Quantum-Enhanced Deep Reinforcement Learning for Central Bank Digital Currency (CBDC) Liquidity Management

Lars Hupel, Makan Rafiee

Abrar Ahmed

The introduction of a Central Bank Digital Currency (CBDC), such as the Digital Euro, presents a critical challenge to commercial bank liquidity management due to the potential for large-scale deposit conversion, with the ECB estimating deposit drains of up to EUR 700 billion under extreme scenarios. This paper proposes a novel Quantum-Enhanced Deep Reinforcement Learning (QDRL) framework to optimize commercial bank funding strategies under CBDC-induced liquidity stress. By integrating a Variational Quantum Circuit (VQC) into the Soft Actor-Critic (SAC) algorithm, the QSAC agent minimizes funding costs while maintaining regulatory liquidity buffers across a five-bank simulation network with interbank contagion dynamics. The paper rigorously addresses the barren plateau problem through theoretical analysis and empirical gradient variance measurements, and stress-tests the framework under NISQ noise models calibrated to IBM Quantum hardware using Zero Noise Extrapolation. Results show an 8.1% funding cost reduction over classical DRL baselines and zero LCR breaches across four calibrated stress scenarios, though the advantage narrows to 3.2% against industry-standard MPC benchmarks. A Quantum Explainable AI analysis further interprets the agent's decision-making, identifying LCR and CBDC flow as the most influential features. While results remain simulation-only pending fault-tolerant quantum hardware, this research positions QDRL as a promising tool for financial stability and regulatory readiness in the evolving era of central banking.

Dhivya Guru

Ved Seetohul, Hamid Jahankhani, Stefan Kendzierskyj +1 more

Mohamed Aswith. H, Densingh Joshua Israel, V.V.Kashimma +3 more

Despite rapid evolution of digital currencies, the demand for secure, fast and scalable cross border payment solution has emerged. At the moment, the frameworks for the existing Central Bank Digital Currency (CBDC) are slow for transactions, lack security, and are not capable of adjusting to the dynamic changes with respect to the exchange rate. This paper therefore suggests a novel Quantum Ledger Based Multi Currency Smart Exchange System (QL-MSES) to counter challenges Quantum computers are capable of, quantum cryptography, quantum ledger technology and machine learning driven prediction models. The real time sound synchronization of global transactions as well as unbreakable encryption are performed in Quantum Key Distribution (QKD) and Quantum Entanglement Protocol (QEP systems in this system. AI-PEX proxies the rest of the world using deep reinforcement learning, and the remainder of the world input is passed through ACMA which dynamically selects the best exchange paths. In addition, the system includes the Homomorphic Encryption and Zero-Knowledge Proofs system, hence allowing the flow of information without jeopardizing the user’s privacy and without violating regulatory compliance. With this evaluation, this proposed QL-MSES is enabled to be faster with transactions, secured with data, and agile in exchanges when markets are volatile. The results of the performance evaluation show that QL-MSES offers higher efficiency and security than a conventional CBDC system. By employing this innovative solution, the applications of CBDC 3.0 can be utilised for cross border remittance, smart treasury management and decentralised finance, a revolutionary structure to adopt CBDC 3.0.

Leonhard Kunczik