arXiv Science⌕ Search

arXiv · 2610.22830

Bystander on the Fence: Sanctions in a Multipolar World

Abstract

We provide a theory of optimal trade sanctions with three strategic players: Home (sanctioner), Foreign (target), and Bystander. Bystander may act unilaterally or ally with either party subject to a participation constraint. We derive three novel results. First, members of a coalition generically set different tariffs either under Stackelberg timing or when transfers are imperfect. Second, as sanctions tighten, Home's welfare can improve even beyond the unilateral terms-of-trade optimum due to strategic interactions. Third, we characterize feasible sanctions targets when Home and Foreign compete for Bystander's support. An alternative implementation arises with competition: Home lets Foreign win, but Foreign's transfer to Bystander implements the sanctions target. We apply our model to sanctions against Russian crude. In a calibration exercise, cooperation among India, China, and Russia best matches observed data. Sanctions could have been roughly 70% more effective had China and India acted unilaterally or had the sanctioning coalition made them a better offer.

Explore related subjects

Keep this discovery

Explore connections, maps & timelines

BibTeXRIS

Jan Morgan, Anchi Xia. 2026-10-16. Bystander on the Fence: Sanctions in a Multipolar World. https://arxiv.org/abs/2610.22830

Cite the original work for its findings. Save a collection to share your selection of sources.

KEEP EXPLORING

Related papers

AI-exposed jobs deteriorated before ChatGPT

Public debate links worsening job prospects for AI-exposed occupations to the release of ChatGPT in late 2022. Using monthly U.S. unemployment insurance records, we measure occupation- and location-specific unemployment risk and find that risk rose in AI-exposed occupations beginning in early 2022, months before ChatGPT was released. Analyzing millions of LinkedIn profiles, we show that graduate cohorts from 2021 onward entered AI-exposed jobs at lower rates than earlier cohorts, with gaps opening before late 2022. Finally, from millions of university syllabi, we find that graduates taking more LLM-exposed curricula had shorter job searches after ChatGPT and higher first-job pay when entering highly exposed occupations. Together, these results---pointing to forces pre-dating generative AI and to the ongoing value of LLM-relevant education---suggest that generative AI may not be the root cause of these labor market hardships and that the launch of ChatGPT is a faulty natural experiment for understanding AI's impact on jobs.

econ.GN↗

Understanding the Mechanism of Altruism in Large Language Models

Altruism is fundamental to human societies, fostering cooperation and social cohesion. While recent studies suggest that large language models (LLMs) can display human-like prosocial behavior, the underlying mechanism that produces such behavior remains unclear. We leverage sparse autoencoders (SAEs) to identify a set of features that contribute to an LLM's altruistic choices in the Dictator Game. We show that intervening on these features substantially shifts the model's allocations and produces similar effects in several other social decision scenarios. Together, these results link the model's prosocial output to internal features that can be identified and causally manipulated, providing a transparent approach to studying the computational basis of decision-making in LLMs.

econ.GN↗

Risk Ceilings and Development Deadlines: Pacing AI under Uncertain Safety Productivity

Can a regulator promise both a risk ceiling and a development deadline when safety productivity is unknown? A ceiling below the final model's unprotected hazard requires a minimum stock of safety knowledge, so both promises hold only if weak research can be ruled out. Learning first and then replaying development, with spare compute in safety, comes close to that minimum. In a calibration with only state risk, constant safety yield, and full knowledge transfer, it needs only 3.3 percent more productivity than the necessary bound. Customer services pay for the guarantee. Rules that fix compute allocation fix dates, not risk.

econ.GN↗