Dynamic Modeling of Adverse Selection in Peer-to-Peer Insurance Pools: Learning, Exits, and Market Unraveling
Type
conference paper
Date Issued
2026-07-01
Author(s)
Syed Taha Qadri
Abstract
Experience-rated pricing is usually viewed as a remedy for adverse selection in insurance markets. As loss histories accumulate, prices become more accurate. This paper shows that in a voluntary peer-to-peer insurance pool, the same logic can become destabilizing before learning has converged. Members have persistent risk types, but the pool sees only coarse entry information and noisy claims histories. Renewal premiums therefore respond to imperfect estimates of risk. The key object is the classification wedge: the gap between what the pool thinks a member will cost and the member’s true expected loss. When histories are noisy, some members are temporarily overpriced and leave, while members whose risks are underestimated remain. The surviving pool can then become underpriced, weakening reserves and tightening future participation margins. The model delivers a local speed-limit diagnostic for experience rating, and shows that repricing is safer when histories are informative, entry scores are accurate, buffers are large, or subsidies dampen the household-paid wedge. Simulations show the full dynamic loop, while USDA crop-insurance data provide suggestive component-level corroboration.
Language
English
HSG Classification
None
Refereed
No
Event Title
American Risk and Insurance Association (ARIA) Annual Conference
Event Location
Orlando, FL
Event Date
2nd-5th August 2026
Division(s)
Contact Email Address
syedtaha.qadri@unisg.ch
File(s)![Thumbnail Image]()
open.access
Name
B5 Dynamic Modeling of Adverse.pdf
Size
1.12 MB
Format
Adobe PDF
Checksum (MD5)
e3a7d757b3304db04578d3270285f181