Exploring the relationship between students´ experiences with online payment methods and financial competencies using a Bayesian nonparametric approach

Abstract Background In recent years, there has been generalized concern in all countries about increasing the younger population’s level of financial literacy. However, there is still little evidence on the key factors explaining financial literacy, as the existing literature is inconclusive about this issue. This paper focuses on exploring whether students' personal experiences using online payment tools contribute to fostering effective financial literacy learning. Methods We exploit the large volume of data provided in the PISA 2018 financial literacy assessment. To address the common problem of endogeneity that is frequently present when using observational data such as the information provided by international large-scale surveys, we apply the Bayesian nonparametric modeling approach known as Bayesian additive regression trees (BART) in this study to simulate a counterfactual analysis. This approach is especially well-suited to settings with multiple predictor variables and substantial sources of unmeasured variation since it has a very flexible model that can account for many covariates (independent variables) and can estimate multiple interactive effects between our main variable of interest and other variables included in the analysis. Results Our findings suggest that our variable of interest does not have a significant impact on the financial literacy performance of the students in our sample. Examining its interactive effect with other indicators, however, we observe that experience in the use of online payment methods does have a definite positive influence among students from more socioeconomically disadvantaged backgrounds.