Exploring the Impact of Behavioural Factors and Personality Traits on Private Pension System Participation: A Machine Learning Approach

This study aims to investigate the effects of personality traits, in addition to basic financial literacy, private pension literacy and behavioural factors on Private Pension System (PPS) participation using machine learning algorithms. The PPS participation model was trained using both random forest and LightGBM algorithms, and the contributions of model inputs in the prediction of pension participation were interpreted using the Tree SHAP algorithms with swarmplots. The data employed in the empirical analysis is survey data collected from the Sirnak province of Turkiye with a sample size of 449. The findings of the study shows that: (i) PPS participation is more likely for females and middleaged people; (ii) High basic financial literacy has a negative impact on PPS participation; (iii) Extraversion is the key personality trait affecting PPS participation; (iv) Advanced pension literacy has more impact on participation than simple pension literacy: (v) Present-fatalistic tendency is key behavioural factor and it negatively affects PPS; (vi) Present-hedonistic, conscientiousness, future-time orientation, and locus of control tendencies increase PPS participation. Furthermore, the distribution of colours in LightGBM has a greater degree of uniformity in both directions compared with the random forest algorithm. Finally, to increase PPS participation, the results of the study suggest the implementation of the following policy measures: Tailored pension literacy programmes can help to increase pension participation rates. Incentives should be created to prevent narrow-minded behaviour and establish a sense of protection and control around PPS, targeting middle-aged individuals and women. Keywords: Private pension system, Behavioural factors, Personality traits, Machine learning algorithms, Tree SHAP