【第６回学生論文コンテストJHPS AWARD受賞論文：審査員賞】
Analyzing the Impact of Subjective Well-Being on Consumption-Insights from Machine Learning Predictions and Econometric Models-
Although previous studies have extensively examined how consumption affects subjective well-being (SWB), whether SWB itself influences consumption remains unclear. This study investigates the relationship between SWB and consumption behaviors using the Japan Household Panel Survey (JHPS) and the Keio Household Panel Survey (KHPS). First, this study uses machine learning tools to determine the importance of well-being and socio-economic variables in predicting consumption by ranking the variables based on their importance. Second, based on the variables selected by the machine learning algorithm, econometrics models for panel data are applied to detect causal relationships running from SWB to consumption. Specifically, we apply the fixed effects model and the dynamic panel model to six consumption categories (total, food, food outside, utility, clothing, and leisure) for the overall sample and subsamples by gender and education. The finding from the machine learning models suggests that age, income, wealth, happiness, and preference variables are influential in predicting total consumption, indicating the importance of well-being and preference variables in consumption decisions beyond the effect of income and wealth. The results of the econometric analysis reveal that happiness positively affects total consumption, and that total consumption increases by 5.8% for a one-unit increase in happiness (ranging from 0-10). When we decompose the effect into sub-categories, we find that food outside of home and leisure consumption are driving the positive effects. Moreover, we show that the effect of happiness differs by gender and education: positive for leisure consumption among females and the lower-educated group, positive for food and negative for clothing consumption among males, and no statistically significant effect for the higher-educated group. In further analysis, we examine whether individuals’ preferences could be in the pathway from happiness to consumption. Our econometric evidence shows that the effect of happiness on consumption is unaffected by including the preference variables, and that people who discount the future more tend to increase current consumption. Our results suggest that subjective well-being and preference measures should be considered when designing economic policies.
