Determinants of household cigarette and alcohol consumption in Indonesia: An islamic economics perspective using Had Kifayah framework with interpretable machine learning and econometric approaches

Deri Siswara (Department of Data Science, Perbanas Institute, Jakarta 12940, Indonesia)
Nafisa Berliana Indah Pratiwi (Department of Statistics and Data Science, IPB University, Bogor, West Java 16680, Indonesia)

Halal Studies and Society
Vol. 3 No. 1 (2026): January 2026

The present study is focused on analyzing the determinants of household cigarette and alcohol consumption in Indonesia through an Islamic economics perspective using the Had Kifayah (minimum sufficiency threshold) framework. The incorporation of nationally representative household microdata from SUSENAS March 2024 enabled the classification of households into Mustahik (zakat-eligible) and Muzaki (zakat-paying) groups based on provincial Islamic sufficiency thresholds. In addition, a Double-Hurdle framework alongside interpretable machine learning (XGBoost with SHAP values) was also adopted. The results showed near-universal smoking prevalence, with Mustahik households allocating a larger budget share to cigarettes despite respective lower absolute expenditure. The income elasticities for cigarettes were higher among Mustahik households. Meanwhile, the alcohol income elasticity exceeded unity, which resulted in the classification of alcohol as a luxury good. The XGBoost framework outperformed probit specifications. Furthermore, SHAP analysis identified urbanization, education, and food expenditure share as the leading predictors of participation. These results collectively supported progressive excise taxation combined with targeted zakat distribution mechanisms consistent with maqasid al-shariah.