The Artificial Intelligence Verification Premium: A Dynamic Model of Automation Opacity, Borrowing Costs, and Firm Value

Kwan Hong Tan
Additional contact information
Kwan Hong Tan: Singapore University of Social Sciences, 463 Clementi Road, Singapore 599494, Singapore.

Post-Print from HAL

Abstract: Aims: This study develops the artificial intelligence verification premium, defined as the incremental borrowing cost attributable to financing-relevant AI risk when automation is insufficiently verified, understood, or governed. It links AI adoption choices to corporate borrowing costs and firm value. Study Design: Analytical model development with longitudinal Monte Carlo simulation, paired policy comparisons, policy-space search, convergence testing, and structural sensitivity analysis. Place and Duration of Study: Singapore; model development, computational experiments, audit, and revision were conducted in August 2026. Methodology: A dynamic firm model links automation intensity, verification intensity, verification debt, retained human oversight capability, stochastic uncertainty, incident risk, debt pricing, and discounted firm value. Competitive lenders price one-period debt through a zero-profit condition with loss-given-default. Five stylised management regimes were evaluated with 20,000 Monte Carlo trajectories per regime over 48 quarters using common random numbers. A 324-policy grid used 4,000 trajectories per policy, followed by convergence tests and sensitivity searches for leverage, loss severity, AI productivity, unmanaged-risk sensitivity, and horizon length. Results: Acceleration-first adoption produced a mean quarterly borrowing premium of 130.56 basis points, compared with 9.99 under balanced assurance and 11.31 under assurance-scaled automation. Mean discounted firm value was 356.82, 408.01, and 416.69, respectively. The best evaluated grid policy combined automation of 0.80 with verification of 0.55, yielding mean value of 417.51. This verification level lies just above the analytical direct-risk threshold of 0.538. Conclusion: AI need not raise financing costs. The sign depends on whether verification and human oversight scale with deployment. For managers, boards, and lenders, the practical implication is that validation coverage, unresolved assurance findings, fallback capability, and other governance evidence can be economically relevant to technology investment and credit assessment. Joint optimisation of automation and assurance can preserve productivity gains while containing a financing penalty.

Date: 2026-09-09
References: Add references at CitEc
Citations:

Published in Asian Journal of Economics, Finance and Management , 2026, 8 (1), pp.935-953

There are no downloads for this item, see the EconPapers FAQ for hints about obtaining it.

Related works:
This item may be available elsewhere in EconPapers: Search for items with the same title.

Export reference: BibTeX RIS (EndNote, ProCite, RefMan) HTML/Text

Persistent link: https://EconPapers.repec.org/RePEc:hal:journl:hal-05745096

Access Statistics for this paper

More papers in Post-Print from HAL
Bibliographic data for series maintained by CCSD ().