Please use this identifier to cite or link to this item: https://hdl.handle.net/10419/336341 
Year of Publication: 
2022
Citation: 
[Journal:] Latin American Journal of Central Banking (LAJCB) [ISSN:] 2666-1438 [Volume:] 3 [Issue:] 1 [Article No.:] 100050 [Year:] 2022 [Pages:] 1-34
Publisher: 
Elsevier, Amsterdam
Abstract: 
We develop an unsupervised methodology to group payments and identify possible anomalies. With our methodology, we identify clusters based on a set of network features, using transactional (unlabeled) information from a systemically important payment system of El Salvador. We first preprocess network features, such as degree and strength, through a principal components analysis we reduce the dimensionality of the newly defined data, then we place the main variables into clustering algorithms (k-means and DBSCAN) to analyze anomalous payments. We then analyze, these clusters using random forest to obtain the main network feature. Our results suggest that the proposed methodology works very well to detect anomalous payments, and it is very important to study the case of El Salvador, because of the recent restructuring of the Massive Payment System in El Salvador (promoted by the Transfer365 project), because the authorities want to increase financial inclusion. This change will make the SPM available to the public, to diversify services and incorporate more participants because, historically, it has operated with only three active participants. We expected that Transfer365 will interconnect the LBTR participants' systems with their banking core, the systems of the Ministry of Finance, and other authorized participants to channel large payment flows. Then, identifying possible anomalies through methodology will enhance risk monitoring and management by payment systems overseers.
Subjects: 
Automated detection
Anomalous patterns
Payment systems
Supervisory analysis
Persistent Identifier of the first edition: 
Creative Commons License: 
cc-by-nc-nd Logo
Document Type: 
Article

Files in This Item:
File
Size





Items in EconStor are protected by copyright, with all rights reserved, unless otherwise indicated.