Join us for an ERIM research seminar
- Speaker
- Coordinator
- Coordinator
- Date
- Thursday 1 Oct 2026, 11:10 - 12:30
- Type
- Seminar
- Location
Mandeville Building - T3-42
Abstract
We study corporate bonds issued to finance artificial intelligence (AI) investments. We find that
credit ratings understate the risk of AI-related bonds issued through special purpose vehicles
(SPVs). These bonds carry higher yields than equally rated bonds, and their yield-to-rating
association is muted. Also, SPV debt is generally not consolidated on the balance sheets of the
parent companies that bear the risk, even though these firms are contractually liable in the event
of default. Capitalizing this debt significantly lowers the parents’ Altman Z-scores and distances
to default. Moreover, adjusting credit ratings to reflect the default risk of AI and SPV issuances
leads to downgrades. As a result, insurance companies, for which holding high-yield bonds is
particularly costly, would likely divest a large share of their holdings of AI-related SPV bonds.
