.3386/w34639.
- Model context: production composed of many tasks whose qualities multiply (O-ring technology).
- Workers allocate a fixed endowment of time across tasks; machines can replace tasks at given quality levels.
- Introducing automation changes how workers allocate time across remaining tasks — a mechanism the authors call "focus."
- Three main theoretical results: incomplete task-by-task substitution logic; discrete, potentially bundled automation decisions; and possible labor-income gains under partial automation.
- Implication: linear aggregation of task-level exposure risks will overstate displacement when tasks are complements.
What to watch next
- Empirical validation that exposure indices overstate displacement in sectors characterized by task complementarity: not confirmed in the source.
- Evidence on whether real-world automation adoption exhibits the predicted discontinuities and bundling behavior: not confirmed in the source.
- Data showing instances where partial automation has raised worker incomes through bottleneck amplification: not confirmed in the source.
Quick glossary
- O-ring technology: A production framework where overall output depends multiplicatively on the quality of many interdependent tasks, so failure or low quality in one task reduces total output substantially.
- Task complementarity: A relationship between tasks where the effectiveness or value of one task depends on the quality or performance of other tasks.
- Exposure index: A metric that aggregates task-level measures of automation risk — often by linearly averaging or summing — to estimate a worker's or occupation's susceptibility to automation.
- Bundled adoption: A pattern where multiple automation decisions are taken together rather than individually, typically because automating some tasks changes the returns to automating others.
- Bottleneck task: A task that constrains overall production or value because its quality or time allocation disproportionately affects output or earnings.
Reader FAQ
What is the main contribution of the paper?
It models automation in an O-ring setting and shows that task complementarity generates non-linear adoption decisions and can alter labor-income outcomes, challenging simple task-exposure measures.
Does the paper claim automation always reduces wages or employment?
No. The authors show that under partial automation labor income can rise because automation increases the value of remaining bottleneck tasks.
Are the paper's conclusions empirically tested here?
Not confirmed in the source.
Should policymakers change how they measure automation risk?
The paper argues that metrics which linearly aggregate task exposures may overstate displacement when tasks are complements, implying a need to consider bottlenecks and time reallocation.
Working Papers O-Ring Automation Joshua S. Gans & Avi Goldfarb SHARE X LinkedIn Facebook Bluesky Threads Email Link WORKING PAPER 34639 DOI 10.3386/w34639 ISSUE DATE January 2026 We study automation…
Sources
- O-Ring Automation
- O-Ring Automation by Joshua S. Gans, Avi Goldfarb
- O-Ring Automation | Request PDF
- Automated assembly of non-rigid objects
Related posts
- Utopian Scholastic: The 1990s–2000s Visual Style Shaping Learning
- The meek did inherit the Earth, at least among ants, New York Times
- How Westlaw and LexisNexis Came to Dominate U.S. Legal Research