His argument: While manufacturers are required to set aside the reserves and show them on their balance sheets, the industry is mostly shooting in the dark in deciding how much to stow away to cover the costs of product failures and recalls.
Davies is attempting to convince auto companies that if they could more precisely know how likely their products are to break, the producers would be able to put less cash into their warranty reserves, he said.
"The industry's auditors and chief financial officers have acted conservatively on this question, and it has made them comfortable. They obviously don't want to be wrong and come up short on their reserves," Davies said. "But we believe they're tying up cash unnecessarily."
Predictive analysis is a growing auto-industry practice of crunching large amounts of field data, service records and market variables to determine the statistical likelihood that a given part or product will require repair work in its first three years of operation.
The industry relies on the reserves as a way of self-insuring, rather than buying specific insurance policies on individual parts in the traditional market. But Davies believes predictive analytics can become a tool to allow insurance companies to create policies on a part-by-part basis. That would allow a supplier to buy only the coverage it needs for a defined period, freeing up cash reserves.