Asian Journal of Probability and Statistics

Customer attrition rate in the Kenyan banks has become a critical concern in the recent past. This is majorly contributed by stiff competition from fintech entities in the country. The advancement in technology and less regulatory requirements are key factors that have led to ease of entry of these fintech entities in financial sector,posing great competition to commercial banks. The increased competition threatens the traditional commercial banks as they are directly eating into their market share. Commercial banks must therefore put in place customer retention strategies in order to retain profitability and operational efficiency. This study presents a simulation based framework for analyzing customer churn rates using ARIMA, a time series technique , within a Kenyan commercial banking context. Due to restricted access to real customer churn data, a simulated dataset that mimics the key behavior observed in Kenyan banking environment such as account activity , transaction volume, dormancy , loan requests , loan repayments etc was generated. The study applied the ARIMA model to the simulated dataset so as to show how temporal variations in bank customer churn can be modeled and forecasted. The results showed that ARIMA model is able to capture trend and short term fluctuations under controlled conditions. The ARIMA model provided a good fit to the generated data. Since the findings are derived from simulated data and not from actual bank churn data , the study only demonstrates the applicability of the ARIMA model in the analysis in situations where access to real data may be difficulty. Future studies should use real data in order to validate actual predictive performance and generalizability of such approaches.