Modelling corporate bank accounts

Fry, John; Griguta, Vlad-Marius; Gerber, Luciano; Slater-Petty, Helen; Crockett, Keeley

Abstract

We discuss the modelling of corporate bank accounts using a proprietary dataset. We thus offer a principled treatment of a genuine industrial problem. The corporate bank accounts in our study constitute spare, irregularly-spaced time series that may take both positive and negative values. We thus builds on previous models where the underlying is real-valued. We describe an intra-monthly effect identified by practitioners whereby account uncertainty is typically lowest at the beginning and end of each month and highest in the middle. However, our theory also allows for the opposite effect to occur. In-sample applications demonstrate the statistical significance of the hypothesised monthly effect. Out-of-sample forecasting applications offer a 9% improvement compared to a standard SARIMA approach.

Citation

Fry, J., Griguta, V.-M., Gerber, L., Slater-Petty, H., & Crockett, K. (2021). Modelling corporate bank accounts. Economics letters, 205, Article 109924. https://doi.org/10.1016/j.econlet.2021.109924

• Journal Article Type: Article
• Acceptance Date: May 20, 2021
• Online Publication Date: May 24, 2021
• Publication Date: 2021-08
• Deposit Date: Feb 4, 2022
• Publicly Available Date: Nov 25, 2022
• Journal: Economics letters
• Print ISSN: 0165-1765
• Publisher: Elsevier
• Peer Reviewed: Peer Reviewed
• Volume: 205
• Article Number: 109924
• DOI: https://doi.org/10.1016/j.econlet.2021.109924
• Keywords: Corporate bank accounts; Fin Tech; Forecasting applications; Machine learning
• Public URL: https://hull-repository.worktribe.com/output/3921100
• Related Public URLs: https://bradscholars.brad.ac.uk/handle/10454/18503