Using Logistic Regression in Studying Factors Associated With Choice of Savings Mechanisms for Households in Kenya

Abstract

Savings plays an important role in the lives of individuals in a country. Moreover, saving using formal means enables them to enjoy benefits such as safe keeping of money as well as easy access to other financial products such as loans. Despite the growth in access of financial services in Kenya, there is a segment of the population that doesn’t make use of such savings products. This study explores the association between financial literacy, access and use and of technology and socio-demographic variables on savings mechanisms of households in Kenya. The study used a secondary data source from the 2021 Finaccess Household Survey. The main objectives of the study was to examine the association between financial literacy, access and use of technology and socio-demographic variables on use of formal savings as well to find a parsimonious model that best defines the relationship. Logistic regression was fitted to evaluate the association between the independent variables and households’ use of formal savings mechanisms. Lasso model was then used to determine variables that were most predictive of using formal savings mechanisms as the most important saving device. A reduced model was then fitted based on the variables that were most predictive of the likelihood of households using formal savings as the most important saving mechanism. The findings indicated that financial literacy, ownership of mobile phones, use of the internet, having disability,gender, highest education attained, religion, income source were significantly associated with use of formal savings as the most important savings mechanisms. The reduced model was significantly better at predicting the likelihood of saving formally as the main savings mechanism compared to the full model. The study recommended that relevant stakeholders should create financial products that are more inclusive towards certain segments of the population. They can also use the model created to identify individuals with the least likelihood of using formal savings methods as the most important one.