00000nam 2200000zi 4500 9.895997 CaOODSP 20221107174206 m o d f cr cn||||||||| 210201e20210201oncd ob f000 0 eng d CaOODSP eng rda CaOODSP FB3-5/2021-7E-PDF Castro, Pablo S., author. Estimating policy functions in payments systems using reinforcement learning / by Pablo S. Castro, Ajit Desai, Han Du, Rodney Garratt and Francisco Rivadeneyra. Ottawa, Ontario, Canada : Bank of Canada = Banque du Canada, February 1, 2021. ©2021 1 online resource (ii, 40 pages) : colour graphs text txt rdacontent computer c rdamedia online resource cr rdacarrier Staff working paper = Document de travail du personnel, 1701-9397 ; 2021-7 Includes bibliographical references (pages 20-21). "This paper uses reinforcement learning (RL) to approximate the policy rules of banks participating in a high-value payments system. The objective of the agents is to learn a policy function for the choice of amount of liquidity provided to the system at the beginning of the day. Individual choices have complex strategic effects precluding a closed form solution of the optimal policy, except in simple cases. We show that in a simplified two-agent setting, agents using reinforcement learning do learn the optimal policy that minimizes the cost of processing their individual payments. We also show that in more complex settings, both agents learn to reduce their liquidity costs. Our results show the applicability of RL to estimate best-response functions in real-world strategic games"--Abstract, page ii. Banks and banking. Reinforcement learning. Payment. Banques. Apprentissage par renforcement (Intelligence artificielle) Paiement. Bank of Canada, issuing body. Staff working paper (Bank of Canada) 2021-7. (CaOODSP)9.806221 PDF 1.36 MB https://publications.gc.ca/collections/collection_2021/banque-bank-canada/FB3-5-2021-7-eng.pdf