Monitoring transmission of systemic risk: application of partial least squares structural equation modeling in financial stress testing

Necmi K Avkiran, Christian M. Ringle and Rand Low

Abstract

Regulators need a method that is versatile, is easy to use and can handle complex path models with latent (not directly observable) variables. In a first application of partial least squares structural equation modeling (PLS-SEM) in financial stress testing, we demonstrate how PLS-SEM can be used to explain the transmission of systemic risk. We model this transmission of systemic risk from shadow banking to the regulated banking sector (RBS) using a set of indicators (directly observable variables) that are sources of systemic risk in shadow banking and consequences of systemic risk measured in the RBS. Procedures for predictive model assessment using PLS-SEM are outlined in clear steps. Statistically significant results based on predictive modeling indicate that around 75% of the variation in systemic risk in the RBS can be explained by microlevel and macrolevel linkages that can be traced to shadow banking (we use partially simulated data). The finding that microlevel linkages have a greater impact on the contagion of systemic risk highlights the type of significant insight that can be generated through PLS-SEM. Regulators can use PLS-SEM to monitor the transmission of systemic risk, and the demonstrated skills can be transferred to any topic with latent constructs.