The factors influencing the adoption of Machine Learning for regulation by central banks in SADC

Abstract

The study investigates SADC central banks' readiness to adopt machine learning technologies with raw data collected through an online survey. Subsequently, the raw data was transformed into modellable data using principal component analysis and further fitted into the proposed logistic regression model design. The data underwent reliability and validity tests, which confirmed that the measurements of the constructs were consistent, reliable, and appropriately represented the intended constructions. Correlation analysis was employed to examine the hypotheses of the model, and multiple and stepwise regression were performed as additional tests of the model.

The results show that IT infrastructure is instrumental in enabling SADC central banks to implement machine learning capabilities. Top management is crucial for implementing ML, but adequate IT infrastructure is also essential. The regulatory environment and IT infrastructure indirectly influence SADC central banks' readiness to adopt ML capabilities, despite top management's direct impact. The derivable policy implication from these results is that working groups among the sampled SADC central banks need to be formed to address the noted shortcomings within IT infrastructure and regulatory-related aspects of this adoption holistically