A survey of Indonesia’s e-commerce sellers finds trust in tax authorities linked to voluntary compliance, while enforcement behaves differently.
How online sellers decide whether to report and pay taxes voluntarily may depend on how much they trust tax authorities, according to a study using survey data from 419 Indonesian e-commerce taxpayers.
The researchers found trust was strongly and positively associated with voluntary compliance. They also reported a weaker role for perceived power and a negative, marginally significant interaction suggesting that enforcement pressure could weaken the trust–compliance link.
Using survey data from 419 e-commerce taxpayers, the study reports: (1) trust in tax authorities is a strong positive predictor of voluntary tax compliance (p < 0.01); (2) perceived power shows a weaker positive association with voluntary compliance (p < 0.10); (3) the interaction between trust and perceived power is negative and marginally significant (p < 0.10), suggesting that stronger enforcement may weaken trust’s positive effect on voluntary compliance.
For enforced tax compliance, the study reports: (4) trust is negatively associated with enforced compliance (p < 0.01); and (5) perceived power is positive and statistically significant (p < 0.01). The study also notes that income, age, and education relate differently to compliance behavior. It does not directly measure digital system characteristics.
The findings point to a governance tradeoff in digital tax collection: building trust may support voluntary compliance, but the way enforcement power is perceived could change how—or whether—that trust translates into higher compliance under enforcement.
The evidence comes from a survey of 419 Indonesian e-commerce taxpayers analyzed with hierarchical regression (including tests of associations and interactions). Because this is observational survey research and the abstract does not describe any randomized intervention or causal design, the results indicate relationships rather than proven cause-and-effect. The study also does not directly measure digital system characteristics (such as how platforms or reporting tools work), which limits conclusions about which digital features might drive compliance.
Social Sciences & Humanities Open · 2026 · DOI: 10.1016/j.ssaho.2026.103361
Authors: Fany Inasius, Agner Lim Miarty, Faradillah Muhammad
Institutions: Binus University
People judged AI-generated food images similarly to real food for healthiness and calorie content, but found them less realistic and were less willing to eat them. The results suggest AI images may be useful for controlled food-perception studies, but they are not interchangeable with images of real food.
A preprint reports measurable patterns for how capabilities emerge and are forgotten during language-model training. The study says emergence depends on several model parts becoming useful together, while forgetting depends on how widely and strongly a capability is stored.
A study in nursing mice identified hypothalamic nerve cells that linked prolactin signaling with bone and fat metabolism, food demand and pain sensitivity. Reducing parts of this pathway lowered food demand and pain sensitivity, but the findings have not been established in humans.
