Journal of Applied
Finance & Research

Abstract

This study formally examines how adaptive artificial intelligence systems can support treasury decision making in mid market firms operating under uncertain macroeconomic conditions. We propose a framework that combines time series forecasting, reinforcement learning, and scenario simulation to recalibrate working capital allocations in near real time. Using a multi year panel of anonymized treasury transactions, we evaluate model robustness across regime shifts, including interest rate inversions and supply chain disruptions. The results indicate measurable improvements in liquidity coverage and a reduction in hedging cost variance relative to static rule based baselines. We further discuss governance implications, the role of explainability for audit readiness, and practical deployment considerations for finance teams that operate without dedicated machine learning infrastructure. The interest of this work lies in providing a rigorous, institutionally grounded blueprint for treasurers seeking measurable resilience under volatile conditions.

References

1. [1]Whitfield, J. A. (2022). Liquidity buffers in mid-market firms: A panel analysis. Journal of Corporate Treasury, 14(2), 88 to 112.
2. [2]Ramaswamy, P. N., & Cole, T. (2023). Reinforcement learning for cash allocation under uncertainty. Quantitative Finance Letters, 9(4), 211 to 229.
3. [3]OECD. (2024). Macroeconomic outlook for small and mid-cap enterprises. OECD Publishing.
4. [4]Bank for International Settlements. (2023). Annual economic report 2023. BIS.
5. [5]Chen, L., & Park, S. (2024). Explainability requirements in financial AI systems. AI & Compliance Review, 6(1), 33 to 57.