Credibility-Weighted Pricing of Autonomous Vehicle Liability Under Operational Design Domain Shift

About

Automated Driving System deployments create a foundational ratemaking challenge: sparse experience, shifting operational design domains, and non-stationary risk across software releases. We propose a hierarchical Bayesian credibility framework pooling across cities, software versions, and territories via a learned ODD-similarity kernel, nesting Buhlmann-Straub as a limiting case. Demonstrated on 648 verified-engaged Waymo crashes across four U.S. metros from the NHTSA Standing General Order database against 116 million matched miles, city-aggregate credibility weights are moderate (0.12-0.46), partial pooling decisively outperforms no pooling, and a power analysis shows the learned kernel's advantage becomes detectable at approximately twelve deployed cities.

Doyeon Jang · 2026

Related benchmarks

Task	Dataset	Result
Out-of-sample crash prediction	City Crash and Exposure Data (Los Angeles held-out)	Predictive log-likelihood-7.43	5
Out-of-sample crash prediction	City Crash and Exposure Data (San Francisco held-out)	Predictive Log-Likelihood-6.95	5
Out-of-sample crash prediction	City Crash and Exposure Data (Total LOCO) (All cities)	Predictive Log-Likelihood-30.17	5
Out-of-sample crash prediction	City Crash and Exposure Data Phoenix (held-out)	Predictive Log-Likelihood-5.95	5
Out-of-sample crash prediction	City Crash and Exposure Data Austin (held-out)	Predictive log-likelihood-9.84	5